Document Sample
                       AND FREIGHTER CONVERSIONS
                                    By Jack Feir
                        President, Jack B Feir & Associates


                                   March 26 – 28, 2001

When we say that an aircraft has retired, it hasn’t used up a life limit or simply worn out.
It didn’t just collapse and die one day. It was retired because of a decision by
management. Its owners decided they would be better off with the airplane out of
service than they would be to keep it flying. This is almost always an economic decision,
not a technical one; and it is good to remember that economics has been called the
“dismal science.”

When we forecast an aircraft’s useful life, we are really trying to forecast that decision by
management, and when we look at a new airplane and forecast a useful life of 25 to 35
years we are really looking off into the unknown. Within that lifetime there could be new
rules for noise and emissions that haven't even been proposed yet, competition from new
aircraft that don't even exist yet, or maybe new technology that has yet to be invented.

With all these dismal uncertainties, you might wonder why we even try to forecast useful
lives and retirement rates.

So why bother to try? Why do any of us care how long aircraft lives are going to be?
Simply because retirement rates are among the big drivers in the economic balance of
supply and demand.

   •   Lessors need to know how long the stream of rental payments will last. They also
       need to satisfy certain IRS requirements for booking useful lives and residual

   •   Aircraft owners, whether they are airlines or lessors, need to know when they
       must buy replacements for their oldest aircraft.

   •   Banks want to be sure their loans are getting paid off faster than the aircraft
       market values are going down.

   •   Manufacturers (and all of their suppliers) must plan their production rates to
       make up for aircraft retirements. Of Boeing’s recent annual forecast of 22,000

                                                                           Jack B Feir & Associates
                                                                                       Page 2

       aircraft to be manufactured over next 20 years, over 6,000 are to replace

   •   Suppliers need to predict the market for spare parts.

   •   Modification companies need to estimate the size of the market for things like
       hush kits and refurbishments. They would like to know how many aircraft will be
       kept in the fleet or live long enough to need the modifications.

Almost anyone who has tried to forecast the airplane market has had to include a
forecast of retirements. I’m going to show you:

   •   The bad news is that most retirement forecasts get it wrong.

   •   The good news is that we are just now beginning to see enough real data to do a
       legitimate analysis and so we should be able to do better.

   •   The bad news it that we will probably still get it wrong.

Good or bad, the fact is that all of the above groups still need to rely on some kind of
forecast as they plan to make investments, develop products, open or close factories and
hire people or let them go.


Forecasting Retirements is Not Easy
First, let’s consider why and how retirement forecasts go wrong.

There was a time when aircraft lives really were quite short. Back in the days before
economic deregulation, the ATA had developed formulas for estimating aircraft direct
operating costs. In 1955 their formula for depreciation was based on a 10-percent
residual after only 7 years. To us now, this sounds ridiculously low, but back then the
only information they had to go on was for aircraft like DC-4s and DC-6s that had
become obsolete almost the day after they were built.

When the formula was revised in 1960 this was increased to 10 years, and by 1967 it was
increased to 12 years. With such short lives, residual values and tradeability were not
high priority items. But what happened was that over the years since then most jet
aircraft stayed in service much longer than these formulas would suggest. Because of the
expectation of longer lives, now there are huge numbers of airplanes being leased or
bought with long-term financing, so a lot of people are betting a lot of money on how
long the airplanes’ lives will be and how much they will be worth between the beginning
and the end.

                                                                         Jack B Feir & Associates
                                                                                           Page 3

The “flip side” of the question of aircraft useful lives is the rate of retirements. I would
like to show you just how devilishly difficult it has been to forecast aircraft retirements,
then we’ll come back to useful lives.

                              History Through 1989




         1950        1960          1970          1980           1990           2000

Go back about 10 years and imagine it is 1990. It is your job to make a forecast of
aircraft retirements over the next ten years.

Probably one of the things you would do is pull together whatever information you could
find on the retirements that had taken place so far, and you might have produced a graph
like the one above. It shows the numbers of jet transport aircraft that dropped out of
service each year up through 1989.

Certainly the numbers bounced around quite a bit, but since 1970 they have tended to be
in the range of 50 to 100 retired per year, and only once did retirements get as high as

For you history buffs, the data in the early 1950s represent the de Havilland Comets that
began to be delivered in 1952. There were several lost in accidents in 1953 and 1954,
which led to many others being taken out of service in 1954

You can see that the numbers of retirements have had their ups and downs, so you
already have some idea about how tough this could be to forecast. You can see that
there were a bunch of retirements in the mid-1970s during the fuel embargo crisis and
also in the early 1980s when the industry had one of its worst down-turns. That down-

                                                                             Jack B Feir & Associates
                                                                                         Page 4

turn also coincided with the phase-out rules for Stage 1 airplanes so that caused some
retirements too.

On the other hand, in the late 1980s there were very few retirements because business
was booming and the airlines needed all the airplanes they could hang onto. Airplanes
that might have been retired were kept in service because new ones couldn’t be built fast
enough, so retirements were very low.

Now imagine it is your job to forecast where this is going. I remember trying to do it a
few times, and maybe some of you have tried too. I hasn’t been easy.

                           History Through 1989








        1950            1960              1970               1980                 1990

One of the things you couldn’t ignore was that there was a sort of “baby boom” of
production back in the late 1960s. The chart above shows the annual jet aircraft
production rates each year from 1950 through 1990. In the late 1960s there was a huge
boom when there had been 500 to 700 aircraft deliveries per year. If you expected these
to have lives of about 25 years, by the 1990s you would expect to see this echoed as 500
to 700 retirements per year. You would also have in your mind the Aloha accident in
1988 that looked like it would spell the end for old airplanes.

So looking forward in 1989 you would probably expect a big peak of retirements coming
up soon, and then settling down at 300 to 400 per year in the long run, probably looking
something like the following.

                                                                          Jack B Feir & Associates
                                                                                         Page 5

                         History and Forecast Compared





        1950       1960         1970          1980                1990    2000

The “connect-the-dots-line” is just such a forecast that was made in late 1989 by one of
the most respected forecasters in this industry. The history bars are the same as before;
I’ve just changed the vertical scale to make room for the expected peak. Like the rest of
us would have done, this forecaster projected a big peak of retirements in the 1990s, and
then expected it would settle down at around 300 per year after that. I have also added
the actual data for the 1990s to show what really did happen through the end of 2000.

I won’t mention any names because I’m not trying to embarrass anyone. I’m just trying
to show how hard it has been to forecast retirements.

How could this forecaster (or any of the rest of us who were making the same kinds of
forecasts) have missed the mark by so much? Mainly because in 1989 there was still not
much to go on as far as real-world retirements. Very few retirements had happened yet.

As the chart shows, what really happened is the retirement rate in the early 1990s was
somewhat higher than it had been in the late 1980s, but still well below the forecast. This
particular forecast was for a total of 3,300 retirements in the 1990s, but the reality was
half of that; a bit under 1,700. Even with the Stage 2 phase-outs and more aging aircraft
in the fleet in the 1990s, not once did the actual retirements get as high as the forecast.

In any case, now we enter the new century with a lot more old airplanes than we
expected, and surely they have to go some time soon. So if you are going to make a new
forecast, I wouldn’t blame you if you started out by putting in a big peak coming up
soon, even though this is the same logic that failed before.

                                                                           Jack B Feir & Associates
                                                                                           Page 6

What you really need to improve your forecast of retirements is a good basis for
estimating aircraft useful lives, and that basis was not available in 1990.


Useful Lives
So let’s go back again to the other side of the problem and talk about useful lives rather
than retirement rates. It has been the inability to predict useful lives that has caused us to
miss the retirement forecasts.

This might be a good time to define what I mean by retirement. Most of my analyses
here are based on raw data from Jet Information Services, and I have used their
definition that retirement occurs when an aircraft is finally taken off the national register.
In many cases the actual working life of the airplane is shorter than that because it may
spend its last few years in storage where it is no longer working for a living, but it isn’t
dead either.

It would be nice if we had good data on when each airplane made its last revenue flight,
which would be the perfect definition of retirement, but that kind of data is tough to
come by. Many of the airplanes parked out at Mojave and Marana might already be
retired, but many others are going to return to service again. Until an airplane is
deregistered we can’t really be sure it has made its last revenue flight and is gone for

So whenever you see an analysis of useful lives and retirements, make sure you
understand the definition, because we don’t all do it the same way. Sometimes people
have considered passenger planes to be retired when they are converted to freighters,
and sometimes a percentage of stored airplanes are counted as retired.

Whatever the definition, we are beginning to get enough data that we should be able to
improve on our forecasts of useful lives.

Survival Curves
One concept I am going to refer to is the “survivor curve.” The first time I worked out a
survivor curve for jet transports Gil Speed thought it was so newsworthy he published it
in Speednews in early 1985. But I wasn’t the first to invent the survivor curve. For a
long time survivor curves in one form or another have been used by insurance companies
and the utility industries. For example, if you search the internet you will find studies by
electric utilities that have used survivor curves to estimate useful lives and replacement
rates for such things as wooden telephone poles in different environments.

                                                                             Jack B Feir & Associates
                                                                                                              Page 7

                                       SURVIVAL CURVE FOR 1960 YEAR OF MANUFACTURE
                                                      All Jet Transports

 % Attaining Specified Age




                                   0      5    10    15     20           25      30   35   40    45
                                                           Age Attained, years

This survivor curve represents the jet transports that were delivered in 1960. Originally
there were 255 of them, and what I did was track each one of them, year by year, to see
how the population dwindled, and then plotted the numbers on a percentage scale. They
reached 40 years old at the end of 2000, and there were only 5 left.

The slope of the curve at any point is the retirement rate in that year. During the first 12
or 13 years the slope is fairly shallow. During these early years nearly all “retirements”
are by accidents, and this shallow slope is actually the accident rate. It was a little less
than one percent per year.

After about age 13 the rate was much steeper and this is the onset of voluntary
retirements going off at about 10 percent a year. Why did retirements begin to accelerate
so much just then? That would have been around 1973, and that is when fuel prices took
off, and by 1974 fuel prices were double what they had been in 1973, and they doubled
twice more after that. These 1960 airplanes were mostly 707s, 720s, DC-8s, Convair
880s and de Havilland Comets. None of these were very fuel-efficient (and besides,
Boeing 747s were taking over the prime long-haul routes), so serious retirements began
in the mid-1970s, and half of these aircraft built in 1960 were gone by about age 18.

By the time the surviving aircraft reached age 25, which would be by 1985, only about
10 percent still survived, and these are the stragglers that have been very gradually tailing
off since then.

So how would you define the useful life of a “typical” 1960 airplane? Technically, you
can’t use their average life because until the last one retires you won’t have all the
numbers you need in order to calculate the average.

                                                                                                Jack B Feir & Associates
                                                                                                                                         Page 8

The definition I like to use is to look for the age at which the population has dropped to
one-half of the initial number, i.e. when the survivor curve drops through the 50% level.
Statisticians would call this the median. Half the planes retired before then, and the other
half will retire after that. For these 1960 airplanes, that figure is between 18 and 19
years. And unlike waiting for the last one to retire before you can calculate the average,
you could have worked out the median in 1980 when those airplanes reached age 20.

The trouble is, if you used that number to forecast the retirements of airplanes built in
later years, you would still get it wrong, and like everybody else you would have
predicted retirements to come sooner than they actually did.

Just as people born today will probably live longer than people born 50 years ago,
airplanes built later have lasted longer than the early ones.

                                       SURVIVAL CURVES BY YEAR OF MANUFACTURE
                                                    All Jet Transports

                                                                               Mfr'd 1970
 % Attaining Specified Age


                                                                                            Mfr'd 1965


                                                                                                         Mfr'd 1960

                                   0    5    10    15     20           25            30              35               40     45
                                                         Age Attained, years

Here I have added the curves for the airplanes manufactured in 1965 and 1970. The
interesting thing is that the initial slope, (i.e. the accident rate in the early years) was
about the same, but the voluntary retirements did not begin until later, and they occurred
over a more extended period of time.

One conclusion you can draw from these lines is that in each case the steep retirements
began to happen because of some external event such as the 1973 fuel crisis, the Stage 2
retrofit rules, or the 1991 recession. So the secret to forecasting useful lives and
retirement rates is as easy as forecasting when crises like these will strike.

The curve for the 1965 models leaped a long way ahead of the 1960 models, and hit the
50-percent level at about age 29 (in the mid-1990s). The 1970 models look like they will
hit the 50-percent level in a couple more years at around 31 or 32 years of age. This is

                                                                                                                           Jack B Feir & Associates
                                                                                                                                          Page 9

not much more of a leap ahead, and my own feeling is that the curves for later years will
probably also show a median retirement life between 30 and 35 years. In other words, I
don’t think future curves will keep marching off the right side of the chart.

                                       SURVIVAL CURVES BY YEAR OF MANUFACTURE
                                                    All Jet Transports
                                                               Mfr'd 1980, 85

                                                                                Mfr'd 1970
 % Attaining Specified Age


                                                                                             Mfr'd 1965


                                                                                                          Mfr'd 1960

                                   0    5    10    15     20            25            30              35               40        45
                                                         Age Attained, years

To round out the story, on the graph above I have added the data for aircraft built in
1980 and 1985. They appear as the lines right at the top that have just now reached age
15 and 20, and the points for the two years fall almost on top of one another.

The only thing we can see for certain is that the initial attrition rate for these newer
models has been much lower than it was for the earlier generations. Why should this be
so? This is probably because (a) these newer jets are easier to fly than the old ones, (b)
better weather radar and ground proximity warning systems help avoid some disasters,
and (c) more pilot training was done in simulators, not in real airplanes.

Back in the early days there were quite a few training accidents. In those days pilots had
to demonstrate such things as their ability to cope with an engine failure at the most
critical moment during a takeoff and the failure of a second engine on the approach to
land. This had to be done in a real airplane. If the pilot flunked the test the result was
sometimes a spectacular crash. Today, in the simulators, if he flunks the test he gets to
try it again until he gets it right. No airplane is wrecked in the process and we end up
with better-trained pilots. Add to that the improvements in various systems, and the
result is that newer airplanes are not being wiped out as frequently in their early years.

Based on their survival curve so far, about all you can say is that the 1980 and 1985
models have had a much lower initial loss rate, and that their average lives will probably
be more than 30 years, but maybe not a whole lot more.

                                                                                                                            Jack B Feir & Associates
                                                                                         Page 10

I said earlier that the good news is we can use recent data such as this to improve our
forecasts of retirements. First, we can use survivor curves to help estimate the median
age at retirement, and secondly we need to recognize that regardless of the median
retirement age, actual retirements will be spread over a wide span of time. As these later
curves show, retirements might start in earnest at age 20 and go almost continuously
until age 40, and there is really no peak year in the middle like you might expect from a
normal statistical distribution. This is perhaps the most significant and most unexpected
result of the analysis. So the “baby boom” of airplanes made in the late 1960s does not
hit us like a sudden tidal wave of retirements in the 1990s. Instead, it comes on slowly
and carries on for many years.

If you really wanted to try an analytical exercise, you could probably make a fairly good
extrapolation of the shape of these survival curves, and then use those to create a
forecast of annual retirements. Maybe I should try that, then seal it in an envelope to be
opened at this conference 10 years from now.


So far what I’ve shown here is aggregated data that includes aircraft of all types. The
data is still rather sparse, but you can slice and dice it into smaller groups, depending on
what you want to study.

I’ll show you the results of three cuts I have made:
        Wide-bodies versus narrow
        Passenger versus freighters
        A specific type of airplane

Does Size Matter?
A question that has come up more and more often in the past few years is whether large
airplanes or small ones will have longer useful lives. In the 1980s we “just knew” that
wide-bodies would have longer lives. Knowing that, we also expected that they would
retain their residual values relatively better. Now we can question that logic.

Suppose you were to think about this in terms of Boeing 747s and 737s which were
1960s designs by the same manufacturer.

Consider the 747:
      Deliveries started in 1969.
      There are about 1,000 in service
      They are in “low-cycle” operations, most still under 20,000 cycles

Now consider the 737s:
      Deliveries started earlier, (1967)
      There are over three times as many in service (about 3,500)

                                                                            Jack B Feir & Associates
                                                                                            Page 11

           They are in “high-cycle” usage, many are beyond 75,000 cycles

Based on this, you would expect that by now there should have been at least four or five
times as many 737s retired as 747s.

                         CUMULATIVE RETIREMENTS
                    Voluntary Retirements of Boeing 737s and 747s







   01/88        01/90     01/92     01/94     01/96     01/98       01/00          01/02

But you would be wrong. This chart shows the cumulative retirements of 737s and 747s
plotted on the same scale (these are just the “voluntary” retirements, not accidents). The
first such retirements occurred in 1988, and by the end of 2000 about 130 747s had been
retired, and just over 150 737s. You can see that they have been retiring in very nearly
the same numbers at very nearly the same times.

What this means is that the rate of retirements for 747s in terms of percentages of the
population has been roughly three times as high as the rate for 737s, and it also means
that 747s as a group are not achieving useful lives as long as the 737s. Ten years ago
most of us were forecasting that the larger planes would have the longer lives.

One thing I noticed in Boeing’s annual forecasts for the past several years is that they are
still forecasting longer lives for wide-bodies than narrow-bodies. For narrow-bodies
manufactured after 1980 they use a 28-year replacement, and for wide-bodies they use
31 years. I think the evidence so far is pretty clear that it is the other way around.

This will be even more clearly the case when the 70 or 80 L1011s stored at Marana,
Mojave, Roswell and Victorville are finally declared dead.

As for the future aircraft at both ends of the size scale, we can only speculate. My own
feeling is that the biggest airplanes like the A380 and the proposed stretched 747 will

                                                                               Jack B Feir & Associates
                                                                                           Page 12

have lives as short as, or maybe even shorter than, the other wide-bodies that are not so
big. We are already seeing this in the case of the high-capacity 747-300s. The problem
will be marketability of used ones after their first and second operators have used them
for fifteen or twenty years. Unless they are great candidates for freighter conversion,
they may not last much longer than that. Even then, freighter conversions will probably
soak up only a fraction of the numbers coming out of passenger service.

As for the very small aircraft like today’s 50-seat regional jets, again we have nothing to
go on, and we won’t have for several more years. For now, I tend to give them shorter
lives than the “full size” narrow-bodies, but that’s mainly because I’m just being
cautious, not because I have any real facts to work with. For the regional jets, freighter
conversions will probably not be a factor to extend their lives at all.


Do Freighters Last Longer?
Now that I’ve mentioned it, is it really true that freighters have longer lives than
passenger models? We can look at the data, but so far the retirement statistics for
freighters are still pretty thin. Partly that is because freighters are only a small fraction of
the total fleet so they should account for only a small fraction of total retirements. Partly,
also, it seems like their lives really are longer, which confirms our intuition, but again
that means we have very little data to look at because so few have retired.

The statistics are so thin that it really isn’t possible to divide the fleet into passenger and
freighter models and plot separate survivor curves. What I did do was calculate the
average age at retirement for the two groups. I did this for four successive time periods
to see if the differences are consistent and if there are any trends.

                Average Age at Retirement
                (voluntary retirements only)

                Year            Passenger        Freighter
                Retired         Models           Models
                1989-91         24.2 yrs         26.5 yrs
                1992-94         24.4 yrs         27.7 yrs
                1995-97         26.9 yrs         31.8 yrs
                1998-00         27.5 yrs         32.2 yrs

No surprise here; freighters that have been retired in the last several years have indeed
lived longer than the passenger models, and I think it is a safe bet that this difference will
persist into future years. By this analysis, the difference is about 5 years and getting
wider. It will probably get even wider as there are many quite old freighters that will
drag up the average when they do finally retire. Still, you need to be careful how you use
this data because it represents the retirement ages for aircraft that were built about 30

                                                                              Jack B Feir & Associates
                                                                                                    Page 13

years ago, so you will have to make a long extrapolation if you want to estimate the age
at retirement for freighters in production today.


The third slice-and-dice I wanted to show is one that narrows down to just one specific
airplane type.

Can We “Prove” that DC-8 Freighters Will Have Unusually Long Lives?
About two years ago I was involved in the analysis to support a proposed securitization
of a number of DC-8 freighters. Normally I don’t talk about projects I have done for
specific clients, but this analysis is in the prospectus that is in the public record, so it is
safe for me to talk about it.

These DC-8s were already about 30 years old, so it was not surprising that the rating
agencies were a bit skittish about how much longer these aircraft would remain in
service, and therefore needed some comfort that the projected stream of lease rentals
could be counted on for enough years for the securitization to make sense.

                                 DC-8-60/70 Survivors
                           Compared to Other Aircraft Mfr'd 1967-1972



                                                                          All Others




        1970    1975      1980        1985          1990         1995     2000          2005
                                             Year End

What I included in my report were these two survivor curves which I have updated here
to the end of 2000. The upper one represents all the DC-8-60s and 70s (there were about
260 of them to start with), and the lower one represents all the other airplanes
manufactured in the same years (1967 through 1972). For the DC-8s there have been
almost no voluntarily retirements at all.

The lower curve shows that nearly one-half of the other aircraft manufactured at the
same time as the DC-8s have now been retired, but over 80 percent of the DC-8s are still

                                                                                       Jack B Feir & Associates
                                                                                        Page 14

working for a living at age 30, so it was fairly easy to make the case that it would be well
past age 40 before the DC-8 fleet would get down to anything near the 50-percent level.

In particular, the DC-8-70s had everything going for them. They were narrow-bodies,
which was good; they were almost all freighters, which was good; and they had the CFM
engines that gave them the low noise and low fuel consumption of much newer airplanes.

  • Until recently, real data on useful lives and retirement rates has been sparse, so
    forecasts were very difficult.

   •   Some pretty good data is now becoming available.

   •   By necessity, the data we are getting is for the previous generations of aircraft, so
       as in any forecast we will have to find a way to extrapolate somehow to predict
       the future retirements of today’s generation.

   •   Unforeseeable events like fuel prices, noise rules and business cycles will always
       make it impossible to forecast retirements with certainty.

   •   The earliest generation of jets had useful lives under 20 years, but this has quickly
       been extended so that 30 to 35 years is a reasonable expectation for the median
       age at retirement for today’s aircraft.

   •   Actual retirements tend to be widely distributed over many years before and after
       the median retirement age. There is no peak retirement age.

   •   As a group, narrow-bodies tend to have longer useful lives than wide-bodies,
       which is contrary to expectations of some years ago.

   •   As a group, freighters (and passenger planes converted into freighters) tend to
       have longer useful lives than passenger planes.

   •   Some particular groups of airplanes, like the DC-8 freighters, will have unusually
       long useful lives.

   •   And a tentative prediction: perhaps DC-10 and MD-11 freighters will turn out
       the match the longevity of the DC-8s.

   •   Most of these conclusions are not truly surprising, but at least now we can put
       some numbers to them.

                                                                           Jack B Feir & Associates

Shared By: