How Interest Groups wth Lmted Resources can Influence Poltcal Outcomes: Informaton Control and the Landless Peasant Movement n Brazl by Lee J. Alston Unversty of Colorado NBER Gary D. Lbecap Unversty of Arzona NBER Bernardo Mueller Unversty of Brasla June 6, 2005 JEL classfcaton: D23, D72, D78 Keywords: Landless Peasant Movement, MST, Interest groups, multprncpal, multtask, land reform. We thank the Internatonal Center for Economc Research (ICER), Turn, Italy for Lbecap s support and also the followng workshops and meetngs where we receved valuable comments: XXIII Encontro Braslero de Econometra Salvador 200, Unversdade Católca de Brasíla 200, Department of Economcs of the Unversty of Brasíla 200, Latn Amercan Meetngs of the Econometrc Socety (LAMES 2002), Insttute for Behavoral Scence - Unversty of Colorado 2004, Colorado Unversty Envronmental and Resource Economcs Workshop 2004.
I. Introducton. In ths paper we examne how an nterest group wth lmted resources (votes and campagn contrbutons) nevertheless effectvely nfluenced poltcal polcy through the control of nformaton to general voters. Voters n turn lobbed poltcans to take actons desred by the nterest group. Our focus s on the Landless Peasants Movement (Movmento Sem-Terra) or MST and ts success n nvgoratng land reform n Brazl. Although we drect attenton to the MST, our analyss can be generalzed to nterest group behavor n other settngs. We analyze the poltcs of land reform n Brazl through a model that takes nto consderaton the constrants faced by the Presdent n determnng how many resources to devote to nterest groups. Although there are well-organzed, wealthy consttuents, large property owners, who oppose land reform, we show how the MST effectvely countered by affectng how voters perceve the government s land reform efforts. In general, urban voters support rural land reform, but snce t does not drectly affect them, they have lttle ncentve to determne whether or not the government s clams of acton are consstent wth actual resource allocaton. Indeed, pror to the md 990s and the effectve rse of MST, despte broad voter backng, there was lttle progress on land reform. Entrenched opposton from property owners and the nablty of voters to montor government polces resulted n assertons of acton wth lttle practcal results. After 993, however, the pattern changed and we analyze how ths occurred. II. Background: Land Reform n Brazl. Brazl has long had one of the most concentrated land ownershp structures n the world. Approxmately 45% of the agrcultural land s held by the largest % of farm owners, and large tracts of ths land are not used. The Gn coeffcent of 0.85 n 985 was the 9 th hghest n the world, only behnd countres lke Panama, Barbados, and Guam (FAO/UN, 2004). In terms of the sze of the populaton affected t s certanly among the The ten hghest Gn coeffcents n the FAO dataset (www.fao.org/es/ess) are Barbados (0.94, 989 data), Paraguay (0.93, 99), Guam (0.88, 987), Vrgn Islands (0.87, 987), Panama (0.87, 990), Bahamas (0.87, 994), Peru (0.86, 994), Span (0.86, 989), Brazl (0.85, 985), Argentna (0.83, 988).
most problematc cases wth more than 4 mllon landless peasants estmated 2, a contngent bgger than the entre populaton of some of those countres (Panama 3 mllon, Barbados 0.265 mllon, Guam 0.66 mllon). Concentrated land holdngs grew out of the Brazlan colonal experence, and snce the 9 th century there have been repeated efforts by the central government to substtute small holdngs for latfunda [Dean, 97:624]. Ths desre was to a large extent motvated by the comparson wth the Unted States where an equaltaran system based on small famly farms was provng to be hghly successful n attractng mgraton and generatng economc growth. In 938 the federal government created the Land and Settlement Dvson whch focused mostly on dstrbutng publc land. In 946 the Consttuton ntroduced the noton of expropraton of a prvate farm f t were not fulfllng ts socal functon, allowng latfunda to be exproprated. After the mltary coup of 964, the government vewed land reform as key to economc development, but even 2 years of a mltary dctatorshp dd not lead to serous changes n the dstrbuton of land ownershp. Wth return of cvlan rule further attempts were made, and ndeed, every new government had a specal land reform program wth ambtous goals that featured promnently n party platforms and electon campagns. But lttle of consequence happened. The Gn coeffcent barely budged. In 960 t was 0.842; n 978, and 998, 0.843. Very large farms of over,000 hectares also contnued to domnate land holdngs. In 940.5% of the farms held 48.3% of farmland; n 960.0% held 44.%; and n 980 0.9% held 45.%, and n 996.% held 45.%. 3 Despte ths lack of acton, the general electorate has long been sympathetc to the noton of land reform, a natural reacton gven such salent nequalty. 4 Ths broad consttuency for land reform, however, s unorganzed, heterogeneous and has only lmted nformaton regardng how much land reform s beng carred out. Rural property owners 2 Ths s the estmate typcally gven by the Landless Peasants Movement. The Lula government s II Natonal Plan for Agraran Reform (Mnstéro do Desenvolvmento Agráro, 2004) puts the demand for land reform at 3. mllon, or 5 mllon f one counts those who have land but n an nsuffcent amount. The Food and Agrculture Organzaton estmated the demand for land reform n Brazl at 2.5 mllon famles n the mdnnetes (Romero et. al. 994). 3 Gn coeffcents presented here are from Incra (200) whch tres to deal wth several of the methodologcal dffcultes n the calculaton of ths ndex. 4 Evdence to back ths statement wll be gven n Secton III. 2
have steadfastly opposed t. 5 Hstorcally, they were well organzed and provded support to poltcans through contrbutons and votes. They were represented by several organzatons, such as the Brazlan Rural Socety (Socedade Braslera Rural) and especally by a large group of Congressmen from varous partes, known as the rural bench (bancada rural), that unted to promote the nterests of land owners and agrcultural producers. As a result untl after 993, the pattern was for poltcans to call for aggressve land reform durng electoral campagns, and for lttle to be mplemented once the electon was over. So long as large landowners could delver more support than could landless peasants, and so long as voters had lttle nformaton on the actual state of land reform, poltcans devoted few resources to t. The pattern began to change when the MST (organzed n 985) took advantage of the new Consttuton of 988 that mandated the federal government to exproprate and redstrbute unproductve propertes and enablng complementary legslaton that was passed n 993. 6 By the md 990s, the MST had honed ts strategy of nvadng unproductve propertes wth elaborate press coverage to demonstrate the plght of the landless poor. Ths new publc relatons effort galvanzed voters and spurred the government to act on land reform. 7 As the numbers of nvasons multpled, urban voters were contnually remnded of the task at hand, and land reform moved to the forefront wth real resources devoted to t. The purpose of ths paper s to analyze how the MST nfluenced polcy makng n land reform so effectvely. It s one of the most successful grassroots movements n the world and s frequently held as a model of nterest group effectveness even though t lacks drect votng power and funds to contrbute to poltcans. 8 We show how the MST has used ts strengths to nfluence the level of nformaton receved by voters. There s a large lterature on the role of nterest groups as transmtters of nformaton (see Austen-Smth 999, Grossman and Helpman 200 and Van Wnden 999 5 See Mueller (998) for an ncdence analyss on whch groups were affected by land reform and econometrc measurement of ther effect on land reform polcy. 6 Although the Land Statute of 964 already provded the legal bass for expropratons, the new Consttuton broadened the scope for the use of ths nstrument, made t easer to use and, most mportantly, sgnaled the dsposton of government and socety that land reform should be pursued wth hgh prorty. 7 For the hstory of the MST and an analyss of ther organzaton, thnkng and mpact see Wrght and Wolford (2003). 8 Noam Chomsky was cted by the Economst (2004) as statng that the MST s the most exctng popular movement n the world today. 3
and 2003 for surveys), however, the flow of nformaton n that lterature s almost always from the nterest group to polcy makers to nfluence ther actons. 9 In ths paper, however, MST s modeled as revealng nformaton to voters, who n turn pressure poltcans for polcy change. In our earler papers (Alston, Lbecap and Mueller,999a, 999b, 2000) we descrbed how MST nvasons generated negatve publcty for poltcans, stmulated broad sympathy toward the landless, and led to further nvasons. We were nterested n explanng the pattern of nvasons and modeled government nterventon as exogenous. Here, however, we seek to explan the level of government acton n land reform by endogenzng government actvty. Ths s done through a multple-prncpal, multple-task prncpal-agent model where the government faces pressure from several nterest groups and the electorate to pursue several dfferent polcy objectves. Each nterest group exerts pressure by provdng the government wth votes (poltcal support or opposton), contrbutons, or by affectng the nformaton asymmetres faced by all nterest groups and the voters concernng the government s true level of effort on each polcy. The choce of whch channels of nfluence to pursue depends on each nterest groups comparatve advantage wth each of these nstruments. We argue that the MST has characterstcs that make t partcularly effectve at nfluencng polcy by ncreasng the electorate s awareness of what the government s actually dong to mplement land reform. The multple-prncpal, multple-task model of nterest group pressure s presented n the next secton. Ths model captures three mportant aspects of the relaton between nterest groups, the electorate and the government: ) the moral hazard that arses from the nformaton asymmetres n the relaton between nterest groups and the polcy makers; ) the exstence of multple groups smultaneously pressurng the government for competng polces; and ) the possblty that some groups may affect the level of nformaton asymmetres between other groups and the government thereby ndrectly changng the equlbrum level of effort on some polces. From the model we derve several characterstcs that make an nterest group more nclned to use the nformaton channel on the electorate rather than contrbutons or votes. Then n Secton IV we argue that the MST 9 Accordng to Reuben (2002), Most of the nformaton lterature has concentrated on the nformaton transmsson between the nterest group and the polcymaker. It has neglected to look nto the nformaton flow among and wthn the nterest groups. Some exceptons are Grossman and Helpman (999, 200) and Cameron and Jung (995). 4
matches those characterstcs and provde evdence that ther actons do have the effect of ncreasng the pressure of the electorate on the government for land reform. III. A Multprncpal, Multtask Model of Interest Group. 0 In ths secton we present the multprncpal, multtask model of nterest group pressure specfcally appled to the case of land reform n Brazl. In the appendx the more general model for n+ groups (n nterest groups plus voters) s presented and can be referred to for detals not shown n the more streamlned presentaton n the text. The agent n the model s the government (Executve) who s n charge of creatng and pursng all polces ncludng land reform. There are three prncpals, whch we denote wth superscrpts m, l, and v, for MST, landowners and voters, respectvely. Each prncpal s nterested n a specfc polcy (task) and would lke to see the government satsfy ther preferences. The tasks sought by each prncpal are denoted wth the followng subscrpts: s for a polcy of expropraton of prvate land and creaton of settlement projects defended by the MST; p for the (non-) reform polces sought by landowners; and q for the reform polces sought by voters. Gven the lmts on the agent s tme and resources, effort expended on one task reduces the level of effort that can be allocated to another. The vector of government effort toward land reform s t ' = t p t q t s, where the prme denotes a transpose. In general the prncpals do not observe the level of effort placed by the government n each task, nstead they observe the outcome of that effort. The vector of outcomes s modeled as x = t + ε, or: x x x p q s t t p q s ε p = t + ε () ε q s where ε~n(0,ω) and Ω s the covarance matrx of the random nose varable ε. Ω s a 3x3 matrx wth prncpal dagonal ω, =, 2, 3, and zeros elsewhere. The larger the value of ω the more dffcult t s for the prncpals to nfer the level of effort t from the observaton of the outcome x. 0 Ths model s based on that by Dxt (996, pg.57-7), whch s a combnaton of the multtask model of Holmström and Mlgrom (99) and the multprncpal model of Bernhem and Whnston (986). Note that the land reform polces desred by the voters can be the same as those pursued by the MST. In the next sectons we wll provde more nformaton on the nature of each groups preferences. 5
Prncpal benefts from polcy outcomes accordng to the beneft functons b, whch can be wrtten as: beneft to prncpal = p b b b x = b ' x (2) p q s x x q s Followng Dxt (996) the cost to the government of drectng effort to all the tasks s modeled as the followng quadratc functon: t ' Ct where C = 2 c c c pp qp sp c c c pq qq sq c c c ps qs ss (3) where the matrx C s assumed postve defnte. If the off-dagonal terms are postve there wll be substtuton amongst types of effort, so that an ncrease n t wll mply a decrease n t j, and vce-versa. If these terms are negatve the types of effort wll be complementary. In the appendx we present the detaled dervaton of the general model. Here we dscuss only the man steps of the model so as to focus on the results. The dervaton s done n four steps, startng wth a frst-best benchmark and addng addtonal elements n each step; () asymmetrc nformaton, () multple-prncpals and () nformaton manpulaton. We wll present and dscuss the optmalty condtons for each of these steps. The benchmark case s one where the prncpals observe the levels of effort chosen by poltcans and addtonally are able to act cooperatvely. As shown n the appendx, the optmal level of effort n ths scenaro s obtaned by maxmzng the sum of the agent s and the prncpals net beneft from ther nteracton n the polcymakng process. Ths yelds the frst-order condton b Ct = 0, so that the frst best level of effort s: t = C - b (4) where C - s the nverse of the C matrx. Ths result smply states that the margnal cost of effort n each task equals the margnal beneft to all partes. The second step s to relax the assumpton of observable effort. In ths stuaton contracts between the prncpals and poltcans must be made contngent on x (outcomes) and no longer on t (effort). As shown n the appendx, a lnear reward scheme s used to stpulate the government s pay-offs gven outcomes x. That s, gven the observed outcomes x, the unted prncpals provde poltcans poltcal support n terms of votes and 6
contrbutons that has a monetary equvalent equal to α x + β, where the αs are the value of the margnal support gven by the prncpals to government effort and β s a fxed payment that can be adjusted to assure the agent s reservaton utlty s at least matched. In ths scenaro the frst-order condtons are: t = C - α (5) Comparson of (4) and (5) shows that the addton of asymmetrc nformaton leads to a substtuton of α for b n those equatons. In the appendx t s shown that the relatonshp between α and b s: b = (I + rcω)α (6) where I s an dentty matrx and r s the coeffcent of rsk averson of the government. Gven that () all elements of C are postve (assumng outweghng substtutablty amongst tasks); () the elements of Ω are postve, because they are varances; () the αs are postve, because the unted prncpals wll not want negatve effort, t must be that b j > α j where j=p, q, s. Thus the government optmally chooses less effort when effort s not observable than n the frst-best stuaton where t s. Ths s the standard second-best story where, as a result of moral hazard arsng from nformaton asymmetres, less effort s realzed n each task. In other words, the ncentves n the case of asymmetrc nformaton are more low-powered than n the full-nformaton case, whch s due to the fact that n the second-best case there s a sharng of rsk between the prncpals and legslators. The thrd step s to allow non-cooperatve behavor among the prncpals Ths nvolves fndng the Nash equlbrum of the game where each prncpal provdes hs own ncentves to the agent and strategcally takes nto account the actons of the other prncpals. Now each prncpal 's ncentve scheme for task j s α + β for j=p, q, s and =l, v, m, whle the total for each prncpal s α x+β. In the appendx we show that the expresson for the total beneft arsng from the Nash equlbrum, addng the beneft of all ndvdual prncpals, s: b = α + 3rΩCα (7) j x j j 7
Ths equaton can be compared to equaton (6), the total beneft that resulted when prncpals were able to act cooperatvely. 2 Rememberng that when α=b and the frst-best soluton s acheved, we can see that wth non-cooperatve prncpals a stuaton s reached that s even further from frst-best than wth unfed prncpals, snce r s now multpled by the number of prncpals. The stuaton s therefore a thrd-best, characterzed by apparent neffcences and low-powered ncentves. In fact the neffcences are smply a drect consequence of the multprncpal multtask nature of the problem. To take the fnal step n modelng land reform poltcs n Brazl, suppose now that each of the three prncpals can nfluence polcy not only through drect ncentves (contrbutons, votes) represented by α, but also by affectng the level of nformaton avalable concernng the government s efforts n each task, that s, by affectng ω pp, ω qq and ω ss. The problem faced by each group then becomes that of decdng not only the optmal level of α j to allocate for each task j, but also how much effort t wll place towards affectng the nformaton avalable to all partes (and especally voters) regardng each of the tasks. Let ths effort by each nterest group = l, v, m to nfluence the nformaton concernng efforts n each task be e ' = e p e q e s. When decdng on the optmal level of e the nterest group wll take two factors nto account. The frst s that effort s costly, where the cost of that effort s represented by the cost functon G (e ). The second s the fact that all other groups may also expend efforts to affect nformaton avalablty, so that the soluton wll be a Nash equlbrum. Let e - be the vector of effort of all nterest groups other than. It s shown n the appendx that nterest group s problem s now to maxmze the followng objectve functon wth respect to α and e takng α - and e - as gven: b ' C α rα ' Ω( e, e ) α α '( C + rω( e, e )) α G ( e ) (8) 2 The change compared to the prevous objectve functon s the cost functon at the end and the fact that the matrx of varances s now a functon of the level of effort by each prncpal to nfluence nformaton. The frst order condtons for the maxmzaton of (8) n extended form are: 2 In the more general case of n+ prncpals ths equaton s b = α +( n+)rωcα. 8
b b b s p q = α + rc ω α + rc ω α + rc ω s p q ss ps qs ss ss ss s = α + rc ω α + rc ω α + rc s s = α + rc ω α + rc ω α + rc ω sp pp qp pp pp pp p p p sq qq pq qq ω α qq qq q α α q q (9) (0) () rω α ss pp qq s rω α rω α α p q s p α q 2 α 2 rω ( α ) 2 ss pp 2 rω ( α ) qq s 2 rω ( α ) q p = G 2 e s = G = G e e q p (2) (3) (4) The frst order condtons (9), (0) and () defne α *, the optmal ncentves by prncpal for each task. These equatons show, as before, that the prncpal wll offer a thrd-best level of ncentve for each task due to the nformaton asymmetres and the exstence of other prncpals who are also provdng ncentves to the government. The frst order condtons (2), (3) and (4) defne e *, the optmal level of effort that prncpal wll place towards affectng nformaton avalablty on each of the n+ tasks. The two terms on the left of each equaton n that system show how much the margnal effort ncreases or reduces the wedge between the frst-best stuaton b = α and the thrd-best stuaton b =α + r CΩα. Those two terms are therefore the margnal beneft from effort e, whereas the term G e s the margnal cost. It s possble to perform comparatve statcs on ths system to see what happens to optmal ncentves for poltcal acton wth a change n the level of nformaton avalable. The drecton of change depends on all parameters of the model. Intutvely, each prncpal wll alter the ncentves provded for a gven task when more nformaton becomes avalable regardng the government s effort Those who beneft from that task wll want to provde more ncentves now that they have a better noton of what they are gettng from poltcans n exchange. Those who oppose the task wll gve fewer negatve ncentves, snce the returns from opposton are reduced. However these reactons may be reversed dependng of the relatve values of the cost and beneft functons, C and b. Whatever the case, each nterest group can strategcally calculate how much and n whch drecton to affect nformaton so as to pursue ts polcy preferences. Implcatons from the model for nterest groups choce of nstrument 9
In order to llustrate the workng of the model for land reform polcy n Brazl, assume that the government s man constrant are voters so that value of α v, the support (or opposton) gven by the voters, s the largest part of the total support receved by the government for all the tasks n ths polcy ssue. Take from (9) the expresson that defnes v α q, the optmal level of ncentves that the voters wll offer poltcans for an addtonal unt of effort on land reform polcy: v b q v = α + (rcωα) q (5) q 3 Because the MST s nterested n task q, t would gan f the voters ncreased ther ncentves to poltcans for that task. At a gven level of nformaton avalablty, that s a v v gven Ω, the voters wll be offerng *, whch s less than the frst best level, the α q dfference between them beng (rcωα) q. 4 v Snce the voters are favorable to task q, - whch s an exogenous parameter - wll be postve. Therefore the MST can gan by puttng effort towards reducng ωqq n Ω so as to dmnsh the term (rcωα) q. 5 Greater ncentves lead to more effort by the government on land reform beng accomplshed, thus beneftng the MST. How much effort the MST wll choose to apply towards pursung ths beneft s determned by the frst order condtons n (2), (3) and (4), whch show the margnal gans and margnal costs of an addtonal unt of effort to affect nformaton. The model n ts general form has each prncpal gvng ncentves α for each n+ tasks and affectng nformaton on each task through e. In addton each prncpal s aware that the others wll also act ths way and takes that nto account when makng hs decsons. The fnal effect on government effort, and consequently on outcomes, thus depends on the net result of all these smultaneous forces. b q b q 3 Where (rcωα) j s the j th element of ths 3 x vector. Note that the voters desred polcy q s the same as the MST s desred polcy s. 4 Note that f there were no rsk averson, r=0, or f there were no nformaton asymmetres, Ω = 0, then rcωα would equal a null vector and the frst-best level of ncentves would be offered. 5 More generally, affectng nformaton may ental ether ncreasng ω jj (obfuscaton) or reducng t (makng truthful nformaton avalable), dependng on the sgns of the elements n α and C snce r and Ω are always postve. Note that α ncludes the ncentve of each prncpal for each task, whch are endogenously determned, and C ncludes parameters that are negatve f another task s a complement to task j and postve f a substtute. Interest group wll assess all ths nformaton and ether obfuscate or provde more nformaton so as to lead the voters to provde a hgher level of *. α j v 0
In practce we would not expect all nterest groups to be able to nfluence nformaton on each task, but rather that each group would have a comparatve advantage n nfluencng partcular tasks. That s, n real applcatons we would expect that the optmal ncentves provded by each prncpal on some of the tasks, as well as the optmal effort expended to nfluence nformaton, to be corner solutons and equal zero. The reason for ths s that t s typcally not easy for an nterest group to be able to affect the level of nformaton, ether to make thngs more transparent or to obfuscate. Dong so often requres specal characterstcs of the nterest group that are hard to acqure, and ndeed, may not be readly purchased or emulated. In some cases, for example, t may be credblty that leads voters to beleve the nterest group s clams about what the government s actually dong. Whether an nterest group wll be successful n pursung ts polcy nterests through the manpulaton of nformaton depends on the characterstcs of the nterest group and the polcy that t s pursung. By solatng some of the elements n the frst order condtons (0), () and (2) we can analyze three such characterstcs of an nterest group: ) The hgher the margnal cost of nfluencng nformaton, G, the lower wll be the optmal level of such effort chosen by that group, ceters parbus. If the margnal cost s suffcently hgh, then t may be above the margnal beneft for all postve level of effort, so that the group wll not try to nfluence nformaton on that task. The fact that some nterest groups pursue ther objectves through contrbutons, rather than manpulatng nformaton may be due to the relatve costs of the manpulatng nformaton beng too hgh. Interest groups that are successful n pursung polcy goals through nformaton manpulaton thus have comparatve cost advantages n these actvtes. ) The dervatve ω ( e, e jj j j e j ) can be nterpreted as the productvty of effort by an nterest group to nfluence nformaton. The more an addtonal unt of effort changes ω jj, the more productve the group and the more nfluence t wll have over polcy for each dollar spent n effort. Low productvty for some groups may be due to ther lack of credblty among voters. Accordngly nterest groups that work through the nformaton channel wll tend to have reputaton advantages and effectve means of gettng notced. ) The ablty of an nterest group to affect polcy through nformaton depends on the preferences of all n+ prncpals, that s b. If voters care strongly about a gven polcy, e j
ether favorably or n opposton, then changes n the level of nformaton they receve can have large mpacts on the government s effort level for that polcy. If they are closer to ndfference however, then pursung that task by nfluencng nformaton wll be less frutful ceters parbus, even for a group well endowed wth the other characterstcs. In the next secton we argue that the MST matches the characterstcs descrbed above and pursues ts polcy objectves by nfluencng the nformaton held by voters. IV. The Informatonal Role of the MST In ths secton we analyze the recent hstory of land reform n Brazl n the lght of that model and show that the MST s methods and characterstcs ft our portrayal of an nterest group that pursues ts goals by affectng the level and qualty of nformaton receved by the other groups and voters concernng the government s actons. In order to understand the mpact of the landless peasant movement t s useful to apply the model to land reform poltcs n Brazl wth and wthout the MST, so as to capture the perod before and after that group became actve throughout the country. In the pre-mst perod the model would have as prncpals the landowners and voters. The landless peasants and rural workers could also be consdered prncpals, but because they lacked organzaton they had lttle power to affect government polcy. The task for landowners was to ether block expropratons or f some land reform had to occur, t should nclude polces that benefted them as well, through government credt and other subsdes. Both of these actons would reduce the resources avalable for actual land reform. The second group, urban voters sympathze wth land reform. 6 Voters often mstakenly vew land reform as costless to them and the country. 7 Although urban voters 6 Several publc opnon polls have been conducted over tme to gauge socety s poston towards land reform. Almeda (998) revews eght opnon polls from 962 to 998, thus coverng a large span of land reform hstory, and shows that there has consstently been broad support towards land reform. These polls were undertaken under very dfferent methodologes and samples, but all overwhelmngly reflect the fact that Brazlan socety has consstently vewed land reform favorably. In 998, for example, a poll conducted by IBOPE revealed that 80% of those ntervewed were n favor of land reform. 7 Expropraton for the purpose of land reform n Brazl s, by consttutonal mandate, compensated at far value, though much of t n Ttles of the Agraran Debt, so that there are hgh costs for obtanng land as well as the expendtures to settle and mantan the benefcary famles. In 2004 the budget for INCRA, the federal land reform agency, was R$ 2.5 bllon (approxmately US$ 833 mllon) though n the end 5.98% of ths was frozen by the central government to contrbute towards prmary surplus targets. In 2004 8,254 famles are clamed by the government to have been settled. 2
support land reform, t s not a central preoccupaton. They naturally are more concerned about ssues whch affect them more drectly such as unemployment, nflaton, health and crme. Consequently, they are only mperfectly nformed on what the government s really dong n terms of land reform. Before the MST became actve, voter nformaton on land reform was essentally what the government presented. The equlbrum resultng from ths stuaton was one where the government announced land reform programs but never really mplemented them. Ths equlbrum changed once the MST s ntroduced nto the model. The MST s a well-organzed nterest group wth a comparatve advantage n nfluencng the nformaton receved by voters. Ths s done through hghly-publczed farm occupatons, marches, nvasons of governmental offces, roadblocks, and accusatons that the government s stallng. Through these actons voters revsed ther vews regardng the government s commtment toward land reform and ncreased ther poltcal pressure for more acton. In ths new scenaro the equlbrum level of government effort for the land reform task s greater than n the pre-mst scenaro. That ths corresponds to realty s suggested by Graphs and 2, whch show that crca 993 the MST became more actve, ncreasng the number of nvasons and occupatons of prvate farms (Graph ). Ths resulted n greater budgets for land reform (Graph 2) whch n turn led to greater numbers of settled famles of landless peasants (Graph ). 8 To llustrate, consder the stuaton regardng land reform after the electon of Presdent Lula n late 2002. The Presdent and hs party stressed land reform, but to avod past nacton that followed electons and that seemed to be occurrng n 2003, MST s leader announced that n Aprl 2004 the movement would ntate a campagn of occupatons that would lead to a red Aprl. Ths threat of volence forced the new government to ncrease 8 Hereda et al. (2002) fnd that n a large sample 96% of land reform settlement projects emerge from some form of conflct rather than government ntatve. They also show that although there are several other groups, the MST s by far the most actve. The drop n occupatons and settlements after 999 can be attrbuted to several factors. By that tme so many famles had been settled that the MST s man focus shfted to pressurng the government to make good on ts pledges of credt to those famles rather than obtanng more land for new famles. Ths s mportant for the MST as t receves -4% of all the credt provded to settled famles. Also, n 999 the economy passed through a severe crss nvolvng a large devaluaton of the real. In the followng years the government successfully dealt wth ths crss by pursung strct fscal restrant (Alston, Melo, Mueller and Perera, 2006), whch severely constraned the governments land reform effort. Fnally, n the year before the 2002 presdental electon, the MST purposefully reduce the number of nvasons so as not to harm the electoral chances of Lula. 3
ts pace of land reform, expropratng 34 farms n Aprl (Ornagh, 2004). Its most mportant effect, however, was to make voters aware that land reform was stll not movng forward. The MST has the characterstcs outlned n the model that underle a comparatve advantage n nformaton control: () low margnal cost of affectng nformaton receved by other groups; () hgh productvty of effort n affectng nformaton; () extreme confguraton of preferences of other prncpals, partcularly voters; and (v) favorable cost relatons (complementartes and substtutabltes) between ts favored task and other prncpals tasks. ) Low margnal cost of affectng nformaton ( G e ). j It s straghtforward that nterest groups that have low margnal costs of affectng nformaton wll pursue more of that strategy, ceters parbus. The MST works closely wth the meda and mantans a flow of newsworthy events. The purpose here s to show that the MST s able to pursue these actons at relatvely low cost. Before dong so, however, t s mportant to note that our argument does not requre that voters approve of the MST and nvasons of prvate farms. What s mportant s that the MST s contnually able to elct press coverage, that voters approve of land reform, and that the MST s actons mpart nformaton to them on the government s efforts towards that end. An mportant characterstc affectng the costs of the MST s strateges s that they are extremely labor ntensve, requrng large contngents of people to be moblzed for long perods of tme. The nvason and occupaton of a farm, and the process of transformng the occuped farm nto an offcal land reform settlement, are procedures that can take several years, durng whch the potental benefcares go through extreme and unrelentng physcal and emotonal stress. An nvason typcally requres the group (generally 30 to 500 people) camp out n neutral terrtory, such as by the sde of a hghway, for long perods, often months, awatng the rght tme to act. The camps are tradtonally composed of cheap black plastc and cardboard tents that have become so much assocated wth the mage of landless peasants. The nvason may go through peacefully or may nvolve volent conflct. 9 The occuped farm s smlarly made up of makeshft tents and lvng condtons are only slghtly better than n the prevous camps. After an nvason there are evcton attempts, 9 From 985 to 2003 there were 3,524 conflcts for land n Brazl, most of whch nvolved an nvason and the subsequent resstance to evcton (Comssão Pastoral da Terra, 2004). A graph of the number of conflcts follows closely the number of occupatons n Graph. 4
ether by the landowner wth prvate mlta or by polce followng a court order for rentegraton of possesson. If there s an evcton, the group generally wll return to a provsonal camp to awat the rght tme to renvade the same property or start over on another one. Ths cycle can be repeated several tmes and many years can go by before progress s made. To mantan order the MST mposes strct dscplne n ts camps, where, for example, no alcohol s allowed, settlers can only leave wth permsson and for lmted perods of tme, and all work s done collectvely. In addton there s constant ndoctrnaton of the settlers on the goals of land reform and other poltcal objectves. The movement has more than,000 schools n ther settlements where they teach ther own currculum despte beng fnanced by the state (Wenberg, 2004). 20 Ths descrpton shows that the means used by the MST to affect the nformaton receved by other partes on the government s land reform effort nvolves moblzng very large numbers of people and convncng them to undergo extreme hardshp for long perods of tme. 2 Practcally all of the nterest group lterature snce Olson (965) and Stgler (97) has recognzed the ablty to overcome free-rder problems and low costs of organzaton as key determnants of nterest group success. MST s ablty to maneuver a large contngent of people for whatever task s necessary to attract attenton, no matter how gruelng, tedous or dangerous, s key to the MST s success. It s the low opportunty cost of landless peasants that enables the MST to control ts members. Most MST members are extremely poor wth lttle to lose and nowhere to return. 22 The lack of alternatves makes them more receptve to accept the hardshps mposed on them by the movement wthout rebellon or deserton. Wth such low opportunty costs, the dstant promse of a pece of 20 Despte the based content of the educaton that s provded n the schools n settlement projects (Karl Marx, Che Guevara and the Chnese revoluton are major topcs), t s nevertheless educaton and chldren that otherwse mght not have had the chance, learn to read and wrte. In 995 the MST receved a prze from UNICEF n recognton of ther work n educatng chldren. 2 From 995 to 2002 423,83 famles were settled n 5,00 offcal land reform projects most of whch arsen from MST occupatons, Heredea et al. (2002). Ths number ndcates that the MST s qute adept are recrutng members. 22 In ths regard Wrght and Wolford (2003: 54) cte a settler s recollecton of hs days n an MST occupaton: We lost what lttle we had when we went to the encampment. We could take lttle even of those few thngs that we owned nto the new encampment, the only thng we took was our (wood-burnng) cook stove. What lttle savngs we had were soon gone, because we were earnng nothng. We had no house nor land to return to, no household goods, hardly any clothng, very few of our tools everythng was lost. And there was no way to go back and be the same person agan to the old neghbors, the frends on the outsde. In other passages the authors also document several postve memores that settlers held from the occupaton days, n partcular the camaradere and the sense of empowerment from partcpatng n the movement. 5
land s suffcently attractve so as to make partcpaton n an occupaton a worthwhle prospect. 23 It s mportant to consder not only the MST s cost of nfluencng nformaton, but also that of ther chef compettor, landowners. As noted by Becker (983) what matters n competton between nterest groups s not absolute but rather relatve pressure. Landowners have consderable fnancal resources that could be used to nfluence the avalablty of nformaton f that proved to be a productve means of affectng polcy. Land owners are well organzed and have overcome the free-rder problem. In the late 980s the landowners even formed a poltcal party, the UDR (Unão Democratca Rural) to fght land reform. What s relevant for our analyss s the margnal cost facng land owners of nfluencng voter nformaton as compared to the benefts they receve from dong so. Margnal cost ncludes the opportunty cost of not usng those resources on drect nfluence, as well as on other productve actvtes or consumpton. It s reasonable to assume that those opportunty costs are consderably hgher for landowners, relatvely, than they are for the landless peasants, where the resources are mostly n the form of tme and effort for whch they have much fewer alternatve uses. ) Productvty of effort n affectng nformaton ( ω jj ( e j, e j ) ) The second characterstc for an nterest group to be successful at nfluencng nformaton receved by other groups s the productvty of those efforts. Even f an nterest group manages to get ts message through t may have no effect f t has no credblty. A hgh value of ω ( e, e jj j j e j ) means that the nformaton asymmetry between the government s actons and the other groups (especally the voters ) perceptons of those efforts can be greatly affected by addtonal efforts of group at reducng or ncreasng that asymmetry. A low value means that those efforts have low payoffs. The dervatve thus measures the productvty of efforts to affect nformaton. Ths s an mportant characterstc as t s one that s dffcult to attan. Even an nterest group that commands resources may fnd that ts 23 Wth the large number of landless peasants that have been settled and gven land n the past ffteen years, the stock of actual landless peasants, that s, those that really have apttude to work the land rather than smply beng poor, has reduced and t may become harder for the MST to recrut n the future. Grazano (2004) argues that there are no more true landless n Brazl and that current MST occupatons are flled mostly wth poor, unemployed people lvng n bad condtons n the ctes. e j 6
nvestments n advertsng brng lttle persuason. They lack credblty or reputaton. Ths s why nterest groups typcally lnk ther objectves wth the broader publc nterest. The MST, by ostensbly helpng the landless poor n a country rddled by an extremely skewed land dstrbuton, has been able to garner credblty, despte ts llegal nature and dsrespect for prvate property, whch otherwse most Brazlans support. As noted by the Economst (997) n 996 the MST won the ultmate accolade: sympathetc portrayal n a prme-tme soap opera on Globo, Brazl s leadng televson staton. By contrast, landowners have found t hard to appeal to publc sympathy. All ther attempts at publcty have had very lttle effect n changng ther mage as wealthy and trgger-happy hoarders of large unproductve tracts of land. Thus far, we have argued that the MST s not only more adept than landowners and other groups at gettng ther message through, but also that ther message s more effectve at alterng the level of pressure exerted by voters on the government. Aware that ther clams regardng the problems wth the government s land reform wll sound self-servng, the landowners prefer to center ther efforts at pressurng the government through ther representatves n Congress, whch s consstent wth what the model would predct for a group wth hgh margnal costs of affectng nformaton and low productvty of effort. ) Extreme confguraton of preferences of other prncpals (b) The thrd characterstc whch the model ndcates that an nterest group should have for t to be successful n controllng nformaton s an extreme confguraton of the preferences of the other actors. If several of the prncpals, and especally voters, feel strongly about the cause pursued by an nterest group, then t has the opportunty to manpulate ther demands on government by alterng the nformaton receved about the government s actons (Ω n the model). If, for example, an nterest group pursues a task that voters approve (abhor) then they can elct more (less) pressure from the voter on the government by reducng (ncreasng) the nose n the nformaton receved by voters on the government s effort. If the voters are ndfferent to, or only mldly nterested n, that polcy, then affectng the nformaton they receve wll have lttle effect on the ncentves they provde poltcans and the nterest group would do better by pressurng through another channel. 7
Voters are sympathetc to land reform and ths s crtcal for MST. To llustrate the mportance of land reform and the poltcal pressure placed on the Brazlan Presdent to mplement t, we estmate a model of Presdental popularty, addng to the usual specfcaton of economc and poltcal explanatory varables another varable that ncludes the number of farm occupatons by MST n the correspondng month. If we show that more MST farm occupatons reduces the Presdent s popularty because they demonstrate a lack of effectve polcy, ths wll be evdence of the MST s effectveness n manpulatng voter perceptons and forcng subsequent government acton. There s a large lterature whch suggests testng the determnants of presdental popularty by regressng measures of popularty, usually opnon poll data, aganst a seres of varables that capture the state of the economy and poltcal events (Prce and Sanders, 993; Edwards, 99; Erkson, 989; Markus, 988; Monroe, 984, Mueller, 973). There are no such studes for presdental popularty n Brazl. Our dependant varable s the percentage of the electorate that fnds the Presdent s performance very good / good or regular (versus very bad / bad and don t know ) n perodc publc opnon polls performed by Datafolha Instuto de Pesqusas (2002). 24 As explanatory varables we use monthly data on nflaton, nterest rate, exchange rate, plus lagged popularty. Unemployment, mnmum wage, and level of sales were tred as well but were not statstcally sgnfcant. In addton to these varables we add another that measures the number of occupatons promoted by the MST and other landless groups n each month. Ths varable proxes the level of actvty of the MST and should capture the percepton of voters regardng the government s effort on land reform. In order to rule out that our results may be spurous, we used only varables that were found to be ntegrated to the frst order 24 Our perod of analyss s January 997 to December 2002. The ntal date s constraned by the avalablty of monthly data on land occupatons (www.cpt.org.br) and the fnal date s the last month of the second Cardoso term. There s data on Presdental approval rates for most months n the perod. When no poll was performed that month, we repeat the value of the prevous month. Usng only very good / good as dependant varables yelds smlar results. In order to use logarthms on nflaton, 0.5 was added to each observaton. Descrptve statstcs are shown below: Varable Mean Std. Dev. Max. Mn. Popularty 66.35.45 40 85 Occupatons 32.09 23.54 2 3 Inflaton 0.98. -0.38 5.84 Exchange rate.88 0.72.05 3.89 Interest rate.68 0.49.02 3.33 8
I(), as most macroeconomc varables tend to be, and subsequently tested for contegraton. Ths allows us to estmate both the long-term (or equlbrum) relatonshp between popularty and the explanatory varables, and subsequently to estmate the short term relatonshps through an error correcton model. Counterntutvely, the value of monthly GDP was found to be statonary n the sample perod and consequently ths varable was not ncluded n the regresson, despte beng a theoretcally mportant predctor of popularty. 25 Note however that the effect of GDP s captured n the other macroeconomc varables. 26 The results are presented n Table. The estmated coeffcents can be nterpreted as constant elastctes as the data are n logarthms. Newey-West standard errors are shown. 27 Lagged popularty s postve and sgnfcant at % showng a strong nerta n presdental popularty wth an elastcty of 0.88. The exchange rate was found to be negatvely and sgnfcantly assocated wth popularty. The sample perod ncludes perods of overvalued exchange rate, pror to January 999, as well as the devaluaton shock that occurred that month. A one standard devaton ncrease n the exchange rate (that s, a devaluaton) decreases popularty from 65.0% to 64,0%. The estmated coeffcent for nflaton s negatve and sgnfcant, wth a one standard devaton ncrease n nflaton leadng to a fall n popularty of -.33 percentage ponts, gven all other varables n ther means. In the same manner nterest rates are estmated to be negatvely related to popularty wth a one standard devaton ncrease found to reduce popularty by -0.76 percentage ponts. Note that the sgn of each of these varables are as expected, as nflaton, nterest rates and devaluatons all reduce voters real ncome. 25 When we ncluded GDP n the regresson, gnorng the unt root problem, the estmated coeffcent was found to be postve and sgnfcant. Its excluson dd not mpact the results sgnfcantly. 26 The relatonshp between macroeconomc varables such as nflaton, exchange rates, nterest rates may rase concerns about mutcollnearty. In ths regard we note that: () examnaton of the correlaton matrx for the sample perod does not ndcate that multcollnearty s severe n the sample used; () usng subsets of the explanatory varables does not alter the result for the occupatons varable; and () multcollnearty affects (ncreases) only the standard errors of the estmated coeffcents and not ther consstency, so that even n the presence of the problem the result for the occupaton varable wll stll be vald. 27 Contegraton s tested through a resdual-based verson of the Augmented Dckey-Fuller test usng approprate crtcal values from Charemza and Deadman (997, Table 3), ncludng an ntercept, at %, N=75, sx explanatory varables, whch are lower bound = -6.0, upper bound = -5.83. Addtonal evdence of contegraton s gven by the Durbn-Watson statstc on the devatons from the regresson n Table, whch was CRDW (6, 7) =.94. As a rule of thumb ths statstc should be near 2 f there s contegraton (Engle and Granger, 987). Fnally, contegraton s also ndcated by a Johansen test whch yelds an egenvalue of 0.337 and a trace statstc of 2.646 whch accepts contegraton at 5%. 9
[Table here] The estmated coeffcent for the number of farm occupatons, our varable of nterest, s negatve and sgnfcant. The estmaton ndcates that a one standard devaton ncrease of the number of farms occuped, wth all other varables at ther means, reduces presdental popularty by.06 percentage ponts from 65.00% to 63.96%. Ths s a strong effect for a non-economc varable that doesn t really affect most (urban) voters well-beng drectly. Ths result s consstent wth our statements that Brazlans are strongly favorable towards land reform. Furthermore, t shows that there really s scope for the MST, through ts occupatons and other actons, to ncrease the pressure that voters put on the government for land reform. The more actve the MST s, the more resources the Presdent dedcates to land reform, not because he wants to placate the MST tself, but rather because voters would punsh the Presdent (loss of popularty) when they perceve that land reform s not progressng as expected. Because the seres n Table are contegrated, they have an error correcton representaton (Engle and Granger, 987) and we can estmate an error correcton model that allows us to analyze both the short term effects of the explanatory varables on presdental popularty as well as the rate at whch devatons from the long term equlbrum are corrected n each perod. 28 The results are shown n Table 2. Past popularty remans the major determnant of current popularty n the short term, wth a postve and sgnfcant estmated coeffcent. Nevertheless, n the short term occupatons stll have a negatve effect on popularty. The mpact s low, -0.025, however t s sgnfcant at %. Interestngly the effect of nflaton n the short term s postve and sgnfcant, even though the long term effect s negatve. Exchange rates and nterest rates are not found to have a short-term effect. The error correcton term s negatve and sgnfcant at nearly %, wth a value of -0.60, whch ndcates that n each perod 60% of devatons from the long term equlbrum are corrected. Ths s a relatvely large coeffcent, whch ndcates that the speed of adjustment of popularty to shocks n the explanatory varables s hgh. Ths suggests that popular opnon about the Presdent reacts quckly to new nformaton of 28 The error correcton equaton s Y t = α 0 - α (Y t- - X t- β ) + β 0 X t + ε t, where Y t s popularty and X t s the matrx of explanatory varables. In ths model α s the rate at whch devatons from the long term equlbrum are corrected, β 0 s the short term effect of the explanatory varables on popularty, and β s the vector of equlbrum relatons between the explanatory varables and popularty. 20