RULES FOR DISTRIBUTING A FREE GO

							National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

RULES

FOR DISTRIBUTING A FREE GOVERNMENT AMONG AREAS OF A CITY
CARL S. SHOUP*

SERVICE

ABSTRACT Just whatallocation does, rule orshould, predominaten the day-by-day i decision Isa typical urbangovernment service- making isa subject thathasbeenrelafire orpolicerotection, p orgarbagerel explored, least y econoat b moval,or f example-being rendered evenlytivelyittle withtheattention mists,ompared c throughout a city? Ifnot, hatalternative thedistribution certain given w of a total f o rules arebeingfollowed? Decisions to are Political have being made continuallythespatial taxation.inthisscientists been on al- moreactive butfor statend a location ofinputs: policemen, the firemen,local ervices a field,intheUnited s as whole, sanitation workers. makesthese Who de- States, remarksmade twenty-five the cisions, whatrulesave and h they adopted? apply: Infact,ittle " l is Thesequestions haveattracted theatten yearsago still - known about distribution government of tion few economists, ofmany more services y location, ofa and b race,religion, inpolitical scientists. Answers havebeenin- come class, r othercategory.he laws o T ferred rom statistical f surveys,utalmost providingortheservice resilentnthis b a i no one seems tohavegone directly tothe respect;f authorizing appropriating the or rulemakers,whoevertheymay be,toas- committeesof legislatures not discuss do certainheprevailing t allocation formulae. it; budgetssubmitted theexecutive ay by s The present paper summarizes themajor a givenservice findings todateand offers ome apprais- nothingabouthow the users. his istobe s distributed among T silence alsand suggestions. reflects inparta social ropensityodisp t criminateovertlyn ways thatare not c i tolerable intaxation" (Shoup, 964, 1 p.383). 1.Distributing Fixed Totalof an a Partialxceptions e tothis statement re a Urban Government ServiceAmong tobe foundinthedispensingfeducation o Users and inprotection against crime. great The T the local level government, of and changesthathave been made inthe disAto a lesseregreeat thestate d level, tributionf education wellknown, o are decisions must be made on how to dis- howeverfarthatdistribution still may be tribute fixedamount ofa certain a ser- from some specific standard. he distriT vice, ayfire s protection, among users. he butionof protection police been a T by has decisionsormallyconcerngeographical lesspublicizedssue, ut a considerable n i b groupsof users, individualouse- literature not h inscholarly journals nd books a holdsand businessfirms.Protection has developed.iscriminationnthesupD i against may be greater less fire or inone plyof public housing, too,has been disneighborhood han in another, not cussed. t but Much, however, remains unusually one neighbor for than for another. known orunresolved,veninthesefields, e The level ofprotection might be made the withrespect totwo basic questions:hat W same fortwo neighborhoods y shifting rules be formulated?"at rulesare b can N some fire-fighting inputsfromtheonearea in factbeingfollowed y thosewho deb tothe other. This move, however,might terminethe distribution urban serof result nan increase ntotal damage vices? i i fire overthetwo neighborhoods ogether,rit t o Some answers tothesequestions ofare might beinconsistent withan equalwork- feredinSections Iand III I below.In Secloadrule, orwith some otherdistributive tion11, alternative fordistributing rules standard. urban services re analyzedand evalua These arerulesthathave been pro*Professormeritus,Columbia University-Cen ated. E terSandwich,NH 03227. posedby thoseworkingin thisfield, or 103

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

104

NATIONAL

TAX JOURNAL

[Vol. XLII

inferred them from statistical by studies of actualdistributions.ome of these S studiesare summarized in Section III. Rules announced by thosewho decide the distributingre,on the otherhand, rare. a The aim of the presentpaper isnot to designatewhich rules are the most appropriate, even to discoverthose most or widely used.Rather,itistostimulatefurther analysis of this largelyunformulatedproblem.We are still the stage at where itseems usefultoraisemore questionsthan are answered. For illustrative purposes,the services of fireprotection and garbage and trash removal willbe given particular attention. Certain topicscuttingacrossthe several serviceshave been ratherneglected, and two of these willbe analyzed below. One is the possible monetary lossto the user ofthe government service through a market processof pricechanges that tend to offset benefitto him from that serthe vice,analogous,in reversefashion,to the shifting f a tax (Section o IV).The other isthe distinction between input and output and the measurement of eitherone, in physical monetary terms (Section or II).

there isthe problem of measurement. Inputs can usuallybe statedreadilyin terms of money. Outputs normally have no money pricesattached to them, and, moreover, are commonly consideredto be more appropriately measured in physical terms. (1)Equal Inputs

The equal input rulewillnot prove acceptableifthe serviceisone of removing something or preventing something from happening. Consider,forexample, a state highway department faced with a varied distribution snow throughout the state. of If equal inputsper mile of affected road were the rule forsnow clearance, either some roads would be impassable in winter,or otherroads would continueto be scraped forsome time afterthe snow had been removed. Similarly, equal input of garbage removal serviceper mile of street would disregarddifferencesn densityof i populationor ofbusiness. Equal input per capitaforthe areas served would be more nearly acceptable, ut would ignoredifb feringpropensitiesoproduce garbage.An t average rich family produces more garbage than an average poor family,and an equal input per capita rule would lead Il. Alternative Allocation Rules eitherto garbage pilingup at well-to-do Some dozen alternative rules for dis- residencesor being snatched every hour tributingan urban servicehave been at poor residences. analyzed in the literature. Some of these Ifthe serviceisone ofpreventing somerules are not generallyacceptable, ut thing from happening, as isthe prevenb neverthelessmay be inforcein some com- tionofdamage by fire, he equalinputrule t munities(Section III). is again unappealing.If the input,per resident, were justenough topreventvirtually all damage by firein a wealthy (A) Equality in Distribution residential district, identicalevelof an l Equal distributionfthe service o per re- input per residentin a poor area would cipient a rule that has a strongintui- allow substantial is firedamage because of tive appeal.Two versionsof the rule, more combustible buildings,greater however, need tobe distinguished. seach crowding and the like. I recipient tobe allocated the same amount For otherthan removal or preventive of input,even though, because of differ- services,qual input per capitahas gained e ing environments from area to area, the considerable support. Kennett (1982)terms result willbe differingutputsper recip- it"the simplest, o probablymost widelyapient?Or shallthe rule be equal outputs pliedand the one currentlyenjoyingjureceivedby each person,even though this dicial support"(p.81);he cites two cases may imply differing nputsper recipient? where the court ruled againstcomplaini Besides the equity aspectsof this choice, ants who showed unequal outputs despite

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

No. 21

DISTRIBUTING

A FREE GOVERNMENT

SERVICE

105

equalinputs. inputmeasures used by a little. The "Outputswould be assessed the by Boyle and Jacobs are noted in SectionIII utilization rateforprograms,both inraw numbers ofclientsnd the intensityfuse a o below. per client. can measure such unitsof One service hours of counseling unitsof as or (2)Equal Outputs methadone prescribed.Outcomes are t Turning toequality ofoutputs, find changesindrug dependencyinclientshat we thatappropriateeasuring unitsaredif- can be attributed othe program.Impacts m t ficultodefine, illustrated garbage are changes in measures ofdrug abuse in t as by removal and fireprotection. the citydue to the program,and might be For garbage disposal could,with measured againstsuch indicatorss drugwe a some difficulty, measure the bulk or related deaths, births addicted to mothers weight ofgarbage removed, and compute and overdosecasesbrought to hospitals. the amounts taken per capitain each dis- Clearlymany factors . . [including] . trict. Would equality fsuch amounts,over availability o fordrugson the street . . . s . p districts, an acceptable be equitystan- can affect uchindicators . ."(Jones, . dard? Ifitwere employed,some house- 162). thispassage, In "output"isused in holdswould be left,very day,with some about the same senseas inthe presentese unremoved garbage.The output desired, say. however,isabsenceof garbage.Ifthisis Rich has alsodistinguished outputsfrom not achieveable,erhaps the standard outcomes,in a manner closer thatused p to shouldbe,an equal amount ofgarbage left by Bradford, Malt and Oates.In street at each place, over allneighborhoods. An cleaning, example, the "scheduleof for isan output, while the "meaalternative standardmight be,equalde- cleaning" layin pick-up ofallgarbage.(We bypass sure ofcleanliness streets"san outof i the problemthatnot allunremoved gar- come indicator. outcome,but not the The bage isequallyoffensive.) output,isaffectedy "characteristics b of As to a preventive service,ikeprotec- the community,"e.g.,ropensity o litter l p t 1982,pp.7 10). The view takenin tionagainst damage by fire,tcan never (Rich, i be known how much damage has been the presentstudyisthatthe scheduleof prevented. Instead, must use a nega- cleaningisan input,and the degree of we tivemeasure:the amount of damage that cleanlinessf streets, output.In any o an doesoccurdespite service. smaller service the The rendered by the government, the thisamount, the higheristhe level the outputisofcoursedependentin parton of ofthe community. The service, terms of output. in Equalityin the characteristics is amount of the service receivedtherefore importantdistinction between those t deliverer an conc exists when an equalamount offiredam- factors hatthe service o and those he age occurs overthe neighborhoods (butfor trol(schedule f cleaning) what time period?). Computing a total cannot(communitycharacteristics). money value of severaltypes of damage "Intermediate utput"isa conceptthat o istroubling,speciallys to damage to seems usefulin some instances.tisoute a I propertyand lossoflife r injury. o put thatservesas an inputto achievethe In some ofthe public-service literature,consumer service desired. Craig (1987) theterm "output"has been used todenote used thisterm in analyzing police distriwhat isheretaken tobe input, what bution. histerminology,nputismeaand In i isconsidered here to be outputiscalled suredby the number ofpolice. "clearThe ttoutcome" "consequence," r "impact." ance rate," percentageof crimes that or o the Bryan Jones (1982)approvingly quotes are considered solved, usuallyby arrests, which issafety, Bradford, Malt and Oates (1969)citing isnot the finalproduct, police patrolunitsas "outputs" (p.156). but ismerely a stagealong the way. The In anotherexample, a drug treatment higher the clearance rate,the more are program sponsoredby a cityhealthde- criminals deterred, nd the greatertherea partment,Jones'definitions seem to shift fore is "safety," measured inversely to

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

106

NATIONAL

TAX JOURNAL

[Vol, XLII

crimes per capita.Craig considersthe clearance rate an "intermediate input," or, indifferently, "intermediate utput"(p. an o 335).The amount of "output"(final output to consumer),safety, ismeasured by subtracting crime rate per capitafrom the a constantterm, thisterm being "slightly largerthan the maximum crime rate"(p. 335). What Craig terms "output"is named "outcome"in Behrrnan and Craig (pp.4243).It seems preferable reserve"outto come" forsubsequent consequences,as in the political scienceliteratureeferred r to above.Alternatively,outcome" might be " used to describedistributionfthe output o among the districts the city: certain of a levelof safetyhere, another levelthere. Behrrnan and Craig so use it, but also, as noted,apply itto Craig's "output." Keeping the two conceptsdistincty using difb ferentterms seems desirable. Equality of levelof serviceover subareas isso difficult defineand measure to when the serviceis one of preventing something undesirable from happening, or removing something undesirable, that we may have to be satisfiedith observw ing grossinequalities, evidencedby the as occurrenceof the anti-goods. Fire protection and garbage removal are good cases in point. Insteadof computing average distance from the firehouse, forexample, we might measure the amount of damage done by fires.fthe number of such fires I per,say, one thousand residentsper year and the amount of damage estimated to be done were observed to be much higher in one censustractthan in any of the others, it couldbe inferred that distributionf the o service, ireprotection,as certainly f w unequal as respectsthis census tractrelativetothe others. general, highlyunIn a equal distributionof the anti-good, geographically,ould be proofof unequal w distributionf the serviceand of a need o to redistribute total the inputofthat serviceamong sub-areas. Attention would be focussed,not on measuring equality of servicedirectly, on spotting but gross inequalities the occurrenceof the antiin good.This iseasierto do, and leadsto remedial actionwhere itismost needed.

This kind of reasoning assumes, to be sure,that equalityof serviceimplies whatever inequality inputs is necesof sary to overcome the neighborhoods'differingdegrees ofpropensityto experience the anti-good.fcertainneighborhoodsare I composed chieflyof fire-trapwellings, d or are inhabitedlargelyby persons who, through befuddlement from drugs or alcohol, through ignoranceof how firedanger may arise, through sheer carelessor ness,are inadvertentarsonists,he input t of fireservice must be greater, per person, than in other neighborhoods, if equalityis to be achievedin the occurrence of the anti-good. hile thisview W may seem based on undue permissiveness or undue willingness overlookindividto uals'destructive onduct,it is probably c more generallyacceptable than a rulethat penalizesthose most likelyto leta fire startand develop.A sub-areathat experiencesdisastrousfires with great lossof life ear afteryearwillsurelyhave more y input of the fireservice allottedo it(at t the expense ofother sub-areas, iftotalinput is not increased), whatever other measures are alsotaken toavert these calamities. Servicesproducingpositive goods rather than preventingthe occurrenceof antigoods may alsorequireinequalityof inputs over areas of a cityto achieveequalityof outputs.Education is an example. Children in certainareas,disadvantaged by theirenvirom-nent, may need more input to achievea given levelof education. (B) Inequality nDistribution i The output of a particular urban servicemight be distributednequally among u districtsor eitherof three reasons:to f mitigateinequality incomes,to meet of unequal needs or preferences, tosatisfy or some more complex socialwelfare function. (1) Unequal Incomes The levelofthe servicemight be made higher in low-income districts, reasoning by analogy with the reversedistribution supportedfortaxation, thus givingmore

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

No. 21

DISTRIBUTING

A FREE GOVERNMENT

SERVICE

107

to thosewho have less. he analogy with tainlypersists. more specificsthe T The i distribution the tax bill of isclear, how- thinggiven,thatis, the smaller the vaever,onlyto the extentthatthe recipient rietyofuses itcan be put to,thegreater in o ofthe service regardsitas equivalent to willbe the disparities preferencesr a cash grantofthe same cost, pointdis- needs forit,among recipientsr among a o cussedbelow. Otherwise,the specificitygroups of recipients. ofthe service hatmakes its t value differ Perhaps the most sensible approach is among recipients rendersthe comparison tostart with thepresumptionthatequity withdistribution thetax bill ifficult. of d impliesequality,nd then modify that a presumptionin the more striking cases. No one isgoingtosuggestthatbecausea (2) Unequal Preferences orNeeds residential sub-area istohave a park with Isan equaldistribution service ofa an swingsand sand boxes,a businesssub-area equitable distribution ifthe recipients (thefinancialistrict?) have the d must differreatly theirpreferencesr need same. But toward the otherextreme,it g in o forthatparticularervice?o, notat all, may be unconvincing argue that res N to say some observers. "Consider, for ex- tiredpersonsneed parks lessthan do ample,the effectsf providingidentical families-thoughagain,if we become o park facilities totwo neighborhoodswhen sufficiently specific-playgrounds with one ispopulatedprimarily elderly by re- swings and sand boxes-needs or preftired personsand the otherconsistsostly erencesmay be seen to differ m sufficiently ofyoung families withchildren. ." . (Rich, to leavethe safehaven of equality. 1982,p.5).Waiving the question ofwhich Thisdifficulty indefining equitable an one ofthesegroupsreally wants,or needs, distribution service ofa thatisveryspethe park more, we are facedwith the cific isappears d ifallthe usersof itreproblem:how much lesstogivetothe one gard the level ofthe service they areregroup (average,er person?) p than to the ceivingas being equivalent a simple to other, achieveequity?Some rough ap- cash grantofthe same amount (amount to proximation may be better than nothing, beingmeasured in terms of cost).hey so T but we may supposethateither group can regarditforthatlevel ofthe service low make a good caseforor againstwhatever enough thateveryuserofitwould,ifthe unequal allocationsproposed. i freeservice were discontinued, purchase The importanceof thisproblem isre- thatamount on themarket at thatcost. duced somewhat by the factthat delib- Supply ofa modest amount ofwater isan eratelyunequal levels service of can be example.Thisfreesupplysimplyreleases, rendered,forthese urban services, only forthe consumer of it, corresponding a as between groups, as among individ- amount ofmoney, which,ifthe waterwere not uals, owing to the quasi-public-good na- notsuppliedfreeofcharge, would spend he tureof the services. Within the sub-area, inbuying water from a private company, once the service being renderedat a or the cityitself. amount of general is The certain level any one personitisalso purchasingpower releasedto the conto beingrenderedto allthe otherpersonsat sumer isthe same forevery consumer, thislevelin thatsub-areagroup.As be- with respect thatamount ofwater that to tween sub-areas,owever,the problem is would be thus purchasedby the lowesth a realone,ifthe group preference need using consumer.The water thatisin fact or forthe service differsarkedly from one supplied m above thislow levelisa specific good,with respect the equityproblem. to sub-areato another. This problemisnotescapedentirely f The hypotheticallternative i a purchasefor thegovernment distributes, partic- a priceneed not even be of preciselyhe nota t ularservice, a sum ofmoney. Among but same good,ifitsatisfies same need. the personswith equalincomes,itmay be ar- In the absenceof policepatrol, housegued,some have a greaterpreference for holderswould presumably spend more on money than others, at leastsome need locks, or burglaryalarms,and the like. In itmore than others. a differenceer- the absence of freefirefighting,hey Yet c t

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

108

NATIONAL

TAX JOURNAL

[Vol. XLII

to be served isto maximize the totaloutput.In fire protection, fire the stationsill w be sitedand the equipment and personnel sodistributed, among sub-areasofthe city, so that the total damage by fire over the cityas a whole willbe minimized.Whether a given pattern of deployment isin fact minimizing totaldamage can be ascertainedby comparing marginal outputsin the sub-areas. Total damage does not need tobe computed forthispurpose.Ifa unit, say $100,000worth, of firefighting resourceswillreducefiredamage more, in (3)SocialWelfareFunction a certainothersub-area,say a censustract, The distribution an be viewed as re- ifit ismoved there,than the increment c flecting social a welfarefunction, in as ofdamage thatwilloccurin the tractfrom Behrman and Craig,and in Craig.They which itis removed, totalcity-wide outassume that the functionreflects wo at- put has not yet been maxiniized,and the t titudes that modify the desireto mini- marginal adjustment should be made, acmize crime over the cityas a whole. The cordingto thisallocation ule. r first isaversionto inequality the safety in When thereare no more such marginal levelsamong districts, the second, and adjustmentsto be made, and totaloutput running counter to the first,s unequal i has thereforebeen maximized, the reconcern as tothe levelsofsafetyfrom dis- sulting distribution fireprotectionill of w trict district. to Why may therebe "more onlyby coincidence equal among allthe be concern"about the levelof safetyin one sub-areas.Residents of the disadvandistricthan in another? Behrman and taged areas willnot be receptive the t to Craig citeas reasons:(a)differences n i reassurancethat totalfiredamage over neighborhoodpolitical support, (b)differthe cityas a whole has now been miniences in tendency ofresidentstoemigrate mized.Indeed, such a deployment of fireunder a given level safety(p.38),and fightingresourcesmay prove quite unof (c) greaterconcern forthe poor (pp.38feasible politically. isconjecture;lThis a 39). most nothingisknown about the political reaction such a distributionf the serto o (4) Unequal Inputs vice.The political cientists, s who have so If the benefitsto the recipient the contributed much to an understanding of rule, servicecan be measured in terms of in- ofthe equal-distribution are not accustomed to dealing with marginal puts ratherthan, as above,outputs we changes in distributiono reach an overt may estimate roughly,at leastforsome ' ofoutput.' Itis, rather, services, how the distribution matches allmaximum level with the recipients'incomes. Musgrave and marginal changes in types of input that attention. n some services I Musgrave (1984,pp. 261 265) find,for have attracted some ofthe example, that with respectto education the consumer himselfsupplies whether toexpendituresin the United States, which Input,and the questionarises they distribute in linewith the distri- taloutput can be increasedby substitut" ing an increment of consumer-supplied bution of studentsby income brackets Input forgovemment-supplied input. For "higher education shows the benefit-toa income ratiorising with income while that example, "municipal trash collectorsnd may be substituted foreach other forlower education. . .declines" 261). citizens (1p. in transporting refuseto the curbsideor to othercollection locations" (Parks, al., et (C) Maximizing TotalOutput P. 188).Such "co-production"as distrih Another reasonable rule fordeploying butionalsignificance, ifthe time and efthe fixedtotalof resourcesover the area fortthus requiredof the service recipient

would surely spend more of their own money on firepreventionand capability of fighting fires. Most urban services are of thistype:substitutes available,p are u to a point, that can be purchased on the market, in contrastto,say,defenseagainst enemies from abroad. Maybe as much as half, even more, of the totalcostof uror ban services given freeof charge representsthe equivalentofa cash grantto the serviceusers.

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

No. 21

DISTRIBUTING

A FREE

GOVERNMENT

SERVICE

109

differsith the environment from district tion,in thisrespect. he amount of effort w T one fireman must exert during the day to district. depends on the kind of district is ashe signed to,and on the number of fellow (D) Equal Work Load firemen available help him. to A fourthallocationule is equal work r If such activities policingand fire as load.Those who are rendering the ser- fightingare allocatedamong city disvice-policemen, for example-are so tricts y thisequal work load rule,the b distributed over the city's sub-areasthat other allocation standardsdiscussedabove they allhave roughly equal work loads. must be discarded-equal or unequal For police,he work loadcan be measured t treatment of the servicerecipients, and in terms of the number of crimes,suita- maximization of output overthe cityas a bly weighted,that a policeman is called whole-unless one or the other happens on to dealwith.This was the decision rule to coincide with the equal work load rule. used inChicago in the 1960s (Shoup,1969, Treviiio Westendarp did in factfindsuch p. 117).It is a standardrarely, ever, a coinciding if fordistributionf policemen o mentioned in today's writingson alloca- in New York City,if, and only if, pothe tionof a service; itdoes not appear, for lice elasticities ofcrime were constantand example, inLineberry's informativechap- equal in allprecincts the city. that of In ter on how the bureaucracies' own deci- event,the distributionf the police o force sionrulesgovern the allocationf inputs among the precincts o would be the same over the sub-areas(Chapter6) (1977). It under the equal work load rule and the seems a natural enough rule for a com- maximum city-wide output rule(1970,pp. missionerto use for equityamong police- 75-92, especially . 80).In general,it p men (notamong servicerecipients). seems unlikelythat the distribution of Ifthe work load ismeasured simply by personnelinputs over sub-areaswill, for the number ofhours spent on the job,then, any service, atisfy s more than one of these indeed,ithas no part to play in the de- standards at once. cisionon how much input is to be asIn thiscontext, policing and firefightsigned to each sub-areain the city.No ing differ one important respect. a in If matter how the personnel is thus de- policeman is transferred from one preployed,allwork loadsare equal ifevery- cinctto another,he not only sharesthe one works, say,seven hours a day fivedays work load carriedby his fellowsin the a week. (We abstract from differentime second precinct, ut alsoreduces the total t b shifts.) police In work, however, the sim- job to be done. The mere presence of anple amount of time spent is evidently otherpoliceman willnoi7nally reduce the thought to be an inadequate measure for number ofcrimes conunitted. such drop No treating policemen fairly. in the totaltask willoccurjustbecause a In some othercityservices,he amount t fireman ismoved from one districto ant oftime spent may indeedbe a fairly good other.(In the precinctthe policeman measure for equal treatment of the em- leaves, the totaltask willof course corployees.Perhaps garbage removal is an respondinglyincrease.)fthe equal work I illustration. Ifsome of the workers have loadstandard isnot beingobserved,itcan more strenuous or more displeasing tasks be reached,in policing,ith a smaller w than in firefightthan others,that factcan be allowed for amount of reallocation in settingpay scales, sincethe relative ing.This may mean that the departure loadsamong workers isnot likely vary from the equal output and maximum toto greatlyfrom day to day. Controlof each taloutput standards can be lessseverein one'swork load is in the hands of a su- policing than in firefighting. pervisor. or policemen,this is not so; F distribution the policemen isdictated of (E) Other Allocation Rules by the crime pattern, ifwork loadsare to be equalized. Fire protection ismore like At leasteightother rulesor guides for police protection than likegarbage collec- distributing n urban service a have been

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

110

NATIONAL

TAX JOURNAL

[Vol. XLII

proposed, inferred or from statistical sur- of urban political analysis" Lineberry, (
veys (seeespecially Lineberry,1977, Ch. 3,and Rosentraub and Thompson, 1982). (1) Racial Discrimination.Ethnic minorities, when grouped geographically within the city, given a lower level are of the servicethan are ethnicmajorities. (2) Income-Class Discrimination. Lower-income classes, when grouped geographically,re given a lower levelof the a service, egardlessof ethnicity. r (3) Power-EliteFavoritism.Those districts n which the "power elite" i liveget a higherlevelofthe service.Power elite" " means those who more or lessdominate the political structure the city. of These three rulesare grouped as the "underclass hypothesis"forthe distributionof urban services. (4)Ecological Influences. isparities n D i servicelevelsare traceableto differing geographical and time-period circumstances. "The physicalproperties a land of parcelmay make itidealforpublicparks, or even for dumps and sewerage treatment plants"(Lineberry, 1977, p. 63). Older districts will be betterservedbecause the construction f librariesnd the o a likedoes not keep pace with the development of new residentialreas,though, a to be sure,facilities the olderneighin borhoods will not be the most modem. Density of populationmay be included under thisrubric;there willbe more crowding ofthe facility inthe more densely populated district. (5) Bureaucracy Decisions.With respectto the services forwhich the initial sitingdecisions are lessimportant than the day-to-daydecisions, geographic for allocation f personnel(police,orexamo f ple,compared to libraries), considerable discretions to how the serviceshallbe a distributed eographically leftto the g is officials the department, and even to in the lower levelsof personnel. What irules do they set?One has already been noted: equal work load among areas.A lessgenerally acceptedrule,more vague, isthat of"bureaucratic convenience"(Lineberry, 1977, p. 66). Generally,however, "The decisional processes"of the "urban bureaucracy. . .remain the terraincognita 1977, p. 64). (6) Complaints -Requests. Some servicesmay be distributed among districts partlyin responseto number and intensityof requestsor complaintsfrom the residents. Garbage collectionay be an m example, as also police protectionand street maintenance. A linkage with socioeconomic statusmay exist.If more requests per capita come from the higherincome districts, application f this reo sponse rule would be, in effect, doption a of a rule for distribution directlyy inb come class(for findingson thispoint,see the summary below of the study in Wichita, Kansas, by Sharp). The complaints-requestsypothesismay h be viewed as part ofthe bureaucratic decisionexplanationof distribution. Percy (1989)has analyzed "one set of bureaucratic decision rules[i.e., which conthat cerns]'demand processing,' that is,the ruleswhich publicserviceagenciesuse to receiveand interpret citizen requestsfor assistance, coupled to rules used to organizeand distributegency response"(see a also Percy and Scott,1985, which containsa 74-page study of "Demand Processingin PoliceAgencies,"based on researchby the authors,using data on calls forpoliceservice FortWorth, Texas, and in Percy, 1985). (7) Taxes Paid. Some servicesay be m distributed oughly in proportion local r to propertytax paid;these would be "property-related" services, includingpolice, fire, and sanitation,n contrastto weli fare,health and education.The amount of the urban servicemight be in proportionto the amount (value) property. of (8) Daytime versusNighttime Allocation.In some services,otablypolice, n and fire, allocationf the resourcesmust an o be made, not only over sub-areas, over but day versus night.The most notableinstance of thisoccurswith respectto the allocation between a businesssub-areaand a residential ub-area.The flow of pers sons from the latter the former in the to morning and the reverseflow in the evening createsa need to reallocate subby areas twice every twenty-four hours.

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

No. 21

DISTRIBUTING

A FREE GOVERNMENT

SERVICE

ill

Treviiio estendarp deviseda method for without assigningthem top importance. W measuring the nightpopulation and the Indeed,the lessthereisof a givenservice day populationin each precinct New tobe distributed, more importantbein the York City,and the night-to-day ratios in come therules fordistributing thatlittle. the 123 precincts went as low as 0.01and Similarly, one may appreciate Hero's ashigh as 1.68(Table2,pp.20-21).This connnentsthatthe distributive studies ail f swing inpopulation rom day tonightmay "to deal with possible f regulatory,yms not requiretwice-a-day reallocation bolic otherftmctionsfservices"p. of or o ( 669) personnel fora service likegarbage dis- and that"thepotentialaluesor benefits v posal, r education,ut itmust have some to citizens political o b of participation are, influenceon the deployment of police by implication, further eroded and politpersonneland fire-fighting personnel, ical passivity (implicitly) is further ensincetheiractivities, unlikethoseofthe couraged," sothat "The urban service litgarbage disposalworkers,must extend eratureimplicitlyuggestsa vitiated s around the clock. understandingof citizenshipith the w This daytime-nighttimeallocation transformation fthe citizen ntoa mere o i problemwill typically involve firstour consumer" (p.673). the f rulesdescribed above (equalor unequal Whatever one'sviewson thesematters, distribution, maximum output, equalwork much benefit be obtainedfrom Hero's can load)and possibly the complaints-re- careful analysis the technical of shortquestsrule, but not the othersix(racial comings of the conceptsand measures discrimination, income-class discrimina- commonly used in the allocational litertion, power-eliteavoritism, f ecologicaln- ature (especially pp. 663-666). i his fluences, bureaucracydecisions, taxes and paid). The bureaucracy, owever, in ar- 111.What Rules Are Being Followed? h riving at itsown decisionrules,may somewhat influence thistime-period disTo findout what rulesare in factbeing tribution. followed service in allocation among areas ofa city, wo methods are available: t the interviewing pproachand the statistical a (F) I'lon-Allocation Rules approach. Quite anotherpointof view on urban service delivery isthatwhich deniesthat (A) The Interviewingpproach A "the'mostsignificant single issueregarding municipalservices . .istheirallo- This approachconsists finterviewing . o cationto sociospatial groupswithin the thosewho setthe rules, the commisat city' quoting [ Lineberry, 977,pp. 12-141" sionerlevel furtherdown in the bu1 or (RodneyHero,1988,p.660). Hero,in his reaucracy.It seems not to have been recentsurvey of the literature this widelyused.An alternative on isto intersubject,oncludes onlythat"The out- view recipients the service,n different c not of i pouring of researchon urban servicede- districts, toget their impressions,fany, i livery. . . does not addressthese con- of inequality service of levels among discernsverywell," ut alsothatit"hasnot tricts, to deduce,from thisinformab and addressedthe questionof why some ser- tion, what ruleisbeingapplied. his apT vicesareprovidedand othersare not,nor proachisnotveryreliable, view ofthe in why the services rovidedareprovidedat difficulty, p forany one recipient, comof particular levels" pp. ( 660-661).He adds paringthe service level inhisdistrict with that"it may matterlittle getswhat' levels 'who ofthatservice otherdistricts. in when italreadyhas been determinedin The direct approach of interviewing earlier stagesof the policyprocessthat thosewho make the decisions n distrio therewillbe little toget"(p.661). bution of a certainservicerequiresa One can agreewith Hero on the impor- knowledge ofwho thosepersonsare.Here, tance of studyingthose earlier stages, the services intotwo groups. fall

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

112

NATIONAL

TAX JOURNAL

[Vol.XLII

In the first roup,exemplified li- findthe decisionmakers allocation g by on of among areas or groups willdebrary service and, to a lesser degree,fire a service protection, relative the level ofservice hat pend on which of threegroups,not just t a district isgoing toreceive willdepend two, the service falls in,according Roto largely an initial on 8iting decision: how ,entrauband Thompson 119821, They conshall thelibraries, orthefire stations, sidered be threeservices s "having a high a allocated spatially?nce thesedecisions potential orthe presenceofdiscretionary O f are made and the structuresuilt, he authorityby service b t deliverers: police, relative ervice s levels in considerable emergency services,nd animal control. are a partdetermined, the distances li- . . .Two services ere considered by the w tohave f in brary usershave to go,and the distances a medium potentialordiscretion serthefire enginesmust covertogetto a fire. vicedelivery: parksand libraries. .Five .. Itmay provequitediticultofindoutwho of the services t studiedwere seen as havhas made the siting decisions. ing a low potentialordiscretionary f acdeliverers . . . :garbage The secondgroup ofservices consists of tionsby service . thosethatare personnel-intensive, so that and brush collection. . street-maintea considerablyargeror smallernumber nance programs . . .allutilityervices l s . protection.. . . of personnelcan be attachedto any par- . . .bus service . .fire d perticular site. Here, not much importance In these fiveservices,elivery-level attaches tothe location ofthe buildingor sonnelactually have few opportunities to otherinfrastructure in supplying used the change service decisionsade atthe prom level" (pp.174-175). service.arbage collection G isan illustra-duction tion,as isprotectiony police.or these b F services,he levels output can be var- (B) The Statistical t of Approach ied, from one area to another, ifnot every day,then every week or month. The patA number of statistical studies urof ternof service distribution among areas ban servicedistribution have been made has not been largely frozenat the time of in the United Statesoverthe pasttwo desiting.he policecommissioneror sani- cades.Most ofthem have been carriedout T tation commissionerisusing one or more by political scientists, usingbivariater o rulesin deployinghis personnel, or,to multiplecorrelation techniques.he few T some degree, simplyleavingthe decisions studiesby economists have employed on distribution the service hisoffi- econometricmodels,allof them being of to deployment. cials,ometimesratherfardown the line. concernedwith police s The interviewer an then start c with the topofficial and,ifnecessary,ork on down w (1)Findings of Political cientists S through the ranks. This isa formidable The earlier indings fthepolitical f o scitask,yet itmay be necessaryifwe areto entists ere tothe effecthaturban serw t answer the basicquestion, What rulesare largely "unpatin in factfollowed thosewho determine vicesare distributed by temed inequalities," determinedchiefly the distribution local of services, these by bureaucratic in decision rules. This view personnel-intensive activities? Lineberry has been challengedby otherpolitical sciisquiteemphaticon thispoint. There are entists " inthe pastdecade,in studies that two remarkable thingsaboutdecision-rulestend to supportraceor income-classrefp in the bureaucraticllocations a ofpublic erence hypotheses. services. first The ishow low in the bureaucratic hierarchydecision-rules are (i)LINEBERRY (SAN ANTONIO) formulatedand implemented.The second San Texas studygives ishow insulated he rule-makers from Lineberry's Antonio, t are external constraints, insulationorti- the former interpretation. data are an f The fied therelative by invisibility ofboththe mainly from 1969 to 1972. rulesand the rule-makers" (Lineberry, San Antoniowas the eleventhcityin sizein the United States. It"containsew f 1977,p. 154). How fardown the lineone must go to blacks, more than halfits but population

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

No. 21

DISTRIBUTING

A FREE GOVERNMENT

SERVICE

113

isChicano [i.e. Spanishheritage].co- familyincome clearly a greaterdisof E are houses than thosescoring nomically, cityisheavilydependent tance from fire the upon its [three]ilitary m bases. . .but is low on the income measure.. . .The data fundamentallya poor one.. . .The City's do not suggestthat fireprotection isd rank in home ownershipisquitehigh,al- tancesare actually closerto neighborthough the houses are of relatively low hoods with advantages of power and stavalue,[and the]housing conditions are tus." Instead,it was the older poorindeed"interms ofovercrowding. he neighborhoodswhere the highestcorreT political "power structure" concen- lations ere found.(Lineberry, is w 117-119). tratedin the Good Government League, Garbage collection not covered in was (,a sortofupper-classolitical p machine," the correlationnalyses. a and most of the candidates supportedby Lineberry's eneralconclusion g forthe he theGGL have been "drawn from high sta- services did coveris: tusoccupations. - .Between 1955 and . "There are numerous explanations or f 1971,fullyeighty-eight ofthe city'sne thedistribution o ofurban services. have We hundred and thirteen censustractshad focusedon a testof severalof theseexno councilmanicrepresentation. . . planations. thetwo major typesofur. Of Ethnically, councilmenhave tended to be ban services, fixedand mobile,we found disproportionately Anglo" (Lineberry, considerable variationnthe former. i Parks 1977,pp. 54 57). This backgroundclearly and libraries to be distributed a tend in makes fora good testof whether the "un- fashionwhich might be characterized as derclass" hypothesesof service allocation "unpatternedinequalities." underThe are valid. classexplanation,owever-neither its h Using the census tract the sub-area, racial, or itspower, nor its socioecoas n Lineberry's study ascertained how the nomic components-did not offer much tracts differed n the independentvari- explanation.o the degreethatan explai T ablesused, and compared those differ- nationfortheseinequalities be found can ences with the differences found in the in the attributesf neighborhoods,he o t distribution ofone and anotherofthe ser- ecological traits seem most important. vices. "On the otherhand, the mobile service The levelofthe fireservices endered we investigated, r police patrol, ffered ess o l within a census tractwas measured by variation among neighborhoods.he city T ascertaining the average distanceof of San Antonio providesremarkably buildings from a firestation. "Counting equivalent allocations manpower from of thefire personnel and facilities available sector sector.his equivalencepersists to T in a particular neighborhood isof little use whether one usespopulation, rime rates, c becausefireunitscan readily 'doubleup' orcalls forservice a denominator.There as in caseofa fireor coverone anotherwhen issimplynot much inequalityheretobe t ( p.142). (Butcomone unit is busy."Data on "mean re- explained."Lineberry, o distribution in sponsetime tofires" as unavailable.he parethe analysis fpolice w T study therefore calculated, "forsixran- some economists' studies, elow). b dom pointsin each census tract,he dist Other researchthat supported"thebutancefrom thatpointtothe nearestfire reaucraticnterpretation urban service i of station," used the mean distance, d- patterns" and a iscited Koehler and Wrightby 2 mittedly very rough approximation "a of son("K-W') (1987). fireprotection." Correlationsere then w computed "between lineardistanceand (ii) MORE RECENT SURVEYS variousattributes census tracts, . . of . m "relying the higherthe correlation, closer the the In contrast,ore recentsurveys, station." on strongerresearchdesigns, . . pro. The "underclass" ypothesesof distri duced no evidencethatbureaucratsare h butionof the service, firedamage min- immune from politics orpoliticians, or that imization, not supported. Citizens n urban resourceallocations the prodwas " i are census tractsscoringhigh on median uctsof bureaucratic ecision d rulesalone"

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

114

NATIONAL

TAX JOURNAL

[Vol. XLII

(K-W, P.83).' now turntoa brief We de- racialor economicversionsof the "underclasshypothesis"verified,hishyt scriptionf fourof those studies.' o pothesis being that"thosewith social and (a)Boyleand Jacobs (New York City) economic advantages use the political The surveyby Boyleand Jacobs(1982) ("B- system toobtainmore than theirshareof J")of certain servicesn New York City publicservices"p.372). i ( yielded findings markedly different from (b)Sharp(Wichita, Kansas) thoseof Lineberry's and other earlier studies.he data were forthe fiscalear Ifdistribution service sguidedlargely T y ofa i 1969-70. B-J attributedhe differencesby relative evelsof requestsand comt l infindings somuch todiffering not char- plaints the variousdistricts, in a collatacteristics the cities tudiedas to dif- eralquestion, s noted above,is:do the of s a ferencesin conceptsand techniques. differences suchlevels in accordatallwith in or First,-J ascertainedifferences in- differences income levels othersoB d in put,thatis, expenditures er capita, ot cio-economic p n status? Sharp (1982,p. 109) differences output. in They considered ex- has exploredthisarea.She notesthe penditures preferableeasure,though widely conflicting the m findings previous of they didnot go deeplyintothe many con- surveys:positive ssociation, a negativeas"parabolic relationship" very (the siderations relevant thischoice(p. to 373). sociation, Second,they employed multipleregres- poor and the very richrequestor comsion analysis, omewhat in contrast s to plainlittle; middleclass, uch),no the m Lineberryand others, who forthe most relationship all. at Sharp'sown data from partused bivariate orrelation. c a random-sample citizen survey in WichThird,they posited a possible as expla- ita, Kansas in 1980 showed a positive er nationdifferences urban tax per capita lationship, in among the districts. (c)Bolotin CingranelB and (Boston) B-J'sfindings contradictedhe convent tional onclusion thatthedistribution Bolotin c " of and Cingranelli 1983), ( studying services sunpatterned" i (p.377). They as- the distribution police of expenditures er p certained capitaexpenditures each capitaamong 145 censustracts Boston, per in in of sixty-two Community Planning Dis- Massachusettsin 1971,concludedthatthe trictsf New York City, o foreach of seven methodology employed largelydeterservices: police, ire, f sanitation, pre-col- mines the findings. hey criticized T prelegeeducation, higher education,uman h vious studies service of distribution for to r resources (welfare) health. and They also having failed distinguishesidential distributed totalof these expendi- from business the districts 210), (p. ortotake turesover the sixty-two istricts, d using accountof differences inneed (p.210)as the percentageof CPD families with in- shown by differences crime rates(they in comes below the povertyline. regres- referred Ifa toViteretti, 1973). sioncoefficient forthisvariable was pos"For our purposes. . .the most imporitive,he service t was being allocatedn tant finding i isthat the exclusion f the o accordwith the "compensatoryhypothe- businesstracts esultsn a meaningfulrise r i sis." in the (negative) predictive valueof the B-J concludedthat the distributionsproportion ofblackson levelof police exwere explicabley the compensatoryhy- penditures.hisfindinglendssupportto b T pothesis or"human services ikewelfare, the racepreference f l element of the underhealth, elementaryand secondaryed- class and hypothesis . .after . controlling for ucation" (P.377),but,in contrast, "tax the effectsf need [crimerate]and eleco contributions capita" per emerged as the toral clout withthe mayor. . ."(p.218). major factor that determinesthe alloca- B-C alsocriticized previousstudiesfor tionamong districts the "property-re-having used "bivariate orrelations of c belatedservices," thatis, "police, and tween various fire, neighborhood attributes and sanitation" (P.377).In no casewere the some appropriate measure of serviceef-

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

No. 21

DISTRIBUTING A FREE GOVERNMENT

SERVICE

115

forts outcomes.. . ." compare among on To neighborhoods "with comparable needs for service.... the easiest way . . . is to employ multipleregression techniques, but thisrarelyhas been done" (p.212).Lineberry,for San Antonio, used multiple regression techniques, but "appropriate measures of need usuallywere not included among the explanatory variables in the models tested"(P.212). 1 In summary, B-C concludedthat 'It is premature to rejectLineberry'sunderclasshypothesis["some groups sufferbecause of theirrace,because of theirsocial status, because of theirpaucityof poor litical powei" (Lineberry, 1977, p.12)]because the internalvalidity most prein vious studieswhich testitissuspect."

first appears to be an increasein the allocation park services Chicago'smiof to norityneighborhoods between 1967 and 1977" (p.82). (2)Findingsof Economists The findingsof economists tend to support the more recent surveys by the political scientists, although,as noted,the economists have restricted hemselves to t police,hile the political w scientists have ranged over a wide varietyof urban services. (i)TREVINO
WESTENDAPP CITY)' (NEW YORK

The pioneeringempiricalstudy of police distribution thatby Trevifio was Westen(d)Koehler and Wrightson(Chicago) darp (1970),which remains unmatched, i In 1987 Koehler and Wrightson ("K-W")5 in some respects, n itsrigorand analytpublisheda re-surveyofpark facilities icalinsights.reviiio's in T findingsare not Chicago.These facilities been studied reproduced here, since they have been had by Mladenka in 1980, and his conclusions summarized recentlyelsewhere (Shoup, were "generallycitedas the single,ost 1988). m significantvidence that urban service e allocations are the product of bureau(ii) FABRIKANT (NEW YORK CITY) cratic ratherthan political decisions" (p. Fabrikant's analysis (1977), importantfor 81). understanding the spillover effect from K-W used the same data for1967 and precinctto precinctof allocation f the o 1977 as had Mladenka, togetherwith data policeforce, willnot be examined here, for1983.They concluded that Mladenka's sinceitdoes not concentrateon distriburesults"are equally attributableo ext tivestandards. ogenous variables[not consideredin his In passing,however, we note thatwhen study]which have been generallyovera policeman ismoved from PrecinctA to looked in research on [urban servicedePrecinctB the crime rate in B declines livery] . ."(P.80).They estimateda . because (i) there are now more policemen regressionmodel to testcompeting explathan before per potential criminal and (ii) nations,and found that "efficiency and some criminalsin B move to A because of politicsre independentlysignificantx- the police a e shift, so,in B, there are still planatoryfactors the distribution in of more policeper potential criminal. The park facilities, equityisnot"(p. while 80). crime rate in A rises for the reverseof The exogenous variableswere population these two reasons.Thus, under spillover, mobilityand the immobilityof certin a shiftof one patrolman to another prepark facilities (p.85)."Focusing considcinctismore effectiven reducing crime i erationon geographically stablepopulathan when there is no tionsubsets, and on facilities which are in that precinct but i I spillover, lesseffectiven reducingtosusceptible short-termredeployment, talcrime in the two precincts to 7 together. significant correlations are found between political variablesand the distri(iii)ENNErr K (NEW YORK CITY) butionof park facilities" (p.85).Therefore,"demographic changes and other Next to cultivatethe policefieldwas environmental factorsaccount for what Kennett (1982),who followedTrevifio's

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

116

NATIONAL

TAX JOURNAL

[Vol. XLII

basicstructure,ut had the advantage of b availability data,not forjustone preof cinct inNew York City, for71 (omitbut tingCentral Park) (p. Kennett, 84). toobtainthe necessarysimplification, assumed thatonly one type ofcrime,robbery, was relevant tothedistribution 82). de(p. He velopeda model of a police service productionfunction and ofthe supplyof offenses (p.83), usingthe clearance rate(see Craig,under Equal Outputs,above), the crime rateper capita, nd manpower disa tribution. Kennett concludedthat "the existing distribution manpower isnot oriented of to the minimization crime nor to the of equalisationf victimisation o ratesacross precincts, does itconforin equalinnor to put per head" (p.89). Nor, evidently, did itreflect desire a forequal work load among policemen.(Kennettwent on to demonstratethat thislaststandardfor distribution ofthepolice forcecouldwell leadtoa minin-dzation ofcrimeforthecity as a whole (pp. 90-93)). Kennett estimatedthat "a move to equalityfrisk(equal o number ofrobberiesperhead ineach precinct)ould lead w toa risein total robberies f24%-most o of the riseoccurring relatively in law abidingmiddle class districts police as are draftedtocutthe crimeratein themore crime-proneareas"(p.94).An inputequality distribution the police of force "would leadto a doublingofthe robbery total,ith the incidence histime inthe w t lessstable areas"(p. 94).

among areas isnot,usually, haphazard, withoutthoughtofgeneralrulestobe followed,though Linebe"y'sfindingsfor parks and librariesere somewhat unw settling ("unpattemedinequalities"). Rules that seem to have been followed,ore or m less, one time or placeor another, in includeequality (Lineberry), inequalityo t assist thepoor (B-C and B-J),taxespaid (B-J), complaints-requests (Sharp),acial r discrimination (B-Ci),nd efficiency a and "politics" (K-W) (seealsothereference to Treviiio estendarp). W Allthesefindings,owever,are inferh encesfrom statistical data.Directstatements by the allocators totherules as they tryto follow, why they select and those particularules, r are,with a few exceptions, at hand. This isthe big infornot mation gap.Researchprojects designedto fill t, i preferablyhroughjoint t efforts of political scientists eocnomists, and seem warranted.

IV. Indirect Payment for a Free Government Service The recipientf a freegovernment sero vicemay have to pay forit indirectly. Those who livein an apartment house locatednear an attractive citypark will probablybe charged a higher rent because of thisamenity.These tenantswill stillbtainthe benefits f the park,but o o not really freeofcharge.They willhave relinquishedhe money value of the bent efits, part or in whole.Thus the free in park benefits, money terms,go tothe in (iv)BEHRMAN AND CRAIG (BALTIMORE) nearby landownersratherthan the users o Behrman and Craig (1987)concluded(p. of the park.This relinquishment f the from the free 39)that, to "distribution as ofpolice and money valueof the benefits isthe analogueto a of safetyfrom crime among neighbor- government service o where the burden ofthe hoodsinBaltimore. . .the local govem- shiftingf taxes, ment does sacrifice some productivityn money payment restsnot upon the imi orderto achievea more equitable distri- mediate payorbut upon some customer or butionofservice outcomes,and . . .un- supplier. Land distant from the new park isnow equal concernispro- oor and pro-young, lessattractive, relativeo land near the t but racially neutral.1,v park,and land valuestheremay well declineaccordingly.enants in these disT (C) Do Service Allocatorspply Rules A tant apartments may ultimatelybenefit forDistribution? from a loweringofrents. termsofdisIn The evidence summarized above sug- tribution,here are two advantaged t geststhat allocationf urban services groups,the nearby landlords o and the dis-

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

No. 21

DISTRIBUTING

A FREE GOVERNMENT

SERVICE

117

tant tenants,and two disadvantaged residents,theythinkaboutitatall, ill if w groups,the taxpayerswho financedthe expecttouse the firehouse uch less m than creation nd maintenanceof thepark,and the park.On the otherhand,the valueof a the distant landlords. the use when itdoesoccurwillbe much Under thesecircumstances,he goalof greater t than thatof use of a park.The equal distribution park services amount offiredamage preventedby the of throughouta cityis a complex one, not fire fighting service willdepend in large readily definable. What therecan be is parton how farthe equipment has to go equaldistribution oftheavailability ofa when the time comes,and through what park foran indirectrice. p traffic conditions. Grosslyunequal treatThis indirectayment fora freegov- ment of neighborhoods, the sense of p in emment service matter thathas been grossly isa unequal responsetimes,probably ignoredsinceitwas analyzedsome twenty doesaffect relative entschargedand, for r yearsago (Shoup,1969,pp.89 93).Nei- the homeowner, land prices. he points T therLineberry,strom nor Rich dealwith raisedabove with respect enjoyment of O to it, and itisnot to be foundin public fi- parks therefore applyto the enjoyment of protection, though perhapsto a lesser nance texts. be sure,the effectf a free fire To o government servicein enhancing prop- degree.The homeowner, renteror busiertyvalueshas been noted.' the other ness firm pays,indirectly, But forthat free sideof thiscoinhas been neglected:rop- service,hough the amount so paid may p t ertyvalues areenhanced becausethe ser- bear little relationothe costofthe sert viceusers (tenants) forcedby the vice. are There are the same two advantaged market to pay forthe"free" good,e.g.,he groups and two disadvantagedgroups as t citypark.They make an indirectay- in the parks case. p ment foruse of the park.Ifthe service is Much the same conclusions applytothe financedby an increasein the property service garbage removal.The effectsf of o taxon thelandlords,nd they recoupthat discriminatory ervice a s among neighborincrease raising by rents, they are unaf- hoods is,however,much more evidentto fected. tenantsare required,s itwere, even the casualobserver. f, The a I forwhatever were to buy the service,hough of coursethey reason,freegarbage removal service t can avoid making thispurchase by mov- discontinued fora certainneighborhood, ingto areasdistant from a park. propertyvalues therewould surelydeWhen accountistaken both of the free cline somewhat. The residents, paying less service and of the propertytax thatfi- rent, couldcontractith private w firmsto nances it, advantaged and disadvan- remove the garbage,a costthat would the taged groups noted above tend to disap- more or lessoffset he money gain from t pear. the lower rents, and they would be more Neglectof the indirectayment fora or lesscloseto where they were before. p freegovernment service has been partly Unequal treatmentin the removal of due tothe factthatthe valueofthe free garbage freeof charge does not therep servicesnever stated i specifically terms forenecessarilyroduce unequal,or at in of money, per unit ofthe service received leastgreatlyunequal,net monetary orof the unitbought or sold(apartment burdens. The difference, thesetwo services, in in space, the example above),in contrast in must be recalled. to taxes,which are attachedto private the importanceofsiting market transactions incertain typesoftax, Once the firehousesave been built,he h t as a percentageof those transactionsr neighborhoodscan be fairly o sure thatthe i as so much per physical unittransferred. patternof inequalityn degreeof service, The indirect ayment isnever statedso ifany,willpersist ver some years. p o The specifically. degree ofgarbage disposal service, nconi c so A firehouse nearby probablyincreases trast,an be changed quickly, thereis thevalue ofland and buildings much as lessassurance, more fear, a mainor of patternor of a does a park,exceptthatthe degree of in- tenanceof the existing creasemay wellbe much less, sincemost change for the worse. Propertyvalues,

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

118

NATIONAL

TAX JOURNAL

[Vol. XLII

however, are probablynot much affected VI. Current Low Level of Interestin by thisincreasedrisk, definedas degree Area Distributionof Urban Services of dispersionf possible o outcomes. The low levelofinterest shown in area distribution ofurban services snotable. i As an illustration most recentMaythe V. Privatization or'sManagement Report forNew York City(September15,1988)isreviewedhere Privatization an urban government Iseealsothe sections f the presentpaper of o and service willaffect sub-areadistribu-on Privatization on the R. Hero surthe on tionof the service at all, greatly, vey of recentliterature thissubject). not or bedepending on the degreeand kind of pri- This New York City reportis cited causeof itsexcellencen allotheraspects i vatization. At one extreme,the serviceis com- ofreporting the city'services. on s Its655 pletely rivatized, government shed- pages relateto the operations the Citp the of dingallresponsibility. Thisisfeasible only y'sagenciesforthe year ended June 30, forthe few services that are person-spe- 1988,as wellas projecting service evels " l (Mayor's letter, 1) forthe p. cific, site-specific, as garbage col- and goals" or such i lection, thatexclusion priceisprac- coming year.Anyone interestedn being so by ticable. The horizontal quity question informedon the great number of decie a disappears intothe more generalone of sionsthat must be made in allocating income distribution. ofthe poorwill largecity'sudget among activities, Some b probe excludedby the pricingmechanism; grams, techniques output, of and the like some of the not-so-poor willexclude willbe rewardedby a study of thisvolthemselves,tospend on otherthings. The ume. The sheer sizeof New York City same remarks, in reverse, apply to stati- guaranteesthe area diversityhat raises t zation, where the government takes over the issuesdiscussed here: seven million inhabitants, millionschoolchildren, one a hitherto rivately upplied p s service. At the otherextreme,the government 1.77millionmotor vehicles enteringand merelyengagesprivate firmsto supplythe leavingManhattan everyday,26,138acres service pricespaid them by the gov- ofparklandwith2 million at parktrees (plus emment, and with distribution sub-area 700,000 street by trees), 32,171 structural in whatever patternthe government fires fiscal988 (an averageof 88 a in 1 specifies. degreeof privatization This re- day)and 91,890total fires (252a day), an moves none of the issuesthe cityfacesin averageof 658 arrests day by the Police a distribution the service. n example, Department-and soon. of A reapparently,sthe contracting-out fire Nowhere does thishighlydetailed i of service Scottsdale in and certainother portstate orcomment on the problem,how to distribute fixedtotal a ofinputfora ArizonacitiesHatry,1983,p.19). ( Not much has been publishedon the givenservice over thesub-areasofthecity. probableeffectsf privatization sub- Certainnew programs were saidto have o on area distribution services.n excep- been started thisor thatsub-area, of A in and tionisHatry (1983, Chapters2-12),who extended to certain otherswhen successevaluateselevendifferentegreesof pri- ful, no reasonsare givenforselecting d but vatization, includes,n the evalua- the initial (seepp.9,10,13 forpolice and i area tion,.potential on distribution effects ofthe and p.229 forsanitation). the 227 pages Of service. The conclusions necessarily oftabularmaterialin the report,nlyone are o brief, rather and conjectural, giventhelack page containsdata broken down by area of basicresearchin thisfield. Itisthe in- (fire, boroughs,foraverage response by between typesofprivatization thatneed time,p.85); one otherpage contains ratmore inquiry, determine the degree to ing data forgroupsof districts, with to but which the government can specify the no identificationthem (sanitation, of p. distributional patternthat the supplier 248). Two charts,p.232 and 234,are area p must follow. charts(sanitation districts), shadwith

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

No. 21

DISTRIBUTING

A FREE GOVERNMENT

SERVICE

119

ingsby level service,ut the texidoes of b Iam indebtedoCharles t E.McLure,Jr., toa and refereeorcommentson an earlier f draft,nd to a a not discuss implications. the Lorraine Eden, A lackofinterestnthedistribution number of colleagues, especially i of RobertL.Lineberry, but StephenL.Percy, StevenRhoads, an urban publicservice isshown alsoin and PaulShoup forassistance inlocatingherele t recentscholarly literature.10 There has vantpolitical contributions, Dieter science and Biehl on studiesnthis i been no treatise focusing thissubject forinformation Europeaneconomists' on sinceLineberry's 977 book.Hero's1988 field. are, 1 'Mere indeed,ome exceptions. respect s With surveyof the political science literature toprotection fromfire, Coulter tal. e (1976,.257), p on urban service delivery shows few ar- concentrating organizational on changestoimprove concludey noting b that"analternative ticlesn thisarea (Hero's i latest-dated ref- Performance, changes isimplicit ... inthe erencesare of 1984). survey ofsome of totheorganizational theleastroductive A findings. .The fact .. that p cities the European scholarlyiterature, l under- tendto havemany fire service organizational chartaken forthe presentpaperby a student acteristics thatonewouldexpect tobeassociated with issuggestive. Maybe a reallocation of Professor DieterBiehl's, with entries highproductivity and isrequired. examFor rangingin time from 1965 to 1987,indi- ofresources capabilities ple, decisionmakers lowproductivity in municipalicates thattherehas been little interestn ties i shouldconsider allocating a larger roportion p of thedistributive aspects ofurban govem- their existing suppression capability tothedensely populatedeighborhoods n withdeteriorated housing." ment servicesil Lineberry 1977, p. 151 152)describes ( p theconcept of"inerementalism": year-to-year the changesin a VII. Some Research Suggestions" department's budgetareusually nsmallincrements i notin largeamountstransA few suggestions to furtherre- and smalldecrements, as f t B pracsearchin thiscomplex field urban ser- ferredromonedepartment oanother.ut this of tice seemstoreflect the rules more ofthegame for vicedistribution as follows. are cont@n@ing departments,ot a consciouseekingto n s First,tshouldprovehelpful fpolitical maximizeoutput i i bychanging inputby increments and scientists economistswould work comparingtheconsequent and incremental changesin jointly nthisarea,meldingtheir i respec- output. 2 SeeLevy eta]. (1974) nd Mladenka(1980) nd, a a tivestrengths.he former have, forin- forabsenceofcontrolariables, T v Antunesand Plum stance, shown expertisein a realistic lee(1977)nd Mladenkaand Hill a (1977). searchformotives and standardsthat, 'Stephen.Percy(1989) oncludes "Urbanpo L c that scientists haveyettoclearly verifynd dema even though they are not generallyac- litical t decision on rules ceptable,ay nevertheless o fartoward onstrateheimpactofbureaucratic m g serviceelivery." d Itisverydifficult "tospecifyow h explaining current distributionof the ur- bureaucratic decision operate oastoinfluence rules s ban services an example isthe "under- organization ( behaviorshich,inturn. w affect service distributions." class hypothesis"). Economistshave made 'Fora summary ofNardulli nd a use ofmarginal analysis ways thatare Stonecash 1981) ofthefindings in ( and Browning, Marshall Tabb and not emphasized by the othergroup.Sev- (1984)seeK-W, pp.83-84. eralothersuch joinable strengths could 'Thefirst sixpagesofK-W present informative an historical surveyofthepolitical literature science on be cited. ofurbanservices 80 85). (pp. Second,beforefurther field work isun- distribution 'Forotherreferences towritingsy economists b on dertaken,conceptualproblemsneed tobe policerior o 1982, p t seeKennett(1982, .95). p further explored,articularly defini- ?A refereeakes the interesting p the m suggestion that ofa w i tions ofinput, output, outcome,equality, addition policeman illalsoresultn a larger proportion crimes of beingreported. Thisidea, foland the like. lowedthrough,mplieshatan addedpoliceman ends i t t Third,the choice made among these bothtoincrease (reported) crimeand todecreaset, i conceptsneeds to be fullyexplainedand thenetoutcomepossibly goingeitheray.Maybe this w ispartoftheexplanation forthepuzzling ina result supported. by thePoliceoundation F inKansasCity, isM Fourth,more use might be made of the studythatallocation ofpoliceas no effect h atallon interviewing technique, learnwhat souri, to crimedistribution, asreportedyLineberry 1977, . b ( p rulestheallocatinggentsareusing or at 138)(fromShoup,1988, a note1). 'TheB C analysis some admitted imitations. has l least profess be using. to Each crimeiscountedasone,regardless nature, ofits and onlypersonalrimesare included, c thatis, com ENDNOTES mercial crimesarenotcovered (BC, p.43;Craig, . p M**l am pleased toacknowledge T d crimes thesupport ofthe 344).he problemofweightingifferent toget Orris eekFund inthepreparation B ofthis paper. a crimetotal sanalyzed n Anderson(1974,p.4 i i p

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

120

NATIONAL

TAX JOURNAL

[Vol. XLII

ments."American Economic Review, December, 1972,891-901. Borcherding, Thomas E.,Winston C. Bush, and Rob ertM. Spann. "The Effects PublicSpending on of the DivisibilityPublicOutputsin Consumption, of Bureaucratic ower,and the Sizeofthe Tax Shar P ing Group,"in Borcherding(ed.), Budgets and Bu reaucrats:The Sources of Government Growth. Durham, N.C.,Duke University Press, 1977,211228. Boyle, John and David Jacobs. "The IntracityistriD butionof Services: Multivariate A Analysis" American Political Science Review,June,1982,76, 371-379. Bradford, .,Malt,R.,and Oates,W. "The RisingCosts D of Local PublicServices:ome Evidenceand ReS flections." National Journal. olume XXII,No. Tax V 2,June 1969,185-202. Cingranelli, David."Race,Politics, Elites: and Testing Alternativessumptionsof MunicipalService A Distributions." American Journal of Political Sct ence, 1981,25,664-692. Coulter, hilip P B.,MaeGillivray,ois, L and Vickery, WilliamEdward. "MunicipalFire ProtectionerP forinancen Urban Areas:Environmentaland Ori ganizationalnfluences n Effectiveness ProI o and ductivity Measures"inOstrom, ed., TheDeliveryf o Urban Services: OutcomesofChange.Beverly Hills, CA, and London,U.K Sage Publications, 1976,231260. Craig,StevenG. "The Impact ofCongestionon Local PublicGood Production."ournal of PublicEcoJ nomics,Volume 32,No. 3, April, 1987,331 353. REFERENCES Dosser,Douglas."Notes on Carl S. Shoup's'Standards forDistributing Free GovernmentalSera Abney, Glenn,and Thomas P. Lauth. "A Compara vice: Crime Prevention."' PublicFinance.Volume tiveAnalysisof Distributional Enforcement and XIX, No. 4, 1964,395-402. Decisionsn Cities." i Journal ofPolitics, February, Eden, Lorraine, nd Melville a McMillan."Local Public 1982,44, 193-200. Goods: Shoup Revisited,"n LorraineEden, ed., i Anderson,RobertW. "Towards a Cost-BenefitnalA Retrospectives Public Finance.Durham, N.C., in ysis of PoliceActivity."ublicFinance.Volume P Duke University Press,1989. XXIX, No. 1,1974,1 18. Fabrikant, Richard."A Long Overdue Comment on Antunes,George,and John J.Plumlee."The Distri- Shoup's'Standards forDistributing FreeGoverna butionof an Urban PublicService: thnicity,o@ E S mental Service:Crime Prevention."' ublic FiP cioeconomicStatusand Bureaucracyas Determi nance, Volume XXXH, No. 1,1977,111-118. " nants of the Quality of NeighborhoodStreets." Gramlich,Edward M. and DanielL. Rubinfeld.MiUrban Affairs Quarterly,977,12,313-332. 1 cro Estimatesof PublicSpending Demand FuncBehrman, Jere R.,and Craig, StevenG. "The Distri tionsand Testsof the Tieboutand Median-Voter butionof PublicServices: Exploration Local An of Hypotheses." Journal of Political Economy, 1982, Governmental Preferences." American Economic 536-560. Review Volume 77,No. 1,March, 1987,37 49. Hatry,Harry P. A Review ofPrivateApproachesfor Bennett,R. J.Geography of PublicFinance:Welfare Deliveryof PublicServices.ashington,DC. UrW underFiscal Federalismand LocalGovernmentFi ban Institute Press.1983. inance. Chapman & Hall,1980. Hero, Rodney. "The Urban Service DeliveryLiteraB sh,Robert L. "Fiscal Equalizationhrough Court T ture:Some Questionsand Considerations." Polity, Decisions: PoliepMaking WithoutEvidence"inOs1988,19,659 677. trom, ed.,The Deliveryof Urban Services: ut- House, Richard.A Review of theFieldof Incidence O comes of Change. BeverlyHills, CA. and London, Studies. h.D. dissertation, P University EastAnof U.K. Sage Publications, 1976,75 102. glia, Norwich,NR4, 7TJ, England.1985 (unpubBloch,PeterB. Equality fDistribution o ofPolice erS lished). Ch.6:"Allocating theBenefits romPublic f vices: Case Studyof Washington, A D.C., WashingExpenditure." ton,D.C.:Urban Institute, 1974. Jones,Bryan.Service Delivery intheCity: Citizene D Bolotin, Frederic and David L. Cingranelli. W. "Eqrnand and Bureaucratic Rule.L-ongmans, 1980. uityand Urban Policy: he UnderclassHypothesis Jones,Bryan D. "Assessingthe Productsof Govern T Revisited." Journalof Politics, 1983,45,209 219. ment" in Rich,ed., Analyzing Urban-Service isD Borcherding,homas E.,and RobertT. Deacon."The T tributwns. Lexington, Mass.,and Toronto,Canada, Demand forServicesof Non-Federal Govern 1982,pages 155-169.

14), Shoup (1988, and p.7) 9See, e.g.,usgrave and MuqgTave (1984),p.472M p 473,and Bish (1976), ho concludes w that"Capitalization f the valueofgovernment programs or costs o intoproperty valuesgreatly reducesthe capacity of the states and the national government to redistrib ute income throughprograms targeted smallgeoto graphicareas"(p. 96;seealsopp 86 90 and 97). Emphasis in Bish,however, is on reductions costs in (taxes)ore than on increasesnlevel m i ofservices. But the latter s clearly ecognized: . .property i r ". with good publicservicesis worth more than property withoutgoodservices . ."(p.86). 10FOr studies "theincome distributional on effects of government outlays" a broad sense, in including, e.g.,ffects the incomesofthosehiredby or selling e on to govermnenta,and referringhiefly national r c to o state-provincial expenditures, seede Wulf (1981). De Wulf does not coverexplicitly distribution an the of urban service among a city'sub-areas. s "This list ofEuropean referencessavailablepon i u requesttothe authorat RR 1,Box 303,CenterSandwich,NH 03227. 12 1 am indebted to a refereeforthe information that "thegrowing literature the 'geography public on of finance'-now. . .a recognizedubdiscipline ges in ography . . ."may have somethingtoadd on this subjectfareadistribution o ofurbangoverranenter s vices; ennett(1980)and thejournal, nvironment B E and Planning C: Government and Policy, cited. are

National Tax Journal, Vol. 42, no. 2, (June, 1989), pp. 103-21

No. 2]

DISTRIBUTING

A FREE GOVERNMENT

SERVICE

121

Kennett, David. "Standards and Proceduresfor the Distribution ofa PublicService: Shoup Revisited." Public Finance, Volume XXXVII, No. 1, 1982, 8097. Koch, Edward 1.(Mayor,City of New York),Brezenoff, tanleyand Gunn, Barbara. The Mayor's ManS agement Report [forfiscal year ending June 30, 19881.New York. Citybooks.Municipal Building, One Centre St., New York, NY 10007,$17.50.Or: Officeofthe Mayor, Office ofOperations, Church 100 St,20th Floor,New York, NY 10007.Sept. 15,1988. Koehler,David H.,and Margaret T. Wrightson. "Inequalityin the Deliveryof Urban Services: ReA consideration the Chicago Parks." Journal of of Politics, February, 1987, 49, 80-99. Lineberry,RobertL.Equalityand Urban Policy: The Distribution Municipal Public Services.everly of B Hills,a. and London, UK.: Sage Publications,977. C 1 Lovrich, Nicholas P. Jr. "DifferingPriorities an in Urban Electorate."ocial ScienceQuarterly,1974, S 55, 704-717. Mladenka, Kenneth. "OrganizationalRules,Service Equality, and Distributionn Urban Politics." i So cialScienceQuarterly, 1978, 59, 192 201. Mladenka, Kenneth R., and Kim Quaile Hill."The Distributionf Urban PoliceServices."ournal of o J Politics, February,1978,40, 112 133. Musgrave, Richard A. and Peggy B. Musgrave. Public FirwnceinTheoryand Pructice. FourthEdition. ew N York: McGraw-Hill, 1984. Nardulli,PeterF.,and JeffreyM. Stonecash.Politics, Professionalism, nd Urban Services:The Police. a Cambridge, Mass.: Oelgeschlager, Gunn and Hain, 1981. Nivola, PietroS. "Distributing Municipal Service: a A Case Study of Housing Inspection."ournal of J Politics,978,40, 59-81. 1 Ostrom. Elinor, ed.The Deliveryof Urban Services: Outcomes ol'Change. BeverlyHills, A: Sage,1976. C Parks,Roger B. "PolicePatrolin MetropolitanAreas Implications forRestructuringthe Police" Osin trom, ed.,The Deliveryof Urban Services: Outcomes of Change. Beverly Hills, CA. and London, U.K. Sage Publications, 976,261 283. 1 Percy,Stephen L. "The Impact of Demand Processing Systems on the Productionand Distribution Urof ban Services," especially pp. 13-16. Paper presented at Midwest Political Science Association, 1985, mimeograph. Percy,Stephen L.LettertoC Shoup, March 16, 1989 Percy, Stephen L. and Eric J Scott.Demand Processingand Performance in Public ServiceAgen-

cies. University, Alabama: Universityof Alabama Press, 1985. Rich,Richard C. "Problems ofTheory and Method in the Study of Urban ServiceDistributions"n Rich, i ed.,Analyzing Urban-Service Distributions. Lexington,MA, and Toronto,Canada, 1982, 3 18. Rosentraub,Mark S. and Thompson, Lyke. "Bureau cratic Discretion and Surveys of CitizenSatisfaction"in Rich,ed., Analyzing Urban-ServiceDistri butions. Lexington,MA, and Toronto,Canada, 1982, 171-184. Sharp, Elaine B. "Citizen-Initiated Contacting of Government Officials and Socioeconomic Status: Determining the Relationshipand Accounting for It."merican Political A ScienceReview,1982,76, 109 115. Shoup, Carl S. "Standardsfor Distributing Free a Government Service: Crime Prevention."PublicFinance. Volume XIX, No. 4, 1964,383 394. Shoup, Carl S.Public Finance. Chicago,IL. Aldine, 1969. Shoup, Carl S. "Distribucion e los Beneficios los d de Servicios Estatales: Equidad Horizontal." Hacienda PublicaEspaiwia. Numero 100, 1986, 115-128. Shoup, Carl S. "Distributionf Benefitsfrom Govo ernment Services: HorizontalEquity."Public Finance.Volume XXXXIII, No. 1,1988, 1-18. Thurow, LesterC. "Equity versusEfficiency Law in Enforcement."PublicPolicy, 1970,28, 451 462. Thurow, Lester C. and Rappaport, Carl. "Law Enforcement and Cost-BenefitAnalysis."Public Finance,Volume XXIV, No. 1,1969, 48 68. Treviflo estendarp, Raul.On the DistributionfFree W o Government Services (Police Protection inNew York City). New York City.Privately printed, une, 1970. J Ph.D dissertation, Columbia University,New York City Photocopies available a price; at inquireof University Microfilms International, P.O.Box 1764, Ann Arbor, Michigan 48106, U.S.A. Viteretti,oseph P. Police,Politics J and Pluralism in New York City: Comparative Case Study.Beverly A Hills, CA: Sage, 1973. Weicher,John A. "The Allocation ofPoliceProtection by Income Class,"Urban Studies, October,1971,8, 207-220. De Wulf, Luc. "FiscalIncidenceStudies in DevelopingCountries: Survey and Critique."nternational I Monetary Fund, Departmental Memorandum 74/24. February 15, 1974. Mimeograph. De Wulf, Luc.'Incidence Budgeting Outlays:Where of Do We Go from Here?" Public Finance, 1981, 36, 55-76.