AMERICAN STATISTICAL ASSOCIATION
- - -
COMMITTEE ON ENERGY STATISTICS
- - -
- - -
Thursday, April 25, 1996
- - -
The Committee convened in the Clark
Room, Holiday Inn Capitol, 550 C Street, S.W.,
Washington, D.C., at 9:00 a.m., Dr. Timothy D.
Mount, Chairman, presiding.
TIMOTHY D. MOUNT, Chairman
BRENDA G. COX
JOHN D. GRACE
GRETA M. LJUNG
RICHARD A. LOCKHART
DANIEL A. RELLES
BRADLEY O. SKARPNESS
G. CAMPBELL WATKINS
C O N T E N T S
Presentation by Jay Hakes 6
Presentation by Yvonne Bishop 34
Presentation by Art Rypinski 43
Presentation by Richard A. Lockhart 61
Presentation by Douglas Hale 84
Presentation by Art Holland 97
Presentation By Timothy D. Mount 112
Presentation by John Cymbalsky 148
Presentation by Erin Boedecker 159
Presentation by Campbell Watkins 168
Presentation by Jerry Coffey 194
Presentation by Inder Kundra 240
Presentation by Greta Ljung 245
1 P R O C E E D I N G S
2 (9:07 a.m.)
3 CHAIRMAN MOUNT: Let's begin this
5 My name is Tim Mount from Cornell
7 This meeting is being held under the
8 provision of the Federal Advisory Committee Act.
9 This is an American Statistical Association, not
10 Energy Information Administration Commission, which
11 periodically provides advice to the EIA.
12 The meeting is open to the public.
13 Public comments are welcome. Time will be set aside
14 for comments at the end of each session. Written
15 comments are welcome and can be sent to either ASA
16 or EIA.
17 Non-EIA attendees should sign the
18 register if they wish to receive a copy of the
19 meeting highlights.
20 In commenting, each member of the public
21 is asked to stand, state his or her name, and speak
22 into the microphone at the podium there. The
23 transcriber will appreciate it. Also, members at
1 the table need to speak loudly, Committee members.
2 We did change the schedule originally,
3 but I think we can go back to the standard beginning
4 where we introduce members of the Committee, and
5 hopefully by the time we've done that, our first
6 presenter will be here.
7 So let's start going around.
8 MR. HAKES: Jay Hakes, EIA.
9 MS. BISHOP: Yvonne Bishop, EIA.
10 MS. MILLER: Renee Miller, EIA.
11 MS. LJUNG: Greta Ljung, MIT.
12 MR. RELLES: Dan Relles, RAND.
13 MR. GRACE: John Grace, Earth Science
15 MR. LOCKHART: Richard Lockhart, Simon
16 Fraser University.
17 MR. WATKINS: Campbell Watkins, LECG.
18 MS. COX: Brenda Cox, Mathematical
19 Policy Research.
20 MR. SKARPNESS: Bradley Skarpness,
22 MR. CHATTERJEE: Samprit Chatterjee, the
23 New York University.
1 CHAIRMAN MOUNT: And members in the
2 public there?
3 MR. GROSS: John Gross, EIA.
4 PARTICIPANT: (Inaudible) EIA.
5 MR. WEINIG: Bill Weinig, EIA.
6 MR. WOOD: John Wood, EIA.
7 MR. CRONE: Tom Crone, EIA.
8 CHAIRMAN MOUNT: Thank you.
9 What else have I got here? There will
10 be a luncheon for the Committee and invited guests
11 at 12:15 in the Lewis Room, and breakfast tomorrow
12 will be in the Lewis Room, Committee members.
13 So I think that we now turn to an
14 important address from the boss of EIA, Jay Hakes.
15 MR. HAKES: Well, first I thought I'd
16 deal with a question that most of you are probably
17 most interest in, and that is why your gasoline
18 price is so high.
20 MR. HAKES: The reason is we have a very
21 cold winter. That put a lot of pressure on heating
22 oil stocks, which were low anyway. There's a lot of
23 uncertainty in the world market because of Iraq.
1 It's not that the Iraq capacity is needed to create
2 adequate supplies. It's that people are reluctant
3 to bring on supplies if they think Iraq is going to
4 come out and the price might go way down.
5 Also there are some refinery problems in
6 California now which are really having a big impact
7 out there, and also demand is rising very rapidly to
8 record levels, partly because as people continue to
9 drive more and more, it's not being offset by
10 vehicle efficiency like it had been in the past.
11 So the combination of all those things,
12 plus the usual spring run-up you get every year
13 anyway, has created a pretty volatile situation that
14 will probably continue for another month or two, but
15 when you average out the whole year, probably we'll
16 still be fairly low by historic standards.
17 Fortunately I did not write a detailed
18 speech out yesterday because if I had, I would have
19 had to change it last night with regard to our
20 budget, but let me start by talking about the
21 electronic revolution, and this is something we've
22 talked about before.
23 It is hard to remember that only two
1 years ago this was sort of an idea in a few people's
2 minds. We did not have a real electronic presence
3 out there, and we now have a Web site that is not
4 fully completed yet, but is a very usable, user
5 friendly vehicle, and it's interesting to watch the
6 usage trends.
7 When we started off last summer, we were
8 getting about 100 people per day coming out of the
9 system. We don't measure it by hits, but we
10 basically measure the number of people, subtract our
11 own employees, and use that as our metric, and we're
12 now up to 800 people a day using and visiting the
13 Web site. That's been a very steady increase month
14 by month, except for December when the holidays, the
15 weather, government shutdowns caused the numbers to
16 go down.
17 But if that trend continues, we see, you
18 know, tremendous usage.
19 Another metric that's a good one for us,
20 I think, is the number of files that are downloaded
21 off of our system. Last summer that was running
22 about three to 4,000 files being downloaded a month.
23 It's now up to about 14,000 data files being
1 downloaded a month.
2 Some of these are people who are
3 converting from hard copy usage to electronic, but
4 many of them seem to be new customers. There's a
5 very good cross-section of people from industry,
6 government, academia, the general public. A lot of
7 users still are not ready yet to do things
8 electronically, but I keep repeating this. A high
9 school student in Boulder, Colorado, who's got a
10 modem has better access to energy information than
11 any individual in the Forrestal Building had two
12 years ago. That's a major revolution.
13 In fact, that student can be in a
14 foreign country, as well, because we do have a fair
15 amount of usage.
16 Our latest product is a CD-ROM that has
17 now gone through data testing, and it's generally
18 available to the public. It sort of duplicates what
19 is on the Internet, but the difference is it's just
20 a lot faster and has more features. It has a search
21 feature on it where you can type in any word or
22 phase, and in 15 seconds it will take you to every
23 page in our current publications where that word or
1 phrase is mentioned with that word highlighted.
2 It's really a tremendous asset for a
3 non-energy specialist. For instance, if you're a
4 staffer on the Hill and you deal with energy in
5 several other substantive areas and your boss asks
6 you to write a speech or find some information on a
7 specialized energy topic, you can now do that in a
8 couple of minutes, and there's really not too much
9 that you couldn't do using that.
10 I've often said, too, that this is
11 tremendous. It saves us all of the embarrassment
12 because I don't care who you are. There's no one
13 that knows all the acronyms and the jargon in
14 energy, but who wants to admit it, you know? So now
15 you've got this CD-ROM, and you whip right through
16 there, and you can find the definition to anything.
17 So this electronic revolution has had a
18 major impact. It is a tribute to a lot of employees
19 at EIA who spent a lot of time on this. It
20 ultimately has very little incremental cost because
21 you can start to prepare your documents digitally to
22 begin with, and so having them available to transfer
23 into this electronic form is important.
1 And also particularly with the Web, for
2 you economists, it's about the only thing I can
3 think of the government can give to people that
4 doesn't have a marginal cost. Once we provide it
5 100 people off of our server, we can provide it to
6 100 million people and the cost to us is the same.
7 That's a new way of thinking because
8 everyone wants us to cost out everything these days,
9 and I'm trying to convince people that information
10 is a free good, ought to be treated as such. After
11 all, the Census Bureau was in the Constitution. So
12 I think the electronic revolution will help us, and
13 frankly, it's a good selling point to members of
14 Congress because it sort of shows if usage is
15 getting that high, it starts to show them that
16 people in their district, in their state will use
18 I can also take the CD-ROM when I'm
19 showing them and type in cities in their
20 congressional district, and, boom, they get all of
21 the energy information about that city if it's
22 mentioned, which it usually is if it's a large city.
23 The second item is quality management.
1 Let me get one prop here. I don't know if Renee put
2 this in the package, but for those of you who are
3 interested, we can certainly get you one by request.
4 This is our application for the 1996
5 energy quality award, and we just finished this
6 document this week. So it's very fresh information,
7 and basically what it talks about is the things that
8 we've done in terms of quality management, the
9 development of performance measures, and the use of
10 cross-cutting teams, how we've addressed the
11 electronic revolution, the process we've gone
12 through on business reengineering, and I think that
13 is information you will find useful.
14 Last year we did such an application.
15 This was the first time the department had had such
16 an award. They brought in external reviewers, and
17 we were one of only two organizations in the
18 headquarters of DOE to receive a quality award. In
19 fact, of the various awards that have been given on
20 various quality topics at DOE, we have certainly won
21 more than any other organization in headquarters.
22 So I think there is a lot of change
23 going on. Part of it has helped produce the
1 electronic changes that we've seen, and I think
2 there will be more.
3 I mean we are in the process of deciding
4 what parts of the business reengineering proposals
5 will be implemented, and there are a lot of changes
6 that are coming at EIA. We expect lower resources
7 in the future. We're under mandates to reduce the
8 number of employees, both contractor and federal.
9 So we're having to change the way we do business.
10 On the personnel front, we are just
11 about meeting our personnel reductions. We're very
12 close. Any week here now we'll meet our personnel
13 reductions for this fiscal year. So all the
14 reductions will be voluntary this year, which is a
15 much better way to do it.
16 Now, on the budget front, there are very
17 recent developments that seem to be positive, but
18 let me see if I can go through the complexity of the
19 federal budget system.
20 We are now in fiscal year 1996, which
21 started on October 1 of 1995, and we still don't
22 have an appropriation yet. We have been funded
23 through this fiscal year by a series of continuing
1 resolutions that provide temporary funding at lower
2 levels than the appropriated amount.
3 We do believe, however, that within the
4 next day or two we will receive an appropriation of
5 approximately $72 million. This now seems somewhat
6 high to us because we've been operating under a
7 continuing resolution at the level of $65 million.
8 So we will be back up to the $72 million level, and
9 life will continue a little bit. People can travel.
10 We've had some promotions frozen and other things.
11 But the $72 million level, we must
12 remember, is a $13 million cut from the $85 million
13 appropriation in 1995. So we have basically
14 undergone a 15 percent cut in our resources.
15 Well, you say: what difference does
16 that make? The Congress directed that half of that
17 cut come out of our forecasting operations. This
18 was done at the very last minute in a small
19 conference committee with no consultation with us or
20 anyone else that I'm aware of, and the argument the
21 committee gave was that they wanted to, quote,
22 protect the data.
23 So this resulted in a 35 percent cut in
1 the amount of resources available in the forecasting
2 area and sort of sent some alarm bells out that we
3 have a job to explain to the Congress what the role
4 of forecasting and modeling is.
5 The Congress itself is a heavy user of
6 our models, but what the difficulty is that the
7 staffers in the Congress understand pretty well how
8 these models are used, how our data are used, but
9 the actual members by the time it's filtered to them
10 may not realize where it comes from or how these
11 calculations are made.
12 And the appropriations process has
13 really changed in the last year or two. EIA has
14 been very popular with the staff on the Hill, and
15 they have sort of looked after our budget, but in
16 last year's budget cutting frenzy, the members
17 themselves got more involved in the small budgets,
18 which they had not done before, and we were
19 competing against the National Endowment for the
20 Arts, Indian education, other items that seemed to
21 be more visible in some ways politically, and so we
22 did not fare well.
23 But we are also going to be cutting out
1 some data series. We're going to be collecting less
2 detail on electric utilities. We'll be collecting a
3 little more information on independent power
4 producers. Some of our technical publications,
5 things that are highly technical, appendices and
6 things like that will be available only
7 electronically. They will not be available in hard
9 We are trying to make these cuts in as
10 painless a way as possible, but it's not an easy
11 situation. We are having to cut. We, for instance,
12 had to cut out our support of the Stanford Energy
13 Modeling Forum, which we've provided some support to
14 for a number of years, and we will be publishing
15 probably in the next few weeks a Federal Register
16 notice listing the cuts that have been made and ones
17 that are being contemplated to invite public
18 comment, and we urge full participation in that
19 because we do want to consult with our users to make
20 sure that the cuts that we make are the least
21 damaging to the work that people need to do.
22 We are consulting closely with Capitol
23 Hill on this budget, and that's one of the reasons
1 I'm going to be slipping in and out during the next
2 couple of days, is I do have some meetings over
3 there talking with people about our 1997 request.
4 EIA's 1997 request is for $71 million,
5 which would be pretty close to a continuation of the
6 $72 million that we apparently now are going to get
7 for fiscal year '96. However, we don't know whether
8 we're going to actually get that amount
9 appropriated. Sixty-six million of it is listed as
10 our budget; approximately five million of it is
11 listed in the efficiency and renewables budget. I
12 think there was an attempt to show as much support
13 for energy efficiency and renewables as possible.
14 It's the same money. It comes directly passed
15 through to us, so there's no ability of EE to
16 influence this in any way.
17 But that's where we stand. We could get
18 a continuation budget if we're successful. However,
19 looking at the steps that will be required to
20 balance the federal budget over the next five years
21 and the pressures on the discretionary spending, in
22 other words, the nonentitlement part of the budget,
23 there's virtually no way that that number can be
1 held for very long, and we do expect further cuts,
2 and we're planning to be able to deal with those
3 cuts, while at the same time trying to persuade
4 people of the value of the programs to the economy
5 and to try to minimize those cuts.
6 Certainly we do have some opportunities
7 for efficiency, but not enough to offset the levels
8 of cuts that are being discussed. We're basically
9 looking ahead to 1998 and '99. On Monday and
10 Tuesday we are spending two full days in strategic
11 planning basically to try to look ahead to what kind
12 of organization EIA will be, what should be the
13 appropriate analysis between data analysis and
14 forecasting. All of those things are important.
15 Should the balance in the future be more or less
16 what it is now? Does it need to change in some way?
17 I think on the data side there are
18 opportunities for automation there where I think the
19 efficiency can be increased, and even with cuts we
20 can maintain a lot of the data. Analysis is sort of
21 a very human activity, less susceptible to
22 automation, and that's more difficult.
23 I do think the model needs to certainly
1 be maintained in a strong way. It will be a simpler
2 model than we might have contemplated a year or two
3 ago. We've had to pull back on some of the
4 development work on the model. We've also pulled
5 back on the PC version of the NEMS, but we are doing
6 some limited work in that area, but so far at least
7 the model is still a strong analytic tool and still
8 more used than ever before.
9 I think the best thing at this point may
10 be, Tim, if it's all right, to take questions.
11 CHAIRMAN MOUNT: Sure.
12 MR. HAKES: And see what comments and
13 questions people have.
14 CHAIRMAN MOUNT: Sure.
15 MR. GRACE: What do you need to protect
16 the data? You made the comment that part of the
17 modeling was to protect the data or protect the
18 figures or something like that. Maybe I
20 MR. HAKES: Well, so that the cuts would
21 not be as great in the data area, the data
22 collection area.
23 MR. GRACE: Oh, I see. Okay.
1 MR. HAKES: I think some of you may be
2 aware the Washington Post actually wrote two
3 editorials last summer about EIA because they were
4 concerned that the nation might lose some of its
5 ability to deal with shortages of supply of fuels,
6 and those editorials particularly emphasized data,
7 and I think some of the people got a little worried
8 about that.
9 I would hope that the next editorial
10 from the Washington Post might mention forecasting,
11 as well. You know, obviously there's a whole set of
12 issues like the restructuring of the electricity
13 area, how you might deal with greenhouse gas
14 policies, all of which will rely very heavily on the
15 NEMS model.
16 MR. WATKINS: I was interested in your
17 comments on the gasoline price issue, and if I
18 understood that correctly, one of the problems is
19 the use of No. 2 fuel oil, and higher requirements
20 for that reduced the build-up of the inventory
21 because you have all of the requirement slate on the
22 use of No. 2 fuel oils. Have I got that right?
23 MR. HAKES: Yes, and secondly, I think,
1 as I understand it, the refineries undergo some
2 maintenance usually when they're switching over from
3 heating oil to the spring gasoline system, and that
4 maintenance was delayed a little bit this year
5 because of the colder winter, and so it's the slate
6 within refinery and the maintenance schedule.
7 MR. WATKINS: The inventory would
8 normally be --
9 MR. HAKES: Right. This is a good topic
10 for somebody to do some very good research on, and
11 we're trying to do some. We went through the whole
12 winter with the inventories for heating oil being
13 below the low point of the historical range. Now,
14 that's an interesting question because a lot of
15 people in industry will tell you we've gotten more
16 efficient handling inventories, and they certainly
18 So maybe it's healthy to have low
19 inventories because it reduces your cost.
20 They also feel they can move product
21 around the world a lot easier. To some extent they
22 can, but, for instance, California, which is a
23 fairly isolated market, takes three weeks to go
1 through the Panama Canal.
2 So what would happen if the whole world
3 had a cold winter? You know, it's not statistically
4 probably that likely, but there's an issue there of
5 whether we're going to get back into a more volatile
6 situation because of the much lower inventories than
7 we have seen historically.
9 MR. CHATTERJEE: You spoke a little
10 about using a simplified forecasting model. That
11 may not be a bad idea. Sometimes forecasting models
12 do as effectively as the more complex models.
13 MR. HAKES: Well, I guess we're a little
14 bit conflicting because one thing you do is you
15 forecast, but you're also doing "what if" scenarios
16 for purposes of legislation and other purposes. So
17 the ability to do "what if" scenarios is somewhat
18 diminished with less detail even if the detail
19 doesn't add to or maybe even detracts a little bit
20 from your accuracy.
21 We did go through a lot of machinations
22 this year on prices of fuels. We found over the
23 years that a lot of the projections seemed to hold
1 up pretty well in terms of patterns of consumption
2 and production, but price has been a variable, and
3 we, like many forecasters, have been on the high
4 side compared to what's happened in actuality, and
5 we've lowered very substantially this year our
6 projections for natural gas, and I think that was a
7 good thing to do.
9 MR. GRACE: Along the same lines, with
10 the increasing support you've received by use of
11 your Web site, and I've been one of the people who's
12 been on there --
13 MR. HAKES: Good.
14 MR. GRACE: -- is the decision to pull
15 back on the PC NEMS perhaps kind of contrary to that
16 philosophy, that if you had greater accessibility to
17 the models via a platform that was more popularly
18 available, that you might be able to build the same
19 sort of support on the modeling side that you're now
20 building on the data side through the electronic
21 access afforded through your Web site?
22 MR. HAKES: Yeah, I mean, that's a good
23 point. I think, you know, when you talk a 35
1 percent cut, you sort of get a little desperate, and
2 you sort of cut what you can.
3 You know, ultimately, too, to the extent
4 that our biggest customer is the Congress, they
5 probably are many years away from having the ability
6 to operate even a PC NEMS. We'll have to run it for
7 them. So in terms of our key customer that's going
8 to make the appropriations decision, it probably
9 wouldn't make a huge difference.
10 I hope we can get that back on track
11 because I think it is part of the vision we have of
12 really sort of allowing people to get in there and
13 use the stuff themselves, you know. With the
14 smaller number of people that we have that's almost
15 a necessity.
16 Now, you know, two years ago we were on
17 a mainframe. We're now down to RISC work stations,
18 and the PC thing would be nice.
19 Yeah, Dan.
20 MR. RELLES: Fifteen percent is a lot,
21 and presumably there should be a lot more pain to
22 describe than you've described here, especially
23 since I think any staffer or any Congressman, if he
1 doesn't hear that there's a lot of difficulty this
2 has caused, will say, "Well, okay. I'll do 15
3 percent next year until we hear about it."
5 MR. RELLES: Now, presumably the
6 voluntary retirements are not going to be able to
7 absorb another 15 percent, but what about the users?
8 Are many of them complaining about the reduced
9 volume or quality or availability of data?
10 MR. HAKES: We don't play the Washington
11 Monument game like many people do in town, that the
12 first thing we'll cut is the most valuable thing we
13 do, and I think we've done a good job of becoming
14 more efficient.
15 I think there has been a lot of pain
16 internally, but I think we've shielded a lot of our
17 customers from it. You know, in the government
18 shutdown, we did not shut down at any point because
19 we had carryover money.
20 The next set of things are going to be
21 very tough. I mean I think things like the
22 voluntary reporting on greenhouse gas, which is one
23 of Art's programs, is going to be tough. We're
1 going to do like oil reserves every other year
2 instead of each year in terms of the detailed
3 analysis. Our consumption studies are going to
4 probably be in four-year cycles rather than three-
5 year cycles. That in and of itself saves sort of a
6 bit of money. Now, is that enough to get people
7 jumping up and down and screaming? Probably not at
8 this point, but, you know, if you, say, reduce your
9 sample size so you can't do regional analysis, you
10 know, again, is the average person in the street
11 going to jump up and down and say, "I want regional
12 analysis"? Probably not. People like you might
13 yell and scream, and that might be a good thing.
14 I guess my preference would be on like,
15 say, consumption studies is to do them less
16 frequently to preserve some detail and some
17 regionality, you know. So that's one way to save.
18 In a sense, the next five million is
19 almost tougher than the first 15 million because
20 it's just the nature of things and how rapidly you
21 have to change, but it's certainly tough.
22 I mean, in the modeling area we had some
23 projects going, say, on the international natural
1 gas model, refinery model that we had spent some
2 money on over the years that we had to stop in the
3 middle of the work and suspend the project, and
4 that's very irrational because you're wasting money
5 you've already spent.
6 But when the Congress sort of targets
7 the cuts that way, you know, we don't really have
8 any choice at that point.
9 MR. RELLES: Yeah. Do you think
10 electronic media are going to be a big money saver
11 for you in the future by not having it published in
12 paper? You'll be able to save a fair amount?
13 MR. HAKES: Well, we spend about $1
14 million a year on paper. So there is definitely
15 some money to be saved there. In a budget our size,
16 that's significant, but I think what it allows us to
17 do is talk about expanding our customer base without
18 increasing cost, and so it saves some money.
19 Now, of course, automation is not just
20 dissemination. It may be collection; it may be
21 processing. There's a lot of things that can be
22 automated. So the savings of automation and
23 computers, in general, probably is much greater than
1 just the cost of saving paper and dissemination.
2 But, you know, to the extent we can do
3 that, we can absorb some cuts. What we're trying to
4 do is set aside money each year, too, for investing
5 in this technology. You know, we're trying to
6 think, well, where do we need to be in three years.
7 What kind of training do we need to get to that
8 point? What kind of technologies do we need?
9 We're going to try to standardize a lot
10 more the software that's being used so that our
11 offices communicate more easily with each other.
12 Maintenance costs go down. We project that getting
13 totally off the mainframe will probably save us
14 about $2 million, net savings.
15 So, you know, there are a lot of things
16 that can be done, and you know, we plan to do those
17 anyway. So if we can keep continuation for a year
18 or two to sort of catch up from last year, I think
19 we can maintain a very high level of service.
20 MR. GRACE: I do a lot of work with the
21 U.S. Geological Survey, and in the last few years
22 they have started to go a direction -- and I don't
23 even know what the acronym stands for -- creating
1 CRADAs. CRADAs are essentially allowing the Survey
2 to take private sector money as income and provide
3 something like proprietary services, but not
4 exactly. It's somewhere in between, but it's a new
5 source of income for them.
6 And especially when I think about, for
7 instance, the reduction in the rate from one year to
8 two of collecting data on oil and gas reserves, is
9 there a possibility within the legislative confines
10 of the EIA to set up partnerships with the private
11 sector to augment your budget in a way that would
12 allow the provision of services that might have a
13 specific client base out there that's willing to
15 MR. HAKES: Well, I think it's worth
16 discussing. I mean we have CRADAs at the Department
17 of Energy in a lot of our technical development
18 areas where there are partnerships with industry.
19 It's very interesting. Those CRADAs are under great
20 attack in the Congress right now.
21 MR. GRACE: Are they really?
22 MR. HAKES: They're viewed as, quote, a
23 subsidy to business. You know, the current rhetoric
1 and ideology is very interesting and how things get
3 I must say that there's a side of me
4 that's much more comfortable making the public good
5 argument and continuing to ask for taxpayer dollars
6 than to get into the issue of proprietary
7 information. You know, Statistics Canada, for
8 instance, copyrights their information. That does
9 create a different ball game.
10 I personally would not like to do that
11 because if you look at the history of energy in this
12 country, one of the big issues in the '70s was the
13 public didn't trust industry data. It's been very
14 beneficial to the industry, as well as to the
15 environmental community, consumers, and the
16 government to now have some data that's viewed as
17 being objective. I mean I don't think anybody
18 really questions our objectivity, and it's now so
19 assumed that people don't talk about it anymore.
20 But there is in this country a suspicion
21 of the oil industry, for instance, which, you know,
22 whether it's justified or not, it's sort of there.
23 So if you start saying, "Okay. Well, we've got
1 these partnerships and they're giving us some
2 money," and all of that, I'm not saying we couldn't
3 handle it and I'm not sort of taking it off the
4 table, but I feel more comfortable arguing for --
5 you know, the government has something. I think the
6 government has the responsibility to provide the
7 base information, and it's very interesting because
8 when the World Bank goes into a lot of the
9 developing nations, they are urging them to set up
10 and EIA type organization, and so we are visited
11 frequently by people from other governments around
12 the world saying, well, how do we do this.
13 The reason is, one, the investors will
14 invest more in a high information system. You know,
15 low information equates to risk. Risk relates to
16 higher rates of return, more reluctance to invest.
17 The other thing the World Bank feels is
18 that public officials will make better decisions
19 about energy if they have data to deal with. Those
20 are all public reasons to have this kind of data.
21 It would be sort of ironic if in spreading our data
22 out, helping other countries set up this system that
23 we ourselves sort of lost it.
1 But that's a question I get a lot. The
2 Paper Work Reduction Act, of course, which is part
3 of the framework under which all of the statistical
4 agencies operate, says that our basic system is to
5 charge only for the dissemination of the
6 information. It costs so much to print a document.
7 That's what the cost of the publication is, not our
8 collection cost.
9 And so to move away from that would be a
10 pretty major change, and I have done some fairly
11 superficial look at Canada, New Zealand, and
12 Australia, and some other countries that have tried
13 to move more into the cost recovery phase, and I'm
14 not sure that they really recover costs at the level
15 that it's perceived that they do. I think, you
16 know, we have several hundred thousand dollars a
17 year that goes into the federal treasury from the
18 sale of our publications. So we at that point
19 recover some cost, but I think information because
20 it has such low marginal cost, it's very difficult
21 to get it to pay for itself.
22 Well, we thank you all. This year is a
23 strange year because I think we have stability on
1 the Committee. I don't think there are any new
2 members to announce, but it's good to see you all
3 back again, and I'm looking forward to the exchanges
4 on this.
5 These critiques that you have done have
6 been very helpful to us, and we really appreciate
7 the help that you're giving us.
8 CHAIRMAN MOUNT: Thank you, Jay.
9 I certainly am glad to hear you saying
10 that this is an important public good that you
11 provide. You know, I really do agree with that, and
12 I think it's truly impressive to see the number of
13 people and the new types of people that are being
14 able to get access to data on energy now through the
16 So I think the next item is Yvonne
17 Bishop to tell us how the EIA responded to the last
19 MS. BISHOP: Well, he said a lot of the
20 things I was going to say.
22 MS. BISHOP: On the last meeting Perry
23 Lindstrom spoke about the renewables in the AEO. I
1 had a great big package from him responding to the
2 comments. I think I'll give it to the persons who
3 made the comments rather than repeat them all here.
4 The difference in the forecasts for
5 renewables in '93, '94, and '95 was queried, and he
6 points out that although the differences are large
7 relative to the specific market, they're small
8 relative to the total energy picture, and the
9 reasons for the differences include not only a
10 dynamic industry, but changes in the model as well.
11 And ethanol and biomass are not really
12 renewable. They are agreed that, say, they would
13 like to use this broad definition because these are
14 treated the same way in a policy perspective.
15 The side cases taken as a whole give a
16 likely range of renewable penetration rather than a
17 one point estimate. They feel that what they're
18 really trying to do is give plausible ranges for
19 what might happen in the future.
20 Doman and Peabody spoke on the
21 international energy outlook. They much appreciated
22 Campbell Watkins' comments. They tried to include
23 as many of them as they can. They've discontinued
1 the use of sensitivity ranges and replaced them with
2 two economic group scenarios, and they have
3 described the method by which the scenarios were
5 The 1996 outlook, which will be here any
6 moment, includes comparisons with the previous
7 year's projections and has -- which is what was
8 suggested -- and has added a lot of new features.
9 They now include exports from the former Soviet
10 Union, China's nuclear expansion, improved estimates
11 of energy intensities, and there's a better
12 electricity demand model, and they're planning on
13 including India in the whole story.
14 The other issues concerned largely data
15 collection and dissemination in a changing
16 environment. As Jay mentioned, we will address some
17 of the issues raised in the strategic planning next
18 week, and it was suggested a new mission statement.
19 We may get around to that.
20 We need to define what is the core data,
21 and we need to consider the issues of integrating
23 In terms of how much we learn from our
1 customers, we've again had a small survey of
2 telephone customers with much the same results as
3 last time. We've mailed questionnaires and included
4 cards in publications for various groups of readers.
5 For the Weekly Petroleum Status Group
6 report, of the 200 who responded, 70 percent have
7 computer capability, and nearly all of that 70
8 percent said they don't want paper anymore. So we
9 took them off the mailing list.
11 MS. BISHOP: We have had many favorable
12 comments spontaneously on the Web site, and we are
13 currently discussing development of a questionnaire
14 for electronic users to get their opinions of the
15 strengths and weaknesses of the current offerings.
16 The Information Products and Services
17 Committee has begun to analyze Internet accession.
18 I got a different number than you. I've got 90,000
19 a month, but anyhow, it's growing very rapidly, and
21 MR. HAKES: That's probably kids.
22 MS. BISHOP: No, sessions, and now
23 totals 400,000. Anyway, whichever number is right,
1 it's getting bigger.
2 They can determine by looking at these
3 the type of user, which files were accessed, the
4 number and order of the files accessed, and the
5 number of return visits. So this will give some
6 indication of what's popular.
7 We were planning to present this
8 analysis in Chicago, but have not been able to make
9 any firm travel plans.
11 MS. BISHOP: On business reengineering,
12 the executive sponsors reviewed the team's plans and
13 implementation suggestions. As of Tuesday, they
14 came out with a summary of their decisions, and they
15 proposed some pilot efforts and some reallocation of
16 resources. No doubt details will be available
18 Meanwhile some of the business
19 reengineering enthusiasts have been thinking about
20 the issues, and two of them will be talking about
21 the more statistical issues today. Nancy Leach will
22 be talking on performance statistics for data
23 process, and Renee Miller on the important, but
1 unresolved, issue of how we measure quality.
2 The issues arose because we recognize
3 that if you're trying to improve the efficiency of
4 data processing, you'd better have some measures
5 about how efficient you are, and the question of
6 quality arose because we had goal of speeding up
7 how quickly we got the information out, and we were
8 anxious that this would not be a cost of the quality
9 of the data.
10 In general, many staff recognized that
11 reduced resources implies a need for training for
12 those who take on more or different
13 responsibilities. Howard Magnus has been around to
14 all of the offices and collected the perceived needs
15 for training and is now trying to arrange seminars
16 and workshops on these topics.
17 I have a list of the topics, and I
18 thought that if I handed out this list of the
19 topics, maybe the Committee would help by suggesting
20 persons who might help us by leading a seminar or
21 workshop in these areas.
22 We've been doing some of it in house.
23 Doug Hale has been organizing a training seminar on
1 electricity pricing based on Schweppe's book on
2 electricity spot prices, and Ruey-Pyng Lu has been
3 conducting introductory statistics courses for those
4 who felt they needed that.
5 And I think that's all that we have
6 today in terms of follow-up. I'll hand those things
7 out now.
8 MR. RELLES: You mentioned telephone
9 surveys and the Internet in the same sentence, and
10 it actually made me think of an experience that I
11 had recently, which I must say was rather annoying,
12 but it might be of use to discuss it. I was on the
13 Internet trying to get some stock market information
14 from Scudder, and before it would let me get into
15 some of the more interesting stuff beyond its home
16 page, it asked me a whole bunch of questions, and I
17 couldn't get any further without answering some
20 MR. RELLES: Now, I mean --
21 MS. BISHOP: That's pretty annoying.
22 MR. RELLES: But actually --
23 PARTICIPANT: Did a broker call?
1 MR. RELLES: But actually I didn't mind
2 it all that much because I really did want to get
3 it, and I figured, well, that's the cost of my
4 getting information, and it was phrased in a neutral
5 way. You need a password to get in there. So,
6 okay, so I'm going to make up a password, and then
7 it asks you your name and where you live and a few
8 things like that, but that might be a device that
9 the EIA could consider using to try to survey some
10 of its customers, especially looking for ways to cut
11 costs. Instead of making phone surveys all the
12 time, you've got 14,000 accesses. If I'm asked to
13 answer some questions once and that registers me
14 with you and you know a lot about me, that could
15 very well be a useful device for collecting that
16 kind of data.
17 I know it goes against the grain that
18 information should be free because I now have to
19 spend some of my time answering your questions, but
20 on the other hand, given the value of the
21 information, I think a lot of people would be
22 willing to do it.
23 MS. BISHOP: The telephones -- sorry.
1 MR. HAKES: I think I can respond
2 specifically to that. On the Web site there is an
3 icon called "feedback," where we encourage people to
4 do it, and we do get some data.
5 Now, the Department of Agriculture has a
6 system that's similar to what you're describing,
7 that you basically are giving it, and it sort of
8 hits you up front when you're identified as a first-
9 time user, and we're going to take a look at that.
10 I do think that you have to have a very
11 easy bypass for it because a citizen does have the
12 right to come into that anonymously, I believe, and
13 I think you would have legal problems if you
14 required them to fill out a form, but I think we
15 could do a better job of capturing that
16 automatically and maybe purging people more than the
17 current system does.
18 MR. RELLES: Yeah, I actually saw your
19 attempt to request information, and I bypassed that
20 very quickly.
22 MR. HAKES: We appreciate your concern
23 about respondent's burden.
2 MR. HAKES: You of all people.
3 MR. RELLES: Hey, I'm counted as one of
4 your users.
5 MS. BISHOP: Still the telephone service
6 is for the people who call in on the telephone. I
7 should have made that clear. We still get thousands
8 of phone calls asking for information, and at the
9 time they apparently say, "Would you mind if we
10 phone you back?"
11 MR. HAKES: The problem is a lot of our
12 phone users are playing the commodities market, and
13 you can tell when you're interviewing them that
14 their mind is elsewhere, and they don't want to be
15 distracted for too long.
16 CHAIRMAN MOUNT: So we move to the first
17 presentation of the results of the greenhouse gas
18 voluntary reporting program by Art Rypinski, Office
19 of Integrated Analysis and Forecasting.
20 MR. RYPINSKI: Okay. Good. We have the
21 opening slide.
22 The greenhouse gases voluntary reporting
23 survey is regrettably not a household word. So I
1 thought that it would be useful to provide you with
2 some background so you can figure out what this
3 curious animal is.
4 Can I have the next slide, please?
5 We made a presentation to the ASA a
6 couple of years ago when we were first developing
7 this program that stimulated a very lively
8 conversation, including Mr. Coffey of the Office of
9 Management and Budget, who argued, if I recall
10 correctly, that if this was not a statistical
11 survey, that OMB if presented with it was under, of
12 course, no obligation to approve it.
13 So I just thought I would come back now
14 and talk a little bit about now that we've completed
15 our first reporting cycle what this program is, what
16 we think we've learned, and you know, where we're
17 going from here.
18 The first thing to note about this
19 program is that it's a statutory program. It's
20 specifically required by Section 1605(b) of the
21 Energy Policy Act, which, no, I don't expect you to
22 remember that particular section of what is, after
23 all, a telephone book size document, but what
1 Section 1605(b) does is it requires that the
2 Department of Energy will, in cooperation with the
3 Environmental Protection Agency, provide a mechanism
4 by which entities -- and that's the word in the law,
5 "entities" -- may report, and this is a quote,
6 "annual reductions of greenhouse gas emissions and
7 carbon fixation achieved through any measures,
8 including fuel switching, forest management
9 practices, tree planting, use of renewable energy,
10 manufacture or use of vehicles with reduced
11 greenhouse gas emissions, appliance efficiency,
12 methane recovery, cogeneration, chloroflurocarbon
13 capture and replacement, and power plant heat rate
14 improvement," and so other things, too, "and an
15 aggregate calculation of greenhouse gas emissions by
16 each reporting entity."
17 The Energy Information Administration is
18 required to develop forms for voluntary reporting
19 under these guidelines developed by the Department
20 of Energy, make the forms available to anyone who
21 wishes them, and put the data received in a
22 database. So that's what we've been doing.
23 The law was passed in October of 1992.
1 The Department of Energy issued its guidelines in
2 October of 1994. In November of 1994, we submitted
3 draft forms for public comment. We submitted final
4 forms to the Office of Management and Budget for
5 review under the Paper Work Reduction Act in
6 February '95. They were cleared in May, the end of
7 May. We spent six weeks with the printer and
8 putzing around with necessary changes, and released
9 them to the public in July of 1995.
10 The first reporting cycle purportedly
11 closed on October 31st, but in practice, the actual
12 reports drifted in mostly after the deadline.
13 The forms development process was quite
14 lengthy, as you would expect from something that
15 involved all of these items that I listed, and
16 essentially the problem we faced was that the
17 Department of Energy, faced with the question of
18 building a structure for this program, answered the
19 question: who can report? Anyone. And what can
20 they report? Anything.
21 So whenever the department came to a
22 binary choice, "we can do it either this way or that
23 way," they always said, "Let's do it both ways.
1 Make it possible to do it both ways."
2 And given this range of activities, we
3 wound up with a fairly complicated form. We worked
4 closely with potential reporters. We worked with
5 the Policy Office of DOE. We worked with the
6 Environmental Protection Agency. We went out, and
7 we developed the forms. We pre-tested them with a
8 number of companies, including Niagara Mohawk,
9 Houston Light and Power, General Motors, New England
10 Electric System, and that was extremely helpful.
11 We wound up with two kinds of forms.
12 There's a long form, which is on the order of 40
13 pages. It's like a 1040. It's divided into
14 sections. So no one reporter has to report all of
15 it. You report the pieces you need to report to
16 cover your particular problem.
17 It comes in four schedules. The first
18 schedule asks you what you want, who you are, what's
19 your name, what's your address, what is your quest.
20 The second schedule covers people who
21 would like to report projects, which are actions
22 which we define following the law as actions causing
23 reductions of emissions of greenhouse gases, and we
1 have ten project types of capture specific
2 information for each project type. Most people
3 don't use very many project types.
4 And then we have a Schedule 3, which is
5 the aggregate emissions and reductions of the
6 reporting entity.
7 As we were developing the forms, and as
8 the guidelines were being developed, in April 1993,
9 President Clinton announced that he was committing
10 the United States to reduce its emissions of
11 greenhouse gases to the 1990 level by the year 2000,
12 and he prepared, "he," the administration, prepared
13 a climate action plan, which describes the
14 mechanisms by which the administration expects to
15 use to get to that objective.
16 Many of those mechanisms are voluntary
17 programs which encourage firms and individuals in
18 the private sector to do things that reduce their
19 emissions of greenhouse gases.
20 We have, since this program was already
21 in train in the statutory program, we've reached
22 arrangements with many of these voluntary programs
23 that they will use the 1605 program as their
1 reporting mechanism. So it becomes a means by which
2 third parties can assess the success of these
3 voluntary reporting programs.
4 We also have two kinds of forms, a long
5 form and a short form. The short form we thought of
6 as being for individuals and households. It's two
7 pages, front and back, and then we have the 40-page
8 long form.
9 We developed an electronic form, which
10 is a visual Basic application, which was widely
12 Next slide, please.
13 This is a list of our 108 reporters.
14 MR. WATKINS: Including you?
15 MR. RYPINSKI: Including my name. I am
16 a reporter, and I can assure you that nothing was
17 more educational about the form than having to fill
18 it out myself.
20 MR. HAKES: How much did you save?
21 MR. RYPINSKI: How much did I save?
22 Five tons. I'm sorry. I admitted five tons. I
23 saved about -- no, I saved five tons. That's right.
2 Next slide. I'll characterize those.
3 Yeah, good.
4 As I said, we had 108 entities. Most of
5 them, 96 of them, were electric utilities. Most
6 utilities were participants in a DOE voluntary
7 program for reducing emissions of greenhouse gases
8 called Climate Challenge. Even the 96 utilities is
9 kind of an understatement of our coverage in the
10 electric utility industry because many of the
11 utility reporters reported as holding companies.
12 For example, Southern Company, Cinergy, flipping
13 back, all of those guys with the "ergy" names turn
14 out to own multiple regulated utilities.
15 We had many investor-owned utilities.
16 We had municipal utilities. We had G&T co-ops. I
17 think that the evidence is that we got somewhere
18 between a half and a third of the emissions of the
19 electric utility sector.
20 At the project level, we received
21 information on 645 individual projects. One
22 submission from GPU, which is a big utility up in
23 Pennsylvania, was 500 pages. So when we say 108
1 reporters, that doesn't really do justice to the
2 volume of stuff we got.
3 Non-utility participation was rather
4 modest. There were three manufacturers, General
5 Motors, IBM, and Johnson & Johnson; two coal
6 companies, including Peabody Holding, which is the
7 largest coal company in the United States; two
8 aluminum companies that were reporting on
9 perfluorocarbon reductions, a landfill methane form;
10 two forestry groups, and two households, including
11 as Dr. Watkins pointed out myself.
12 Next slide, please.
13 Okay. How did they report? Basically
14 about two-thirds of them used the long form. About
15 one-third used the EZ form.
16 What we discovered is that we thought of
17 the short form as being a mechanism for households
18 and voluntary groups. It turned out that it sorted
19 out that it actually was a mechanism for voluntary
20 burden reduction for people who wanted to report,
21 but didn't want to put in a lot of efforts.
22 So one of our short form reporters, for
23 example, was Pacific Gas and Electric, which is a
1 large firm, and one of our long form reporters was
2 yours truly, who's a household.
4 MR. RYPINSKI: So among the long form
5 reporters, most of them -- it actually divides up a
6 little more than half -- some half, two-thirds
7 reported on the missions of their entire
8 corporation, and about one-third reported just on
9 project, which are individual actions.
10 Nearly everybody who filed the long or
11 actually not nearly everybody; about two-thirds or
12 three-quarters of the people who filed the long form
13 used the electronic form. Most people who used the
14 short form used paper. It's like the burden
15 associated with flipping through those diskettes and
16 firing up the software approximated the burden of
17 filling out the form itself.
18 Next slide, please.
19 As you would expect, for a process that
20 turned out to be dominated by utilities on the
21 project side, most of the projects were -- that's
22 interesting. That's a different graph than I had
23 with my -- never mind. Don't worry about it. No,
1 no, no, it's right. It's just earlier data. That's
2 embarrassing. The results kind of changed.
3 Most of the projects were energy end use
4 and power generation and transmission. We got a lot
5 of forestry projects, carbon sequestration projects,
6 76 on that slide, and then there was what for me
7 were the most interesting projects, though not the
8 most frequent, is distribution of other activities,
9 including particularly a lot of landfill methane
10 capture projects, gas pipeline methane capture
11 transport projects, and halocarbon reduction
13 We also had a tenth category in our
14 forms for "other," which was for projects we hadn't
15 thought of, and it turned out we got quite a few
16 "other" projects. The most frequent "other" project
17 was coal ash recycling by electric utilities, which
18 was something I didn't know about, and basically
19 coal ash substitutes for limestone in cement
20 manufacture, but doesn't need to be calcite. So it
21 reduces the CFC emissions.
22 By gas, the number of projects and
23 reductions reported were dominated by carbon dioxide
1 and carbon sequestration, but we got a fair number
2 of other gases.
3 Let's see. Next slide please.
4 MR. WATKINS: I have one question.
5 MR. RYPINSKI: Yes, sir.
6 MR. WATKINS: Gen. tran. is generation
7 and transmission --
8 MR. RYPINSKI: Yes.
9 MR. WATKINS: -- of electricity?
10 MR. RYPINSKI: Of electricity, yes.
11 MR. WATKINS: Okay. But the transport
12 is the transportation of natural gas mainly or --
13 MR. RYPINSKI: No, it's motor vehicles
14 essentially. Motor vehicles, yeah. The pipeline
15 projects fall under coal gas CH4.
16 As you can see, I did not submit this
17 slide to OSS for approval, and I was regrettably
18 constrained on space or ingenuity or some
19 combination thereof.
20 We had 40 entity reports, which is
21 people reporting the emissions of their entire firm.
22 Those emissions covered about 1.2 billion tons of
23 carbon dioxide. That's about 20 percent of the U.S.
1 total, and included most of the largest emitting
2 firms in the United States.
3 The biggest single emitter turned out to
4 be General Motors. We will come to this in a
5 minute. They claimed responsibility for the
6 emissions of their entire fleet of vehicles on the
9 MR. RYPINSKI: Which came to some 350
10 million tons, and they noted that because the
11 average MPG of the vehicles they're putting out on
12 the road is steadily improving, that this number has
13 come way down from 1990.
14 Interestingly, they didn't claim
15 formally a reduction. They just wanted to make this
16 point, which they did.
17 Most of the other --
18 MR. WATKINS: They didn't take people
19 who drive?
20 MR. RYPINSKI: Well, General Motors
21 claims that if you're driving a GM car, even if
22 you're driving more, you're emitting less. I
23 shouldn't comment on that.
1 Most of the entities who reported used
2 what we called in 1605 speak a modified reference
3 case. It means that emissions were higher than they
4 were in 1990, but lower than they would have been
5 had they not undertaken a number of good deeds. The
6 most common good deed, the big number good deed was
7 improving the availability of nuclear power plants.
8 Twelve companies reported reductions on
9 what we call, once again in 1605 speak, a basic
10 reference case, which says that emissions were lower
11 than 1990, and these firms were actually
12 concentrated in New England, Niagara Mohawk, New
13 England Electric System, Public Service Electricity
14 and Gas, which is in New Jersey, and Northeast
16 Most projects were associated and most
17 of the reports were associated with these firms
18 participating in various voluntary action plans
19 associated with the President's Climate Change
20 Action Plan. I noticed the largest ton reductions,
21 and at the entity level I think we got about 60
22 million tons of carbon dioxide in aggregate claimed
23 reduction, though I would be most loath to add those
1 numbers up because of the differences in definition
2 -- yes, I see. I'll move along swiftly.
3 From TVA, Duke Power, and Niagara Mohawk
4 improving availability of their nukes.
5 Next slide, please. That's been there
7 We learned an awful lot. The biggest
8 single accounting issue was how do you treat
9 wholesale electricity transactions. Electric
10 utilities tended to claim emissions. It's very
11 interesting. Consumers of electricity tended to
12 feel that they were responsible for their
13 electricity consumption, and electric utilities
14 tended to feel that they were responsible for the
15 emissions arising from providing electricity to
16 their customers.
17 But then the question of who's
18 responsible for wholesale electricity transactions.
19 Different reporters came down all over the map on
20 that one. Some said, "We're responsible only for
21 our plants, period." Some said, "We're responsible
22 for our plants, plus our purchases." Some said,
23 "We're responsible for our plants, plus our
1 purchases net of sales," and there are all of these
2 various arguments.
3 There was a lot of discussion during the
4 design of this program about would people gain in
5 the system, and what we learned is that these
6 reports were so complicated that with a couple of
7 exceptions, people didn't gain the system. They
8 worked hard. They had enough trouble doing the
9 first report without doing five to figure out which
10 one made them look best.
11 Also, reporters tended to be very
12 conservative in their claims and in their
13 accounting. They tended to follow the view "if in
14 doubt, leave it out." If we're not sure about our
15 claim, we won't make it. So for the most part, a
16 lot of the gaining problems turned out to be
17 nonexistent. They turned out to be self-enforcing.
18 There was a lot of discussion during the
19 design about fuel cycle effects. Reporters that
20 they didn't know about fuel cycle effects.
21 We discovered that reporters required
22 substantial assistance with calculating their
23 emissions of greenhouse gases, even if they were
1 very sophisticated reporters. It turns out that
2 operationally, when we were designing the system, we
3 had this notion that these reports would be filed by
4 electric utility planning offices which would have
5 sophisticated generation planning models to run.
6 What we found out was for the most part they were
7 filed by environmental or environmental compliance
8 offices, and one of the problems internally was
9 getting the data out of their operating arms and
10 understanding it before they could report it to us.
11 The concept of emissions for individual
12 projects turned out to be very problematic in the
13 database. If you have a power plant whose heat rate
14 you improve so that it emits less, in effect, it
15 affects the emissions of every plant on your system
16 because the improved heat rate means that it jumps
17 up in the merit order, and other plants go down.
18 People did all sort of things to capture this
19 concept of emissions, and they weren't very
20 comparable across companies or projects.
21 I have a couple more slides, but I think
22 I'll just do one more. What we were planning to do
23 next were the dates here, which are already out of
1 date as we spiral out of control here.
2 Data validation took far longer than we
3 anticipated in large part -- partly because the
4 reports straggled in, partly because the reports
5 were far more complicated than we had imagined in
6 our worst nightmares, and partly because, as I
7 alluded to earlier, the people needed a lot more
8 help than we anticipated.
9 We're working on a database, including a
10 public use database, so that we can distribute this
11 information to the public. This slide says April.
12 That was what I thought in March. Now I think May.
13 We're in the throes of writing, which is
14 what I have to go back to when I finish here, a
15 report on the data. I think that was due to the
16 administrator a couple of days ago. We're still
17 struggling with it, and that will probably be out in
18 June now.
19 We'll be releasing the 1995 forms in
20 June, and then the second reporting season will
21 close in October.
22 I guess that's where I'll end, and let
23 my discussant, whom I terribly skipped in sending
1 him materials in advance, for which I apologize,
2 have a shot.
3 Dr. Lockhart.
4 MR. LOCKHART: I'm being introduced now,
6 I have only two transparencies and only
7 a very few remarks. I have been struggling since
8 1993, when Art Andersen made a presentation to this
9 Committee, to understand what this survey is for,
10 and I was pretty critical of the idea in 1993, and
11 I'm still critical of the basic concept.
12 So I put up some questions and answers.
13 Is a voluntary survey a good idea? I
14 don't think so, and I haven't changed my mind,
15 except that I was impressed on this slide. I put
16 this caveat at the bottom: "except when the outcome
17 is nearly a census." So if everyone fills it out,
18 then the fact that it's voluntary, the non-response
19 bias is presumably negligible, and I gather that we
20 have a very high rate of return from the utilities.
21 So at least in some sections we're getting a
22 reasonable supply of information.
23 Is EIA doing a good job of analyzing the
1 pitfalls of a voluntary survey? I think the answer
2 seems to be yes. That last question, there were two
3 households that reported. That's the other
5 MR. RYPINSKI: No, the other household
6 was Carter Lewis.
7 MR. LOCKHART: Oh, was it?
8 MR. RYPINSKI: Yes, Moorhead P.S. is --
9 I'm sorry. Moorhead Public Service is a little
10 muni. utility.
11 MR. LOCKHART: All right. I got it
12 wrong. I was looking to try to figure out who the
13 other household one was.
14 In 1993, we were warned in advance there
15 would be baseline problems, that is, they couldn't
16 decide whether to do this modified reference case or
17 basic reference case kind of baseline, and I see
18 both have been done, and that does seem to be a
20 Government budget cuts, which are
21 relative to imaginary future budget increases, are
22 not entirely -- not universally seen as credible,
23 and that may be -- I just don't know whether it will
1 be seen as credible here either.
2 We were warned about other problems with
3 few graphic scope, that is, people reporting
4 emission reductions that didn't really occur inside
5 the country because they were forestry projects that
6 were happening outside the country.
7 We were warned about problems with
8 forestry and land use baselines, and we were warned
9 about this problem of many people taking credit, and
10 the example of GM taking the blame but also the
11 credit for all of the improvement in their fleet
12 mileage performance so that we'd get from
13 individuals the same claim that they've made
14 reductions if they bought more fuel efficient cars.
15 You can't count that kind of thing twice.
16 My last two questions on the slide are
17 connected with the "is a voluntary survey a good
18 idea." I don't think that they're generally a good
19 idea because I don't see how to make use of the data
20 in any further activity. You don't want to add it
21 up. So I'm wondering who the users of such a
22 database will be.
23 I'm concerned that the answer is
1 politicians with an ax to grind and a point to make.
2 I'm wondering what possible use can be made of the
4 That's really all I have to say, but I
5 do want to congratulate you on what I think is a
6 very good job of looking at the pitfalls. I have
7 the idea of reading or hearing the speech and
8 reading the documentation that, in fact, the EIA's
9 job here is to certify individuals as having, yes,
10 achieved a reduction rather than maintaining a
11 database, which is sort of what I'm, as a
12 statistician, used to commenting on.
13 CHAIRMAN MOUNT: Some comments from the
14 Committee? Bradley.
15 MR. SKARPNESS: It would have been
16 useful if we could have had one of these long forms
17 or short forms just to see exactly, you know.
18 MR. RYPINSKI: I'm sorry. I didn't know
19 to even bring them.
20 MR. SKARPNESS: Yeah, I know, but if I
21 could have seen them.
22 MR. RYPINSKI: I can get you one by
1 MR. SKARPNESS: I may fill it out or
2 look at it.
4 MR. SKARPNESS: The other thing is that
5 I used to work for TVA, and I see that they have
6 this largest tonnage reduction claim when they get
7 their nuclear reactors running, and when I was
8 there, there was the potential for that, but of the
9 four reactors, none of them were running most of the
10 time, and so I'm wondering. There is a claim there,
11 but what is actually happening?
12 In other words, when you have one of
13 these humongous reactors on line, you can really
14 shut down a lot of the coal facilities, but when
15 they aren't on line, they're running gung-ho, and
16 you're not getting the savings that you potentially
17 could. So it's sort of like a disconnect of what
18 actually is the reductions compared to what the
19 claims are. So that's where I think there is maybe
20 a little bit of a disconnect that's related to, you
21 know, what is actually the data telling us.
22 Thank you.
23 CHAIRMAN MOUNT: Greta.
1 MS. LJUNG: Yeah, I think that it's a
2 good idea if you gave everybody on the Committee a
3 form and we all filled it out. It would increase
4 your response rate by 1,000 percent.
6 MS. LJUNG: You said you have two
7 individuals reporting?
8 MR. RYPINSKI: Yes, that's right.
9 MS. LJUNG: That would increase it 1,000
11 CHAIRMAN MOUNT: Brenda.
12 MS. COX: Oh, in Arthur's defense, I
13 have to say we have seen a form before, but I don't
14 remember what it was. I thought it was just last
15 year. Yeah, the program was presented before.
16 I really think that this issue was
17 bagged last time, and it was bagged this time, and I
18 kind of understand why, but I really think you
19 should grapple with the difficult issue of why in
20 the world is this study being done. I mean it's not
21 to accomplish a statistical objective. I hope it
22 isn't because as a statistical objective goes,
23 you're going to have trouble defending it, other
1 than maybe just to gain some information that might
2 allow you to do a valid study later.
3 In other words, you are gaining
4 information, et cetera. The last time that we
5 spoke, I just assumed that Congress was trying to
6 encourage widespread voluntary efforts to reduce
7 emissions, and maybe that's the purpose, but I think
8 you really need to grapple with it because it
9 certainly in these days of budget cutting and not
10 being able to do things that we know are valuable
11 and important, you need to say why is this being
13 The data deficiencies are obvious, not
14 just the voluntary ones. You really don't have a
15 target population definition, and you don't have
16 definitions for the units that report, and they're
17 vastly different. So as a statistical database, you
18 just can't do anything with it other than kind of
19 discuss its attributes the way you did.
20 I'm actually impressed, by the way. You
21 did a good job of a difficult task, but you come
22 back to why is this being done and should it
23 continue being done.
1 CHAIRMAN MOUNT: Samprit.
2 MR. CHATTERJEE: I think one reason, I
3 think, which Brenda said, is probably right. It's a
4 consciousness raising act, you know, across the
5 industries as a whole. I think the information
6 gathered may not be just applicable directly to get
7 an estimate, but I think it gives you the gross
8 aggregate figure in which we might be able to see
9 the direction and trend in which this is going and
10 also to get some kind of very broad estimate. You
11 know, we're not talking about standard errors or
12 anything of that, but at least a ball park figure.
13 I think that's one thing which this
14 particular form might be useful. Another thing is
15 basically, as you say, making people more aware of
16 generating greenhouse gases and some attempts to
17 reduce them.
18 MR. LOCKHART: I would like to ask
19 Arthur explicitly whether you think that the EIA's
20 role is principally asserted by that reductions have
21 been made.
22 MR. RYPINSKI: I do not. In fact, I go
23 back to the language of the statute which mercifully
1 I didn't quote in full, but it says persons
2 reporting under this subsection shall certify the
3 accuracy of the information reported. EIA does not
4 certify the accuracy of the information reported,
5 and you'll notice that some of my apparently casual
6 language was, in fact, exquisitely phrased. I said
7 "claims," that General Motors "claims"
8 responsibility for most of the vehicles in the
9 United States.
10 Read my lips.
12 MR. LOCKHART: But let me turn to the
13 your function, the function of the agency in
14 collecting the data. Let me try it a different way.
15 Is the establishment of guidelines which will make,
16 provided they are followed, the individual claims
17 more credible, that is, your guidelines carefully
18 prepared enable the companies that report to assert
19 that they have made calculations in a sensible way
20 so that they can claim more credibility having
21 received reductions?
22 MR. RYPINSKI: It is my hope that the
23 design of the forms enhances, does things that
1 enhance the credibility or, at the minimum, the
2 transparency of the information reported. That was
3 one of our objectives, to make it clear, as clear as
4 possible, what was reported and what people were
6 We do not assert that these claims, as
7 an agency, that these claims are accurate, and with
8 the various accounting definitional issues, I don't
9 think the problem is people are claiming things that
10 are not true. I mean, you know, there's a statute
11 that says, "You shalt not make false statements on a
12 government form," and it has criminal penalties. It
13 doesn't matter what the government form is, and that
14 was never my sense that the reporters were making
15 statements that were not true.
16 What does happen is that there is no
17 definition of who owns emissions, and there's
18 nothing resembling a set of generally accepted
19 emissions accounting principles that firms can adopt
20 and produce reports that are consistent even to the
21 degree of, say, financial statements, and we, in
22 fact, know, those of us who have worked with
23 financial statements, know that the comparability of
1 financial statements in detail is a little tricky
2 despite all of the effort expended on it.
3 So the real problems are not that the
4 claims aren't true. It's that what one company
5 claims, another observer might not -- one observer
6 might find it credible under one theory of
7 accounting, and another observer might find it
8 unbelievable under a different theory of accounting.
9 Our mission is to make it clear what it
10 is they claim so that people can form judgments on
11 those questions.
12 CHAIRMAN MOUNT: Cal, welcome.
13 MR. KENT: Thank you.
14 CHAIRMAN MOUNT: So if you want to joint
16 MR. KENT: Just to make a comment, and
17 that is I find this discussion amusing in that it
18 almost takes me back four years ago when we were
19 talking about this particular piece of legislation,
20 and I think if we weren't on tape I could be more
21 explicit about the discussions that went on about
22 this particular "why are we doing it," and basically
23 this is one of the things that the "why are we doing
1 it" is because this was a compromise between the
2 people who wanted a certification program so that we
3 could give out gold stars or green trees or whatever
4 the badge was going to be; those that wanted to use
5 it for regulatory purposes to say, "You've got to do
6 it by so much"; and those that wanted to be able to
7 claim that there were certain reductions as a
8 fallout of this particular piece and be able to
9 point that this particular piece of legislation had
10 had so much positive impact on the environment so
11 that in international conferences, and so forth, we
12 could, rather than lying directly like the Europeans
13 do, that we would be able to have a statistical
14 shroud that we could drape ourselves in when we made
15 our outrageous claims for how effective we were
17 So that was basically why it came into
18 being. The problem is that every one of the issues
19 that you guys have just raised was discussed in
20 detail, and basically what the writers and the
21 promoters and the pushers on both sides of the aisle
22 of this legislation said is, "We don't care. This
23 is at least something we can point to," and the
1 result is it will be used for mischief. I can
2 assure you it will be used for mischief.
3 And when it is used for mischief, I can
4 assure you as in the past EIA will be blamed.
5 CHAIRMAN MOUNT: Campbell.
6 MR. WATKINS: It seems to me a key here
7 is going to be how are you going to write up and
8 present this material. I saw on the issue paper
9 that you have, in fact, a report. Have you given
10 any thought to how you're going to or do you have
11 any insights on how you're going to present the
13 MR. RYPINSKI: Yeah, actually there are
14 a number of people here who are anxiously awaiting
15 that report: my bosses.
16 We propose to in the report -- basically
17 the report will probably follow the tone and the
18 spirit that will follow very much the way that I
19 have presented it, which is to play it straight and
20 say this is what people are claiming, and to discuss
21 the problems as fully as we can.
22 MR. WATKINS: But wouldn't you want to
23 go a bit beyond that and make some kind of at least
1 qualitative comments on what it all means.
2 MR. RYPINSKI: Well --
3 MR. WATKINS: Otherwise I think it will
4 be a bit misleading to say, "Well, here is what
5 people claim." I think you would have to alert the
6 public to what the problems are with this sort of
8 MR. RYPINSKI: Well, I expect that we
9 will, that that is one of the things that is an
10 objective in the report. As always, preparing these
11 documents requires a bit of art.
12 MS. LJUNG: I like Samprit's idea of the
13 awareness, or making people aware, but the problem
14 seems to be that nobody is aware of the survey. Two
15 households responding, obviously the people probably
16 don't know about it.
17 CHAIRMAN MOUNT: Jay?
18 MR. HAKES: I'd like to make a couple of
20 One is to put this in context, I think
21 you'd have to go back to the Clean Air Amendments of
22 1990 where we have regulated sulfur emissions
23 seemingly quite successfully at much less cost than
1 was anticipated and pretty sizable reductions.
2 The system that they used was a capping
3 system where you say there's such-and-such limit on
4 sulfur emissions, and people trade emissions. That
5 has worked well enough that it's likely to be the
6 model for any further emissions. So you would have
7 a system where carbon emissions are capped and then
8 traded back and forth.
9 One of the problems with that system is
10 the issue if the baseline is the point at which the
11 legislation is passed, there is no incentive to
12 reduce emissions before that point because you don't
13 get credit. In fact, the people who voluntarily
14 reduce emissions are disadvantaged because they've
15 already taken the easy steps.
16 So contemplating that there is a chance
17 that the Sulfur Act would be a model for greenhouse
18 gas reductions, the major emitters would like some
19 system to allow them to claim credit, and while the
20 methodology is a little "loosey-goosey" at this
21 point, it is transparent as could possibly be made
22 so that if one decided how one was going to do this,
23 you could make adjustments within the systems.
1 I think the value of this system -- it's
2 an uncomfortable situation for EIA to do this in
3 some ways because it's not statistics work as we
4 know it. On the other hand, I think it's work that
5 EIA is probably better prepared to do than anybody
6 else is.
7 One of the things I would urge you to
8 look at this as is a definitional effort. I mean
9 before you measure something, you have to define it,
10 and if you look at the last two years and our
11 knowledge of what greenhouse emissions are, how the
12 rank of one emittant compares to another, what
13 measures are used to mitigate them, we know a lot
14 more now than we did two years ago.
15 So that if you're going to measure it in
16 the future in a way that is statistical, at least we
17 have some definitions on which to base those, and
18 Congress, if they're going to pass a law, would have
19 at least some definitions and some information on
20 which to base that law. Most of the environmental
21 regulation in this country has been based on pretty
22 good guesses, you know, educated guesses about
23 things and hoping that it would work out if there is
1 some chance that there is legislation in this area,
2 that that would be worked out.
3 I would also say that the nuclear
4 capacity is going up very rapidly. We just issued a
5 report within the last two weeks, and it's gone up
6 very year in recent years and is a fairly major
7 driver in the electric system.
8 The other is I think this electronic
9 reporting system that Art has developed can be a
10 model for us of how we ought to be user friendly in
11 data collection and other areas, but it is not true
12 that Art is being considered to head the IRS.
14 CHAIRMAN MOUNT: Well, thank you.
15 So I think that that -- have you got one
16 more? Sorry. I didn't see. You wanted to address
18 MR. RYPINSKI: I wanted to address very
19 briefly --
20 CHAIRMAN MOUNT: We are running a little
21 bit over time. So --
22 MR. RYPINSKI: Okay. I'll bang through
1 Some of the comments. First, on
2 geographic scope, we got a small number of overseas
3 forestry projects. Nearly everything that was
4 reported was domestic. The overseas forestry
5 projects actually were quite interesting, but by no
6 means did they dominate the landscape.
7 Forestry and land use turned out to be
8 for the most part that we saw in the forms through
9 careful forms, and the forestry numbers turned out
10 not to be that large, in part, because the forms
11 were designed carefully to prevent people from
12 summing 150 years of savings into the first year.
13 The next question is what is this for.
14 There are a couple of useful purposes that I would
15 suggest to you that I think are important. The
16 first one is that we have today a situation in which
17 the President has made a major national commitment,
18 which is to stabilize emissions of greenhouse gases,
19 and he has defined a mechanism, which is a set of
20 voluntary programs. If people are interested in
21 assessing the effectiveness of that mechanism and
22 linking back the claims made for voluntary programs
23 to what companies are actually doing, this database
1 is an extraordinary, even a unique way of doing
3 So I think that this will ultimately
4 make it possible to assess the voluntary programs
5 and point a realistic way.
6 And the second thing is that in it there
7 is an array of all sorts of interesting information
8 about what I would characterize as unusual projects:
9 landfill methane reduction, perfluorocarbon
10 emission reduction, fly ash recycling. And I think
11 there's a lot for us to learn. Certainly there are
12 things to be learned for the national inventory in
13 the database. There are things to be learned in
14 there, in fact, about very low cost approaches to
15 controlling emissions of greenhouse gases in areas
16 where emissions are uncontrolled because there's
17 never been a reason to do it.
18 So I think there's a significant social
19 interest in that now. With that I'll shut up.
20 CHAIRMAN MOUNT: So some of us are going
21 into caffeine deficit. One brief comment from
23 MR. SKARPNESS: Brief. You do aggregate
1 these numbers, okay, and as a consequence, I think
2 that's where we have a problem because there's no
3 definitions or there's no standard way of
4 calculating these reductions.
5 MR. RYPINSKI: That chart is pretty good
6 because it's number of projects.
7 MR. SKARPNESS: Yeah, but this is what
8 people are going to use. Once you aggregate them,
9 then they're going to start using these.
10 Is there a way -- I mean in
11 nonattainment areas, there are air monitoring data
12 being collected. Maybe if you could compare, you
13 know, aggregate some of those numbers where it's
14 done somewhat independently, you know, with using
15 air monitors, compare trends there that have gone
16 over the years with what's going on, what's being
17 claimed here. That might give some credence to
18 these numbers. I don't know.
19 MR. RYPINSKI: Well, I don't think the
20 problem is the numbers that are claimed are untrue.
21 MR. SKARPNESS: No.
22 MR. RYPINSKI: If there's an issue, it's
23 not going to be that the monitor says X and the
1 reporter claims Y. The problem is going to be that
2 the circle the reporter draws around his project
3 will include things that some people will think
4 should have been out or in.
5 MS. GUEY-LEE: Can I ask --
6 CHAIRMAN MOUNT: Can you speak at the
7 microphone, please?
8 MS. GUEY-LEE: I'm sorry.
9 CHAIRMAN MOUNT: And state your name.
10 MS. GUEY-LEE: My name is Louise Lee. I
11 worked with Arthur on the project.
12 I think that when you look at the
13 reporting, what you saw, we know what the aggregate
14 statistics in the U.S. is, and it supports what was
15 reported. A lot of our reports were about
16 improvements in nuclear availability. We know from
17 the national data that that was a trend in that
19 So I guess if I characterized this, the
20 reports, you know, you shouldn't think that they
21 were not credible. They were reported within the
22 rules. For example, General Motors was a
23 manufacturer. They invested in improving the
1 process of developing cars that were more efficient,
2 and within the rules of the game, they claim they
3 reduced indirect emissions. So now that sounds a
4 little better when you look at it that way. They
5 were within the rules.
6 Does that help?
7 MR. SKARPNESS: Well, nobody's claiming
8 that they're not doing it in a legal, you know, way,
9 but there is a truth out there that as a
10 statistician we're after that truth.
11 MS. GUEY-LEE: Okay.
12 MR. SKARPNESS: Okay? And so that's --
13 well, we don't see the disconnect between what the
14 numbers are and maybe how they relate to, quote,
15 unquote, the truth, and that's --
16 MS. GUEY-LEE: Well, one of the ways
17 that we know we can improve is to make more
18 efficient appliances, more efficient cars. GM did
20 MR. SKARPNESS: I agree with you.
21 MS. GUEY-LEE: And they reported it.
22 MR. SKARPNESS: Okay.
23 CHAIRMAN MOUNT: So I think that this
1 discussion should continue over coffee.
3 CHAIRMAN MOUNT: Let's try and
4 compromise and have a ten-minute break because we've
5 got an important topic on restructuring the electric
6 utility industry after the break.
7 (Whereupon, a short recess was taken.)
8 CHAIRMAN MOUNT: So this session on
9 restructuring of the electric power industry has two
10 papers. The first one, "An Analysis Agenda for a
11 Restructured Industry," by Doug Hale, Office of
12 Statistical Standards.
13 MR. GRACE: Tim, I understand there's a
14 little problem of hearing in the back from people
15 making comments at the table.
16 CHAIRMAN MOUNT: So, members of the
17 Committee, you have to speak into the microphones.
18 You're not doing a good job, and you will be graded
19 on the second session.
21 CHAIRMAN MOUNT: Those that pass will be
22 allowed to go to dinner, which is the other things
23 that I should have announced, that we plan to have a
1 dinner for the Committee at Le Rivage, the usual
2 location, at 6:30, but we need to have a number
3 count of how many people will be able to go, and I
4 think the best thing is to decide by the lunchtime
5 and make sure you tell Tracy.
6 Okay. So now you're on, Doug.
7 MR. HALE: All right. Thanks, Tim.
8 My name is Doug Hale. I'm the author of
9 the paper on electric power industry restructuring.
10 The world is changing all around us.
11 Last week they started dancing at Baylor University.
13 PARTICIPANT: They didn't when I was
15 MR. HALE: Yesterday the FERC announced
16 new open access rules for the entire U.S. domestic
17 or U.S. electricity industry. Unless things have
18 changed for the worse, the far worse, we all know
19 where the dancing is going to lead up. We don't
20 know where the electric pricing is going to end up,
21 and so the Administrator has given me an opportunity
22 to think about this for a while with some of my
23 colleagues at EIA, and I want to present kind of
1 where we are in our thinking.
2 I must say that there is a huge debate
3 about how this is all going to shake out. On the
4 one hand, you have people like Paul Joskow claiming
5 in different ways that the benefits from
6 restructuring and price competition may, in fact, be
8 Joseph Stiglitz yesterday at the
9 announcement said that we got enormous economic
10 benefits from the deregulation of the
11 telecommunications and natural gas industries. This
12 is the first comprehensive step in a similar
13 transformation of the electric industry.
14 First slide.
15 What I've tried to do in this paper is
16 present a conceptual framework for thinking about
17 the restructuring of the industry. I've described
18 EIA's current projects that relate to electric power
19 restructuring, and I've made some suggestions about
20 a prospective analysis agenda.
21 This paper has been widely circulated,
22 including to the Department of Justice, Federal
23 Trade Commission, and others, for the purpose of
1 getting suggestions about alternative approaches and
2 different suggestions for projects. I would
3 certainly like your ideas in that area, but more to
4 the Committee, I'm concerned with learning about
5 ongoing work in the area.
6 I was at a Department of Justice-FTC
7 closed seminar Tuesday dealing with the merger-
8 acquisition guidelines and their application, and in
9 the course of that, they had invited people from
10 Enron, Berkeley, several of the private consulting
11 firms to present research that's not been published,
12 and it's very clear to me that there's a lot of work
13 going on that we will have to take advantage of if
14 we're going to have an effective analysis program in
15 this area.
16 I'm particularly interested if the
17 Committee has access or knowledge of ongoing
18 empirical work dealing with such things as nodal
19 pricing, distribution of nodal prices,
20 approximations to nodal pricing, elasticities under
21 demand -- of demand under spot pricing, anything you
22 may have on transmission economics, and the metering
23 controlled costs associated with something closer to
1 real time or closer to nodal pricing.
2 Most of the economic analysis of this
3 industry is based upon network models, and we at EIA
4 have not had much experience with them. So we're
5 very interested in any information you have about
6 small scale network models applicable to the
7 electric power industry that we might use, if
8 nothing else, as learning devices at this stage.
9 The Justice Department is also
10 extraordinarily interested in them, and they have
11 already suggested informally joint projects with us.
12 Finally, any information you might have
13 on ongoing work to quantify the price quantity
14 impacts of restructuring in the U.S. markets would
15 be very much appreciated. We're learning more and
16 more about what went right and what went wrong in
17 the U.K., excellent empirical papers in the Journal
18 of Political Economy and other places talking about
19 the number of competitors, that there are too few.
20 There isn't anything quite ready yet in the U.K.,
21 even though there have been a lot of preliminary
22 pricing experiments in the last few years.
23 The next slide presents the organization
1 of the paper. The paper is kind of long, so let me
2 just run you through it.
3 The first four sections basically deal
4 with the conceptual framework, the first question
5 being: why do we regulate? Why are we stopping
6 regulation of the usual sort?
7 The second deals with the benefits of
8 restructuring. What is there? Are the benefits
9 real or are they not?
10 We then look very briefly at three
11 archetypical restructuring proposals. How do we get
12 from here to there is the issue. Even if there are
13 benefits from competition, how do we move from the
14 state we're in to this better world?
15 Finally, an analysis issue is: is there
16 a there there? Again, it's not at all clear. How
17 much of a change there's going to be depends upon
18 empirical facts. It depends upon the equilibrium
19 outcome of this evolution, and also our ability to
20 quantify it is very much limited by the data we
22 The final part of the paper deals with
23 the analysis agenda, again, the first phase being
1 what we're doing now and the latter phases of a
2 prospective nature.
3 I want to speak just a second about
4 benefits because it's the benefits that set the
5 parameters on how big the changes might be.
6 Do you want to put that up?
7 If you look at the literature, most of
8 the benefits are said to come from five areas.
9 First are improvements in operating efficiency and
10 the allocation of system-wide resources, including
11 the closing of uneconomic facilities.
12 The second thing, which can't be over-
13 emphasized, in my opinion, is that by moving to
14 something closer to nodal pricing or marginal cost
15 pricing or flexible pricing, you are now able to
16 make beneficial trades that simply were not possible
17 under average cost pricing. That changes
18 everything. Okay. That changes when and how the
19 plants -- you know, when demands appear, how plants
20 are run, what emissions are like.
21 The third thing is an asset reevaluation
22 at market values. Utilities are valued by the
23 market, those publicly owned utilities, but you
1 cannot get a value for individual facilities apart
2 from book values. The market will make it very
3 clear as to the values attributable to transmission
4 versus generation versus better control, better
5 coordination, all sorts of information features that
6 are simply lost now. It may lead to new products,
7 new ways of delivering and contracting for
9 And finally, the restructuring is said
10 to lead to appropriate signals for long-term
11 investment, perhaps more investment in storage
12 capacity, perhaps more investment in incremental
13 changes to upgrade our ability to transmit power.
14 I don't want to go into the analysis
15 issues as such. They're long; they're tedious; and
16 they're written in "economese," I guess, but I do
17 want to mention a couple of things on the empirical
18 analysis issues.
19 What I'm talking about here is how big
20 are the potential benefits. How much might we get
21 out of restructuring? And that's why we look at
22 these sets of questions.
23 So the question then is: are the
1 utilities actually minimizing the cost of the
2 services they provide? Are they operating the
3 proper facilities in the proper sorts of ways?
4 If, in fact, the great benefit from
5 competition is prices will get closer to either
6 nodal or marginal costs, then how much do those
7 prices vary during the day over seasons, by
8 location? If they don't vary, then there's less
10 Is demand really elastic over the
11 distribution of nodal prices? Can you really reduce
12 demands 15, 20, 30, 40 percent for a few hours and
13 then have them met at other times when there's
14 plenty of capacity? If you can't, the restructuring
15 benefits are lessened.
16 Even if there are large benefits, and
17 most people, I think, believe there are, what are
18 the actual magnitudes of the metering, the
19 communication, control, and contracting costs
20 associated with moving to the new world? Are they
21 large enough to swamp the potential benefits?
22 And finally, what are the opportunities
23 and costs of changing transmission capability? The
1 FERC strategy for making the industry competitive is
2 based upon transportation, getting competition
3 through transportation. Whether or not and how or
4 under what terms you can improve your transportation
5 system is key to how effective this is going to be.
6 Right now, EIA is running four projects,
7 three of which made it on this graph, one of which
8 didn't. So let me start with the important project
9 that was inadvertently left out.
10 The first one deals with the status of
11 FERC and PUC regulations with the status of new
12 entrants into the industry and takes a preliminary
13 look at transportation capabilities, of
14 opportunities for expanding transportation
15 capabilities. This is ongoing now in the CNEAF.
16 The second project, which is prospective
17 and has just been started in a preliminary way,
18 deals with the efficiency of individual generating
19 stations. This is an attempt to use NERC data to
20 look at what used to be called the X efficiency of
21 the operations of plants. How well that project
22 succeeds depends upon the data. Right now we don't
1 And finally, there are two modeling
2 projects. The first, which Art is going to talk to
3 you about a lot, deals with simulating marginal
4 costs, pricing, plus adjustments to marginal costs
5 under a regime where you're assuming elasticities
6 and a lot of other things you would like to know
7 instead of having to just assume.
8 And another project deals with the
9 effect of removing regulatory constraints on how our
10 models would view investment over time.
11 I think these projects, on the whole,
12 are going to be very informative and successful.
13 I'm quite excited about Art's in particular, but I
14 think what we're going to realize after they're done
15 is that our database is still a bit thin, and so I
16 think for prospective projects the first thing we're
17 going to want to do is get a better handle on how
18 much the prices vary and how much transition
19 metering costs might really amount to.
20 I think we're going to find out that
21 with these networks, the costs and effects of
22 alternative ways to increase transmission capacity
23 is at least as important, if not far more important,
1 than generation capacity at least for the next ten
2 or 15 years.
3 I think also we're going to discover in
4 a very profound way how difficult it is to estimate
5 demands with the information we have available.
6 There's never been anything like the spot market or
7 see anything approaching a spot market for
8 electricity in this country. Ergo, there is no
9 data, at least local data, or very little.
10 The other thing we're going to have to
11 start worrying about is scenario construction. By
12 the time these projects are done, we will have had
13 basically three years under a slowly moving, perhaps
14 slowly moving change to a restructured environment.
15 We should be in a better position to talk about
16 where this is all going to end up, what the
17 equilibrium outcomes might be like, but right now
18 it's too early, but maybe then. In order to do our
19 analysis, we're going to have to make some guesses.
20 I think finally we'll have to do what
21 we're going to have to do at every juncture in this
22 project, and that is to reassess where we are. This
23 is clearly a one step at a time learning process.
1 I just want to speculate a little bit as
2 to what would come next, and I think in this Phase 3
3 we would start emphasizing doing serious demand
4 estimates. Right now, as I say, we are just making
5 up elasticities. In the foreseeable future I think
6 we will be able to sort through which assumptions
7 are sensible and which are not, but we're still far,
8 far from making serious attempts to estimate
9 elasticities in the context of this sort of new
11 I think we're also going to have to face
12 the network issues head on. I am proposing that we
13 build three, a couple perhaps toy network models
14 maybe representing California, the Northeast and the
15 Midwest to try to get some feel for how much
16 difference the network effects make for the
17 forecasts and the analysis and the analytical
18 results we get using our standard tools.
19 If those effects are large, I suggest
20 then we adjust our models for these network effects
21 or try to. That may not be possible. In the
22 distant future, perhaps we will, you know, replace
23 what we have now with a more network based approach
1 to electric prices. I don't know. My guess is I
2 will have retired by the time we get to that. So we
3 will find out.
4 I'd like to return to my questions one
5 more time. This is an incredible area of analysis
6 and activity. EIA cannot do it all, should not do
7 it all. We have to find targets of opportunity
8 where we can add particular value to the ongoing
9 debates, and one place to start in making those
10 sortings out is to get a better idea of the work
11 that's going on.
12 So once again, anything you all can do
13 in your own particular knowledge working in the
14 research organizations to inform us of what's going
15 on would be a great help.
16 And with that, thank you very much.
17 CHAIRMAN MOUNT: Thanks a lot, Doug.
18 We go to the second paper, "Forecasting
19 Electricity Prices in a Competitive Environment,"
20 Art Holland, Office of Integrated Analysis and
22 MR. HOLLAND: Good morning. I'm Art
23 Holland. I appreciate the opportunity to speak with
1 you this morning.
2 I'll be describing the method that we're
3 developing in the Office of Integrated Analysis and
4 Forecasting to model the price of electricity
5 generation services under competition.
6 We're up to Number 2, Doug.
7 I've been asked to review the questions
8 for the reviewers, and they are: what analysis,
9 data or modeling products do decision-makers in
10 industry and government need as they face the
11 uncertainty of the future electric power industry?
12 How can we best use the integrated nature of NEMS to
13 serve them?
14 We are in the process of developing a
15 model of a quasi-spot market pricing mechanism. Are
16 108 pricing periods enough to simulate a spot
18 And the next step is to develop a
19 contract pricing mechanism, or maybe to develop a
20 contract pricing mechanism. Should the contract
21 price be the average spot market price plus some
22 insurance premium? Should this premium be based
23 upon the volatility of the spot prices, assumptions
1 regarding risk aversion and the avoidance of
2 transaction costs? Are there other factors that we
3 should consider?
4 Now, the restructuring of the electric
5 power industry is not just an electricity issue.
6 Fortunately the Office of Integrated Analysis and
7 Forecasting is able to use NEMS for this analysis
8 effort, and the strength of NEMS is in its unique
9 systems integration feature.
10 Now, let me point out before I go any
11 further that what I'm going to describe to you today
12 is not integrated into NEMS yet. It's in the test
13 phase. We're doing it off line, and the purpose of
14 this phase is to determine how to integrate it into
15 the NEMS framework, what pieces to use, and how to
16 hook up all the wires.
17 On the left in this diagram of NEMS,
18 you'll see the supply components; on the right, the
19 demand components; and in the center, the conversion
20 components and the system integration piece. What
21 NEMS allows us to do is gain insights into how
22 changes in the electric power industry will affect
23 other U.S. energy markets, like coal and natural
2 So with that, I'll try to describe
3 something about what we're doing and how we're
4 calculating these competitive prices.
5 Now, there are at least two, and maybe
6 three, components. Again, this is the test phase,
7 and keep in mind that this is generation services
8 only. Transmission and distribution are very
9 important, but we're going to save those for later.
10 The first component is the energy
11 component. Now, the energy component is based on
12 the short run operating cost of the last plant
13 dispatched. There are 108 dispatching periods per
14 year in each region in NEMS.
15 Now, some people believe that in an
16 over-capacity situation like we are in right now,
17 this is the only determinant of prices. Now, we
18 anticipate that most of the times that will be true,
19 but what that means in our algorithm is the
20 reliability component, which is next, and that may
21 not be the best name for it, and I'll talk a little
22 more about that, will be small at least on an
23 average annual basis until this over-capacity
1 situation is relieved.
2 Now, the reliability component may
3 better be called the market clearing component. I
4 have to give Doug credit for that name, and it's
5 based on two general calculations or values. First
6 is the value that consumers place on reliability,
7 reliability specifically of generating supply, and
8 the second thing is the reduction in unserved energy
9 that's contributed by each kilowatt of generating
10 capacity in the region.
11 Let me repeat that and talk about it a
12 little bit because it's a slippery concept if you're
13 new to it. That's the reduction in unserved energy
14 contributed by each kilowatt of generating capacity.
15 Now, unserved energy, think of that as a bad thing.
16 That is the extent to which demand exceeds supply,
17 and anyone that knows anything about electric power
18 knows that that can't happen. So this is an
19 expected value that we're calculating in the model.
20 Now, one of the questions that I'm asked
21 about this component is: how will it be integrated
22 into the prices that consumers see? There are two
23 possibilities that I can envision.
1 First would be an independent system
2 operator in that type of a world enforcing some
3 system reserve requirements.
4 Another way that it may get in there
5 would be through contracting mechanisms where
6 consumers purchase greater levels of reliability
7 and, therefore, pay a higher price for their power
8 on a per kilowatt hour basis.
9 The third component -- I'm going to talk
10 more about the reliability or market clearing
11 component before we're done -- the insurance and
12 convenient component, if that's used, will be for
13 end use consumers who enter into service contracts
14 to avoid the risks and transaction costs at the spot
15 market. Now, I don't know that we're going to do
17 One of the purposes of the current test
18 phase is we're going to look at the results we get
19 and then make a determination of this convenience
20 insurance component.
21 Now, the first component that I talked
22 about, the energy component of price, is on the
23 overhead. Now, I know you can't see the numbers,
1 but I wanted to show you this to give you an idea of
2 the range of costs that we're looking at that are
3 translating into prices, and the costs are color
5 Now, there are 108 rows in this graph.
6 Those are the 108 dispatching periods per year in
7 NEMS. Now, each of those dispatching periods is
8 characterized so that we're trying to capture the
9 entire year in load characteristics that exist on a
10 time basis throughout the year.
11 The columns, there are 13 columns.
12 These are the 13 regions that the model uses for its
14 Wherever you see white on the chart is
15 where the cost or the short-run operating costs of
16 the last plant dispatched, which is the price
17 setting plant for this component, is less than two
18 cents per kilowatt hour. Where you see yellow, the
19 short-run operating cost of the last plant
20 dispatched is between two and four cents a kilowatt
21 hour, and everywhere you see red, it's greater than
22 four cents a kilowatt hour.
23 This tells us that we're in the ball
1 park of expectations. These are the kinds of
2 numbers that we expected to see for the last plant
3 dispatched to meet load throughout the year.
5 Now, the more contentious piece. The
6 reliability or market clearing component has three
7 steps basically. There are four up here, but one or
8 two are done in the model simultaneously.
9 The first step is to calculate unserved
10 energy for the region for each of those 108 time
11 slices for which plants are dispatched in the model.
12 Now, again, unserved energy is not a good thing.
13 That's the difference between supply and demand when
14 demand exceeds supply.
15 I've been asked why would demand ever
16 exceed supply, and of course, it can't. The idea is
17 that it will not exceed supply if it's priced to
18 reflect shortages properly, and that's what we're
19 endeavoring to do with this component of price, and
20 that's why market clearing component might be a
21 better term. It raises the price of electricity
22 when demand approaches the capacity limits in order
23 to communicate through prices that capacity
2 The unserved energy is calculated based
3 on the capacity of each generating plant in the
4 region, expected planned enforced outage rates for
5 each generating plant, and the hourly loads for the
7 Then what we do is we increment capacity
8 a small amount, maybe ten megawatts, and do it
9 again. This shows us the change in unserved energy
10 for a change in generating capacity, and the result
11 of those two calculations gives you the reduction in
12 unserved energy that you get for each megawatt of
13 capacity in the region, and the assumption is that
14 every megawatt of capacity that's available
15 contributes to the same degree in that reduction of
16 unserved energy.
17 If you take those numbers, take the
18 total capacity in the region and an assumed value of
19 unserved energy or, on the other side, the cost of
20 unserved energy, multiply those together, and divide
21 by sales, it gives you the reliability or market
22 clearing component of price.
23 I think some sensitivities, and to show
1 you some preliminary results, might help to clear
2 that up a little bit. Now, this was in the early
3 stages of the model. We didn't have the 108 time
4 pieces. All we had is an average annual number, and
5 as I said, in the current situation of over-
6 capacity, we're going in assuming that that
7 reliability or market clearing component will be low
8 because demand should very infrequently approach
9 your capacity limits in the region.
10 Now, this is New England, and I selected
11 New England because if you look in the far right-
12 hand column, you'll see that with the original data
13 on an average annual basis, the reliability
14 component of price we were getting was three and a
15 half cents per kilowatt hour, which is way too high.
16 That's out of the ball park, and something was
17 wrong with that.
18 So we did some sensitivities. The first
19 thing we did is we doubled the number of generating
20 plants in the region and halved the capacity of
21 each, which means you have the same capacity, but
22 it's broken up more finely. That should lower your
23 calculations of unserved energy, which means an
1 increase in capacity will have a smaller effect on
2 that unserved energy number, which means your
3 reliability component of price or your market
4 clearing component of price should be lower when you
5 do that, and sure enough, it dropped from three and
6 a half to 2.2 cents a kilowatt hour, but that's
7 still too high.
8 Then we started looking at some of the
9 data that was feeding into the model, and we saw
10 that in the original data, the availability rates
11 that were assumed for nuclear power plants were 65
12 percent. That's way too low. It's more reasonable
13 to plug in a number like 80 percent. So that's what
14 we did, and bang, that number dropped from three and
15 a half cents to .7 cents a kilowatt hour. So
16 clearly the model is very sensitive to the
17 availability of your large base load plants, which
18 it should be.
19 Oh, let's skip the next one, Doug.
20 Now, what I'd like to do is show you
21 some of the early results. We've just started
22 getting some what I think are credible results out
23 of this model this week. So this is very early.
1 We've got a lot more analysis of the results to do.
2 So please don't run out and quote me on these, and
3 that's why I'm using 1995, so that we can't call
4 them forecasts, and we also know something about the
5 industry in 1995, and we don't know what's going to
6 happen later.
7 Now, the assumptions in these results
8 you're seeing are the value of unserved energy is
9 $12 a kilowatt hour, and assumed elasticities of
10 demand are minus .15.
11 Now, I have two graphs which are lined
12 up here. The top graph -- and the reason why
13 they're lined up is so you can see how supply and
14 demand relationships, which are on the top,
15 translate to prices, which are in the bottom graph.
16 The top line, the dark blue line, are
17 your seasonal capacity numbers. Those are adjusted
18 for your maintenance schedule and for your firm
19 trades. Under that, the lighter kelly green are
20 your hourly demands. Now, keep in mind that the
21 peaks that you see in demand -- and those are broken
22 up, as you can see, under that by season -- we don't
23 know when those peaks are going to occur. They
1 could occur any time during the month, but we want
2 to get them in there so that they're represented in
3 the model.
4 Under that, you'll see the red line or
5 the red spikes are the reliability or market
6 clearing component of price. You'll notice that
7 those line up whenever your green demand lines start
8 getting very close to your blue capacity lines.
9 Under that -- and these, by the way, the
10 three lines on the bottom are stacked. So the top
11 red line shows you the spot price of electricity at
12 that point or what we're calculating.
13 The green line is the energy component
14 of price based on the short-run operating costs of
15 the last plant dispatched, and you'll see that's
16 fairly flat in ERCOT. I should have mentioned this
17 is Texas, for those of you who don't know the North
18 American Electric Reliability Council Regions.
19 The bottom line is your transmission and
20 distribution component of price, blue. Now, as I
21 mentioned earlier, we're just using the NEMS numbers
22 for that, which for distribution and transmission
23 means that that's the average imbedded cost. So
1 it's going to be flat throughout the year.
2 Now, if you take those green, red, and
3 blue lines and you average them out through the
4 year, your T&D component, which is the same as in
5 the regulated cases, 1.2 cents a kilowatt hour; your
6 energy component is two cents a kilowatt hour; and
7 the reliability or market clearing component
8 averaged throughout the year is .1 cents. That
9 comes out to 3.4 cents per kilowatt hour. This is a
10 very early number, and there are still some things
11 that we need to figure out how to get in there, but
12 that does compare to a 1995 price of six cents a
13 kilowatt hour for this same region coming out of the
14 cost of service regulation method that's in NEMS
15 now. So we are getting big drops in the price of
16 electricity using this method.
17 Another reason I wanted to show you,
18 again, very early, preliminary results that probably
19 still have some problems with them, this is for the
20 MAAC region. Now, MAAC, for those of you who aren't
21 familiar with NERC, is Pennsylvania, New Jersey,
22 Delaware, Maryland region. The lines are the same,
23 seasonal capacity, under that the demands. But
1 you'll notice the difference here is that we're
2 getting a rougher, a higher volatility energy
3 component of price. So the reliability or market
4 clearing component of price isn't kicking in as
5 much. That energy component is enough to drive the
6 demands down away from the capacity limit. So we're
7 not getting that red piece in there as strongly.
8 And, again, a price comparison,
9 regulated cost of service was eight cents in NEMS in
10 '95. This totals to 4.6 cents.
11 And, finally, no discussion of
12 electricity restructuring would be complete without
13 looking at California, and I don't know if we should
14 get any insights from this or not, but I was struck
15 with how flat and how low the price of electricity
16 in California is from this graph. There's very
17 little volatility in the price, which suggests they
18 have certainly in excess of capacity there, and
19 they're able to use their cheaper generating plants
20 to meet their demands.
21 Again, a comparison. I was able to pull
22 nine cents per kilowatt hour out of the Electric
23 Power Monthly. That's all of California. This
1 region isn't exactly California. There are some
2 slight geographic differences, and there may be some
3 modeling inconsistencies, but 9.6 is what came out
4 of NEMS. So you've got nine to nine and a half
5 cents as a regulated price, and these totaled to 4.7
6 cents per kilowatt hour.
7 Doug, you might want to put the
8 questions back up.
9 Again, thank you very much for the
10 opportunity to speak with you this morning, and I
11 look forward to your comments.
12 CHAIRMAN MOUNT: Well, there wasn't a
13 rush from the Committee to be a discussant for these
14 papers. So I hope there are going to be some very
15 insightful comments after my rather bland remarks
16 are over.
17 I think it's fair to say that it's much
18 too early to be able to answer the questions that
19 have been posed. However, I think that the overall
20 approach to this problem is correct. Basically
21 people are going to be interested in prices and
22 demand. Are prices really going to go down? Who
23 benefits? What's going to happen to sales?
1 In many parts of the country with high
2 reserve margins, the need for new capacity is not
3 immediate, and the problems about investment
4 incentives then can be looked up as a sort of next
5 issue that ought to be addressed.
6 So I have four issues that are
7 mentioned, but I think are more important maybe than
8 they are in the paper. The first one is regulatory
9 burden. I don't think it's a surprise that the
10 areas of the country that are interested in
11 competition are the Northeast and California.
12 There's an awful lot to regulatory burden. Some of
13 it is very positive, shall we say doing things about
14 the environment, but some of it really, certainly in
15 New York, has been a way of essentially taxing
16 people indirectly using utility companies, and this
17 has led to frustration from major customers of
18 electricity who think that they ought to be able to
19 get power a lot cheaper, and I think that the answer
20 is that they probably can.
21 But there's one small component of the
22 regulatory burden, and that's the collection of data
23 that also is very important for this. So that I
1 think that EIA needs to make a major review of the
2 types of data that are being collected to answer the
3 most obvious question that one's going to get from
4 the Hill about what is actually happening and are
5 prices going down.
6 So I think that given the financial
7 pressure that the utility industry is facing, having
8 a serious review of data collection is sort of
9 politically sensible. I think that given the fact
10 that we're going to be looking at new forms of
11 financial markets, being able to gather this type of
12 information electronically is a real possibility.
13 So the burden will not be as great as it might be
14 without that sort of mechanism.
15 But I think that the most important
16 thing is that EIA is going to need new types of
17 information on prices that are not being gathered at
18 the moment, and I think that what we have to do is
19 to try to avoid getting into the predicament that
20 there was with natural gas prices that a lot of the
21 market was missed, not covered.
22 So I really congratulate EIA for sort of
23 looking at these issues and trying to grapple with
1 them early at a time when it is not obvious what
2 exactly is going to happen.
3 So the second issue that I want to talk
4 about is restructuring rates that customers pay, and
5 I suppose here one might look at a parallel with
6 what happened in the airline industry. It costs me
7 about as much to go and visit my mom in England as
8 it does to make a business trip to come and visit
9 you folks down here. I also have a vast array of
10 different pricing systems and different "thises" and
11 "thats" and I contract with a broker, my travel
12 agent, to figure out what in the heck is going on
13 and tell me what to do.
14 But I think that the bottom line is that
15 the reason that I pay a high price to come to visit
16 Washington is because I live in a small town,
17 Ithica, and the reason that I don't pay much to
18 visit my mom is that I'm going on a big trunk line,
19 the Atlantic run, and I'm paying very competitive
20 rates, and I think that's basically what it's going
21 to be, that large industrial customers are going to
22 do pretty well or better than they are now, but I'm
23 not too optimistic how I'm going to come out of this
1 as a residential customer.
2 So the something that's very important
3 about the electric industry and that the nature of
4 the product makes it possible to use price
5 discrimination in a way that is completely
6 unrealistic for other types of products, so that we
7 can have prices that vary by day. We can have
8 prices for individual customers. We can price the
9 demand from individual customers so that you pay
10 different amounts for different quantities
11 purchased, and so essentially multi-part tariffs
12 that can be approximated by two-part tariffs.
13 But the ability to come up with a very
14 new, innovative way of pricing electricity is really
15 substantial, and therefore, just assuming that
16 getting information about the average price paid is
17 really inadequate for understanding how these types
18 of price changes are likely to affect demand.
19 I think that it's important to talk
20 about the sort of vision of what the institutional
21 structure of pricing is likely to be. Most of the
22 discussion about competition focuses currently on
23 generation, and the general view is that some sort
1 of real time spot market will emerge for generation.
2 It seems to me that packaged around
3 that, the same way that my travel agent protects me
4 from all of this stuff, there are going to be a vast
5 number of brokers representing suppliers and
6 purchasers. There are going to be many new forms of
7 financial derivatives, forward markets, et cetera,
8 et cetera, that are going to be important, and it is
9 the prices that customers pay that determine their
10 demand and the way that they pay those prices so
11 that customer charges are different form energy
13 With regard to the actual structure of
14 the industry itself, it seems to me that we're going
15 to have a competitive generation system, and then
16 we're going to add what I've called a regulated
17 wedge, which maybe covers transmission and
18 distribution, but clearly the utility industry is
19 very interested in maintaining the ability to
20 recover money for stranded assets, and how much
21 they're going to be able to do that is really the
22 political issue that hasn't been determined.
23 But I think that that ability, if it
1 exists, to be able to extract more than a
2 competitive price for electricity will depend on a
3 sort of quasi-regulated system existing for some
4 part of the industry, and the most obvious one is
5 for transmission, and I think we're going to end up
6 with the sort of bizarre situation where
7 transmission charges cover the cost of nuclear power
9 So I think that this really says that
10 trying to figure out what's going on is going to be
11 extremely difficult, and I don't think that there's
12 any easy way to say what type of data are required.
13 However, I think that there is a
14 difference between the sort of analysis that we're
15 doing at the moment at Cornell and the sort of
16 analysis that's being done at EIA.
17 A lot of the complication in the
18 presentation this morning is trying to come up with
19 the equivalent of a capacity charge. Everybody
20 knows what the energy charge is, the spot market
21 price. I think that there's a good argument that
22 maybe if there is an effective competitive market,
23 that capacity charges are not going to be needed as
1 a separate item. The system that the U.K. used, I
2 think -- I don't know what the polite word is, but
3 "hokey" would be a good one --
5 CHAIRMAN MOUNT: So the bottom line,
6 it's not clear to me that worry about how one ought
7 to measure capacity charges is important at this
8 time. I think that something that could be done now
9 with NEMS is to just assume that somebody is going
10 to get a hit somewhere. There's going to be some
11 stranded assets, and you're not going to have to
12 recover as much revenue.
13 My guess is the big advantage of that is
14 going to go to the industrial sector, and therefore,
15 an interesting question is if, in fact, rates get
16 lower in the industrial sector, what happens to
17 demand. Does this deal with some of the high
18 reserve margin problems that we have?
19 You know, this is a very sort of
20 straightforward analysis with the existing model
22 Then I have one technical argument with
23 Doug over his use of nodal pricing, point-to-point
1 pricing. Certainly the people that we've talked to
2 who are trying to set up a market in California for
3 the Mercantile Exchange think that zonal pricing is
4 the best that we'll be able to get to, but I think
5 that the distinctions of these two are technical.
6 Clearly one has to take into account
7 capacity constraints in some form to avoid the
8 problems that the U.K. had.
9 Given all of these complications with
10 rates, my guess is that the best data that exist are
11 currently at EIA, and I really hope that that stays
12 the way, that EIA is able to get the data that they
13 need in order to be able to monitor this very
14 complicated situation that we're getting into.
15 So the third point deals with customers
16 with large loads, and this is somewhat of a side
17 issue in a way, but I think that the utility
18 industry got into a problem rather like computing
19 groups that hung onto the mainframe and didn't get
20 into distributed computing; that the utility
21 industry did not realize that there were very real
22 efficiency gains from using combined cycle turbines
23 and, therefore, having a small distributed system
1 rather than large plants.
2 There are probably a few people that
3 still believe that we ought to be running an energy
4 system based on plutonium, but my guess is that that
5 view is dying out and that the potential advantage
6 of having this distributed system is that it makes
7 things like cogeneration a lot more feasible, and so
8 that the hope that existed under PURPA really can
9 manifest itself with new technology.
10 However, the customers that are most
11 likely to be able to do this are the ones that are
12 going to get all the breaks with competitive prices,
13 and so I personally think that there's a real
14 tension between trying to do things more efficiently
15 and worry about greenhouse gases, et cetera, et
16 cetera, and the fact that the large customers are
17 the ones that are going to be able to negotiate
18 lower rates in the systems that are being considered
19 at the moment.
20 And the final point I have is nuclear
21 power. Clearly the nuclear industry has made
22 tremendous improvements in the last couple of years
23 so that the many nuclear power plants are now
1 competitive in an operating sense, but there are
2 still large amounts of debt in the rate base of many
3 utilities, and this is a major stranded asset or
4 strandable asset, if you're working for the utility
6 So there's an awful lot of restructuring
7 of the industry going on, and it seems to me that
8 it's very important that EIA keep track of what's
9 going on to the nuclear industry. I think that the
10 important thing about the nuclear industry is that
11 this is not a question of just paying off what's in
12 the rate base.
13 I tried to think of an analogy for it,
14 and the best I could come up with was alimony, that
15 basically there's a real commitment to pay money to
16 these plants into the future and decommissioning and
17 the monitoring of those plants is something that
18 will make companies that have those liabilities
19 uncompetitive compared to companies that don't have
20 those liabilities.
21 So the ability to keep a regulatory
22 wedge so that you're able to recover costs for that
23 is one way of dealing with it, but the bottom line
1 is that I think it's very important that we don't
2 see a restructuring of the industry so all of the
3 nuclear pieces go bankrupt, and then I'm not too
4 certain who owns them.
5 But I think that the importance here is
6 that in my view the federal government cannot cease
7 to be heavily involved in dealing with the nuclear
8 industry, and that includes paying for a lot of the
9 costs that are going to come in the future.
10 Thank you.
11 So now we open it up to the floor.
13 MR. WATKINS: I have several comments.
14 So maybe I'll come up to the podium.
15 CHAIRMAN MOUNT: That's fine.
16 MR. WATKINS: First of all, a few
17 comments on Doug's paper and then one or two on
19 This is a very difficult exercise, of
20 course, but few in Doug's paper -- one of you
21 mentioned about Paul Joskow's estimates that, in
22 fact, electricity deregulation would not generate a
23 lot of benefits. I'm not quite familiar with the
1 paper of his that you're referring to, but it seems
2 that I'm more with your kind of results, that, in
3 fact, there would be substantial benefits.
4 I say that because the mere fact that
5 we're talking at all about a stranded asset problem
6 or perception of it is an indication that there are
7 a lot of assets out there that are uneconomic. If
8 they're uneconomic, then with a competitive system
9 they're not going to be built in the first place.
10 If they're not going to be built in the first place,
11 they don't get in the rate base or whatever, and so
12 your prices should be lower, and significantly
13 lower, other things equal.
14 It may be, however, that Joskow's
15 comments are more directed about what may happen
16 under regulation when, in fact, you have incentive
17 regulation and things like that. So if you do that
18 and if you take account of the fact that maybe some
19 of these stranded assets have reached the point in
20 their life in the regulated system where the
21 depreciation schedule is such that the contribution
22 of those, the cost of services diminishing, perhaps
23 that's how he gets this kind of result.
1 Doug, you mentioned the point about spot
2 elasticities and the fact that you're really data
3 short on that. Two suggestions there.
4 What we're really talking about is peak
5 shifting in that context, is to look at
6 international data where, in fact -- and, of course,
7 the U.K. would be the prime one -- where you do have
8 some data that may be useful.
9 And, secondly, there is quite a
10 literature on time of day pricing even with
11 regulation, and you may be able to use that.
12 A third point, I had a question for you
13 on this question of how pool delineation. How are
14 you going to handle that? I mean how do you
15 delineate what is the power pool in a regional
16 locational sense?
17 On your steps where you talked about --
18 on the latter part of your paper, pages 11 and 12,
19 the various step, when I read those steps I thought
20 that you have to or my guess would be that what you
21 had as a portion of Step 2, which has to do with
22 data collection, was more important than Step 1 that
23 had to do with the NEMS model in that I think if
1 you're going to grapple with this problem, your
2 collection of the actual data that will be emerging
3 or the mechanisms you have to collect those data --
4 I would put that as Step 1, not as part of Step 2.
5 A comment Tim made about capacity
6 allocation and the regulated wedge for transmission,
7 and it's frequent to think of that as being in the
8 kind of natural monopoly category, which is why you
9 may want to have regulation.
10 I think you can go a step further than
11 that, however. If you really had total
12 deregulation, and it's a step that natural gas is
13 yet to go to, that is, where rather than just, in
14 effect, renting capacity rights on the pipeline and
15 in this case an electric transmission facilities,
16 you, in fact, acquire property rights on the system,
17 and you had a capacity release market. You wouldn't
18 need to have regulation even of transmission
20 My final comment on Doug's paper is that
21 my guess is that if you do have electricity
22 deregulation, we're going to be surprised about how
23 pervasive it is and how quickly it takes place in
1 that you do have the natural gas industry to look
2 at, and everybody has been surprised at the extent
3 to which a competitive market has emerged in a way
4 and to a degree that hadn't been anticipated.
5 If you look at the recent studies or
6 papers, say, by DeVany and Walls on the way in which
7 all the prices, the natural gas prices across the
8 system from one end of it to the other, from Canada
9 right through to Louisiana, are linked and he degree
10 to which they're linked, that has been a surprise
11 that that has happened that quickly.
12 Now, dealing with Art's paper, I had one
13 question, and I will, notwithstanding our Chairman's
14 caveats have a stab at answering some of your
15 questions here, but these 108 pricing periods. Were
16 you talking about are they different prices at
17 different times of the day over different days or
18 the year or they're just different days in the year?
19 It wasn't clear from your paper just what these
20 time slots represented, the 108.
21 MR. HOLLAND: They were selected to try
22 to capture the full range of load characteristics
23 throughout the year.
1 MR. WATKINS: Okay. So they vary by
2 time of day.
3 MR. HOLLAND: They vary by time of --
4 MR. WATKINS: That's seasonal and time
5 of day.
6 MR. HOLLAND: Correct.
7 MR. WATKINS: Within the seasons. Okay.
8 MR. HOLLAND: Seasonal and time of day.
9 MR. WATKINS: Okay. I understand.
10 MR. HOLLAND: And they vary in the
11 number of hours, as well.
12 MR. WATKINS: Right. Well, on your
13 Question 1 about what analysis stage while modeling
14 products, I jotted down three things here. One is
15 price formation. The second one, I would suggest,
16 is plant utilization, and I'm including in that
17 investment in new plants, and the other one I jotted
18 down was this question of inter-regional trade that
19 may emerge, as you rightly pointed out, to a much
20 greater degree with deregulation, that aspect.
21 On your Item 4, I think there's too much
22 focus here on the spot price. With a deregulated
23 market, you could have a much greater array of
1 contracts, of different terms and conditions.
2 Tim was talking about the airline
3 industry and, you know, the great array there.
4 You're going to have that probably happen.
5 Also, if you have futures markets that
6 do develop, you may be able to sign up equivalently
7 for prices one, two, three, four years in advance
8 using those kinds of mechanisms. So I think your
9 concern with the spot price is too great. I think
10 you're going to have a greater range.
11 The discussion of the energy prices and
12 your market clearing mechanism, I wondered why that
13 wasn't simpler to think of in terms of just the
14 distinction between short run and long run marginal
15 costs, but I'm not sure I really understood your
16 simulations in terms of your methodology.
17 Could you put up the first graph again?
18 Was it New England? I mean your reactor results.
19 You know, you had the ones you had just done this
21 MR. HOLLAND: The ones with the spot
23 MR. WATKINS: Yeah, and the components
1 of the price.
2 MR. HOLLAND: The first one was ERCOT.
3 MR. WATKINS: Now, I mean, I'm not sure
4 I understand your methodology. The way I saw it
5 described in the paper because there's a gap between
6 the capacity and the peak demands, ostensibly both
7 the ex ante and ex post demand is satisfied. So why
8 is there a red component to the charge at all in the
9 context of your methodology?
10 And then if there is a red component in
11 there, I don't quite understand some of the
12 relationships. For example -- I'll have to point
13 here -- this one is almost past here. So why isn't
14 this --
15 MR. HOLLAND: Actually you should go to
16 the second season. That's the first price in the
17 second season. So you should look at the second
18 position of the top blue line.
19 MR. WATKINS: Oh, all right. So it's
20 not -- what I was going to say is suppose they were
21 aligned. You would expect this one to be much
22 higher. The gap here is squeezed. So maybe when
23 you align it, perhaps you get that pattern when you
1 align it, but my main question was I don't quite
2 understand how the calculations emerge if I
3 understand your methodology properly.
4 I think that's it.
5 CHAIRMAN MOUNT: Cal?
6 MR. KENT: Most of the things that I was
7 going to say have already been covered, but let me
8 just stress a few things that I think need to be re-
10 The first one is to thank Doug for a
11 very insightful paper. As an economist, I
12 appreciated reading it very much, and what you're
13 attempting to do with that, I think, is extremely
15 I did come away though with the feeling
16 that it's going to be very difficult for EIA to
17 respond even to your paper, much less to this
18 environment, because we don't know yet what it is
19 that we need to know, and that's very hard to plan
20 or to model or to even figure out what data it is
21 until we know what it is that we need to know, and I
22 didn't get the feeling that we were certain enough
23 in our knowledge of what was going to be expected,
1 what questions were going to be asked, that we
2 really knew what it was that we ought to be out
3 there looking for.
4 The second thing is I would reemphasize
5 what Campbell just said, and that was I think there
6 was an overemphasis in Art's paper on spot markets.
7 I think that the interesting play is going to get
8 to be the futures markets, which I think are going
9 to develop very, very quickly in this area, and that
10 that may alleviate some of the problems, Art, that
11 you talked about as soon as a future market
12 develops, and I certainly think that they are going
13 to be more important, as they have proven to be in
14 other energy sources, particularly petroleum
16 The next thing is just to make a general
17 comment. In one sense natural gas has kind of led
18 the way, and as I was reading between about four
19 o'clock this morning as I was flying in from Los
20 Angeles the paper on natural gas, it surprised me
21 that some of the questions that were being asked
22 there about what sort of data issues were being
23 raised by deregulation I think are going to be
1 exactly some of the same issues that you all are
2 going to be facing, and there may be quite a bit
3 more out of the natural gas deregulation that you
4 can mind for indicating to you or indicating what
5 direction you should go, certainly not to the degree
6 that took place in natural gas.
7 The stranded costs have occurred in
8 natural gas, and I'm babbling now. So let me see if
9 I can clarify that point.
10 The deregulation of natural gas has led
11 to stranded costs. They are certainly not of the
12 magnitude of the major nuclear power plants, but
13 particularly into East Coast or eastern area gas
14 producers, such as those in West Virginia, have had
15 terrible losses due to stranded costs that they have
16 not been able to recover because of the distribution
17 system that exists in the East as opposed to that
18 that is now in the South and certainly to the West.
19 And so I think that what you're going to
20 have is that since historically any time an industry
21 has become more competitive, whether it's through
22 the natural entrance of new competitors or
23 deregulation, there are always going to be stranded
1 costs that were there and that were only justified
2 because of a monopoly position, either a natural
3 monopoly position or a regulated monopoly position.
4 And I think that the other remark that
5 was made is that we are going to be surprised at how
6 fast we're going to see competition; that this
7 stranded cost issue may very, very quickly almost
8 get itself solved just because of the fact that the
9 market is going to force the issue much more quickly
10 than we may even be able to comprehend it, and my
11 suspicion is that the end result is going to be as
12 historically it has almost always been, and that is
13 to say to the people who are stuck with the stranded
14 costs, "You get to eat them."
15 And I think if you take a look at the
16 prices of shares of certain utilities today, you can
17 begin to see that to an extent the market has
18 already anticipated that result, and so I may be
19 saying that this whole issue of stranded costs may
20 be one that you're not going to have to worry about
21 modeling as much as you think because it's just
22 going to happen. There are going to be some losers.
23 It's going to be quick and merciless for their
1 stockholders, and then after that the market will
2 sort its way out.
3 And I may be wrong on that, but that's
4 just my suspicion of the way things are going to
6 Then the last comment, and I'll shut up,
7 is I think as Doug's point as made in his paper. I
8 think there's going to be a tremendous amount of
9 interest in what us economists call the transactions
10 costs, and I think this gets back to what Joskow was
11 talking about in his paper, if I remember it.
12 There's going to be phenomenal transactions costs
13 involved here, particularly in the rather extended
14 transition period we're going to go through, and
15 those transactions costs are going to be much more
16 interesting than the generating costs that we talked
17 about in the past, and that may very well be where
18 some of the efficiencies are lost, is just through
19 these transactions costs.
20 And you can go back to Coase's original
21 work in the area and everything like this, and I
22 think that a lot of the anomalies that you may see
23 developing in this market and are probably already
1 developing in the natural gas markets are due to
2 these transaction costs that we tend to overlook and
3 we tend not to model.
4 So that's my whole comment.
5 CHAIRMAN MOUNT: Thank you.
6 The rest of the Committee is silent. I
7 am amazed.
8 So how about members of the public back
9 there? Anybody would like to comment? Would you
10 like to respond?
11 MR. HALE: Oh, yeah, sure. Just first
12 thank you all very much for your comments.
13 First, just a couple of things.
14 Campbell and I will be talking some more this
15 afternoon, I'm sure, but there are a couple that are
16 worth talking about now.
17 First, the emphasis on spot markets,
18 spot pricing, I agree with Campbell that what the
19 consumers are going to see are not the spot prices,
20 okay, but the spot prices to the extent that you get
21 something like that happening is the grist for the
22 mill. I mean that is what all the futures
23 contracts, all the deals, all the packages are based
2 If you don't get the spot market prices
3 right, you should go home now because the rest of it
4 is just -- you've got nothing to work with. So
5 that's the emphasis on spot.
6 We're not all that optimistic. Well, we
7 find that real hard right now to even think about
8 getting the spot market prices right without going
9 to the next level of the packaging that the
10 entrepreneurs are doing right now.
11 Second, I think everybody has a lot of
12 priors about how this restructuring is going to go.
13 You know, as a personal opinion, I believe that
14 we're going to see a huge amount of efficiency gain.
15 I think we're going to see de facto deregulation
16 very, very quickly for industrials and commercial
17 and people who are lucky enough to be close to
18 borders between utility districts.
19 I think you're going to see more
20 attempts on the parts of utilities to insure their
21 monopoly position as best they possibly can be
22 mergers, acquisitions, joining up with natural gas
23 companies, joining up with the cable companies,
1 anything they can to control various aspects.
2 And I also think that what you're going
3 to see is, in fact, you won't see nodal prices.
4 You'll see something zonal, but we'll get to that in
5 a second.
6 I believe all this stuff, but when I
7 look at the data I got and I look at my models and
8 my analytical capability, I'm not really persuaded
9 that I've got other than a religious case to make at
10 this stage. I believe that the answer I've given is
11 the right one. I haven't seen it yet.
12 MR. WATKINS: Religious must be perfect
14 MR. HALE: Once again, I started out by
15 saying I know where dancing leads. I don't know
16 where this leads for sure.
17 Okay. I really don't see any conflict
18 whatsoever between environmental protection and
19 restructuring. I think that there's a real chance
20 that by pricing electricity much closer to its
21 social cost you're going to see much better resource
22 use, and I think you're going to see the old, ill
23 controlled facilities shutting down much quicker. I
1 think you're going to see electricity used the way
2 it should be, and I think the environmental benefits
3 could be substantial.
4 I think this is one where the
5 environmentalists are very, very wrong, and I hope
6 that I'm right on this one.
7 The issue of power pool delineation is
8 critical. It's the name of the game. I don't know
9 what to say. I mean the size of the market
10 determines how well this is going to work, and if
11 the market is expanding, the larger it is, you know,
12 the better for all of us.
13 There was a point made that perhaps in
14 the analysis agenda Step 2 should actually be Step
15 1, and I think the argument I'm making or the fact
16 is Step 1 is already underway, and what I was trying
17 to do in Step 2 is argue that we should return to
18 some of these data issues just as quickly as we
19 possibly can.
20 I'm not saying we shouldn't be doing
21 what we're doing, but the data issues are not going
22 to go away, and they are basic material for coming
23 to any conclusion at all or contributing to this
1 debate at all.
2 Joskow, I'll either get the proper
3 citation and quote or I'll send you a note
4 recanting. How's that?
5 So I really appreciate your comments and
6 suggestions on the paper, and I'm looking forward to
7 talking to you all privately on more technical
9 Oh, one other thing I do want to
10 mention. In looking at how spot prices are
11 calculated, it becomes very, very clear to you that
12 a lot of the pricing algorithms and a lot of the
13 dispatch operations that are done now by engineers
14 are done based on rules of thumb, and computer
15 software and control software that's good enough.
16 Why? Because making the software better, making the
17 calculations better, getting your approximations
18 more accurate doesn't make you any money in an
19 average cost pricing world. It's a wash.
20 Now I think there are going to be
21 substantial incentives for much better
22 approximations of the physical constraints, the
23 flows throughout the system, substantial incentives
1 to go from zonal, you know, to finer and finer and
2 finer zones because if you have a difference between
3 the accounting cost and the economic values, that's
4 a license to steal, and they're going to do it.
5 So that's my last pitch on that.
6 MR. CHATTERJEE: Can I ask? This
7 question is for Doug.
8 MR. HALE: Yeah.
9 MR. CHATTERJEE: Do you envisage the
10 effect of deregulation to be many small, efficiently
11 run utility companies, or do you see large mega
12 utility companies?
13 MR. HALE: I think you're going to see a
14 lot of gorillas, you know, fighting over each
15 other's territory a lot at the fringes. I don't
16 think you're going to see atomistic competition. I
17 just don't.
18 But, you know, the key elements to me,
19 at least, are what do they do with the transmission
20 systems, you know, and control. It has very little
21 -- and control has to do with what gets dispatched,
22 who gets turned on and off, and what conditions,
23 that sort of thing.
1 I think at least the control part is
2 inherently monopolistic and will remain that way and
3 will be regulated. Transmission I'm not so sure of,
4 at least long distance, you know, the high power
5 transmission. I'm not so sure what that is.
6 So I would imagine you're going to have
7 fairly large firms.
8 MR. CHATTERJEE: The difference in spot
9 pricing, of course, was diminished at the efficiency
10 of the transmission.
11 MR. HALE: Absolutely, absolutely. So
12 that's our best bet, I think. I think you're going
13 to have a lot of gorillas, you know, big firms, but
14 if we can get the transmission system working
15 properly, I think there's going to be very effective
16 competition amongst large firms. I just don't see
17 small firms being able to complete. I mean they
18 can't even afford the software to do the
19 calculations I don't think.
20 MR. WATKINS: I think, Doug, you should
21 draw a distinction between the generation and
22 distribution and transmission. There's greater
23 likelihood of a small firm or a more atomistic
1 competition in generation.
2 MR. HALE: Oh, absolutely. Yeah, I'm
3 sorry. See, I think the real question has to do
4 with transmission and control of the system. I
5 think control of the system is probably going to --
6 you know, the ISO or whatever you call it is going
7 to end up viewed as some sort of natural,
8 indivisible, you know, kind of monopoly that you're
9 going to have to have regulation. Transmission to
10 me is an open game still. I just don't know.
11 MR. WATKINS: I mean in generation the
12 efficiency of small turbines and the change
13 associated with that is very strong.
14 MR. HALE: No, I agree. I think in
15 generation you're going to have opportunities for
16 small firms, but I don't think generation is where
17 the action is going to be in this market. I think
18 you're right, but that's not where you're going to
19 make your money, in my opinion. We'll see.
20 MR. HOLLAND: I had just a couple of
21 comments I wanted to make about some of the things
22 that were said.
23 This issue about the emphasis on spot
1 prices, there are a couple of reasons for that. I
2 agree completely that most consumers are probably
3 not going to see those spot prices. They're going
4 to see contracts, and one of the things that we're
5 doing with the spot pricing mechanism or the spot
6 pricing algorithm is to try to calculate what those
7 contract prices are going to be.
8 There's a possibility we may even use
9 the spot pricing algorithm to develop a range of
10 prices over which contracts could exist. Given what
11 assumptions are done in the particular modeling run,
12 you may even think of the energy component as
13 defining the floor of the contract prices and the
14 reliability or market clearing component and
15 possibly the insurance component as defining the
16 ceiling of those contract prices, depending on how
17 much reliability different consumers want to
18 purchase from their generation suppliers.
19 A second thing or point for the spot
20 pricing is if you reduce the price of electricity by
21 restructuring the electric power industry and
22 nothing happens to your demands, what have you
23 accomplished? Possibly some of your big bangs are
1 going to come from load leveling that will change
2 your capacity planning decisions and also could
3 change the way you operate your system.
4 When Larry Ruff spoke at the NEMS
5 conference, he talked about a system where as the
6 limits of your capacity start to be reached, you
7 need some mechanism to jack the price up so that you
8 start getting people who don't value electricity as
9 much off of your system. That keeps you from having
10 to decide where you shed your loads.
11 It also is a good load management -- I
12 think pricing is possibly the best load management
13 system, but the energy component doesn't give you
14 enough of that. You have to have some other way to
15 calculate how much you increase your prices. So
16 this algorithm is trying to do that.
17 We talked about that it was mentioned
18 that there was no capacity. Some people are using a
19 capacity chart which is based on the cost of
20 building the capacity, but that's doesn't get at
21 what the demand -- what your consumers' value is,
22 how much they're willing to pay, what the market
23 will bear is, and that value of unserved energy is a
1 way to estimate or guess what the consumer or what
2 the market will bear.
3 That has to be a varying input
4 assumption because everyone has a different value
5 that they place on electricity during a blackout. A
6 good example might be to look at what a restaurant
7 owner in Georgetown might be willing to pay for
8 electricity on Friday night if he's got a dining
9 room full of customers and the lights go out, and
10 then you might look at a college student with their
11 boyfriend or girlfriend on Friday night sitting on
12 the sofa in front of the television and the lights
13 go out.
15 MR. HOLLAND: They are very different
16 amounts that they would pay to bring the lights back
17 on. So that's what that value of unserved energy is
18 trying to get at.
19 So what's all that I wanted to say about
21 CHAIRMAN MOUNT: Anymore comments from
22 the Committee? From the public?
23 So I thank you both very much, and we
1 are adjourned and we will meet back here at two
3 (Whereupon, at 12:22 p.m., the meeting
4 was recessed for lunch, to reconvene at 2:00 p.m.,
5 the same day.)
1 AFTERNOON SESSION
2 (2:05 p.m.)
3 CHAIRMAN MOUNT: I'd like to reconvene
4 the meeting of the ASA Committee, and there will be
5 two presentations on the residential and commercial
6 demand models in NEMS.
7 The first one is going to be John
8 Cymbalsky, Office of Integrated Analysis and
10 MR. CYMBALSKY: Okay. Thank you, Tim.
11 Is it okay if I talk like this for the
12 transcript? Okay.
13 As Tim said, my name is John Cymbalsky
14 with the Office of Integrated Analysis and
15 Forecasting. I've been with EIA doing residential
16 demand modeling for almost seven years now. That
17 covers, I think, seven AEOs I've done since I've
18 been here.
19 Today's presentation, I'm going to talk
20 about the NEMS residential model and especially
21 focus on the projections from AEO '96.
22 To give those of you who are not
23 familiar with NEMS a little overview of what it
1 does, okay, the NEMS energy model has 11 supply and
2 demand models, a macroeconomic activity model, and
3 an emissions model. This is a general equilibrium
4 model, and it iterates over the years to come to a
5 partial equilibrium for the energy supplies/demand
7 Now, talking about the residential model
8 in particular, it's not an econometric model. It's
9 an energy engineering economic modeling structure.
10 It's very data intensive. It has information about
11 the housing stock, the equipment stock, appliance
12 efficiencies. We have technology choices. We have
13 building shell efficiencies, and all of these
14 numbers vary by the nine Census divisions and three
15 building types.
16 We also model at the end use level, and
17 we have about seven end uses, including lighting,
18 heating, cooling, water heating, and the like.
19 Now, for AEO '96 and all AEOs that I've
20 been involved with, we run generally three cases for
21 the AEO. We have the reference case, which I'm
22 going to call the business as usual, and what that
23 means is that we assume that policy will remain
1 unchanged as it exists today for the next 20 years.
2 We also have the low and high world oil
3 price cases where we perturb the oil price
4 projections and see what the response in the energy
5 markets are.
6 We also do low and high economic growth
7 cases, and we perturb GDP and see what happens,
8 again, with the energy markets in these scenarios,
9 and for the last two AEOs, we've also used stand
10 alone technology cases for both the residential and
11 commercial models. In these cases, what we did is
12 we made assumptions about technology, whether it's
13 going to be a high technology case, which is the
14 best technology available, or frozen technology case
15 where we freeze technology at its 1995 levels and
16 then we want to see what happens, you know, when you
17 make different assumptions about technological
18 progress in the future.
19 Okay. My first graphic here is just
20 showing historical and projected residential primary
21 and delivered energy. You can see the primary
22 energy line versus the delivered energy line has
23 been diverging over the past, say, ten years, and we
1 project this to continue, and this is just
2 increasing electricity penetration in residential
3 market, and when you convert that to primary, you
4 can see that the lines diverge even more.
5 Okay. This graph now shows primary
6 energy intensity for historical and projected
7 residential energy demand. You can see that on a
8 per household basis, energy consumption per
9 household has been coming down for the past 20
10 years, and also -- and this is due to efficiency
11 standards and the like -- and our forecast you can
12 see also has it slightly declining, but the thing to
13 note here is if you look on the square foot basis
14 and the per household basis, you can see the square
15 foot basis, it's declining more in the forecast, and
16 that's due to the fact that we're projecting bigger
17 households in the future in terms of the physical
18 square footage.
19 Okay. This graphic shows historical and
20 projected primary fuel shares in the residential
21 sector. From the graph you can see that in 1970
22 electricity and gas almost had the same share in the
23 residential sector, and in 1994 you can notice how
1 different the fuel shares have changed. We have an
2 increasing electrification, and again, this is
3 primary energy. So electricity has, you know,
4 dominated the market in terms of market penetration
5 in the last 24 years, and we project this to
6 continue for the next 20 years or so.
7 Now, if we look at the end use energy
8 for the residential sector, you can see that space
9 conditioning, which is heating and cooling, is the
10 largest end use in the sector, has been and will
11 continue to be in the projection period.
12 You can see that the other category is
13 all of the other categories that aren't listed, and
14 this is mostly miscellaneous electric appliances.
15 You can see that grows enormously in the projection,
16 and this is due to the fact that they're all
17 electric, and there's no efficiency standards for
18 most of these appliances. So as they penetrate the
19 market, they just increase over time.
20 And another important fact here is
21 refrigeration as an end use, even with the addition
22 of the number of refrigerators in the stock in the
23 past up till now, you can see the actual consumption
1 by refrigerators has declined, and this is due to
2 the federal energy efficiency standards for
3 refrigerators in both 1990 and 1993.
4 Okay. This next slide shows housing
5 growth in the pie, in the projection period from
6 1994 to 2015, and what we did with the pie is we
7 wanted to break out and show you where in the
8 country the housing growth is projected to be in the
9 next 20 years.
10 You can see that the South has 47
11 percent of the growth in housing in the projection
12 period, and to show you, you know, what's going to
13 happen with fuel use for the next 20 years, it's
14 important to know regionally what fuels are being
15 used, for space heating especially since it's the
16 biggest end use, and you can see that in the South,
17 which has 47 percent of the builds, it also has 56
18 percent space heating for electricity and only 43
19 percent for gas.
20 Now, nationally -- well, for the rest of
21 the nation the trend is different. It's 68 percent
22 for gas and only 24 percent for electricity. So
23 this is very important in the modeling to know that
1 the region in the country where the houses are being
2 built is very important in terms of which fuels will
3 be used in the future.
4 MR. LOCKHART: Is that primarily the
5 heat pumps?
6 MR. CYMBALSKY: Yeah, heat pumps mostly
7 are in the South. That's right. So most of the new
8 construction in the South are using heat pumps.
9 Whereas the rest of the country tends to go with gas
11 MR. LOCKHART: Okay.
12 MR. KENT: Is that the delta that you're
13 talking about there or is that the stock that you're
14 looking at?
15 MR. CYMBALSKY: Right. It is the
16 accumulation of all the new stock from 1994 to 2015.
17 MR. KENT: But it isn't anything -- I
18 mean what you're saying is 2015, this is where we're
19 going to be. It's not just talking about what the
20 change is?
21 MR. CYMBALSKY: It says that between '94
22 and 2015 if you build 20 million homes, of those 20
23 million that's the percentages.
1 MR. KENT: Okay. So it is the change.
2 MR. CYMBALSKY: Right, correct.
3 Yeah, and in the South the reason they
4 use heat pumps obviously, they have lower heating
5 loads than in the North. The fuel price is less
6 important. So, you know, they'll go for
8 Okay. Again, here this is just going to
9 reestablish the growth in the households that I
10 mentioned in the last slide, and you can see that in
11 the middle, the south Atlantic, east south central,
12 and the west south central are the three big growers
13 out of the nine Census divisions, the south Atlantic
14 especially because that's already the biggest Census
15 division in terms of households and energy use, and
16 you can see that's going to grow the biggest, and it
17 already has the most houses. So that's a major
18 impact in the forecast.
19 Okay. This is the energy intensity
20 change from '94 to 2015 as a growth rate: primary
21 fuel consumption per household, and again, you can
22 see refrigeration on a per household basis,
23 refrigeration as an end use we project to decline
1 over three percent per year in intensity, and again,
2 "other" is all of the miscellaneous electricity end
3 uses which do not have federal efficiency standards
4 associated with them. So they basically will grow
5 as they penetrate into the housing market.
6 This graph just shows to give you an
7 idea of the intensity for each of the three housing
8 types that we have in the model in 1994. There's
9 really no surprise here. Single family, you know,
10 they're the biggest houses; they use the most
11 energy. Multi-family, they, you know, tend to be
12 apartments and smaller. So they use the least.
13 Okay. As I mentioned before, we had
14 three side cases that we ran or two side cases that
15 we ran relative to the reference case in this year's
16 AEO, and this graph is just going to show over time
17 the difference from the reference case that
18 different technology assumptions -- what impact they
19 have on consumption in the sector, and you can see
20 that if there were on efficiency improvements
21 relative to 1995, we would use about one and a half
22 quads more by 2015 relative to the reference case.
23 But if we employed best technology
1 choice in the next 20 years, there's a potential for
2 over three quads of savings in the sector by 2015.
3 And when we run these cases, basically
4 the cases are run such that economics are not a
5 factor in the decision making for the technologies
6 for the side cases, but what I've found interesting
7 and people always ask is, "Well, what if you looked
8 at the investment costs for these technologies? How
9 much would they cost and what would be the
10 incremental investment needed for the savings?"
11 So what this graph does is try to show
12 you what level of investment in certain appliances
13 is in the AEO reference case and then how much more
14 it would cost to get the savings that I just
15 mentioned in the previous slide.
16 The top line shows in the AEO reference
17 case what Americans are paying for the fuel costs
18 for these appliances, and you can see it's around
19 $80 billion in 1991 dollars, and then the next line
20 down is the capital investment associated with the
21 AEO reference case. So you could see that Americans
22 are paying somewhere around $30 billion for this
1 And then relative to the base case, we
2 want to look at the best technology case and ask
3 ourselves, well, how much incremental investment and
4 how much incremental savings we would get from
5 running the case, and you can see the red line. In
6 the first years you need a lot more investment to
7 get the fuel cost savings, and then around 2005 you
8 can see that the lines intersect, and by the end you
9 have more fuel cost savings than you do investment.
10 Now, there is no discounting of the cash
11 flows here. So this is just undiscounted money.
12 And the last slide is to look at this
13 same thing, but on an end use level for electricity
14 and gas. So you could see that the important thing
15 to note on this is the bottom graph -- excuse me --
16 the top graph shows that in electric water heating
17 it's the only end use where the savings is greater
18 than the investment, and this basically is because
19 of the heat pump water heater that is out there and
20 is a viable technology. It's not be purchased much
21 now, but the fuel cost savings associated with using
22 this electric heat pump water heater far outweigh
23 the investment needed.
1 So you can see by all of the end uses
2 sort of what would be needed in terms of investment
3 to get the savings.
4 Okay, and I think I'm out of time. I'll
5 turn it over to Erin.
6 CHAIRMAN MOUNT: The next presentation
7 is going to be by Erin Boedecker, Office of
8 Integrated Analysis and Forecasting.
9 MS. BOEDECKER: I'm John's counterpart
10 on the commercial side of the house, and so I
11 project energy demand in the commercial sector,
12 mainly dealing with building energy.
13 Today I'd like to look at the trends
14 that are shown in AEO '96 as far as commercial
15 energy use and intensity, looking at both primary
16 energy, which includes the electricity losses from
17 production and distribution, and also delivered
18 energy; look at what types of fuels are used and how
19 those fuels are used. then I'd like to look at some
20 of the factors that contribute to the trends, look
21 at the end uses in more detail, look at different
22 areas of the country, and also look at the different
23 types of businesses that are portrayed in the model.
1 And finally, I'd like to present the
2 future energy savings that are projected in AEO '96
3 in the reference case, and also show the potential
4 for additional savings if the commercial sector were
5 to change the way that they make their investment
7 In the commercial sector, we're mainly
8 talking about building energy, and so for our
9 intensity measure we typically use BTUs per square
10 foot instead of per capita or some other measure.
11 Over the forecast, you can see that
12 energy intensity is projected to decline, and as far
13 as primary energy, this reverses the historical
14 trend for the last 20 years. In the model, this
15 decline is due to efficiency gains represented
16 through standards, such as the Energy Policy Act;
17 voluntary programs, such as EPA's Green Lights and
18 Energy Star Programs; and also efficiency gains in
19 electricity generation.
20 This year the shift in focus from
21 delivered energy to primary energy has emphasized
22 the commercial sector's reliance on electricity, and
23 this dependence is expected to continue just like it
1 is in the residential sector.
2 I'll explain a little bit about this
3 slide before I talk about it. The numbers far to
4 the left are total commercial energy consumption.
5 The top of the slide is for 1994 and the bottom is
6 for 2015. The pie charts show us energy consumption
7 by fuel, and here you can see that electricity is
8 expected to maintain its share at 73 percent. The
9 right-hand side columns show you the energy
10 consumption by end use.
11 Now, the fuel shares are projected to be
12 stable, but the composition of end uses is projected
13 to change slightly.
14 As you can see, office equipment, which
15 includes PCs, copiers, faxes, that type of
16 equipment, and also the other use category, which
17 includes such emerging technologies as medical
18 imaging technology and telecommunications equipment,
19 are both expected to continue to penetrate the
20 commercial market.
21 We made more changes to the model this
22 year to more accurately reflect the popularity and
23 advances in these areas.
1 On the other hand, the percent of total
2 energy consumption and also energy use per square
3 foot, as is shown on this slide, both decrease for
4 uses that have reached market saturation, such as
5 space conditioning, water heating, and lighting.
6 These declines reflect the efficiency
7 gains that I spoke of earlier, and also in water
8 heating, they reflect the shift from electricity to
9 natural gas.
10 The efficiency gains are also expected
11 in office equipment, but these gains are
12 overshadowed by the project growth.
13 This slide shows fuel shares by end use,
14 and this more clearly reflects the shift in water
15 heating from electricity to natural gas as a fuel.
16 Again, it shows the sector's reliability on
17 electricity, and it shows that the changing end use
18 composition does play an important role in trends,
19 but there are also other factors that affect energy
20 consumption, and we'll see whether they reinforce
21 the national trend or counteract each other.
22 Just as John looked at different areas
23 of the country, we will for the commercial sector,
1 too, and energy use in different areas of the
2 country is affected by both climate and
3 demographics. The commercial sector does operate at
4 the Census division level, which there are nine of,
5 and the south Atlantic division is the largest
6 sector. It covers the East Coast from Maryland to
7 Florida, and it's also projected to grow the
9 The only division in the commercial
10 sector that is not expected to grow is New England.
11 Looking at energy intensity by area of
12 the country, it's expected to decline in all areas.
13 However, there are two areas of interest on this
14 graph. If you look at the east south central
15 division, which includes Mississippi, Alabama,
16 Tennessee, and Kentucky, it shows very little change
17 in intensity relative to the other areas of the
18 country, and New England shows declining use in
19 energy per square foot even though there's no growth
20 in commercial floor space.
21 This can be explained at least in part
22 by trends exhibited by another contributing factor,
23 which is building or business type. In NEMS we have
1 11 commercial building types, and energy use per
2 square foot does vary by business category. Food
3 sales, service, and health care have the highest
4 intensities, while warehouse space has the lowest.
5 Remember in the last slide that energy
6 intensity decreased very little in the east south
7 central division. It turns out that health care
8 related space in this division is expected to grow
9 at twice the national rate, and it happens to be one
10 of the higher intensity end uses and also is
11 projected to decline in intensity the least.
12 At the same time, revisiting New
13 England, while no growth is expected overall,
14 warehouse space, which has the lowest energy
15 intensity, is projected to increase close to one
16 percent per year over the forecast period.
17 This slide just projects floor space by
18 building type kind of for completeness since we
19 showed it for the Census divisions. You can see
20 that the mercantile and service category, which
21 contains everything from retail businesses to auto
22 repair, starts with the largest share of floor space
23 and is also expected to show the highest rate of
1 growth. It retains a much greater share of floor
2 space than most of the energy intensive businesses
4 The next slide shows the considerations
5 that go into choosing equipment in the commercial
6 sector, which in turn affects the energy use. We've
7 already looked at the AEO '96 results by end use,
8 Census division, and building type. Another factor
9 that certainly affects energy use is the payback
10 period consumers require to cover their investment.
11 In NEMS we have a distribution of six
12 implicit or observed discount rates ranging from
13 just under 20 percent, which is roughly a five-year
14 payback period, to a rate representing those who
15 only consider capital costs in their purchase
17 The effects of commercial sector
18 discount rates can be observed in the technology
19 scenarios for AEO '96. Just as in the residential
20 sector, for the commercial sector we ran two stand-
21 alone cases with the 1995 technology case assuming
22 no efficiency advances after that year. The
23 efficiency gains represented in the reference case
1 compared to that case represent 2.4 percent of
2 energy savings by the year 2015.
3 However, the potential for much greater
4 savings exists in the high technology case where
5 commercial consumers would choose only the most
6 efficient technology available, regardless of the
7 cost. If the commercial sector started to base
8 investment decisions more on energy use than the
9 cost of equipment, an additional 12 percent savings
10 could be expected by 2015.
11 However, the price paid for energy by
12 the commercial sector is only expected to rise one
13 tenth of one percent per year over the forecast
14 horizon. So many changes would have to occur before
15 that potential could be realized.
16 Okay. That's the end of my
17 presentation. It was requested that we come up with
18 a few questions to focus discussion. Since this was
19 an informational briefing, we had a tough time. I
20 did pull a few out.
21 The first is more specific. The
22 presentation focused on energy use per square foot
23 as the measure of intensity in commercial buildings.
1 What additional measures of intensity should be
2 investigated and would add some type of insights or
3 give a better picture of energy use, if there is a
4 better picture?
5 And the second question is up for grabs.
6 Is there a statistical method of any type that
7 would reconcile the inherent differences between a
8 short-term, econometric type model and a long-term
9 engineering economic simulation model when your
10 forecasting years overlap?
11 This is something that we keep grappling
12 with and would like help with if there's any help
13 out there.
14 Thank you very much.
15 CHAIRMAN MOUNT: So the discussant,
16 Campbell Watkins.
17 Thank you very much, both of you.
18 MS. BOEDECKER: Thank you.
19 MR. WATKINS: I don't know whether
20 there's any help out there, but you'll be pleased to
21 know that I did at least anticipate the second
22 question in some of the remarks I want to make.
23 Both these models are very comprehensive
1 and complex and contain a lot of fine detail, and to
2 my mind or at least to my knowledge are the best
3 models available to capture a lot of the things that
4 you're trying to capture at this time.
5 As you just indicated in one of your
6 questions there, there are a mix of what I'd call
7 the engineering and econometric approaches, and one
8 can but hope that the outcome of this marriage is
10 It reminds me a little bit of a story
11 about an Irish playwright of the first half of this
12 century, George Bernard Shaw, and he was popular on
13 the circuit at that time. They used to invite all
14 of the famous and the literati to weekend
15 gatherings, and one of the persons who used to host
16 these events was a person called Lady Ottiline
17 Morrell, and she became very enamored of George
18 Bernard Shaw and said, "Well, you know, really we
19 should get married because look at what talented
20 children we would have. They'd have my looks and
21 your brains."
22 And he said, "Well, madame, what if the
23 child had your brains and my looks?"
1 So I can but hope that the outcome here
2 is useful to us. I cannot delve into all the fine
3 details of the model. So what I'm going to do is
4 focus on several design aspects.
5 Concerning particularly the interaction
6 between the engineering and econometric approaches,
7 I'm going to focus on the models themselves rather
8 than your projections because, after all, the
9 projections flow from the models, and if I focus my
10 attention on the models, that is more valuable, I
12 I'm going to talk mainly about the
13 residential model. However, several of the comments
14 I have with respect to that carry equally over onto
15 the commercial mode, and I will, however, finish
16 with one or two comments on the commercial model
18 Let me deal first of all with the
19 consistency between the econometric and engineering
20 aspects, and I want to discuss just briefly three
21 aspects. One is the price elasticities; secondly,
22 the issue of technological change; and, thirdly, the
23 issue of reversibility.
1 As I read the description that I have of
2 the models dealing first with price elasticities,
3 the space heating and cooling energy consumption are
4 assumed to be the components that are affected by
5 prices in the short run, and the short run price
6 elasticity of minus 0.15 is currently employed.
7 That to my mind raises five issues:
8 firstly, whether other end users, such as water
9 heaters, dryers, and washers, would not be affected
10 as well; whether the impacts are confined to just
11 own price effects or whether they include cross-
12 price effects; whether the short run elasticity is
13 very strictly speaking short run in being restricted
14 to intensitive (phonetic) use of the existing
15 stocks; how the longer run dynamic impacts of price
16 changes are accommodated; and also the origin and
17 nature of the short-run price elasticity used.
18 Now, typically a short run elasticity
19 picks up not just changes in intensity of use in the
20 short run and whether it's one year or whatever, but
21 we'll talk about one year, but also it does pick up
22 the first period impact on replacement and new
23 appliance stocks.
1 Let me just by illustration put up --
2 Richard, if you could just put up a simple model
3 here, this is just purely for illustrative purposes.
4 I'm not suggesting this is the kind of model you
5 would want to use, but it's just to illustrate a
6 point I want to make.
7 This is a simple dynamic, double
8 logarithmic model where the variables are expressed
9 in logarithmic terms, and that being the case, the
10 coefficients of the model represent elasticities.
11 So the short-run elasticity that you see there for
12 price would be the coefficient B. You can show that
13 the long-run elasticity will be given by the ratio
14 of B divided by one minus C.
15 C is analogous to the retention effect
16 or maybe in certain models it can actually be one
17 minus the rate of depreciation. So that if C were
18 zero, that means all of the adjustment is in the
19 first period, and so the long-run and short-run
20 elasticities are the same.
21 The larger C is, the longer the period
22 of adjustment to a given price change.
23 Now, the point I want to make is that
1 you have a short-run elasticity which for the sake
2 of argument we'll assume is B, is minus 0.15. You
3 have a price change either up or down. You plug
4 that into your model for the given category of
6 Now, suppose that higher price or lower
7 price is sustained in the next period. If I read
8 the model description correctly, you wouldn't
9 register any change induced by price because there's
10 been no change in the price between Period 1 and 2.
11 However, if you think of the dynamics of the model,
12 what you do is you have another impact from the
13 first period price change, which is how you get all
14 of these dynamic impacts to asymptotically become B
15 over one minus C eventually as the thing works
16 through, if the price is maintained at that level
18 Now, what is not clear to me is whether
19 that kind of dynamic impact is imbedded in the
20 model. If so, it's not clear from the descriptions
21 that I have read as to just how it does emerge.
22 The direction of the impact, of omitting
23 the dynamic impacts, if, in fact, they are omitted,
1 is in the direction of reducing the impact of price
2 changes either up or down, but we'll see in a moment
3 another comment I have is maybe an offsetting
4 element to that.
5 The short-run elasticity reference that
6 I have for how do you get the minus 0.15 was taken
7 from a report done by Carol Dahl for EIA about a
8 couple of years ago. I checked the reference there,
9 and it said that that was the elasticity for natural
10 gas across all sectors. So if my reference is
11 correct, the question that arises is to what extent
12 is that appropriate for the residential sector
13 rather than the average across all sectors, and
14 secondly, the extent to which the natural gas
15 elasticity is appropriate for other fuels.
16 As an order of magnitude, when I see
17 that kind of number, it doesn't kind of trigger
18 something in my mind that says, "Hey, that's too
19 high," or too low. It's just I'm trying to
20 reconcile your source, apparently source, cited for
21 that elasticity and the component of consumption to
22 which you're applying it.
23 Technological change. As I understand
1 the model, these are largely treated autonomously.
2 An example here would be how he treated the building
3 shell efficiencies that gradually improve over time.
4 The concern I would raise is the interaction
5 between that and any econometric elements of the
6 models, and that is because if you're using
7 econometric estimates of, say, price elasticities,
8 they themselves are influenced by changes in energy
9 efficiency over time.
10 I mean it's no surprise that interest in
11 energy efficiency became stimulated by the
12 substantial increases in prices in the 1970s and
13 early 1980s. So to the extent to which any
14 econometric elasticities are affected by changes in
15 energy efficiency and, therefore, embody them, and
16 then you have an autonomous technological change
17 that also is depicting improved efficiency, there
18 could be some form of double counting there.
19 And if that were the case, that would
20 tend to over-estimate impact. So it would be the
21 opposite side of the coin to the way that I've
22 suggested, a lack of attention to the long-run
23 dynamic aspects may underestimate.
1 Reversibility. That issue arises if you
2 have an increase in energy prices followed by a
3 decline in energy prices or the other way around.
4 Are these responses reversible? In other words, if,
5 say, the price goes up and then returns to its
6 original value and nothing else changes, do you get
7 back to consumption at the original value?
8 There are various reasons why that would
9 not be so. Irreversible technology, and maybe you
10 capture that in model specification; mandated
11 efficiency standards that don't revert, are not
12 responsive to changes in the conditions. There are
13 new operational modes, become imbedded. There's
14 changes in what is terms bounded rationality,
15 whereby, say, in the 1960s nobody cared too much
16 about energy budgets, and in the 1970s they became
17 very concerned about those, and when prices went
18 back down in real terms to figures that are not far
19 off what they were in the late '60s, early '70s,
20 nevertheless, people are still very aware of energy
21 costs and, therefore, would not revert to their
22 original behavior.
23 The other aspect is that just because
1 prices have switched direction in terms of price
2 expectations, people don't necessarily think that's
3 going to be the case throughout. So expectations
4 may affect things as well.
5 These are all reasons why you're not
6 going to necessarily get reversibility. I don't
7 know what sort of dynamics are imbedded in the
8 models, but maybe by experimentation you can see
9 whether, in fact, this imbedded assumption in the
10 model that assumes reversibility. If so, I think
11 it's something that should be adjusted.
12 Let me make some other comments. On the
13 residential model, there's two housing vintages
14 where you kick off from. There's pre-1991 and post-
15 1990. For the future you keep track of the start by
16 vintage, but as I read the description in the
17 residential model -- there's a different aspect in
18 the commercial model which I'll come back to -- you
19 seem to treat the 1990 stock on a kind of average
21 If that is the case, it seems to me you
22 could by a process -- as long as you have data on
23 housing starts, you could kind of backtrack and get
1 the historical vintages of the housing start.
2 You mentioned that the equipment
3 retirement rate in your model depends on a linear
4 decay function. I assume that the way that function
5 operated is it operated between the minimum and
6 maximum lives that you have, and you weren't
7 implying that you'd have an attrition from the first
8 year. That would be like a flat function and then a
10 All equipment in the residential sector,
11 except that for space heating, seemed to be replaced
12 by the same equipment types, although that may well
13 have improved efficiency, but I think that
14 assumption is a bit restrictive in that water
15 heaters, washers, dryers, and ranges could all
16 switch as well.
17 The real discount rate that you used of
18 20 percent in life cycle costing for choosing among
19 technologies to my mind appears a bit high, but that
20 may have been picked based on observations about
21 consumer behavior, and I'd be interested in your
22 comment on that.
23 I'm not sure I fully understood the
1 relationship between the macro model and these
2 components, both the residential and commercial
3 sector in that as I understand it energy prices are
4 exogenous to the two residential and commercial
5 models. What I'm not clear on is the way in which
6 changes in the residential and commercial energy
7 demand aggregates, which of course are a sizable
8 proportion of the total aggregate, react on the
9 price projections.
10 You've got around very neatly in the way
11 the problem of what I call simultaneity in
12 identification, but I wasn't clear on the
13 relationship between aggregate price formation and
14 the demand in the individual sectors.
15 I also think there's a hidden assumption
16 in the model, and you touched on this in your
17 presentation, John, that changes in mandated
18 efficiency standards always act to reduce energy
19 demand. That's kind of intuitively appealing, but
20 there are quite a lot of references in the
21 literature to why the impact of mandated efficiency
22 standards may be muted, and that is because if you
23 increase the efficiency of an appliance in a
1 mandated way, what you do is make the output BTUs
2 cheaper for the user because he doesn't need to buy
3 as many input BTUs to get the same level of service.
4 So you have reduced the price of output
5 BTUs. That being the case, some of the efficiency
6 impacts on energy consumption can be muted and even
7 conceivably exceeded by the fact that people will
8 use the appliance more intensively or they may buy a
9 bigger appliance.
10 So I think that indicates there's some
11 caution to be used in assessing the impact of
12 changes in efficiency standards. It depends on how
13 your model operates.
14 Now, let me finish by making a few
15 comments about the commercial sector that are solely
16 related to it. A lot of the comments I've made on
17 pricing dynamics, elasticities, et cetera, carry
18 over equally to the commercial model.
19 The first one that's really commercial
20 specific is cogen electricity. That is included, at
21 least for commercial establishments, in the
22 commercial aggregate data. My suggestion would be
23 it would be preferable for you to exclude that and
1 put it in the power generation sector.
2 Discount rates, I see those as treated
3 in a much more elaborate way than in the residential
4 sector. You've got 11 consumer time preference
5 premia over risk free rates. The question that
6 brings to my mind is why that sort of or why several
7 premia might not also apply in the residential
8 sector, and there is some literature on that.
9 Floor space vintaging. As I read the
10 description, your pre-1989 floor space are
11 calculated by what you call back-casting, and I
12 think maybe that is the kind of technique that could
13 be applied to the residential sector if it hasn't,
14 in fact, been done because what you've done is back
15 off from 1989, your earlier vintages, by using the
16 floor space data that you have for earlier years as
17 additions to floor space.
18 Energy price projections. I saw in the
19 description where you're doing the full life cycle
20 analysis that you depend on what you call foresight
21 routines, that kind of myopic or reductive or
22 perfect foresight. What I didn't understand was
23 that you already have price projections from the
1 macro model, and there seems to be a distinction
2 between the price projections that you're using for
3 the life cycle analysis and the price projections
4 that may flow from the macro model. I would have
5 thought you were going to make them consistent.
6 A final comment is on DSM links. I see
7 that the commercial sector model does make specific
8 reference to DSM linkage. No such treatment is
9 advertised for the residential sector, although I
10 may have missed it in the description, but if, in
11 fact, it is not included, then I think it should be
12 because particularly in, for example, the
13 electricity sector, the impact of DSM programs is
14 certainly noticeable.
15 CHAIRMAN MOUNT: Thank you, Campbell.
16 So anybody from the Committee who wants
17 to add to Campbell's comments? Dan.
18 MR. RELLES: These aren't exactly adding
19 to Campbell's comments. They're taking a different
21 There's a lot of statisticians here, and
22 we'd be guilty of malpractice if we let you talk
23 about the year 2015 and we just sat here and didn't
1 raise the specter of uncertainty. I want to say a
2 little bit about that.
3 I know, on the one hand, you do deal
4 with uncertainty by varying your base cases, and you
5 said that up front, oil prices, GDP, and so forth,
6 and in fact, the one time I tried to run NEMS on the
7 Web, I was also given some choices. Do I like these
8 data values? Do I want to change them?
9 So I appreciate that, but there are
10 other sources of uncertainty that don't get
11 reflected in these projections. Coefficients have
12 gotten estimated. Surveys have gotten compiled and
13 models estimated based on those, and I guess I'd
14 like to see some discussion in these things about
15 uncertainty or even see NEMS provide the capability
16 of providing some kind of uncertainty measures.
17 I don't think it's feasible to think
18 about reprogramming NEMS to do that. I appreciate
19 that it's awfully hard to communicate uncertainty in
20 a bar chart, but the kind of uncertainty I'd be
21 willing to live with would be if I could run NEMS;
22 if I could plug my assumptions in and rum NEMS five
23 or ten times, having it each time vary the
1 parameters a bit and letting me, you know, select my
2 output and just try to see how it varies over those
3 different ten iterations.
4 That idea of varying things ten times
5 and looking at it has gotten well accepted into the
6 statistical literature. Concepts like multiple
7 imputation in the end basically say, "Go ahead and
8 re-impute this database five or ten times and see if
9 there are major changes."
10 So it's a fairly well accepted notion,
11 and I guess I'd like to put a plug in for instead of
12 having it ask me do I want to change this data
13 value, put a plug in to have it ask do you want me
14 to change all of my parameters by amounts suggested
15 by random variation, and I would most certainly say
16 yes to that because I really have no idea of what
17 the uncertainty is when it comes out, when out comes
18 a projection.
19 MS. COX: Just to remark on his remark,
20 and that is in terms of what he was saying about or
21 what was said about modeling uncertainty, it is a
22 good point for some of the things that I do when I'm
23 just guessing on the size of parameters or what
1 effects might be, et cetera. I'll go in sometimes
2 and say, "Well, okay. I thought the cost ratio was
3 going to be this, but suppose it's ten percent more
4 or 25 percent more."
5 If you don't see a whole lot of
6 reaction, you feel pretty good, but if it's bouncing
7 all over the place, then you start wondering and not
8 feeling a lot of confidence in the results.
9 CHAIRMAN MOUNT: Well, I think it's very
10 important that you both raised this issue of
11 uncertainty since this is a favorite topic of this
12 Committee, and I think it has got lost.
14 CHAIRMAN MOUNT: And I think it's
15 important to bring it back again to the forefront.
16 PARTICIPANT: I'm a little surprised it
17 wasn't on the agenda.
19 CHAIRMAN MOUNT: But there's a point
20 that I would like to raise about the use of end use
21 models. In particular, the fact that you are using
22 a relatively high discount rate in order to
23 approximate what is actually going on suggests that
1 people are not making energy efficient decisions
2 when they make appliance choices, and I wondered
3 whether you could say a little bit more about this
4 particularly for new construction.
5 Is it because, for example, the people
6 who are making the decisions are not the people who
7 are actually going to use the buildings or are there
8 other reasons?
9 I think that with a model like this, the
10 possibility exists unlike many econometric models to
11 actually address this type of issue and to say how
12 well are we making energy appliance and equipment
13 choices now, and in particular, it's surprising how
14 much penetration is being -- of electricity into
15 space heating and things like that, given the prices
16 that we pay for it in the Northeast particularly,
17 where heat pumps don't work very well.
18 So ny other Committee members want to
20 Anybody from the public want to make a
21 comment? Would you like to respond?
22 MR. CYMBALSKY: I don't know where to
23 begin with all of those comments. I'll begin with
1 the one I can answer.
3 MR. CYMBALSKY: The discount rate
4 question has been around forever. So why don't we
5 go for that one first?
6 As Campbell mentioned, he said the
7 commercial model segments its population into 11
8 different discount rates. Six? And his comment
9 was: why doesn't the residential sector do the same
11 And tying into what Tim said, why do we
12 have these high discount rates? Well, what we do in
13 the residential model is we have one segment of the
14 population. That's it. We don't segment it by
15 owners or leasers or builders. We have one, but
16 what we do have is good shipment data from the
17 appliance manufacturer. So we actually know what's
18 being purchased on the market.
19 I'm going to agree with Tim and say,
20 well, if you have the new construction, they tend to
21 do things differently than someone who is going to
22 live in the house and they stock their own
23 appliances. They purchase higher efficiency goods.
1 That's definitely true.
2 I think the problem we have is
3 estimating different discount rates in these
4 different classes. So we used one discount rate per
5 appliance and said, well, the average of the whole
6 residential sector is, for instance, 69 percent for
8 Does that mean everyone has a discount
9 rate of 69 percent? No, but when a builder builds a
10 house, he doesn't care what the efficiency level is.
11 So he throws in a bad one. A consumer going out
12 may buy a better one, but on average, you get the 69
13 percent. So that's what we do there.
14 Uncertainty? I'm certain that I'm
15 uncertain about pretty much everything you said
16 there. We did have a project on uncertainty, and I
17 think with budget considerations that actually did
18 not go through in our plans.
19 You know, to address his point about
20 uncertainty, you can take the residential PC models
21 and perturb a lot of different things, not just this
22 data input set. I think you're referring to the
23 spread sheet where you have different housing levels
1 maybe or different prices.
2 You can change a whole lot of things.
3 It doesn't have a switch to just say, "Change them
4 all by ten percent." You know, that capability may
5 be a little bit tough to do, and I'm not sure how
6 the model would respond to something like that. It
7 may blow up; it may work. I've just never done it,
8 so I don't know.
9 Let's address the housing stock issue in
10 terms of the vintages. There are data that are
11 available through RECs that segment the existing
12 housing stock into different vintage levels. When
13 we were developing the model, we decided to only use
14 the two vintages, basically what existed the year of
15 the last survey and whatever else is going to be
16 built in the forecast period. So we have two
17 vintages and we keep those separate over time.
18 We can segment the '90 stock, you know,
19 by different vintages. I don't know what that would
20 really buy us in terms of our projections. That
21 would be very data intensive. You could see how
22 many pieces of data we already had, and then you
23 would just segment all of those numbers again by how
1 many vintages you would want in the existing
3 The things that would come to mind, you
4 could probably track age of equipment a little bit
5 better. You can track how efficient the building
6 shell is for different vintages of housing a little
7 bit better. Yeah, it's a fair concern. I'm not
8 sure in terms of the amount of data needed in the
9 model. It would just expand the dimensions by, you
10 know, probably four or five.
11 The reversibility issue, I think this is
12 the last one I'll talk about. Maybe Erin wants to
14 The reversibility of all the equipment
15 purchases, they are irreversible, and you know, once
16 you make a decision to improve your housing shell,
17 for instance, you don't make the decision to take it
18 back out. So once there's a price response, say, to
19 increased insulation in your home, once you do that,
20 it's done forever no matter what else happens to
21 price after that time period.
22 And I guess I'll end up with my comments
23 on talking about what we call the rebound effect,
1 and you mentioned, Campbell, that, well, if you do
2 these efficiency gains, your outputs go down, so you
3 should have some sort of rebound effect. That is
4 captured in the model now, in fact. So I think the
5 version of the documentation maybe didn't include
6 that, but we do have the rebound effect in the
8 MS. BOEDECKER: I'll be very brief, but
9 to continue what John was just talking about, the
10 commercial model, too, did just incorporate a
11 rebound effect for this year. So we do take that
12 into account to a certain extent.
13 And as far as reversibility goes, as
14 long as a piece of equipment meets a standard, the
15 next time that person goes out to buy equipment,
16 they can go back to a less efficient piece of
18 Also, as in the residential sector, if
19 they make shell improvements, those improvements are
20 there to stay, and they won't rip the house back
22 I'll just address one more thing since
23 John touched on most of the other things. On the
1 price projections, noticing that the forecast prices
2 used in the life cycle cost calculation were
3 different than those that we get from the macro
4 model for our consumption calculations, the forecast
5 prices used in the life cycle cost are supposed to
6 represent the consumer trying to make a forecast
7 into the future, what they expect the prices will
9 The prices we get from the macro model
10 we get every year. So if we are running an
11 integrated run of NEMS, those fuel prices that we
12 get from the macro model may change from one year to
13 the next, depending on how demand and supply does
14 change. So they will not be, quote, unquote,
15 foresight prices until the year of the model run.
16 CHAIRMAN MOUNT: You've got one follow-
17 up here?
18 MR. GRACE: One simple comment, maybe
19 not simply; simple in theory, perhaps more difficult
20 to implement. We've had a lot of success in the
21 uncertainty area in dealing with these deterministic
22 models to build -- I'll say simply build -- a shell
23 around the model that allows either in a more
1 sophisticated sense drawing in a Monte Carlo from
2 distributions or simple up ten, down ten, up five,
3 down five, but you don't end up having to screw
4 around with the model guts. You're just building a
5 big shell around it that says: go to this array.
6 Draw from some values. Maybe it's a distributional
7 array or maybe it's just some nominated array, but
8 the mechanics of doing it aren't as daunting as they
9 might seem if you're thinking, "Oh, gosh, now I've
10 got to make this whole general equilibrium model
12 And you get pretty much the same
13 effects, at least in stability and some sort of idea
14 about the uncertainty surrounding it, and they
15 should propagate forward in time.
16 CHAIRMAN MOUNT: So we've had a proposal
17 to change the plan slightly and have a break now
18 before Jerry's presentation. So let's have a 15-
19 minute caffeine shot and be ready for Jerry.
20 (Whereupon, a short recess was taken.)
21 So first I've got an announcement for
22 Committee members who are going to dinner this
23 evening, that we were not able to get into Le
1 Rivage, and we're now going to go to 701
2 Pennsylvania Avenue. This is not the same place.
3 It's 701 Pennsylvania Avenue. We have the private
4 dining room, and we have to order before seven
5 o'clock to get the theater meal. So be there on
6 time, please.
7 MS. COX: To get what?
8 MR. HAKES: The cheap prices.
9 MS. COX: Oh.
10 CHAIRMAN MOUNT: The cheaper prices,
11 less expensive prices, and there's a very important
12 announcement for your TV pleasure this evening, that
13 Jay Hakes is going to be live on CNN at ten o'clock
14 to tell us why gasoline prices are going up.
15 MR. HAKES: Any help would be welcome.
17 CHAIRMAN MOUNT: He will be soliciting
18 comments at dinner.
19 MR. LOCKHART: Will it be the same story
20 as this morning or a different story?
22 CHAIRMAN MOUNT: So the first
23 presentation now is an update on confidentiality by
1 Jerry Coffey from the Office of Management and
3 MR. COFFEY: Thank you.
4 As some of you know, I've been working
5 on bits and pieces of this for most of my
6 professional career. So sometimes I give quite a
7 long speech, but I'll avoid that today and give you
8 the quick version, which is the version that focuses
9 on the successes.
10 I'm going to talk about three things
11 today, two of which are now public, one of which
12 should be public some time in the next 48 hours, I
14 On Wednesday of last week, Alice Rivlin,
15 the Director of OMB, transmitted to Congress a bill
16 called the Statistical Confidentiality Act, which I
17 personally have been working on off and on for about
18 18 years. This bill does a number of things, and
19 I'll be happy to go through some of them.
20 Really, you all have copies of this.
21 Rest assured that this is the result of many, many
22 person-months of haggling with our attorneys, and
23 there are lots of subtleties in the language of both
1 the bill and the administrative order which bears
2 some thinking about, and I'm sure as you look at
3 these things and read them closely, there may be
4 questions that will occur to you.
5 What I would like to highlight are some
6 of the issues that are already being bobbled by some
7 of the pundits. The bill itself has a statement of
8 findings and purposes, which hasn't changed that
9 much over most of the time I've worked on it. We
10 really are after the same things, and in fact,
11 recently I pulled out some very old documents and
12 made copies of them for Cathy Waldman, the Chief
13 Statistician, to remind her just how long some of
14 these concepts have been around.
15 I had a page from the 1971 report of the
16 Commission on Federal Statistics, another page from
17 the -- actually pages from two different reports
18 issued in, I think, '78 and '79, one by the Privacy
19 Protection Study Commission and the other by the
20 Paper Work Commission, where you could find language
21 that looked like we had copied it verbatim, and in
22 fact, we did in some sections of this bill,
23 particularly the piece that was written by the Paper
1 Work Commission, which was the last of those three.
2 The privacy issue dealt with both
3 statistics and research and had some complexities in
4 it that we don't have to worry about when we're
5 dealing strictly with statistical agencies.
6 There are a series of definitions. Most
7 of them are fairly straightforward. A couple of
8 them I would note. We do use an unusual definition
9 of "agency." It includes most of the things that
10 most laws treat as agencies, but in addition, covers
11 some unusual agencies, like some of the groups in
12 the national laboratories are included in this, and
13 there's an explanation of why we did that in the
14 analysis of the bill.
15 The other concept over which we have
16 haggled endlessly for several years now is the idea
17 of an agent. What this really means, there are a
18 lot of words here. There are other words in the
19 administrative order. There are explanations in
20 both places. What an agent is in the simplest
21 terms, it's anybody who does anything helpful for a
22 statistical agency and whose behavior can be
23 controlled at least to the extent of assuring
1 information security.
2 The reason we have such a broad
3 definition is that statistical agencies have come up
4 with ways of solving this problem of how do we
5 handle contractors or how do we get people with
6 specific skills in that may help us work on a
7 particular project in so many different ways that we
8 needed a very broad definition in order to capture
9 them all.
10 Why do we want to capture them all? We
11 are proposing in this bill to make agents with this
12 broad definition subject to the penalties of the
13 Trade Secrets Act. For those of you who may not be
14 familiar with the Trade Secrets Act, it was actually
15 derived from a piece of tax law, an old BEA statute,
16 and one other which I'm not quite sure of many, many
17 years ago, and essentially it establishes criminal
18 penalties for employees or officers of federal
19 agencies who make unauthorized disclosures. It is
20 probably the one most general piece of the criminal
21 code dealing with disclosures and the penalties for
22 making inappropriate disclosures.
23 There are some other changes we are
1 going to make in that statute which I'll get to in a
3 But the concept of agent didn't really
4 authorize anything new. In the case of the Census
5 Bureau, they have a section in Title 13 that allows
6 them to sort of swear in people as if they were
7 employees. They use it to bring people in from
8 universities and lots of other folks and essentially
9 impose employee discipline on those people.
10 There are other statistical agencies
11 that are dealt with in this bill who have many other
12 different strategies. There are licensing
13 strategies. Certain kinds of contracts are written,
14 and the flavors of contracts are quite wide in their
16 There are in a few cases situations
17 where a federal agency actually pays the salary of
18 someone operating out in the state and supervising
19 employees of a state agency. This is a NASS
20 agriculture invention which they've had a long time
21 to think that one up, but it works very well. I
22 don't know if anybody else could come up with a
23 scheme like that any time in this century, but is
1 has worked very well for NASS.
2 The responsibilities of the statistical
3 data setters, which are the agencies that are
4 defined in this act who have special authorities to
5 get data for exclusively statistical purposes from
6 almost anyone. Their responsibilities are laid out
7 there. They're pretty straightforward.
8 The agencies themselves are named in
9 Section 4 of the bill. There's Bureau of Economic
10 Analysis, Bureau of the Census, Bureau of Labor
11 Statistics, the National Agricultural Statistics
12 Service, Department of Agriculture, the National
13 Center for Education Statistics, National Center for
14 Health Statistics, the Energy End Use and Integrated
15 Statistics Division of the Energy Information
16 Administration, and a very late-comer, the Division
17 of Science Resources of the National Science
18 Foundation. This was one of the very
19 late changes that we made to the bill in the last
20 couple of weeks. We've had a very interesting time
21 in the month of April.
22 The agencies that are named here are, as
23 I say, authorized to acquire information for
1 exclusively statistical purposes from almost anyone
2 else in government. There are a few exceptions
3 which are in the bill. Particularly national
4 security information is out of bounds. Information
5 which is controlled by a statute that itself sets
6 specific limits on how statistical information may
7 be used also have to be observed in any kind of
8 arrangement for sharing data.
9 The main body, the main policy of the
10 bill is in Section 6, confidentiality of
11 information, and this is where we have some of the
12 phrases that go back for decades. The basic policy:
13 "data or information acquired by a statistical data
14 center for exclusively statistical purposes shall be
15 used only for statistical purposes. Such data or
16 information shall not be disclosed in identifiable
17 form for any other purpose without the informed
18 consent of the respondent."
19 There are some slight changes in wording
20 that came along very recently in there. There are
21 provisions in here for situations that arise in
22 almost all of these agencies where information is
23 collected for both statistical and other purposes.
1 In those cases what we want the agencies to do, the
2 policy that we're establishing is that distinctions
3 of that type have to be made on the record by rule
4 before the data is collected. You can't collect the
5 data for exclusively statistical purposes and tell
6 all of the respondents that's what you're doing and
7 then change your mind later. You've got to do it on
8 the record before the data is collected.
9 This also contains some of the
10 exceptions, kinds of information that cannot be
11 disclosed. It also has a fairly elaborate procedure
12 for data sharing agreements, and this has a long
13 history, and every word has been fought over at one
14 time or another. I encourage you to read it. I
15 doubt if I could explain it in less than an hour.
16 One of the pieces of sub-text that you
17 should understand and which you don't always see
18 explicitly in the language of this bill, there are a
19 number of areas where this bill has been written to
20 dovetail with provisions of the Paper Work Reduction
21 Act. In some cases, we actually restate policies of
22 the Paper Work Reduction Act, one of them being
23 Section 6, Subsection F of Section 6, which
1 guarantees that any restrictions on the use of data
2 in law travel with the data to any other agency.
3 An interesting sidelight here is when we
4 originally picked up the language from the Paper
5 Work Reduction Act, I discovered that there was a
6 glitch in it. It didn't really work. So we fixed
7 our version, and in the last stages of the
8 reauthorization or the 1995 amendments to the Paper
9 Work Reduction Act, the House and the Senate decided
10 we were right, and they changed the Paper Work
11 Reduction Act to match this bill, though they didn't
12 even know this bill existed.
13 There is a section on coordination
14 oversight. Because a lot of this approach is
15 procedural, we're setting up ways of sharing
16 information, ground rules for sharing information,
17 permitting agencies to find the best answers for how
18 they manage information in this kind of an
19 environment. A lot of this we're going to have to
20 invent as we go along.
21 One of the things we learned over a
22 decade and a half is that we cannot write
23 prescriptions that solve all of the problems and
1 remain consistent with all of the statutes that are
2 on the books already. We tried that in 1983, and we
3 had a bill about this thick, and nobody could read
4 it; nobody could understand it; and certainly nobody
5 supported it.
6 So there is a process in here for
7 coordination of oversight. The main engine of this
8 will be the Director of OMB. There are a number of
9 tasks there, including the task of reviewing and
10 approving any rules adopted by any of the agencies
11 that are affected by this bill.
12 Any agency that donates information for
13 exclusively statistical purpose to one of the named
14 statistical data center, and of course, the
15 statistical data centers themselves, may, in fact,
16 be writing policies as regulations for how they are
17 going to manage this. OMB has the task of looking
18 at these and assuring that they are done in a
19 consistent fashion.
20 What this gives us is the opportunity to
21 look for the kinds of problems that have gotten in
22 the way for a long time when an agency that has a
23 solution that works reasonably well for them, but it
1 prevents you from doing something that's very useful
2 somewhere else, and we felt this was absolutely
4 Another important principle is generally
5 rules discovering disclosures of information that
6 are authorized by this act are promulgated by the
7 agency that originally collected the information.
8 The presumption here is the agency that had the
9 mandate to get the information in the first place
10 probably best understands what the problems are with
11 the data, what the sensitivities of the respondents
12 may be, what problems they would have if somebody
13 made a mistake and disclosed something that
14 shouldn't be disclosed. So they are the ones
15 responsible. They have the primary responsibility
16 for drafting these regulations.
17 Each agency writes rules for the data
18 that it collects. This was a very important
19 principle for winning over the Treasury Department,
20 who had always looked at this as this is a way for
21 OMB to rewrite all of the Treasury Department's
22 rules, but that's not what we intended ever.
23 A good bit of the text of the bill is
1 then in the series of conforming amendments in
2 Section 9. There are a couple of real short ones
3 for the Department of Commerce covering the Census
4 Bureau and BEA, one very long one for the Department
5 of Energy, beautifully drafted. I was told before I
6 ever got started in this that there were some people
7 in the General Counsel's Office of Department of
8 Energy who were probably the government's best
9 drafts-persons for legislation, and I was certainly
10 impressed with the job that they did. We only made
11 really one change in what they sent us, except for
12 adjusting the indentations and things of that sort,
13 and that was another one of these fire fights that
14 occurred actually after the bill had been sent
15 forward for Alice to sign off on. We made one
16 slight change affecting the Department of Energy
18 We then have some new conforming
19 amendments. These weren't in there a couple of
20 years ago, weren't in the version that we circulated
21 last year, one for the Department of Health and
22 Human Services and a very short one for the
23 Department of Labor.
1 Any of you who have seen earlier
2 versions of this, the conforming amendment for the
3 Department of Labor has gone up and down and up and
4 down several times. We had a three paragraph
5 version that we wrote with some members of Janet
6 Norwood's staff years ago. Shortly before we went
7 out for review on the previous iteration of this,
8 which was about a year and a half ago, Labor said,
9 "We want to change it," and they took the three
10 paragraphs out to about three pages.
11 They came in again shortly before we
12 were going to circulate it this time and they wanted
13 to restore some of the stuff we had taken out last
14 time. This went round and round. Finally we
15 reminded Labor subtly that, in fact, no conforming
16 amendment was needed for the Department of Labor and
17 that if necessary we could go forward without one,
18 and after some discussions of some very subtle
19 changes in the main language of the bill, which
20 solved some problems their solicitor was concerned
21 with, we ended up with an amendment that's one
22 sentence long.
23 The other new feature here, and this was
1 another late riser, was the amendment for the
2 National Science Foundation. A year ago, when we
3 talked to NSF, they really weren't prepared to
4 become part of this policy effort. However, they
5 went back and started thinking about it. We told
6 them, well, maybe the first round of amendments a
7 couple of years from now we can do something.
8 They went through the process. They
9 spent a lot of time with their General Counsel. As
10 a result, when we went out for comment on the bill
11 about a month ago, their comment came back a
12 completely coordinated, conforming amendment to the
13 section of Title 42 that deals with the National
14 Science Foundation, very well written, well put
15 together, completely coordinated with the General
16 Counsel, and a request that we add the Science
17 Resources Division to the list of statistical data
18 centers, and they had done such a beautiful job that
19 we said, "Fine. That looks great to us."
20 There is one other important conforming
21 amendment that's in Section F disclosure penalties
22 on page 14 of the handout. Section 1905 of Title
23 18, this is the Trade Secrets Act. What we have
1 done there, and this looks a little different from -
2 - oh, my goodness, this is not even the right
3 version. Oh, no, that's all right. Okay. I have
4 to be careful with this one because we changed the
5 language on the recommendation to the Department of
6 Justice in the last three days before we went
7 forward. Ah, yes, okay. This is the correct one.
8 Initially what this law said, it imposes
9 penalties on officers or employees of the
10 government, and there's also a special category in
11 there which are agents of the government, which I
12 think it was put in there by a unit of the Justice
13 Department for certain kinds of agents that they
14 have that they wanted added to this.
15 And what we are doing is simply adding
16 this broad agent concept that we've defined for
17 these statistical data centers as another form of
18 agent who can be punished as if they were officers
19 or employees, and then the other thought we had was,
20 well, this is really old law. It says you can only
21 be fined $1,000 for something like this, which isn't
22 very much these days. So we sort of poked around
23 and originally proposed, well, we ought to increase
1 it to 10,000.
2 When Justice came back, they said,
3 "Well, why don't you just make it conform to the
4 current sentencing standard for what this is, which
5 is a Type A misdemeanor?"
6 I said, "Well, okay. What do you want
7 to do?"
8 They said, "Well, you look in this
9 section and look in this other section, and by
10 simply tying the fine to the title in which this is
11 placed, you have the effect of tying it into
12 whatever gets done with the sentencing standard."
13 I said, "Well, what is the sentencing
15 Well, it's $100,000, which probably
16 makes a lot of sense considering what we're actually
17 talking about here, but it kind of made a lot of our
18 agonizing debates over whether it's going to be
19 10,000 or 15,000 look kind of silly in hindsight,
20 and this was, in fact, the issue we came back to EIA
22 I called Larry Klerr after the bill had
23 been sent forward through coordination in OMB. It
1 was halfway to the Director's office, and I got a
2 call from LRD and said, "Hey, there are two of these
3 conforming amendments which still have this $10,000
4 in there. Do you want to keep them or not?"
5 And I said, "Well, I don't know." In
6 the case of the National Science Foundation, they
7 had different reasons for establishing a $10,000
8 fine. In the case of the EIA amendment, they had
9 basically just lifted the language that we were
10 using at the time that they wrote the amendment, and
11 I called Larry and said, "Hey, go talk to your
12 General Counsel and see if you want to take this
13 ride with us up to 100,000 on the fines."
14 And 20 minutes later Larry had it all
15 coordinated and came back and said yes, and we sat
16 there on the phone and I got the LRD guide and wrote
17 in the new language. I then made replacement pages,
18 went up and caught the package over in the
19 Director's suite, and put the new language. It was
20 that kind of week.
21 Finally, there was one other very
22 important contribution that of all people the
23 Department of Defense made. In Section 10, effect
1 on other laws, we have always had a section in here
2 which acknowledged the relationship of this law to
3 the Paper Work Reduction Act, particularly Section
4 3510, which is the basic piece of law in the federal
5 government that controls exchanges of information
6 between agencies. It sets the general rule, and
7 this reinforces the fact that this still applies,
8 though the way this new law is written, it sets some
9 new boundaries. It's drafted so as to work with the
10 language of the PRA.
11 The suggestion that we got from the
12 Department of Defense was that it may still be
13 useful. It's sort of "belt and suspenders" type of
14 stuff, but it may be useful to reiterate the fact
15 that what we have written as policy on
16 confidentiality in Section 6 is intended to
17 constitute a (b)(3) exemption under the Freedom of
18 Information Act.
19 We went around a little bit as to
20 exactly how to word it, but everybody agreed. Even
21 Justice agreed that this was okay to do. What's
22 happened and the reason this is put in there is that
23 there are a number of courts when they look at FOIA
1 cases who believe because the way FOIA is written
2 they do not have to look at the legislative history
3 of an act in making a determination as to whether
4 the (b)(3) exemption applies, which is why you have
5 sometimes very elegant language which clearly states
6 restrictions that would qualify a law for the (b)(3)
7 exemption, and then somewhere down there there's
8 another little piece of verbiage that says, "This is
9 a (b)(3) exemption in case you didn't notice." So
10 that's basically what was added here.
11 The analysis probably does a better job
12 than I've done in the last few minutes of explaining
13 what all this stuff is about. There is a companion
14 piece that has not yet gotten out of the agencies.
15 One of the things you will notice if you are a
16 follower of the various iterations of this bill over
17 the years is there is no amendment to the tax code
18 in the main bill.
19 What happened here -- well, let me back
20 up a little bit. Over the years, the Census Bureau
21 and Internal Revenue Service have taken turns
22 blocking us from getting this thing out of the
23 administration. During the Bush administration, we
1 came up with a version of the bill that essentially
2 took all of the chips away from the Census Bureau.
3 It didn't get out of the administration, but the
4 Census Bureau has been very cooperative in helping
5 us do a reasonable version of this bill ever since.
6 On the recommendation of a lot of
7 people, we were encouraged to make the bill more
8 ambitious than the one in the Bush administration,
9 and as a result, we tried to come up with a new
10 version of the amendment to the tax code in the bill
11 that we circulated a year ago. Treasury still
12 didn't like it, but shortly after that bill was
13 circulated, there was a high level meeting involving
14 the Commissioner of Internal Revenue, Deputy
15 Secretary of Treasury, and Sally Katzen, my boss'
16 boss' boss, where we came to an agreement on some
17 principles for what might constitute an agreement on
18 an amendment to the tax code.
19 What we were offering them was basically
20 some changes in language that reduced at least
21 theoretically the amount of information that would
22 be disclosed for statistical purposes, but disclosed
23 kinds of information that were much more useful than
1 under the current language of the tax code.
2 After seven or eight months of
3 negotiation back and forth, IRS came up with a form
4 of this that we could all live with, and this is now
5 -- this was also circulated the same week as our
6 bill and has, to my understanding, now been
7 completely revised to reflect all comments from all
8 agencies and is back at the Treasury Department
9 awaiting signature by their General Counsel before
10 it goes to Congress, but I can't really show you
11 copies of that until it becomes official.
12 If anybody wants to talk about, you
13 know, what the changes are, I'll be happy to talk to
14 you after, but I do want to mention one more piece
15 of this strategy which started later, but finished
16 earlier than either of the others, and that was our
17 administrative order which really deals with a lot
18 of the same concepts that are in the bill.
19 One of the things that we decided to do
20 when we were having difficulties many years ago with
21 getting agencies to agree on a strategy for changing
22 the law was we decided to look at how much we could
23 do without changing the law, and EIA was, in fact,
1 an important player in determining how far we could
3 There were some events that some of you
4 in this room will recall in which there were some
5 disclosures that were being requested. There was a
6 lot of pushing and shoving going on within the
7 executive branch that was causing us all kinds of
8 problems, and they really put a fine point on the
9 extent to which in some cases Congress and the law
10 is not the problem; it's the agencies and the way
11 they behave to each other that is the problem, and
12 that's something we ought to be able to get a handle
14 As a result of the case that was made
15 all the way up to the White House by EIA, some of
16 the President's key advisors suggested that maybe we
17 should have an executive order that tells all of the
18 executive agencies to behave with respect to this
19 particular problem in the future. We couldn't do
20 anything about that immediately. Well, this was in
21 the Bush administration, and we only had a few
22 months after that before the Bush administration was
23 no more.
1 But the idea was planted, and we started
2 drafting little things and looking at things. What
3 could we do that would work with the strategy that
4 we were pursuing with the bill?
5 On January 29th of this year, we put out
6 for comment an order, not an executive order, but an
7 order providing for the confidentiality of
8 statistical information. Some of you may be aware
9 that last year we circulated a proposed executive
10 order. After a lot of discussion with our General
11 Counsel, we determined that we could actually do
12 more under our existing authority, which goes back
13 to the Budget and Accounting Procedures Act of 1950
14 and has been since buttressed by an executive order,
15 a congressional requirement to delegate the
16 authority granted to the President which is
17 reflected in an amendment to that executive order.
18 What it all boils down to is we have a
19 strong case that orders adopted under the
20 authorities that exist have the force of law. If
21 any of you doubt this, there is one spectacular
22 case, which is Statistical Policy Directive No. 3.
23 Most of the statistical policy directives sound like
1 advice and are taken as advice. This one is not.
2 This is the directive that controls the release of
3 principal economic indicators. It has been treated
4 as law by six Presidents. It is one of these really
5 strange situations where lowly working stiffs like
6 me write something, go out and talk to agencies
7 about it, come up with a policy, and even the
8 President feels bound by it.
9 Even though it is based on authority
10 assigned to the President, because of the way it's
11 operated over the years and also because of the good
12 sense of a lot of presidential advisors who
13 understood why things need to be done this way, we
14 have a perfect record of compliance all the way up
15 to the President, and as our General Counsel pointed
16 out, you don't get that with a lot of laws.
17 So the point he made, one of the other
18 points that was made is that executive orders
19 frequently change with administrations, which is not
20 the sort of stability that we were seeking. We
21 really want, after all these years of living with
22 informal versions of this policy, we really want to
23 make something permanent happen, and that's what the
1 order is all about.
2 Tactically also, the order gave us an
3 opportunity to put some of these principles out on
4 display for the public, out where Congress and
5 congressional staff would see them, would have a
6 chance to think about them, to talk about them, to
7 debate them before we go to Congress with a very
8 similar strategy that requires them to take action
9 and to pass legislation, and this has generally been
10 very effective.
11 A number of statistical agency heads and
12 people who have a great deal of concern with the way
13 the government does its statistical work testified
14 as a hearing in March, along with Sally Katzen.
15 Cathy Waldman was there, but she was basically a
16 resource for Sally, and that was one of the things
17 that clearly had made an impression on Congress.
18 They had looked at what we were trying
19 to do in the confidentiality order. There was broad
20 agreement with the principles that we were trying to
21 pursue there.
22 Some important differences between what
23 we do in the bill and what we do in the order. I
1 mentioned before that the bill is a (b)(3) statute.
2 the reason that the order has a lot more verbiage
3 in it about how you get things done and a lot of
4 required reviews on the part of the agency as to
5 what statutes require you to do and what policy
6 options are open to you is because the order does
7 not assume that you have a (b)(3) FOIA exemption.
8 What it is doing is, among other things, making a
9 significant change in the way FOIA policy operates
10 in the case of some designated agencies.
11 If any of you are followers of the way
12 Freedom of Information Act law has developed over
13 decades now, one of the principles had been that
14 determinations as to what may be disclosed are
15 almost always deferred until a request for
16 disclosure is made. If you think about it for a
17 minute, you can see this really makes a hash of a
18 confidentiality pledge. How can you go out and tell
19 a respondent that this is not going to be disclosed
20 because it is very important. We know it's
21 important to you and we want you to feel confident
22 the government is not going to splash this
23 information all over the place. It's going to treat
1 you with respect and respect both your
2 confidentiality privilege and your privacy rights,
3 and yet at the same time, try to observe a policy
4 that says we're not going to decide until somebody
5 asks us for it.
6 So part of what we have crafted in the
7 order is a required change on the part of some
8 designated agencies in the way they deal with FOIA
9 policy. We require these agencies to make these
10 determinations in advance, which flies in the face
11 of most FOIA regulations, flies in the face of a
12 couple of executive orders if you get right down to
13 it, because FOIA regulations and executive orders
14 have all predominantly dealt with the case of
15 administrative information. Neither the Privacy Act
16 nor FOIA have ever really dealt adequately with the
17 problems of statistical agencies and particularly
18 the concept of statistical confidentiality, and
19 we're trying to carve out some area where we can
20 make sense of this and make it work.
21 So this went out in January. The
22 comment period ended the end of March. I have some
23 extra copies here. If any of you want to see it and
1 you don't get a copy, I can give you the Federal
2 Register citation.
3 The comment period is over, but since
4 I've been busy writing the last minute changes to
5 the bill, we're going to be working on the comments
6 for a while yet, and if any of you get some
7 inspirations, my E-mail address is in the notice.
8 It's probably easier if I tell you what it is
9 because the particular font I use here, the lower
10 case Ls and the ones are identical. The E-mail
11 address if you have any comments on any of this, and
12 I'd be happy to see them, is Coffeyjay, C-o-f-f-e-y,
13 underscored Jay, at sign, A1.dop.gov.
14 As I say, the formal comment process is
15 over. We are still, however, talking with people
16 who submitted formal comments, and if any of you
17 have real inspirations, I'd be happy to look at them
18 because there are many, many issues that I'm going
19 to be writing up papers on that came in in the
20 course of the comment period, and any insight you
21 can provide me as a personal favor I would greatly
22 appreciate. We can't treat them as formal comments,
23 but I'd be glad to have them.
1 That's about it. How well did I do?
2 Probably behind time. How much time do you have
3 left for questions?
4 CHAIRMAN MOUNT: Does anybody want to?
5 Yes, Cal.
6 MR. KENT: Is the idea that the order
7 and the law go together or is the order seen as a
8 separate act that will not require the law to be
10 MR. COFFEY: Both. We designed the
11 order so that we could get substantial benefit from
12 it whether or not Congress passes the bill.
13 MR. KENT: Whether or not the law
14 passes. Okay.
15 MR. COFFEY: We also designed it to be
16 consistent and to work effectively with the bill if
17 Congress chooses to pass it in the way it's been
18 proposed. Some of the things you see in the bill
19 you will eventually see in the guidance for the
20 order. I don't have as big a writing job as it may
21 appear because I'm just going to lift whole chunks
22 of definitions and other things from one straight
23 across to the other.
1 If you do get a copy of the order, by
2 the way, there's one glitch. There's one thing we
3 missed in all of our interminable debates with our
4 General Counsel. Somewhere in here there's a phrase
5 that says about agents something about "supervision
6 and control." It should have been "supervision or
7 control," which makes a big difference.
8 MR. KENT: And can I just ask one more
9 question as a follow-up? Why is this not applied to
10 all of EIA?
11 MR. COFFEY: Good question. It was a
12 tactical decision made some time ago. The first
13 question was could we apply it to any part of EIA.
14 The choices that we made in this bill
15 were often influenced by what was perceived to be on
16 the edge of controversy or over the edge. In the
17 case of EIA, as you well know, changing the
18 disclosure rules that apply generally to EIA has
19 been a meat grinder since the 1970s.
20 It appeared to us, at least, that even
21 among the people who were dead opposed to changing
22 the way information about energy producers is
23 handled, it's to get the big boys, you know, the
1 anti-oil company mentality, which everybody is aware
2 of, but that has clearly -- in some of the actions
3 of Congress that has not applied to small fry, to
4 households, to small businesses when you're talking
5 about retail gasoline franchises. Congress has from
6 time to time written in special protections for the
7 interest of these not so big boys.
8 We basically seized on that and tried to
9 look at the largest chunk we could define that would
10 avoid that other controversy which had been so
12 The other issue was, and what we have
13 looked for in the designation of each of these
14 agencies, we are looking for units who, in fact, can
15 do something useful with data sharing, and the
16 Energy End Use Consumption division clearly in their
17 work with several other agencies has laid the ground
18 work for doing some very effective data sharing.
19 They've had to be very imaginative in many cases in
20 how they did things, but clearly there was a very
21 real potential there which we could easily explain
22 to Congress if anybody asked us, and that was the
23 other practical consideration.
1 MR. KENT: Right, but this particular
2 bill then would not deal with the issue that we had
3 in the famous White House meeting?
4 MR. COFFEY: No, there are other ways to
5 deal with that, particularly once we get everyone's
6 attention with this order. The way disputes over
7 data sharing are supposed to be managed in the
8 government, most of the time agencies talk to each
9 other and decide either you're going to give
10 somebody some data or you're not, and agencies
11 actually have quite a lot of latitude in how they
12 make these choices.
13 The law that applies in these cases is
14 Section 3510 of the Paper Work Reduction Act, which
15 authorizes agencies to make disclosures that are not
16 inconsistent with existing law. The way that
17 language is written, each of the people in this
18 discussion makes a determination as to what is
19 inconsistent with law. If they can't agree, there's
20 no agreement.
21 Congress then provided for another step
22 to resolve such differences, which is that the
23 Director of OMB may order exchanges of information
1 if he or she determines that it is no inconsistent
2 with the law. Once again, this is an independent
3 determination. This is the way we've interpreted it
4 anyway, along with our General Counsel. If OMB can
5 sit down, they don't have to listen to what anybody
6 else has said in their regulations or what their
7 policies are. They look at the law and make a
8 determination as to whether or not this is
10 Now, one of the dilemmas we've had in
11 this is that without either the confidentiality
12 order or the confidentiality law, there's no
13 distinction in many instances between statistical
14 uses and non-statistical uses. So if you read 3510,
15 it seems to say, well, OMB should be deciding that
16 statistical agencies should be central data
17 collection agencies for the Federal Trade Commission
18 or, you know, anybody else who has an interest in
19 similar information.
20 One of the things we do in both the
21 order and the statute that we're proposing is we
22 finally build into law the concept of functional
23 separation that's been pushed for 25 years now so
1 that when OMB looks at these determinations under
2 Section 3510, we can legally and rationally
3 determine that we may have a central collection
4 agency for statistical uses and another central
5 collection agency for administrative uses of very
6 similar data, and it would not be inconsistent
7 because here we would have provided for the basis
8 for making such a distinction, the functional
9 separation policy.
10 MR. KENT: So the order would take care
12 MR. COFFEY: You can do it under the
13 order. You can do it even more explicitly under the
14 bill for any agencies covered in the bill.
15 MR. SKARPNESS: I've got a sort
16 question, a point of information. So you're saying
17 there's even parts of EIA that aren't going to be
18 covered by this?
19 MR. COFFEY: Right now all of EIA has
20 very little legal back-up for its confidentiality
21 policies. Yeah, I tried very hard with some very
22 ingenious regulations some --
23 MR. SKARPNESS: So across the street
1 here is DOT.
2 MR. COFFEY: Okay.
3 MR. SKARPNESS: You know, were they part
4 of or did they decline getting involved?
5 MR. COFFEY: DOT also got into the game
6 late, in part because DOT has doubled in size and
7 substantially changed its functions in the last six
8 or eight months.
9 MR. SKARPNESS: Oh, yeah.
10 MR. COFFEY: We have been in constant
11 communication with the people in the BTS, Bureau of
12 Transportation Statistics. If they had been farther
13 along --
14 MR. SKARPNESS: They might have.
15 MR. COFFEY: -- this would have been an
16 opportunity to do this. At best now we hope things
17 will settle out, and maybe the first time we get a
18 chance to go back and amend this and add another
19 statistical data center and show Congress how
20 wonderful this whole strategy is working, BTS is
21 high on the list of the next set of agencies to be
23 MR. SKARPNESS: One last thing.
1 MR. COFFEY: Yes.
2 MR. SKARPNESS: There's been talk about
3 a statistical clearinghouse concept, you know, sort
4 of a central location where, you know, this data is
5 cleared. Does that fit into this in any way? I
6 know that each agency --
7 MR. COFFEY: Well, are you talking about
8 a consolidation of agencies?
9 MR. SKARPNESS: Well, no, of just data
10 more or less.
11 MR. COFFEY: I'm not familiar with that
13 MR. SKARPNESS: Okay.
14 MR. COFFEY: I am familiar with a bill
15 that's been discussed on the Hill.
16 MR. HAKES: You're talking about a
17 central place where you can go for data?
18 MR. SKARPNESS: Yeah.
19 MR. COFFEY: Oh, oh, that's a different
21 MR. HAKES: We can create that create
22 that electronically.
23 MR. SKARPNESS: Right.
1 CHAIRMAN MOUNT: A virtual one.
2 MR. SKARPNESS: Yes, where it fits
4 MR. COFFEY: That's happening.
5 MR. SKARPNESS: Okay.
6 MR. COFFEY: Yeah, that's happening
7 right now.
8 MR. SKARPNESS: I mean is this
9 facilitated or you know?
10 MR. COFFEY: It should have happened
11 already, but you know, there's been a lot of back-
12 and-forth in the White House about when they're
13 going to roll it out and who's going to be there.
14 MR. HAKES: It's kind of encouraging
15 because it's going to be a fairly prominent part of
16 the White House Home Page. So a person coming into
17 the federal system will see the statistical
18 resources early in the process without getting much
19 into the system, which should increase the
20 visibility of federal statistics.
21 MR. COFFEY: Right. Our staff and the
22 statistical agencies have all been working on this
23 one for a while. As I say, we sort of expected that
1 it was going to happen before now, but for a number
2 of reasons it hasn't quite been scheduled yet, but
3 it's about ready. It's going to happen soon.
4 Any other questions?
5 MS. COX: I as invisible.
7 MS. COX: I think this might come -- as
8 I understand it, what you've handed out here only
9 affects the named agencies.
10 MR. COFFEY: The act itself names
11 statistical data centers, yes.
12 MS. COX: Right. Now, for the order, is
13 that -- I haven't seen the order. Can that impact
14 on -- yes, I would like a copy of it.
15 MR. KENT: Have you got another copy?
16 I'd like one.
17 MR. COFFEY: I have a few more. If I
18 don't have enough I can also give you the --
19 MS. COX: Such as the Bureau of
20 Transportation Statistics or maybe another group
21 that's concerned about the confidentiality for their
22 survey participants.
23 MR. COFFEY: Right.
1 MS. COX: I did work on a survey once
2 where survey data that had been -- they had said
3 they were going to preserve their confidentiality,
4 but they had no legal right to do so, and it was
5 subpoenaed under -- I can't remember. I think this
6 was under EPA.
7 MR. COFFEY: Which agency?
8 MS. COX: They had a subpoena and had to
9 release data.
10 MR. COFFEY: This was EPA?
11 MS. COX: I think it was EPA.
12 MR. COFFEY: Yeah, I'm familiar with
13 that situation.
14 MS. COX: And it was a very sad
16 MR. COFFEY: Yeah.
17 MS. COX: But they had no way of
18 preventing it.
19 MR. COFFEY: We went through this in
20 great detail with both EPA and the Justice
21 Department when there were proposals to create a
22 Bureau of Environmental Statistics. One of the
23 issues, and it is also an issue at BTS, is whether
1 this agency has a mandate to collect data for
2 statistical purposes, and we asked both EPA and the
3 Justice Department to consider whether this is
4 something that should be added because all of EPA's
5 current data collection authorities all derive from
6 regulatory statutes.
7 The issue we put to EPA and Justice was:
8 does it make sense? And we cited some very
9 important examples where, in fact, there were
10 pitched battles within EPA because they had to, in
11 fact, guarantee very strict confidentiality and
12 guarantee that certain things were going to be used
13 only for statistical purposes in order to get some
14 things done.
15 There were a couple of surveys where
16 clearly they would not have happened, would not have
17 had good results unless they had done that. Every
18 one of them was a bloody battle with the General
19 Counsel at EPA because the general principle that
20 they wanted to observe was anything we get is
21 available for enforcement purposes, and the point we
22 made was that if you say that, you're not going to
23 get anything from these kinds of surveys.
1 So we won the little battles. What
2 happened ultimately though when we talked to Justice
3 about this was they raised a number of important
4 questions. Since Justice has a fair chunk of their
5 resources invested in prosecuting environmental
6 violations, they've got a division over there that
7 spends all of their time doing this.
8 The question that arises in that kind of
9 an environment is: how can we assure that
10 information that needs to be used to prosecute an
11 offense is not tainted by the terms under which it
12 was collected? And it's purely a pragmatic kind of
13 problem. It's can someone who doesn't like what
14 we're doing come in and say, "Well, you didn't
15 collect this correctly because I thought it should
16 have been used for only statistical purposes," and
17 even though the agency may have said otherwise, it's
18 a real hassle. It is something that attorneys have
19 nightmares over, and basically we just didn't want
20 to fight with Justice any further on that,
21 particularly since the bill wasn't going anywhere.
22 Transportation has some of the same
23 problems. They've inherited two regulatory
1 mandates, along with staff. If you look at what
2 they're doing, a lot of the reason for the
3 regulatory collection has disappeared. They haven't
4 done nearly the kind of job that EIA did in the
5 early '80s to sort out what we need for statistical
6 purposes from the legacy of regulatory data
7 collections. There's still a lot of people in
8 transportation who still think the same way about
9 what their information requirements are.
10 They don't quite know how they're going
11 to handle that. Clearly they've got some staff work
12 to do. There is some additional refinement they're
13 going to have to do to their organization. Right
14 now it's an agency that's in flux, and the one thing
15 we cannot do is go to Congress with promises. We're
16 going to raise enough controversies as it is. We've
17 got to go to Congress with agencies who are entirely
18 credible as statistical agencies.
19 CHAIRMAN MOUNT: When is the order going
20 to be sent out?
21 MR. COFFEY: The order? When I have
22 time. I'm basically the resource for both of these.
23 We have two other permanent staff and a chief
1 statistician in our shop, but this is my baby, both
2 of these.
3 CHAIRMAN MOUNT: Like a month are we
4 talking about?
5 MR. COFFEY: A month, possibly longer.
6 I have promised the agency heads some issue papers
7 on some of the extremely interesting comments that
8 we got. We don't anticipate a lot of changes in the
9 order. You can see that one of the reasons is we
10 have to have certain relationships between the order
11 and the statute in order to make this whole strategy
12 work, and those things we're going to be very
13 reluctant to change unless there's somebody on the
14 Hill who wants to make a similar change.
15 What we anticipate is going to happen on
16 the order, we have gotten comments now. We're
17 probably going to go back to the statistical
18 agencies with some issue papers on a number of the
19 comments, a number of the issues raised, hash these
20 out further, and then basically make whatever
21 hopefully small changes will need to be made to the
22 order itself, and then go to the next important
23 step, which is going to be to write the guidance
1 that implements the order.
2 The order is written in fairly general
3 terms. There are a lot of things in here that we
4 flesh out in the next stage of things, which is how
5 we're going to do this. This says what we're going
6 to do. The next step is how we're going to do it.
7 MS. COX: But who's affected? Would EPA
8 come under this order?
9 MR. COFFEY: They could within 60 days,
10 any time we can to do it. Once this order is in
11 place, we can amend that list at the end with a 60-
12 day public comment period. We've got to have reason
13 to do it. We've got to have whatever.
14 What was really needed in this case and
15 what we asked the agencies to do is look at several
16 things. Do you have authority to collect
17 information for statistical purposes? No point in
18 being on the list if you don't.
19 Second, do you have the wherewithal to
20 provide security for this information, to make sure
21 it doesn't get used for other purposes unless
22 there's some statutory reason for doing that? And
23 in that case, you've got to go through some other
1 steps to make sure that the public knows exactly
2 what you're doing.
3 And all of these agencies, we believe,
4 passed these tests. There's still a bit of question
5 about -- we don't have a final opinion from Chief
6 Counsel at IRS over some questions on the authority
7 of SOI Division, but we're pretty sure that BTS will
8 pass muster. All these other agencies are going to
9 pass muster.
10 Under the order it is very easy for us
11 to extend the scope of this, the effect of this
12 order to other agencies by simply adding to the
13 list. If you look at the general terms of the
14 order, it doesn't even say it has to be just a
15 particular set of agencies. We've grafted that on
16 to say, well, the following agencies are the ones we
17 intend for this to apply to at this point in time.
18 CHAIRMAN MOUNT: Are you going to be
19 here afterwards?
20 MR. COFFEY: Unfortunately I've got a
21 command performance back at the office in about 40
22 minutes. I wish I could stay around longer. I am
23 likely to be back. Because I've got to go this
1 afternoon, I'm likely to get back tomorrow morning.
2 It was a choice of one or the other. So if you
3 have some more questions you want to think about
4 tonight and talk tomorrow, maybe at the break we can
5 discuss any other things you're interested in.
6 CHAIRMAN MOUNT: Thank you very much,
7 Jerry. That was a very important topic, and I
8 remember it was one of the first topics that was
9 discussed when I joined the Committee a number of
10 years ago.
11 So we should move on now to the last
12 subject for the afternoon -- I can't even say it --
13 pollution control experiments. I've obviously been
14 here too long this afternoon. Inder Kundra from the
15 Office of Statistical Standards.
16 Sorry for squeezing your time.
17 MR. KUNDRA: I'm Inder Kundra. It's a
18 pleasure to talk to you.
19 The purpose of my talk is to update the
20 Committee concerning the experimental economic
21 project undertaken by EIA to analyze the effects of
22 environmental regulations of energy industries.
23 In 1993 we informed the Committee that
1 we were in the process of replicating the results
2 using a software built by Arizona and using a
3 uniform set of parameters. That's what I'm going to
4 report here.
5 The Clear Air Act of 1990 instituted
6 variable emission permits for sulfur dioxide. Each
7 permit with the industry was given to emit one ton
8 of sulfur dioxide. These permits were the industry
9 could purchase or sell or even keep these permits
10 for future use. That's called banking.
11 And by 1993, a central market was
12 established for trading these permits with the
13 provision that to auction about 2.8 percent of the
14 annual allocation by revenue to discrimination
15 auction (phonetic), this kind of auction does not
16 generate any money for the central authority.
17 The Energy Information Administration
18 contracted the University of Colorado and Arizona
19 for two experimental institutions that mimics the
20 salient features of the sulfur dioxide market to
21 develop and document transportable software that
22 implements the experimental institution and propose
23 to replicate by pots and pots (phonetic), propose to
1 replicate high pluses (phonetic) of possible future
2 testing with additional experimental and field data.
3 The results were presented to the
4 Committee in 1990 and 1991. In '91, following the
5 recommendation of the Committee that EIA should pay
6 attention to the statistical details, we analyzed
7 the Colorado experiment and presented to the
8 Committee with a finding that there was no crossover
9 between the replications, and the Colorado
10 experiment was not subject to any experimental bias.
11 At that time the Committee stressed the
12 need for replicating these experiments.
13 Subsequently we funded the Universities of Southern
14 California and Mississippi to replicate these
15 experiments by using Arizona software and design and
16 by using a statistical experimental design Arizona
18 To test the applicability we adopted an
19 experiment, a two-by-two pictorial design with two
20 supplix (phonetic) within each cell. This design
21 consisted of two forms, high and low, and two
22 technologies, old and new. The high forms, the
23 forms I have indicated the amount of the BTU
1 produced and the technologies, I have indicated the
2 amount of sulfur dioxide emission per BTU, and with
3 the condition that each from the high forms we
4 allocated about eight -- there were two figures. I
5 mean six -- 12 figures for the experiment, one
6 through six and seven to 12. One to six would be
7 allocated eight permits, and we have seen that eight
8 permits for the high forms and four from seven to
9 12, and for the low firms, in those two -- as is
10 obviously from here, that we did deliberately
11 introduce variations taking the firms and the
13 And to start the experiment, we started
14 in the beginning with about $20 for each student,
15 and those students then made their profits by
16 selling the permits. They were the subjects within
17 the firms, made profits either by redeeming the
18 permits at the pre-assigned redemption values or by
19 selling the permits to other firms in each of the
20 transaction periods. The -- either in a revenue
21 neutral, sealed, bid discriminatable auction or in a
22 double auction. That is a centralized exchange
23 where a public bids us and gets prices.
1 The experiment was done, repeated about
2 eight times in each of the universities
3 independently, and in each of the replications, we
4 had different students.
5 The total profits which was made by
6 these students in each of the experiments at the end
7 of the 12 periods were used to determine if the
8 observed differences in estimated profits in
9 universities and application to other subjects,
10 simply things were obtained for determining as the
11 basic parameters to use for making some sort of
12 distinction whether the relation was -- whether we
13 would replicate experiments or not.
14 Next slide.
15 As is obvious from here, at 18
16 universities we could see that there was no
17 difference. Most independent poll, even -- within
18 applications, within universities of different
19 technologies, of different subjects within the given
21 What we concluded from here that we were
22 able to replicate those experiments without any
23 problem, without any bias in here left.
1 I just want to add another slide where I
2 have to thank Dr. David Bellhause from Canada. He
3 suggested an experiment like this, that we should
4 have done the analysis with different firms if we
5 see an obvious infraction between the firms and
6 these, but my main problem was that we were taking -
7 - test the only applicability, and the firms were
8 taken within the universities and all of those
9 things. We did it.
10 It is not something you say whether you
11 do this or that way. Still we can come to the
12 conclusions that we can replicate those experiments
13 and maximum variations which was formed was between
14 the forms, between the technologies within the
16 And I am very thankful to him because I
17 did talk to him once or twice on the phone.
18 That's all I can say about this.
19 CHAIRMAN MOUNT: Thank you.
20 The discussant is Greta Ljung.
21 MS. LJUNG: So I'll be just making a
22 couple of comments on the experiment, and then I'll
23 discuss the analysis, and the last transparency that
1 David had provided covers some of the topics that I
2 was going to mention with regard to the analysis.
3 And the first transparency is just a
4 summary of the experiment. We have a two-factor
5 experiment. The factors are per type and technology
6 type. We have a two-level design. Each factor has
7 two levels. We are running the experiments in two
8 locations, and the experiment is replicated eight
9 times in each location.
10 And the output variable we're interested
11 in is the total profits realized by each
13 The objective of the experiment as I
14 read it is to study the effects of the two factors
15 and also to test the hypothesis that the results can
16 be replicated, that the results are sort of
17 insensitive to the location and also to the subject
18 who was used in the experiment.
19 Now, the two-level design that's used
20 here is, of course, very economical. It includes
21 the fewest factors possible, possible to study the
22 factor effects. The disadvantage is that we can
23 only study linear effects, while if there be
1 nonlinear relationships between our output variable
2 and the factors, we would not get information on
3 those nonlinearities.
4 And in the present setting, a potential
5 concern is that we have only four combinations of
6 the two factors. We have very small experiments.
7 Within each cell there are only two subjects. So we
8 have a total number of participants in each trial
9 here is only eight, and so a question arises: are
10 the results that we get with only eight subjects
11 really representative of the market where there
12 would be many more players?
13 Now, as far as the replications, we have
14 a fairly large number of replications. Eight is a
15 fairly good number. Different students are used in
16 different replications. I think slightly better
17 design here might have been to use -- in each
18 location only use like four sets of students and
19 replicate the trials for those. That would have
20 given us a slightly better estimate of experimental
21 error perhaps and might have given us slightly
22 different information.
23 Now, the tests that are performed show
1 no differences between replications, and that to me
2 says that the experiments if tightly controlled, the
3 same instructions given, somehow those results are
4 not all that exciting. I mean it would have seemed
5 it would have been a little more interesting to try
6 to introduce a little more noise into the system
7 here and to see how sensitive or if the results are
8 sensitive to changes like changes in parameters,
9 maybe use different types of -- different
10 instructions or different training or different sort
11 of subjects, not just these who were all business
12 majors, which are fairly uniform, just introducing
13 some changes instead of just replicating everything
14 under constant conditions could have given us a
15 little more information.
16 Now, when we talk about replicability,
17 the tests that are done in the paper basically focus
18 on the average if you compared the two locations.
19 The tests that are done focus on the average
20 shortage realized at each location. So there's just
21 a comparison of mean values. One might be
22 interested also in comparing the factor effects.
23 Are the factor effects the same when we replicate
1 the experiment or did we get different results? So
2 we might want to compare just the -- we can do the
3 analysis or we can think in terms of a mean. The
4 observed -- for each location is the mean plus the
5 firm effect, plus the technology effect, and then we
6 had possible interaction between the two factors,
7 the replication effect, and then error.
8 I've calculated those parameters
9 separately for the two locations and compare not
10 just the overall means for the two locations as done
11 in the paper here, but we would also be interested
12 in are the firm effects the same in the two
13 locations and are the technology effects the same,
14 and we are also comparing two-factor interactions
15 for the two locations.
16 Now, if we compare the firm effects for
17 the two locations, what we are really looking at is
18 the university by firm interaction. So it becomes
19 an interaction effect with the university. And the
20 same, we have an interaction effect with technology
21 if we compare the 138, actually half of the
22 difference between 138 and 151. Half of that
23 difference is the university and technology
1 interaction. Half of the difference of the two
2 factor interaction becomes the three factor
3 interaction, the interaction between the two factors
4 and the university, and I think all of those
5 quantities are of interest.
6 Now, if we wanted to write out a model
7 that includes location, we would then -- that's
8 written down on the bottom part there. We have an
9 overall mean.
10 Three main effects, if we treat the
11 university effect as a fixed effect, we have three
12 main effects. We have now three interaction
13 effects, three two-factor interaction effects and
14 one and then the three-factor interaction effect,
15 replication effect, and the random error.
16 And the replication effect here, we are
17 not interested in making inferences about those
18 specific student groups that we used, but rather we
19 would treat those as a random sample from a much
20 larger population of students. So the interaction
21 effect should be treated as a random effect rather
22 than a fixed effect.
23 Now, the analysis that we do would then
1 include sums of squares corresponding to those
2 terms, and that was done at the last transparency or
3 table that David had provided. David also included
4 interactions between the factors and replicates.
5 That was the last one that we showed.
6 This is then the table that was included
7 in the paper, and when I first looked at that, a
8 sort of striking feature, I thought, was the 16
9 degrees of freedom for firms and 16 degrees of
10 freedom for technology, and this is a little
11 strange, given that we only had two levels of each
12 factors. There should really be one degree of
13 freedom instead of 16, and the reason we have 16
14 here is that the analysis assumed a nested
15 structure. So we assume that the replicates are
16 nested within universities, and then the treatments
17 are nested within replications.
18 But that is sort of not quite the
19 correct assumption because if we assume firms nested
20 really in replications, we would actually have 16
21 different firms. That's what the nesting would
22 imply, and the same for technologies, 16 different
23 technologies. Each replicate would basically use a
1 different level of the factor.
2 I would note that the levels stay
3 constant across the experiment. There is no
4 nesting. Those factors are crossed. So that is
5 something that needs to be changed here.
6 And then I think we have degree of
7 subjects within cells, and we have an F ratio. F
8 ratio for that particular component, and then we
9 have an error term. Now, the error term in this
10 particular case, since we are already subtracting
11 out the cell variability, typically the cell
12 variability would provide the error, and sum of
13 squares -- the error, sum of squares in this
14 particular table probably comes from the
15 interaction, from omitted interactions and the
16 interactions probably in this case are the
17 interactions between firms and technology. That
18 would give us 16 interactions, 16 degrees of freedom
20 Now, for testing main effects and
21 interactions, my preference would have -- no, no,
22 for the analysis down here -- if you are interested
23 in comparisons between firms and technology, my
1 preference would have been to combine and not break
2 out within cell variability from the error, combine
3 those two into a single error estimate.
4 Now, David's analysis looked quite good.
5 So basically what we need to do is include main
6 effects and interactions, and that was all done. So
7 I think his suggestions there were very, very
8 useful, and that's probably the right way to go
9 about the analysis.
10 Because in the analysis that's done on
11 this particular page, the F ratios all have the same
12 error mean squared, but because the very experiment
13 was done, it's not quite clear that we'll have the
14 same error mean squared for all F statistics, and
15 that's something that one should pay attention to,
16 as well, in the analysis.
17 But it's true of course that although
18 it's said the results don't really change, but we
19 still like to have, even if the results are not
20 affected, we still like to have sort of a sound
22 CHAIRMAN MOUNT: Thank you.
23 From the Committee? Bradley.
1 MR. SKARPNESS: I think this brings up a
2 good point in that when you're doing these kinds of
3 analyses that you include a model, and then things
4 are clarified a little more, exactly what's going on
5 and how you're going to do the analysis, and that
6 would help here.
7 And the other thing is that I was also
8 struck a little bit by the way he broke up the sums
9 of squares here, and in David's approach, I don't
10 think he has a nested effect in there, what he
11 showed. Okay. You do, but this is still not sort
12 of the standard way that we sort of -- you know,
13 from the model you would break up your different
14 components of variability here.
15 And then the other thing I was
16 wondering, you know, you do have this fixed
17 university effect. You know, it could be a random
18 effect, too. You know, you're really not
19 generalizing this. I know it's not, but --
20 MR. KUNDRA: It's not.
21 MR. SKARPNESS: It would be nice to sort
22 of say, well, we did this across universities or
23 something, but that's just an aside. It is a fixed
2 CHAIRMAN MOUNT: Richard.
3 MR. LOCKHART: Well, I'd like to put in
4 a word for multivariate analysis in the form at
5 least of asking a question to reveal that which I
6 don't understand. The eight students are in an
7 experiment together. When you take instead of eight
8 students two in a high tech., in a large, old
9 technology firm, and two in a large new technology
10 firm, they're all in one. They're run together.
11 MR. KUNDRA: Right.
12 MR. LOCKHART: And they buy permits
13 effectively from each other.
14 MR. KUNDRA: That's true.
15 MR. LOCKHART: So that there are the
16 eight responses that are obtained in a single
17 replicate of the experiment. It would seem to me to
18 be at least potentially correlated.
19 I mean the analysis of variance that's
20 been discussed by David and Greta is one of the
21 univariate ways of -- there are univariate versions
22 of MANOVA and then there are multivariate versions
23 of MANOVA, and the data don't actually show much
1 sign of correlation --
2 MR. KUNDRA: No, they don't.
3 MR. LOCKHART: -- between the students
4 within a cell, if you know what I mean.
5 MR. KUNDRA: That was the reason that I
6 did it that way, because we wanted to test whether
7 the students in the cell or not, as well as reasons
8 because we had -- but this was the reason. That's
9 one of the reasons we wanted to test whether there's
10 a change between students or not, and that was the
11 first problem that was raised by Dr. Bishop, to see
12 whether we can test whether there's publishing
13 (phonetic) for students or not. That's the way it
14 was designed to test the students.
15 I understand what she's saying, that we
16 should have not nested, but the question here was
17 that nested or not nested, the main things was we
18 wanted to test where the replications are. We can
19 repeat the experiments and all that. We have them
20 or not.
21 And then one of the other factors is
22 also the error -- to test the question that she
23 raised, whether there is interaction with the
1 squares or not. That's why we did that way, but I
2 mean, they would have suggested the same thing for
3 both of these interactions between the firms and --
4 I mean the firms as the -- and the interaction
5 between the firm and technology and all those
7 I had a discussion with him. First he
8 was taking the university, and he was not even
9 taking the replication test, not nested. Then I
10 discussed with him, and he changed it. So it was a
11 good discussion with him.
12 But I do realize what you are saying,
13 that we should have not used nested because nested -
14 - the purpose was to see whether we can really get
15 what we're getting.
16 CHAIRMAN MOUNT: I do have one question.
17 Bill Schulze at COLNOW (phonetic) has recently set
18 up an experimental lab and has just introduced the
19 Arizona software. So I'm sort of interested if you
20 have evidence about the form of auction in terms of
21 which auctions work best in terms of coming up with
22 efficient prices in this kind of situation.
23 PARTICIPANT: Double.
1 CHAIRMAN MOUNT: Double election.
2 MR. KUNDRA: We did that.
3 CHAIRMAN MOUNT: I mean it's pretty
4 clear, right? Yeah.
5 So are there any more comments?
7 MR. WATKINS: Just so I understand
8 Greta's comments, on the nesting if you had eight
9 different technologies within the nest -- what was
10 the other factor you mentioned that also should be
11 broken out?
12 MS. LJUNG: You have firms and
14 MR. WATKINS: Yeah, firms and
15 technology. So that then if you did that, then your
16 interaction effects would proliferate, if you do all
17 of those combinations, right?
18 MS. LJUNG: Well --
19 MS. BISHOP: Basically we wanted to know
20 whether the change of university affected any of the
21 conclusions that you would have drawn had you only
22 done it in one university, and that was the basic
23 purpose of doing it.
1 CHAIRMAN MOUNT: So have you got a
2 comment, Samprit?
3 MR. CHATTERJEE: Yeah, I think the same
4 kind of thing. Since other factors were balanced
5 with regard to the universities, that's the
6 breakdown. So it is a perfectly balanced layout.
7 So the university is one replication and --
8 MS. BISHOP: Except that they're
9 different students.
10 MR. CHATTERJEE: Yes.
11 CHAIRMAN MOUNT: Any comments from the
13 MR. KUNDRA: I just want to stress the
14 point again we were not suggesting -- that was not
15 our purpose because we had deliberately introduced
16 the variations into those firms and technologies.
17 That's how we started.
18 CHAIRMAN MOUNT: So before we formally
19 close, I just want to ask the Committee: are you
20 going to be here for lunch tomorrow? We do have
21 some Committee business, and who's not going to be
22 here for lunch?
23 MR. RELLES: I have to leave right after
1 I give my discussion.
2 MS. COX: I'll be here, but we can't
3 talk on and on.
4 CHAIRMAN MOUNT: I thought that it would
5 be nice if we could break now. Right? That's what
6 I was assuming. Maybe we could meet -- could we
7 meet at 8:30 tomorrow morning for breakfast and you
8 try and be there a little bit earlier so we're ready
9 to talk at 8:30 so we can get Dan's input?
10 MS. COX: I have to leave at 2:30.
11 CHAIRMAN MOUNT: Yeah, so let's do that
12 then. The Committee try to get down about 8:15 so
13 you have time to eat, and we'll just talk for a
14 while at 8:30 before the meeting at nine, and I
15 think it's next door, I assume, in the Lewis Room.
16 And to remind you that we're going to go
17 to 701 Pennsylvania Avenue for dinner. We can't get
18 in at Le Rivage, and for people who want to walk,
19 we'll meet in the lobby about five past six. Five
20 past six in the lobby. Who's going to walk? You're
21 going to walk?
22 Who's walking? Greta, are you going to
23 walk? I just want to know how many people to look
1 for. Right, you're walking.
2 MS. BISHOP: Where to?
3 CHAIRMAN MOUNT: To 701 Pennsylvania
5 MS. BISHOP: Well, I have a car in this
6 building. How far is it?
7 MR. KENT: It's right across the Mall.
8 CHAIRMAN MOUNT: I assume it's right
9 across the Mall.
10 MR. KENT: It's right across the Mall.
11 MS. BISHOP: Well, I think I'll drive
12 over if anybody wants a ride.
13 CHAIRMAN MOUNT: Would the offer of a
14 ride make you change your mind?
15 Five past six. That will give us time
16 to make it.
17 (Whereupon, at 4:51 p.m., the Committee
18 meeting was adjourned.)