POLITICAL RENTS, PROMOTION INCENTIVES, AND SUPPORT FOR A NON-DEMOCRATIC REGIME

Transcription

1 ECONOMIC GROWTH CENTER YALE UNIVERSITY P.O. Box New Hven, CT CENTER DISCUSSION PAPER NO. 882 POLITICAL RENTS, PROMOTION INCENTIVES, AND SUPPORT FOR A NON-DEMOCRATIC REGIME Vlery Lzrev Yle University nd University of Houston Mrch 2004 Notes: Center Discussion Ppers re preliminry mterils circulted to stimulte discussions nd criticl comments. I would like to thnk Aimee Chin, Steven Crig, Pul Gregory, Timothy Guinnne, Mrk Hrrison, nd Gvin Wright for helpful comments nd suggestions. This pper cn be downloded without chrge from the Socil Science Reserch Network electronic librry t: An index to ppers in the Economic Growth Center Discussion Pper Series is locted t:

2 POLITICAL RENTS, PROMOTION INCENTIVES, AND SUPPORT FOR A NON-DEMOCRATIC REGIME Vlery Lzrev Deprtment of Economics, Yle University New Hven, CT Vlery.Lzrev@yle.edu Abstrct This pper nlyzes the economic foundtions of non-democrtic politicl regime, where the ruling bureucrcy cptures rents through collective control over stte property nd job ssignment. The model developed here yields the equilibrium in the politicl lbor mrket, where the ruling bureucrcy buys services nd politicl support of ctivists recruited from the working popultion. The underlying implicit contrct requires tht the incumbent bureucrts retire fter certin time to llow for deferred promotion of ctivists into rent-pying positions. The mjor implictions re tht the stbility of non-democrtic regime is consistent with high-income gp between the rulers nd the rest of the popultion, strengthened when government pursues n ctive investment policy, nd is not directly ffected by public goods provision or the rte of economic growth. The results of econometric nlysis of pnel dt from former Soviet sttes for the period of confirm the predictions of the model. Keywords: non-democrtic regimes, bureucrcy, hierrchy, politicl support, promotion incentives, implicit contrct, Soviet Union JEL codes: D72, J45, N44, P30

3 For millenni, utocrcy ws the prevlent form of politicl orgniztion. Despite the dvncement of democrcy during the lst two centuries, the mjority of the world s popultion still lives under regimes tht rnge from outright dicttorship to oligrchy, embellished with nominlly democrtic institutions. Moreover, during the twentieth century, number of industril nd mny developing countries reverted to uthoritrin regimes fter periods of significnt democrtic development. Resilience of non-democrtic regimes, recurrence of dicttorship, nd conditions tht enble trnsition to democrcy re therefore reserch problems tht need to be ddressed by economists. Economic nlysis of non-democrcy nd politicl trnsitions is reltively new nd growing re tht ws probbly prompted by the tide of democrtic trnsitions in the developing world nd former socilist countries in the lte 1980s erly 1990s. Theoreticl models developed thus fr cn be divided into two groups. The first group extends Levithn interprettion of government (Geoffrey Brennn nd Jmes Buchnn, 1980) into the re of non-democrtic regimes by ssuming the rulers who use their power to mximize net revenue rised by txing the dominted popultion (Mncur Olson, 1993; Mrtin McGuire nd Olson, 1996; Boz Moselle nd Benjmin Polk, 2001) or to protect their welth from being redistributed (Dron Acemoglu nd Jmes Robinson; 2000, 2001). The second group ssumes tht politicl power is vluble per se nd the rulers objective is to mximize the extent of power or probbility of remining in power (Herschel Grossmn nd Suk Je Noh, 1994; Ronld Wintrobe, 1998; Yi Feng nd Pul Zk, 1999). All mentioned works differ in the selection of relevnt vribles nd predictions of their models, sometimes dimetriclly. Acemoglu nd Robinson (2000), (2001) rgue tht high inequlity is the mjor driving force of democrtiztion, supporting this sttement by the 2

4 fct tht the extensions of frnchise in West Europen countries coincided with periods of the historicl peks of inequlity. In Feng nd Zk (1999), democrtic trnsition is likely to occur when reltively welthy middle clss hs high enough demnd for civil liberties (which is sort of luxury good) nd cn fford destructive uprising. This pproch implies tht the stbility of uthoritrinism is consistent with high inequlity. Another impliction is tht economic growth undermines non-democrtic regime. At the sme time, McGuire nd Olson (1996) shows tht secure rtionl dicttorship promotes economic growth. Wintrobe (1998) lso suggests possibility of positive reltionship between economic growth nd the politicl power of dicttor. The ltter work, however, focuses on the loylty of the popultion bought by the dicttor in exchnge for public goods nd on repression s the mjor determinnts of regime stbility. A similr pproch is dopted in Dmitriy Gershenson nd Herschel Grossmn (2001), where the loylty of the prt of the popultion, coopted into the rnks of the ruling elite, is substitute for repression. Mrk Hrrison (2002) explins the collpse of the Soviet regime by the incresing costs of coercion nd decresing reputtion of the rulers. The three ltter works imply tht stble non-democrtic regime is chrcterized by high levels of public expenditure (which includes but is not limited to police expenditure). Econometric studies of the determinnts of the stbility of non-democrtic regimes nd democrtic trnsitions re lmost non-existent. One exception is Feng nd Zk (1999), which finds, in greement with its theoreticl predictions, positive effects of low inequlity nd high levels of eduction nd economic development on the probbility of democrtic trnsitions in developing countries. Another cross-country study, Robert Brro (1999), yields generlly similr results with respect to the propensity for democrcy, which is mesured 3

5 by subjective index of electorl rights. Finlly, Schnytzer nd Sustersic (1998) studies the prty membership in the republics of former Yugoslvi, which they view s the mesure of loylty, nd find some evidence consistent with the theory of Wintrobe (1998). Two mjor limittions re chrcteristic to some extent of the literture reviewed bove. First, most of these works ssume explicitly or implicitly tht the non-democrtic regime is imposed upon n economy bsed on mrket nd privte property. Correspondingly, two of the three empiricl studies mentioned bove exclude socilist countries. Second, they consider the ruler ( dicttor or ruling elite) s sort of socil singulrity, n entity tht is completely isolted from the rest of the society nd hs no internl structure. Only Gershenson nd Grossmn (2001) nd Moselle nd Polk (2001), llow for connection between the ruling strtum nd the generl popultion. 1 However the mjor tendency of lrge prt of the twentieth century ws the rise of the regimes tht gve more power to the stte thn ever before in modern history nd either destroyed the institution of privte property completely or mrginlized independent entrepreneurship. These regimes, often designted s totlitrin, spred throughout the world in dimetriclly different ideologicl pckging: rcist/fscist (Germny, Itly, Spin) communist (USSR, Chin, Cub, nd mny others), ntionlist (e.g. Indonesi, Irq), or theocrtic (Irn). They ll shre number of common fetures tht provide strk contrst to the principles of free-mrket democrcy. The most importnt fetures include: hierrchicl politicl orgniztion, open for entry from below; bureucrtic control over the economy, 1 Moselle nd Polk (2001) ddresses primrily the issues of nrchy nd primitive predtory stte nd focuses on n individul s choice to become bndit, member of ruling clique. 4

6 often bcked by direct stte ownership of the ntion s cpitl stock; nd significnt populr support rised through the chnnels of the ruling prty or other similr institutions. Most of these regimes hve been forcefully removed s result of militry defet (from Nzi Germny in 1945 to Irq in 2003) or overthrown by populr revolutions (Estern Europe in ). The remining regimes of this kind, such s Chin, North Kore, or Irn, re still notble plyers in the world scenes nd the direction of their chnge is n open question. The first nd the most lsting regime in this row, the Soviet Union, deserves specil ttention becuse of extreme fetures of its politicl-economic orgniztion, its influence on the world politicl development in s, nd its rpid nd peceful demise in 1991 tht occurred despite its seemingly uncompromised coercive power nd cused little turnover in the higher tiers of economic mngement nd government. 2 The Soviet experience rises questions bout the sources of stbility of non-democrtic regimes nd the limits of their sustinbility, s well s the potentil for endogenous institutionl chnge. In prticulr, the end of the Soviet regime suggests possibility of the ruling bureucrcy inititing the chnge. The explntion of these developments cn be sought for in the politicl-economic exchnge between the rulers nd the popultion. Non-democrtic regimes, both totlitrin nd trditionl dicttorships, re conventionlly viewed s bsed on coercion. Voluntry support for the government is often dded s prtil substitute to coercion. Support is costly: the rulers buy it with public goods or trnsfers. If the rulers vlue only power per se s in Grossmn nd Noh (1994) nd Wintrobe (1998), then they cn be expected to evolve in the long run into welfre-mximizing benevolent dicttors. If the rulers ttempt to protect their 2 Reltively high stbility in the rnks of economic mngers nd bureucrcy hs been lso estblished for most Est Europen ntions (Best nd Becker, 1997). 5

7 welth from forceful redistribution (Acemoglu nd Robinson, 2000) or destruction (Feng nd Zk, 1999) by revolutionries, then they hve to give up its politicl power. Both pproches cpture certin fetures of non-democrtic regimes but cnnot explin the durbility of numerous predtory regimes tht extrct rents from the popultion nd re chrcterized by high inequlity. The ssumption of unchllenged power of the dicttor in McGuire nd Olson (1996) simply bypsses the problem of the regime stbility by leving no ctive role to the popultion. For rtionl, net-revenue mximizing ruler, who only uses politicl power s n instrument of revenue-collection, public expenditure is simply tx refund tht mkes the popultion better off, while reducing politicl rents. 3 Associted trnsction costs will normlly mke the net welfre effect strictly negtive. If we follow consistently the positive perspective nd llow for economic rtionlity of ll involved gents, then we hve to look for less costly, in terms of power nd rents, mechnisms of rising support. Hierrchicl politicl orgniztion cretes one such mechnism tht llows for borrowing services nd politicl support of ctivists in exchnge for deferred promotion into of the ruling strtum. In this pper, I develop model of hierrchicl regime tht is bsed on n implicit promotion contrct. Stbility of this type of regime is consistent with lrge income gps between the rulers nd the rest of the popultion, nd ctive government investment in physicl cpitl, while the rte of economic growth nd provision of public goods hve no direct effect on the support for the regime. 3 Defense nd criminl lw enforcement seem to be importnt exceptions. However, these cn be considered s prt of the costs ssocited with revenue production, insofr s they protect the tx bse from rivl predtors (foreign ggressors nd domestic bndits). 6

8 The pper is orgnized follows. Section I introduces the notions of politicl lbor mrket nd promotion contrct in hierrchies s mechnism of politicl-economic exchnge between the ruling bureucrcy nd the popultion. A forml model of n implicit contrct between the rent-mximizing bureucrcy nd the creer-seeking ctivists is developed in Section II. This section lso nlyzes the chrcteristics of the equilibrium in the politicl lbor mrket nd estblishes the limits of regime s sustinbility. Section III discusses the institutionl frmework of the Soviet politicl lbor mrket nd describes the dt tht cn be used to estimte the effects of vrious economic prmeters on the ctivist recruitment. The test of the proposed model using the pnel dt from former Soviet republics is presented in Section IV. Section V concludes. I. Politicl-economic exchnge in hierrchy The most importnt feture of mny, if not most, non-democrtic regimes, overlooked by the existing literture, is tht the rulers re not single dicttors or isolted cliques but rther top segments of complex hierrchies. There is constnt turnover within the ruling strt. A medin dicttor cn hrdly keep hold on power for more thn decde, lthough fmous exmples of longevity, such s Joseph Stlin in the USSR or Joseph Tito in Yugoslvi, might hve creted misperceptions on tht prt. A non-democrtic regime, dicttoril or oligrchic, is politicl nd often lso n economic monopoly, but prdoxiclly it cretes no impssble brriers to entry to the ruling strtum on the personl level. This feture cretes the possibility to rise support in wy tht is consistent with the rtionlity of politicl- 7

9 economic gents: by dmission of volunteers to the lowest rnk of the ruling hierrchy with possibility of subsequent promotion to higher-pid positions. The support obtined in exchnge for the promise of deferred promotion brings bout no losses (trnsfers to the popultion). To the contrry, it cn increse politicl rents: by recruiting gents from the popultion, incumbent rulers provide n incentive for spirnts to the rent-pying positions to volunteer extr effort, for exmple, in the form of supervisory services tht elicit more lbor effort from the working popultion. In other words, the rulers cn exchnge obligtions for support in the ntionl politicl lbor mrket. For the purpose of trctbility, politicl hierrchy cn be thought of s composed of two lyers. The first lyer, bosses, consists of the holders of governing positions (prty bureucrcy proper). The second, lrger one, ctivists, consists of the spirnts to these positions. 4 The bosses enjoy slries nd benefits well in excess of the ntionl verge, while the ctivists retin ordinry jobs. The ctivists re required to py dues, work more, nd render dditionl services, in prticulr, supervisory services. The incumbent elite the bosses enter the implicit contrct with ctivists, ccording to which the ltter py dues in extr lbor nd supervisory services (s well s in money, e.g. prty membership fees) in exchnge for the prospect of promotion. As long s the ctivists re content with the terms of the contrct they lso provide politicl support to the incumbent rulers. In the short run, the costs of this rrngement re borne by the ctivists. Forml dmittnce to the prty does not 4 Forml ttributes of the two groups vry cross politicl systems nd cn chnge over time. For exmple, in the erly period of the Soviet regime, the distinction between the cndidtes new recruits on probtion nd the full prty members drew the forml line. As the prty ws growing, the clss of ctul ctivists cme to include prt of rnk-nd-file full prty members. However, the reltive positions of the two groups remined essentilly the sme. 8

10 mke them better off immeditely. Their positions yield lower utility thn tht of n ordinry worker. Moreover, the probbility of being promoted in the future my be quite low, becuse the demnd for supervisory gents typiclly exceeds the number of rent-pying positions. However, high inequlity, which is the esier to mintin the poorer the country, mkes pursuit of bureucrtic creer gme with high stkes. 5 The bureucrcy bers its shre of costs, too. First, it hs to protect its rents ginst potentil rivls nd overcome resistnce of the working popultion to redistribution of ntionl income. This requires permnent coercion of some sort tht is costly. Second, the incumbent hs to repy debt to those ctivists, who hve fithfully performed their duty, by promoting them into boss positions. Unless sustined economic growth cretes sufficient number of new positions to stisfy the ctivists demnd, the incumbent bosses hve to repy the debt by retiring. As long s the bureucrcy is collective proprietor, bureucrt s benefits re lrgely ex officio. The possibilities to ccumulte personl welth re nrrow nd therefore post-retirement pyments re negligible in comprison to bureucrtic rents. Therefore, retirement is gloomy prospect for bureucrt, especilly if he internlizes the utility of his offspring. Incentives for incumbent bureucrts to enter the contrct re determined by returns to the ctivists service nd the cost of these services the rents forgone due to erly 5 Wht Soviet propgndistic literture used to sy bout worker joining the prty cn be interpreted s: more effort, dditionl duties, promotion to position in the prty bureucrcy or industril mngement in the future. For exmple, Soviet sociologicl review of 1970s purports: Once you become communist, you ssume voluntrily n dditionl hevy duty to led the others. A chrcteristic creer pth of new working clss prty recruit is described in the following mnner: foremn student in n engineering school hed of the plnning deprtment in lrge enterprise. The next step would be further up the ldder of industril mngement or to n entry-level position in the prty bureucrcy. Eighty percent of prty bureucrts of tht period followed this creer pth 9

11 retirement in complince with the contrct. Without the contrct the bosses expect to sty in office longer, possibly indefinitely long (if they re hereditry utocrts or privte proprietors), but the dditionl rents produced by the ctivists services re lost. The outside option of competitive regime bsed on privte property rights lwys exist potentilly. Under certin conditions in the politicl lbor mrket, it my become more profitble thn the hierrchicl regime, prompting the rtionl bosses to initite regime chnge. The effectiveness of the loyl-service-for-promotion exchnge depends on the extent to which bureucrcy is cpble of controlling the sources of income nd, therefore, the pths of upwrd job mobility. Communist sttes of the twentieth century, by estblishing ner monopoly on the ownership of productive cpitl, creted the most fvorble conditions for such control. In the Soviet Union, nomenkltur system of job ssignment, 6 run by the ruling prty, provided n institutionl mechnism for wrding promotion tickets in exchnge for loyl service. Discussion in this pper focuses on the economic incentives for both bureucrts nd workers-ctivists nd the determinnts of demnd nd supply of ctivist services under nomenkltur-type rrngements drwing mostly from the Soviet historicl experience. The Soviet Union is representtive for lrge clss of politicl-economic structures tht rely upon the nomenkltur-type exchnge between the incumbent rulers nd (Rbochii; pp ). 6 M.S.Voslenskii (1984) populrized the use of term nomenkltur s synonym to the Soviet ruling bureucrcy itself. The nrrower mening of ppointment control is dopted in this pper. 10

12 the promotion-seeking ctivists. 7 Although some of the results my not be pplicble to ny non-democrtic regime, studying the Soviet Union hs two mjor dvntges. First, the dominnt stte ownership nd fusion of politicl nd economic dministrtion produces single hierrchy. The policies with regrd to wges, promotion, job ssignments, etc. re sme or similr in the spheres of dministrtion nd production, cross industries nd regions. Therefore, the Soviet-type politicl-economic system on the whole cn be regrded s one enormous corportion. This simplifies the nlysis nd llows pplying the methods developed for the study of provision of incentives in firms. 8 Second, the high degree of centrlized bureucrtic control resulted in ccumultion of dt by governmentl sttisticl offices. Such ccumultion cn hrdly be expected under looser regime, in prticulr in countries where bureucrtic rents come lrgely from corruption nd/or where promotion is bsed on fmily ties nd clientelism. Moreover, relevnt Soviet dt re becoming incresingly vilble. Some of them re used lter in this pper to test the predictions of the model. 7 Stte property nd commnd economy re not necessry prerequisites for nomenkltur-type system. Any governmentl intervention in the economy ssigns resource lloction power to bureucrcy, nd therefore llows cpturing politicl rents through corruption (Andrei Shleifer nd Robert Vishny, 1993). Indonesi is n often-cited exmple. In generl, ll uthoritrin regimes estblish some sort of system of bureucrtic control over ccess to high-income positions beyond the borders of the public sector (licensing of businesses nd regultion of ccess to higher eduction). The militry-bureucrtic pyrmids of Qing Chin nd Imperil Russi in eighteenth century, medievl Ctholic theocrcies nd monstic orders re just few of numerous historicl nlogs to the nomenkltur. Militry dicttorships, supported by the hierrchy of rmy commnd, re lso hierrchicl politicl regimes. 8 See Cnice Prendergst (1999) nd Irene Vlsecchi (2000) for recent surveys of the literture on promotion-relted incentives. 11

13 II. The model 2.1. Bsic ssumptions Let us consider popultion with the size normlized to unity tht consists of two groups: the bosses nd the workers. The bosses incomes re essentilly politicl (bureucrtic) rents. Individul rents re identicl nd equl R. The bosses retire fter T b yers in office. The rest of the popultion, the workers nd the retired bosses, ern uniform wge, W, which is substntilly lower thn the politicl rent: W <<R. A prt of the working popultion, the ctivists, provide dditionl services tht benefit the bosses by incresing their rents: R = f(n ) > R 0 > 0, where f(n ) is the ctivists rent-production function such tht f (N ) 0 nd f (N ) 0; nd R 0 is the bureucrts potentil per cpit rent in the bsence of ctivists. 9 R 0 nd W re positive nd ssumed exogenous. The bosses nd the ctivists comprise smll frctions of the totl popultion: N b << 1, 10 N << 1. No specific ssumptions re mde with respect to the rtio of the numbers of bosses nd ctivists. Activist service requires extr effort on the prt of the workers who choose to enlist s ctivists. No immedite compenstion is offered for the service. Ech ctivist volunteers unit of extr effort. However, the workers re heterogeneous in their idiosyncrtic distste for ctivist service the disutility of the extr effort tht the ctivist service clls for. This disutility hs n income equivlent of ω i, which is n inverse mesure of n individul s predisposition (or bility) for ctivist service. It is distributed within the popultion with c.d.f. 9 R 0 mesures the bureucrts own productivity in rent-collection. R 0 is generlly greter thn zero, since rtionl worker will not fight corruption or resist predtory txtion beyond the level tht equtes the losses form the ltter nd the costs of enforcing the rule of lw t the mrgin. 10 For ll prcticl purposes, the bosses cn be ssumed not to exceed one percent of the popultion. 12

14 Z(ω ) such s Z(0) = 0. The popultion is homogeneous with respect to ll other behviorl prmeters. In prticulr, ll gents re risk-neutrl nd discount the future exponentilly t the rte r. Following the pproch outlined in the previous Section, I ssume tht the terms of nd the returns to the ctivist service nd consequently the number of ctivists t ny given moment in time re determined by n implicit contrct between the bosses nd the ctivists. This contrct offers the prticipting ctivists possibility of promotion into boss position fter T yers of service nd limits the bosses tenures to T b yers in order to fcilitte promotion. 11 The probbility of promotion is π. Therefore, t the time the contrct mtures, πn ctivists become bosses, while (1 π) N rejoin the rnks of ordinry workers. All gents mximize the expected utilities of residul life-time incomes. The contrct stisfies the prticiption constrints s long s () n ctivist s disutility is compensted by expected rents fter the end of service nd (b) extr rents due to ctivist services compenste incumbent bosses for the loss of rents fter retirement. In the bsence of contrct-bsed exchnge between the bosses nd the working popultion, the former never retire (T b is infinite) 12 nd cquire the rents equl to R 0, while the ltter ern W forever. There is no 11 Insted of ssuming fixed ctivist effort, vrible disutility of service, nd rndom promotion, it could be ssumed tht the workers heterogeneity in bility trnsltes into differentil productivity in the cpcity of n ctivist, in the spirit of the rnk-order wge tournments (Lzer nd Rosen 1981). In the ltter model, entry-level employees cn compete for promotion to higher-wge level by exhibiting vrying levels of effort, nd the winners the top N performers get promotion with probbility 1. Both incentive schemes hve similr effects s long s n individul worker hs little informtion on the ctions of the others nd the distribution of bility within the popultion. The implictions would differ with respect to the properties of the bosses: under my pproch, verge bility of the bosses equls the popultion verge, while wge tournments should produce n elite in terms of bility. Discussion of this issue goes beyond the scope of this pper. 12 Infinite tenure mens prcticlly tht the bosses re ble to bequeth their positions t their discretion. 13

15 outside option for worker: the only wy to chieve level of income exceeding W lies on the bureucrtic creer trck nd it requires ctivist service. As fr s the contrct design is concerned, the bosses behve s single entity the representtive boss. The contrct designed by the incumbent boss is offered to every worker who is not nd hs never been before n ctivist. 13 Although the boss dicttes the terms of the contrct to the ctivists, he cnnot force worker to enlist s n ctivist nd hs to choose the contrct terms in nticiption of known voluntry response from the workers, given the chosen vlues of T, T b, nd π; nd the exogenous wge, W. 14 The optiml contrct is subgme-perfect equilibrium in the boss-ctivists strtegic interction where the boss is the prime mover. The contrct is life-long. Once written, it is supposed to be non-renegotible. However, the contrcts offered to successive cohorts of ctivists my differ Supply of ctivists The choice problem fcing n individul worker involves comprison of two lifetime income profiles. The first ssumes permnent sty in the ordinry worker position nd receiving certin income. The second consists of the period of costly ctivist service of 13 The ccurcy of the ssumption of no reentry into the rnks of ctivists depends on the extent of fctionl struggle within the ruling prty. Swings in the power struggle my let previously purged ctivists restrt their creers. Thus, for exmple, lmost ll surviving victims of the Culturl Revolution in Chin ( ) were eventully rehbilitted (Lee 1991). 14 Theoreticlly, single boss, who combines politicl nd economic power, could endogenize wge in the present context. However, the division of responsibility within the ruling bureucrcy mkes it prcticlly impossible to lign the contrct design by politicl leders with wge setting decisions by production mngers (plnners), even if the ltter belong to the strtum of bosses in terms of my model. Mxim Boycko et l. (1995) rgue tht the seprtion of control rights from csh-flow rights under centrlly-plnned economy ws the mjor source of economic inefficiency of the Soviet system. In the theoreticl frmework of this pper, it results lso in politicl inefficiency by preventing the designers of the promotion contrcts from controlling wge. This issue is further discussed in Section

16 durtion T, uncertin promotion to the higher-income boss position therefter, nd retirement fter T b yers in office if promoted respectively. Worker i mkes the choice to become n ctivist if the expected income long the bureucrtic creer pth exceeds the income s n ordinry worker, given his personl bility: 15 (1) T 0 T + Tb rt rt ( W ω ) e dt + π Re dt + (1 π ) i T T + Tb T We rt dt T + Tb 0 We rt dt Inequlity (1) yields the cutoff level of disutility ω tht determines ctivist prticiption: rtb 1 e ω. rt e 1 * (2) = π ( R W ) The supply of ctivists the number of workers for whom ω i < ω holds under given contrct is then the left til of the distribution of the disutility from the ctivist service: * (3) Z( ω ) N s =. Since the ctivists constitute typiclly only reltively smll proportion of the totl s popultion, N << 1, Z cn be pproximted with liner function so tht (3) becomes, N = ζω, where ζ is constnt. In ddition, let us mesure time in units of the term of ctivist service, T, so tht T Then the supply of ctivists is expressed s: 15 Post-retirement income flows re not included in the expression, since they re identicl for ll gents by ssumption. 16 This pproch is justified by the fct tht the ctivist service is typiclly institutionlized (for exmple, s probtionry period for new prty member or the length of militry service for recruit) nd its durtion is fixed t lest in the short run. By contrst, the tenures re seldom regulted explicitly nd its ctul length fluctutes s result of policy chnges or politicl perturbtions. Note tht chnge in the term of ctivist service ceteris pribus trnsltes, by rescling the time xis, into n increse in the rte of future discounting nd decrese in the length of boss tenure in n equl proportion. 15

17 s rt (4) N = ( R W )( 1 e b ) πκ, where κ = ζ /(e r 1). It is cn be esily verified tht the supply of ctivists increses in the expected rents (boss rent 17 nd tenure, nd the probbility of promotion) nd decreses in the vlue of the next best lterntive for worker (workers wge): s s s s N N N N (5) > 0, > 0, > 0, < 0. R T π W b 2.3. The bosses problem At the core of the representtive boss s choice problem is the trdeoff between dditionl rents provided by the ctivists nd the limittion of tenure tht the provision of incentives for the ctivists implies. The bosses, entering the contrct with the ctivists, seek to mximize their residul life-time rents: (7) R = f ( N ) T b 0 e rt dt To chieve this gol, they choose the probbility of promotion, π, nd the length of tenure, T b, tking into ccount the workers response expressed in the form of the supply of ctivists (4). 17 Boss rent, R, s introduced in Section 2.1, is function of the number of ctivists. However, n individul worker hs no informtion on the outcomes of the current nd future recruitment cmpigns t the time he or she is mking the decision to become n ctivist. Therefore, R in the expression for the supply of ctivist should be interpreted s n exogenously determined expecttion of rent, which does not generlly stisfy R = f(n ). This identity should hold in the long-run equilibrium, but since the regimes of the type discussed here re reltively short-lived, the long-run equilibrium my never be reched. 16

18 An dditionl sttionrity constrint to their problem results from the necessity to blnce the inflow of promoted ctivists nd retiring bosses: (8) N b π N =. Tb To simplify further nlysis, let us combine the two constrints, by plugging (8) into (4) nd rerrnging the terms: 1 rtb (9) N [ N R ( e )/ T ] 2 = κ, b 1 b where R = R W is the boss premium. Herefter, the combined constrint (9) is referred to s the fesible supply of ctivists. The representtive boss s problem is then: (10) mx f ( N ) T b T b 0 e rt dt subject to (9). The bosses objective function (7) cn be chrcterized by the lines of equl levels of residul life-time rents in the (T b, N ) plne isorents. The optiml solution to the problem (10) n equilibrium in the regime s politicl lbor mrket is ttined t the point of tngency of n isorent nd the fesible supply curve in the (T b, N ) plne tht corresponds to constrint (9). Replcing the left-hnd side of (7) with n rbitrry constnt, integrting the expression, tking logs, nd rerrnging term obtins n lgebric expression for the isorent: 1 rc (11) N = f rt 1 e b. 17

19 The isorent (11) is downwrd-sloping nd convex curve. It behves pproximtely s hyperbolic curve T -α b with α > 1 n the vicinity of T b = 0 (since f(n ) is concve function) nd pproches horizontl symptote t N = f -1 (rc) s rt b pproches infinity. The derivtive of fesible supply (9) with respect to boss tenure is: 1 rt b rtb rtb ( e ) + T ( ) b 1 e re dn = N b R T κ b dt b Rerrnging terms nd substituting (9) into the expression bove obtins: 1 ( b ) < 0 dn N 1 b (12) = T + r( e 1) rt dt b 2 for ny T b > Therefore, fesible supply is downwrd-sloping curve in the (T b, N ) plne. Its mximum vlue is reched t T b = 0 nd equls: mx (13) N = κn r R b which sets the upper boundry on the number of ctivists under given regime prmeters. 19 The fesible supply curve is lso convex, but its curvture is systemticlly lower thn tht of n isorent. 20 This gurntees the existence of unique interior solution to problem (10). A typicl configurtion of n isorent nd the fesible supply constrint is presented in Figure 1. The first-order condition to the problem (10) is: 1 b rtb The term in brckets, T + r( e ) < 0, cn be rerrnged to obtin rt b + 1 < e rtb, which holds for ny T b >0 by the properties of the exponentil function. 19 By the L Hopitl rule, lim (1 e rtb ) / T b = re rtb = r. In fct, the mximl sustinble number of T b 0 ctivists flls short of tht given by (13) nd is determined by the bosses prticiption constrint discussed lter in this in section. 20 Proof of this sttement cn be obtined from the uthor. 18

20 dn (15) f N ( T ) dt b rtb ( )( 1 e ) + f N ( T ) b ( ) b r e rtb = 0 Substituting the expression for dn /dt b from (12) nd rerrnging terms obtins: (16) N ( Tb ) f ( N ( Tb )) 2rTb = rt f ( N ( Tb )) e 1 rtb tht determines the optiml tenure nd, implicitly, the optiml probbility of promotion. Representtion (16) of the first-order condition seprtes conveniently the effect of the boss tenure (on the right-hnd side) from the productivity of ctivists services (left-hnd side). To nlyze the comprtive sttics of the equilibrium in the politicl lbor mrket, let us consider the two sides of (16) s the functions of N, denoting the left-hnd side expression X(N ) nd the right-hnd side Y(N ) = Y(T b (N )). The inverse fesible supply function, T b (N ), cn be obtined from (12). The solution to (16) is therefore the point of intersection of the two curves X(N ) nd Y(N ). Under this pproch, the position of X(N ) depends exclusively on the properties of given f(n ), while the position of Y(N ) is ffected by the sme prmeters s the supply of ctivists. It cn be shown tht Y(N ) hs the following properties: Y (N )>0, Y (N )>0, Y(0) = 0, it hs verticl symptote t N = κnb r R, nd its slope is mx proportionl to the inverse of N mx. 21 An immedite impliction is tht the equilibrium number of ctivists is ffected by exogenous chnges in bosses rents nd wges in the sme wy s the supply of ctivists nd increses in the number of bosses: * * * N N N (17) > 0, < 0, > 0. R W N b 21 Proof of this sttement cn be obtined from the uthor. 19

21 These three derivtives cn be jointly interpreted s positive effect of expected rent the product of the boss premium nd the probbility of promotion on the equilibrium number of ctivists Production technology Further nlysis requires specifiction of the ctivists production function, f(n ). I rgued erlier in this pper tht ctivists serve primrily s lbor supervisors. Activistssupervisors, whose function is to elicit higher level of effort from fellow workers, cn be considered s sort of lbor-ugmenting technology. Then the rent-production function cn be represented s: (19) f(n ) = F{K, (1+N ) L}, where K is cpitl, L is lbor, is productivity prmeter. F(K,L) is the production function of the economy such tht: F K >0, F L >0, F KK <0, F LL <0, F KL >0. Under these conditions, the left-hnd side of (16) becomes: (20) X(N ) = N F L L/F. Obviously, X(N ) >0, X(0) =0. In ddition, X (0)>0, nd X(N ) reches the mximum t certin point, which my or my not lie within the rnge of dmissible vlues of N, to the left of min(1, κ N b r R ). To verify the ltter properties, let us re differentite X(N ) with respect to the number of ctivists: dx dn L 2 F 2 L [ ] ( F ) = F + LN L F (21) = ( F + LN F ) F LN ( F ) L LL L F 2 L LL F 2 20

22 At N = 0, (21) collpses into LF L /F 2 nd therefore X (N ) >0 t low levels of N.. Since the fctor in round brckets is negtive nd the second term increses in bsolute vlue with n increse in N, X (N ) < A typicl configurtion of X(N ) nd Y(N ) for this specifiction is presented in Figure 2. The number of ctivists in equilibrium is determined by the intersection of the two curves. A chnge in the prmeters tht leds to n increse in N mx = κnbr R cuses rightwrd shift of Y(N ) nd consequently n increse in the optiml number of ctivists. The position of X(N ) is ffected by the chnges in lbor nd cpitl employed in the economy. To determine how, let us consider the derivtives of X(N ) with respect to cpitl nd lbor. The position of X(N ) depends on the elsticity of output with respect to lbor input F L L/F. If the production technology hs unitry elsticity of substitution between lbor nd cpitl, F L L/F does not depend on fctor proportions. 23 If the elsticity of substitution is less thn unity, then F L L/F increses nd X(N ) curves fn out s the cpitl-lbor rtio increses. Since less-thn-unitry elsticity of substitution is the only prcticlly relevnt possibility on the mcroeconomic level, 24 we cn expect: * * N N (22) > 0, < 0. K L 22 The second term my exceed F L to the right of some point N^. In tht region, X(N ) bends down. Alterntively, X (N ) my remin positive s N. In both cses, X(N ) is concve. 23 X(N ) = N (1 α)/(1+ N ) in the cse of Cobb-Dougls production function, where α is the cpitl shre. 24 Numerous empiricl studies show tht the elsticity of substitution between lbor nd cpitl is typiclly less thn one for most modern economies. In prticulr, this pplies to the Soviet economy. See further discussion in Section

23 An importnt impliction is tht government investment in physicl cpitl clls for incresing numbers of ctivists. The intuition behind this result is tht, with substntil degree of complementrity between effective lbor input nd cpitl, more ctivistssupervisors hve to be hired to elicit dditionl effort from workers to mtch n increse in cpitl. The opposite is true with respect to n increse in the lbor force, since ctivistssupervisors re techniclly substitutes for crude lbor input by ssumption. Note lso tht the mrginl effect of n increse in the cpitl-lbor rtio on the mgnitude of F L L/F, nd consequently on the position of X(N ), curves diminishes s the cpitl-lbor rtio increses. This mens tht the link between investment policy nd the recruitment of ctivists is going to weken with the ccumultion of cpitl in the economy Bosses prticiption constrint nd regime chnge I now turn to the discussion of conditions tht mke the equilibrium in the politicl mrket unttinble. An lterntive to the regime with collective ownership nd rottion of the ruling bureucrcy is the one tht is bsed on privte property rights nd lcks the support of ctivists. Rtionl bosses will choose not to renew the contrct with the ctivists if the ltter option promises n bsolute dvntge over the former. Let us ssume tht the bosses hve time horizon of T m, which is determined exogenously nd reflects the perceived stbility of the regime. T m my be infinite (if the current regime is believed to be everlsting) nd in ny cse: T m > T b. The bosses choose to enter into promotion contrct with the workers t moment T=0 if the rents ccumulted over the period of tenure, T b, re expected to exceed those in the bsence of ctivists support (R 0 per period of time for T m periods): 22

24 Tb T m rt (24) f ( N ) e dt 0 0 R e 0 rt dt. Integrting, tking logs, nd rerrnging term yields (25) R0 rt b ln 1 1 e f rt m. ( ) ( ) N The numertor of the frction in (25) lies between 0 nd 1. Under the ssumption tht the ctivist services contribute significntly to the bosses cpbility to collect rents, f ( N ) >>1, the bosses contrct prticiption constrint pproximtes s: (26) N f 1 R 0 1 e rt rtm b, which is hyperbolic curve. Expression (26) determines the lower boundry of region in the (T b, N ) plne, where the contrcts cceptble for the bosses re locted. The point of tngency between the prticiption constrint curve nd fesible supply curve is the boundry optiml contrct. At this point the bosses re indifferent between mintining the contrct nd the chnge of regime (Figure 3). Its loction is determined by the productivity of ctivists cptured by the prmeters of the function, f, nd the bosses perception of the regime s durbility, T m. In prticulr, it follows from (26) tht insecure geopoliticl environment nd/or internl instbility, by decresing T m, depress the prticiption constrint curve nd expnd the region of cceptble contrcts, dding flexibility to the regime. Conversely, declining thret of ggression or uprising cuses the region determined by (26) to shrink, thus mking it more 23

25 likely tht the bosses choose not to enter in the promotion contrct nd bndon collectivist regime. Another theoreticl possibility is reduction in the effective rte of future discounting, r, due to expecttions of sustined economic growth. Higher expected rtes of growth reduce r, nd consequently increse the rgument of f -1 in (26), shifting the prticiption constrint curve upwrds. This implies tht bureucrtic regimes my be incomptible with permnent high rtes of economic growth. 25 It is not only the chnge in the bosses expecttions tht, by ffecting the loction of the prticiption constrint, cn cuse regime chnge but lso downwrd shift of the supply curve. Rel wge increses re of prticulr importnce in this context. As ws noted erlier, wge setting in most centrlized commnd economy is mtter of economic decisionmking. This process is institutionlly seprted from the design of optiml contrcts for the politicl lbor mrket. The industril bureucrcy my find it beneficil to increse wges to stimulte higher productive effort on the prt of ordinry workers, especilly when the productivity of ctivists-supervisors does not yield desirble results. Pul Gregory (2003) rgues tht the efficiency wge considertions were not foreign to the Soviet ledership nd their influence on the economic policy ws notble. Therefore, economic efficiency nd politicl support re generlly contrdictory objectives. While pursuing the former, bureucrts-mngers cn lower boss premiums. This depresses the supply of ctivists to the point when the bosses prticiption constrint cn no longer be stisfied. 25 This is unconditionlly true only if ll the bureucrts nd workers expect their rents nd wges respectively to grow t the sme constnt rte, which is restrictive ssumption. However, low growth rtes re typicl of most historicl non-democrtic regimes. The rtes of Soviet GDP per cpit growth seldom exceeded three percent per yer, nd higher rtes of growth did not lwys trnslte into significnt increses in per cpit consumption (see Gur Ofer, 1987). Therefore, growth rtes per se never ppered to chllenge the stbility of the regime in the USSR. The future development of the PR 24

26 Finlly, the bility of the ruling bureucrcy to rise support in the politicl lbor mrket depends on its bility to control the pths of upwrd income mobility. Thorough control is never possible. Two mjor venues led to higher incomes tht re, t lest in prt, independent from the bureucrtic control: higher eduction nd the shdow economy. In smll nd primitive economy, this problem cn be contined. A more complex economy requires, on the one hnd, more humn cpitl. This brings bout opportunities for higher incomes for politicl professionls. On the other hnd, there is more spce for illicit economic ctivity, especilly in services. Both undermine the regime by crrying off potentil ctivists, who respond to economic incentives outside of the stte-controlled economy. Therefore, modern economic development produces downwrd pressure on the supply of ctivists nd sets the nturl limits to the growth of hierrchicl regimes. III. The Soviet politicl lbor mrket: Institutionl frmework nd dt sources The dtset I use to test the model is pnel of nine sttes (republics) 26 of the former Soviet Union. The dt cover the period of , roughly coinciding with the ledership of Nikit Khrushchev. This period is chrcterized by rebounding influx of ctivists (incresing dmission of cndidtes to the prty) fter trough of the mid-1950s. Economic development of the period is chrcterized by the reltively high rte of growth (bout 4% per yer) nd ctive investment policy, lthough there ws mrked decelerting trend. There were no significnt politicl nd economic shocks during this period, nd the institutionl Chin my inform us on the impct of sustined growth ion the stbility of bureucrtic regime. 25

27 setting remined lrgely unchnged. Neither purges in the ruling prty nor mssive recruitment cmpigns occurred during tht period. Therefore, equilibrium promotion contrcts should dominte throughout this period. The primry gol is to show tht the chnges in the number of communist prty cndidtes ctivists in terms of the model cn be explined by chnges in the vribles tht cn be ssocited with certin prmeters of the model. In the overview tht follows I describe the institutionl fetures of the Soviet politicl lbor mrket nd the vilble dt tht cn be used in the empiricl nlysis Institutionl setting Hierrchy. The Soviet communist prty hierrchy ws lmost n exct mtch to the hierrchy of dministrtive (territoril) units. The ltter include (top down): union, republic, oblst, district/independent city/urbn district, primry prty orgniztion (PPO). The Russin Federtion, the lrgest Soviet republic, hd no seprte prty structure, its oblsts dministrtions reported directly to the union (ntionl) government nd oblst prty committees directly to the prty Centrl Committee. Independent city is reltively lrge city subordinted directly to oblst nd is not ffilited with ny district. Urbn districts existed only in lrge cities nd were comprble in popultion to rurl districts nd most ordinry independent cities. PPOs were typiclly ssocited with industril enterprises. Most primry prty orgniztions hd no pid officils on top. Secretries of PPOs were pid only in the lrgest enterprises. 26 No relible dt is vilble for the remining six republics. 26

28 The number of bosses nd verge boss slries by republic re the functions of the number of territoril units in every hierrchicl level. Chnges in dministrtive (territoril) structure trnslte utomticlly into cretion nd destruction of prty bodies nd, consequently, chnges in the number of pid positions for prty officils, i.e. the number of bosses in terms of the theoreticl model. Territoril structure ws subject to frequent reshuffling. Districts were split nd merged every yer; independent sttus could be given to city nd withdrwn yer fter. This implies tht chnges in territoril structure provide for substntil cross-sectionl nd intertemporl vrition in the number of bosses nd slries. Prty membership. The rules of the Soviet Communist prty (s in other communist countries) specified tril period for new members. During this time, new prty recruits were titled cndidtes. Upon pssing the cndidte review successfully, they becme full prty members. Although the probbility of promotion into the full membership exceeded 90%, only smll shre of the ltter ctully remined on prty creer trcks nd ws ble to rech position in the prty bureucrcy. The ltter constituted round one percent of totl prty so tht the probbility of promotion into the bosses (in terms of the model developed in this pper) ws of the order of The remining mjority of rnk-nd-file full prty members enjoyed minor non-pecuniry benefits of membership (such s softer punishment in cse of criminl prosecution, preferentil right to occupy certin types of jobs, etc.) nd typiclly remined in the prty until deth. The proportion of cndidtes in the totl prty membership declined from bout eight to five percent over the period of on verge. There ws, however, substntil vrition in the rte nd even the direction of chnge cross republics. The reduction ws 27

29 lrgely due to incresing numbers of full prty members, resulting from rpid mortlity decline in this period. At the sme time, the rtio of prty cndidtes to the totl lbor force fluctuted round 1:100 without ny significnt trend. Benefits of prty membership. Prty membership ws normlly prerequisite for ppointment to top mngeril position in ll spheres of the economy or for pursuing creer in government: civil dministrtion, economic control, or prty. A position of leding prty worker (prty bureucrt) ws of the utmost importnce. There is no indiction tht prty membership per se did py, but prty bureucrcy did receive substntil rents in the form of high slries nd fringe benefits. A mjor prt of the prty orgniztions expenditure, ccording to ntionl prty budgets (RGASPI), ws gered to provide benefits to pid prty officils, the reminder being used to cover opertionl expenses nd to finnce propgnd cmpigns. Slries of pid prty officils constituted only minor portion of their rents. Fringe benefits (helth nd child cre subsidies, reloction pckges, etc.) nd non-monetry rewrds, such s free housing finnced from prty budgets, constituted more significnt prt of their rel incomes Avilble dt Prty membership. Numbers of cndidtes nd full members re vilble on the ntionl level for the whole period of the existence of the Soviet communist prty. Republic level dt re vilble for (UFFA). Significnt lcune in the time series restrict the dtset to only nine republics: Armeni, Azerbijn, Belrus, Estoni, Georgi, Ltvi, Tjikistn, Turkmenistn, nd Uzbekistn. In post-stlin yers, the cndidte tril period ws close to one yer, so the number of cndidtes for given yer equls pproximtely the 28

30 number of new ctivists. As demonstrtes, there is significnt vrition in the rtes of recruitment cross the republics, lthough the end of the period is mrked with convergence, probbly due to incresing pressure from the centrl prty ledership. Prty budget, vilble from rchivl sources (RGASPI), records totl expenditures of the Centrl Committee (ctully, including number of ffilited centrl bodies) nd territoril orgniztions. The ltter re the ggregte numbers for ll dministrtive units below the ntionl level. The brekdowns by prticulr republics nd oblsts re vilble only for 1962 nd No dt for lower-level units re vilble. Notwithstnding its limittions, this source is indispensble nd, to the best of my knowledge, hs never been published, let lone used for nlyticl purposes. Prty bureucrcy. Numbers of bureucrtic positions nd slries re vilble for most yers from 1940 to 1965 seprtely for ech hierrchicl level. Brekdowns by republic re vilble for 1950, 1962, nd Comprison of ntionl ggregtes of prty pyrolls nd totl expenditure by territoril prty orgniztions shows tht pyrolls constituted n lmost fixed percentge of prty budgets: 5.8 ± 0.3% in Since no budget dt on ny units below the ntionl level is vilble, I use the dt on slries (verges by republic nd by level in the hierrchy) s proxy for bureucrtic ernings/rents. The sme sources provide office size dt (the number of bosses per dministrtive unit). Slries vry significntly (30%) with the level in the hierrchy but exhibit very low vrition cross republics 2-3% within ech hierrchicl level. Therefore, the territoril structure of republic (numbers of units in ech level) is likely to be the most importnt determinnt of vrition in the prty pyroll nd, consequently, in the verge slries by 29