8/17/2019 Andres Velasco Economia
1/257
Volume 5 Number 2
200Spring
Journal of the Latin American and Caribbean Economic Association
Lora and Olivera on Electoral
Consequences of the WashingtonConsensus
Suominen and Estevadeordal
on Rules of Origin in Trade Agreements
Majnoni and Powell on Basel II
and Bank Capital Requirements
Echeverry, Ibáñez, Moya,
and Hillón on Reforming Public
Transport in Bogotá
Kaplan, Martínez González,
and Robertson on Wages afterDisplacement
8/17/2019 Andres Velasco Economia
2/257
E D I T O R
Andrés Velasco
LATIN AMERICAN AND CARIBBEAN
ECONOMIC ASSOCIATION
BROOKINGS INSTITUTION PRESS
Washington, D.C.
2005
Volume 5 Number 2 economíaSpring
Journal of the Latin American and Caribbean Economic Association
8/17/2019 Andres Velasco Economia
3/257
Articles in this publication were developed by the authors for the biannual Economía meetings.
In all cases the papers are the product of the authors’ thinking alone and do not imply endorse-
ment by the staff members, officers, or trustees of the Brookings Institution or of LACEA, or of
those institutions with which the authors are affiliated.
Copyright © 2005
LATIN AMERICAN AND CARIBBEAN ECONOMIC ASSOCIATION
www.lacea.org
Published by
BROOKINGS INSTITUTION PRESS
1775 Massachusetts Avenue, N.W., Washington, DC 20036
www.brookings.edu
ISSN 1529-7470
ISBN 0-8157-2079-3
For information on subscriptions, standing orders, and individual copies, contact Brookings Insti-
tution Press, P.O. Box 465, Hanover, PA 17331-0465. Or call 866-698-0010. E-mail brookings
@tsp.sheridan.com. Visit Brookings online at http://bookstore.brookings.edu.
Brookings periodicals are available online through Online Computer Library Center (contact the
OCLC subscriptions department at 800/848-5878, ext. 6251) and Project Muse (http://muse.jhu.edu).
Authorization to photocopy items for internal or personal use or the internal or personal use of specific clientsis granted by the Brookings Institution for libraries and other users registered with the Copyright ClearanceCenter Transactional Reporting Service, provided that the basic fee is paid to the Copyright Clearance Center,222 Rosewood Drive, Danvers, MA 01923. For more information, please contact CCC at 508/750-8400. Thisauthorization does not extend to other kinds of copying, such as copying for general distribution, or for creatingnew collective works, or for sale. Specific written permission for such copying must be obtained from the Per-missions Department, Brookings Institution Press, 1775 Massachusetts Avenue, N.W., Washington, DC 20036,fax 202/797-6195; e-mail [email protected].
8/17/2019 Andres Velasco Economia
4/257
Editor’s Summary vii
EDUARDO LORA and MAURICIO OLIVERA
The Electoral Consequences of the Washington Consensus 1Comments by Sebastián Galiani and Ernesto Dal Bó 46
ANTONI ESTEVADEORDAL and KATI SUOMINEN
Rules of Origin in Preferential Trading Arrangements: Is All Well with the Spaghetti Bowl in the Americas? 63Comments by Pablo Sanguinetti and Alberto Trejos 93
GIOVANNI MAJNONI and ANDREW POWELL
Reforming Bank Capital Requirements: Implications of Basel II for Latin American Countries 105Comments by Patricia Correa and Philip Brock 141
JUAN CARLOS ECHEVERRY, ANA MARÍA IBÁÑEZ,
ANDRÉS MOYA, and LUIS CARLOS HILLÓN
The Economics of TransMilenio,a Mass Transit System for Bogotá 151Comments by Mauricio Cárdenas and Andrés Gómez-Lobo 189
DAVID S. KAPLAN, GABRIEL MARTÍNEZ GONZÁLEZ,
and RAYMOND ROBERTSON
What Happens to Wages after Displacement? 197 Comments by Naércio Menezes-Filho and Omar Arias 235
2005
Volume 5 Number 2 economíaSpring
Journal of the Latin American and Caribbean Economic Association
8/17/2019 Andres Velasco Economia
5/257
LATIN AMERICAN AND CARIBBEAN ECONOMIC ASSOCIATION
The Latin American and Caribbean Economic Association (LACEA), or
Asociación de Economía de América Latina y el Caribe, is an internationalassociation of economists with common research interests in Latin
America. It was formed in 1992 to facilitate the exchange of ideas among
economists and policymakers.
Membership in LACEA is open to all individuals or institutions
professionally concerned with the study of Latin American and Caribbean
economies. For membership information, please visit the LACEA website
at www.lacea.org and click on “join LACEA.”
O F F I C E R S
President
Mariano Tommasi, Universidad de San Andrés, Argentina
Vice President
Andrés Velasco, Harvard University and Universidad de Chile
Past Presidents
Sebastián Edwards, University of California–Los AngelesGuillermo Calvo, Inter-American Development Bank and University of Maryland
Nora Lustig, Universidad de las Américas, Puebla
Albert Fishlow, Columbia University
Secretary
Ariel Fiszbein, World Bank
Treasurer
Sergio Schmukler, World Bank
E X E C U T I V E C O M M I T T E E
Nancy Birdsall, Center for Global Development
François Bourguignon, World Bank
Raquel Fernández, New York University
Francisco H. G. Ferreira, Pontifícia Universidade Católica, Rio de Janeiro, and
World Bank
Nora Lustig, Universidad de las Américas, PueblaJosé A. Ocampo, United Nations
Guillermo Perry, World Bank
Carola Pessino, Universidad Torcuato Di Tella, Argentina
8/17/2019 Andres Velasco Economia
6/257
Carmen Reinhart, University of Maryland
Andrés Rodríquez-Clare, Inter-American Development Bank
Jaime Saavedra, World Bank
Cristina T. Terra, Fundação Getúlio Vargas
ECONOMIA
Editor
Andrés Velasco, Harvard University and Universidad de Chile
Editorial Associate
Jennifer Hoover
Managing Editor
Magdalena Balcells
Editorial Board
Rafael Di Tella, Harvard University
Eduardo Engel, Yale University
Francisco H. G. Ferreira, Pontifícia Universidade Católica, Rio de Janeiro, and
World Bank
Carmen Pagés, World Bank
Roberto Rigobón, Massachusetts Institute of Technology
Andrés Rodríguez-Clare, Inter-American Development Bank
Roberto Steiner, International Monetary Fund
Miguel Urquiola, Columbia University
Sebastián Edwards (ex officio), University of California–Los Angeles
Mariano Tommasi (ex officio), Universidad de San Andrés, Argentina
ECONOMIA PANEL FOR VOLUME 5
Mauricio Cárdenas, Fedesarrollo-Colombia
Gerardo Esquivel Hernández, Colegio de México
Ronald Fischer, Universidad de Chile
Ilan Goldfajn, Pontifícia Universidade Católica, Rio de Janeiro
Eduardo Levy Yeyati, Universidad Torcuato Di Tella, Argentina
María Soledad Martínez Pería, World Bank
Francisco Rodríguez, Instituto de Estudios Superiores de Administración,
VenezuelaNouriel Roubini, New York University
Jaime Saavedra, World Bank
Ernesto Schagrodsky, Universidad Torcuato Di Tella, Argentina
8/17/2019 Andres Velasco Economia
7/257
AUTHORS AND DISCUSSANTS
Omar Arias, World Bank
Philip Brock, University of WashingtonMauricio Cárdenas, Fedesarrollo-Colombia
Patricia Correa, Super Intendencia de Bancos de Colombia
Ernesto Dal Bó, University of California–Berkeley
Juan Carlos Echeverry, Universidad de los Andes
Antoni Estevadeordal, Inter-American Development Bank
Sebastián Galiani, Universidad de San Andrés
Andrés Gómez-Lobo, Universidad de Chile
Luis Carlos Hillón, Universidad de los Andes
Ana María Ibáñez, Universidad de los Andes
David S. Kaplan, Instituto Tecnológico Autónomo de México
Eduardo Lora, Inter-American Development Bank
Giovanni Majnoni, World Bank
Gabriel Martínez González, Inter-American Conference on Social Security
Naércio Menezes-Filho, University of São Paulo
Andrés Moya, Universidad de los Andes
Mauricio Olivera, Inter-American Development Bank
Andrew Powell, Universidad Torcuato Di Tella
Raymond Robertson, Macalester College
Pablo Sanguinetti, Universidad Torcuato Di Tella
Kati Suominen, Inter-American Development Bank
Alberto Trejos, Instituto Centroamericano de Administracion de Empresas INCAE
8/17/2019 Andres Velasco Economia
8/257
Editor’s Summary
Nearly two decades after a wave of policy changes swept through Latin
America, economic reforms continue to be the focus of much dis-
cussion. Critics claim that the promarket reforms have failed todeliver economic growth, and that the time has come to try something else.
Advocates claim that the reforms were never given a fair chance—too lit-
tle was done, often too late. Complete the reform process, they claim, and
growth will come.
Both sides do agree on one point: Latin America seems to be suffering
from reform fatigue, and another wave of reforms is unlikely to happen any
time soon. Certainly not in countries led by left-leaning populists, such as
Argentina’s Néstor Kirchner or Venezuela’s Hugo Chávez. The reformmomentum has even stalled in countries led by promarket conservatives—
Mexico under Vicente Fox and Colombia under Alvaro Uribe are two
examples. If such reforms are now unpopular in many quarters, did the
politicians who initially adopted them bear a political cost? Was the Wash-
ington Consensus electorally bad for friends of Washington? That is the
question studied by Eduardo Lora and Mauricio Olivera in the lead article
of this, the tenth issue of Economía.
Lora and Olivera analyze the outcome of sixty-six presidential elec-tions and eighty-one parliamentary elections in seventeen Latin American
countries from 1985 to 2002. Their general conclusion is striking: reform-
ing parties and politicians were rewarded electorally only when reforms
involved macroeconomic stabilization and a sharp reduction in inflation;
otherwise, their reforming zeal cost them dearly at the polls. Economic out-
comes do matter for electoral outcomes. Lora and Olivera find that the
incumbent’s party is rewarded in presidential elections for reductions in the
inflation rate and in legislative elections for increases in the growth rate.
Changes in unemployment and income distribution, however, do not appear
to influence voters’ behavior.
vii
A N D R É S V E L A S C O
8/17/2019 Andres Velasco Economia
9/257
What is even more surprising is that, at the polls, policies matter irre-
spective of their results—that is, their effects on growth or inflation. Elec-
torates seem not to like reform policies of the kind applied in Latin
America in the 1990s. In a regression with electoral outcomes on the right
hand side, reform indexes have a negative and significant effect, even when
the authors control for changes in inflation and growth. The point esti-
mate of the effect of policies on electoral results implies that the incum-
bent’s party typically lost 15 percent of its vote in presidential elections on
account of the average amount of promarket reforms introduced during
its term. More aggressive reformers (say, those reforming one standard
deviation above the mean) sacrificed 27 percent of their vote on account of promarket reforms. Statistically, this seems to be a very robust result for
presidential elections.1
Moreover, lying about one’s true intentions does not seem to be a good
electoral strategy. Several Latin American politicians—including Fujimori
in Peru, Menem in Argentina, and more recently Gutiérrez in Ecuador—
first ran as opponents of the Washington Consensus, then followed ortho-
dox policies. The paper shows that a candidate that said one thing on tax
policy and then did another was, on average, punished more severely atthe polls. Campaign promises do not seem to matter for the effect of other
policies on voting behavior.
These results raise two kinds of questions. For academics, the issue is
why inputs (policies) matter and not just outputs. Is it ideology, pure and
simple? Or is it that because outcomes represent an extremely noisy signal
of politicians’ competence, the choice of policies conveys some informa-
tion that voters find useful? For policymakers, the question is political:
what has to change in Latin America before ambitious reforms become fea-sible again? Are all large-scale reforms out of the question, or only those
that carry the Washington Consensus label? Both sets of questions remain
very much open.
The unpopularity of the reforms does not mean, however, that policy is
frozen everywhere. Trade is one area in which reform has not stopped dead
viii E C O N O M I A , Spring 2005
1. The total effect of reforms on electoral outcomes is the sum of two effects: a direct
effect that runs from policies to votes and an indirect effect that runs from policies to eco-
nomic outcomes to votes. The first is typically large and negative, while the second is pos-itive insofar as the reforms lowered inflation and stimulated growth. The figures given
correspond to the total effect—that is, after the positive indirect effects have been taken into
account. The direct negative effects are much larger. See the paper for details.
8/17/2019 Andres Velasco Economia
10/257
on its tracks. Liberalization agreements, of both the bilateral and regional
kind, continue to be signed, though at a less frenzied pace than a decade
ago. Nearly fifty deals have been forged in the Americas since 1990s. But
this veritable “spaghetti bowl” of overlapping and sometimes contradictory
agreements has costs as well as benefits. One cost, studied by Antoni Este-
vadeordal and Kati Suominen in the second paper of this issue, results from
the rules of origin applied.
At heart, the matter is simple: if Brazil gives Paraguay preferential
access to its market, Brazilian policymakers want to make sure that the new
imports entering Brazil are, in fact, made in Paraguay and not in a third
country attempting to benefit from Paraguay’s preferential status. But whatsounds simple in theory becomes devilishly complicated in practice. What,
precisely, is a Paraguayan good? Goods often have imported inputs, and in
few or no items is 100 percent of value added likely to originate in
Paraguay. Where, then, should a country draw the line?
Rules of origin attempt to settle the issue, but in doing so they face many
pitfalls. If the set of rules is too stringent, then the bulk of Paraguayan
goods may be left out of the Brazilian market. Indeed, such rules can be
used as protectionist devices that effectively undercut trade preferences andcontradict the avowed liberalizing intent of free trade agreements. Another
problem is that rules of origin are almost inevitably complex (the paper
identifies several kinds, each with its own subcategories). Applying them
can be very costly, especially for the poorer economies in the region.
Estevadeordal and Suominen offer three main conclusions. First, putting
stringent rules of origin into an agreement makes it more politically feasi-
ble, since the rules can be used as a tool to pay off protectionist interests.
Second, there is evidence that restrictive rules of origin undercut the liber-alizing potential of free trade agreements. NAFTA is a particularly egre-
gious example of this, with many Mexican goods subjected to rules that
verge on ludicrous. It is unfortunate, therefore—and this is the third con-
clusion of the paper—that the NAFTA model of rules of origin is increas-
ingly being used in other agreements in the region. This does not bode
well for free trade in the Americas.
Not all is lost, however. NAFTA-type rules are at least precise, and they
leave less room for arbitrary application than do other types of rules of
origin used in earlier agreements. Moreover, the growing homogeneity of
rules that follow the NAFTA model simplifies the life of customs officials
and lowers transaction costs. Last, and most important, the NAFTA rules of
Andrés Velasco ix
8/17/2019 Andres Velasco Economia
11/257
origin have what trade experts call lenient facilitation devices. In English,
this means that the rules themselves include ways to reduce their restric-
tiveness. A key aspect is diagonal cumulation, which allows countries tied
by the same set of origin rules to use products that originate in any part of
the common rules-of-origin zone as if they originated in the exporting
country. Therefore, argue the authors, the rules of origin in a future Free
Trade Area of the Americas—if one ever materializes—should not be all
that restrictive. One can only hope they are right.
Financial regulation is another area in which policy is changing, as a
result of both internal needs and international changes in standards. The
1988 Basel Accord on bank capital—the so-called Basel I agreement—isnow in place throughout the region, and discussion has shifted to whether
and how Latin American countries should apply Basel II. It is widely
accepted that bank capital ought to be regulated, but how to do so remains
open to debate. The simple approach of Basel I divides assets into very
broad risk categories and establishes an 8 percent minimum capital require-
ment for risky assets. The potential for arbitraging one’s way around this
simple rule has grown, however, as risk management becomes more
sophisticated. In response, Basel II goes well beyond simple quantitativerequirements, proposing two basic approaches: the standardized approach,
which uses external credit rating agencies together with a table that maps
those ratings directly into capital requirements; and the internal ratings-
based (IRB) approach, in which the banks themselves estimate their cus-
tomers’ default probability (without relying on external rating agencies) and
then use a particular formula to determine capital requirements as a func-
tion of the estimated default probability.
The third paper in this issue, by Giovanni Majnoni and Andrew Powell,focuses on a key aspect of Basel II application. Many emerging markets
do not have many (or any) external rating agencies, so the standardized
approach may not be applicable. The internal ratings-based approach, in
turn, is complex, and its application and supervision may stretch limited
supervisory resources. Majnoni and Powell suggest a simplification of the
IRB approach that could be used as a transition arrangement. In their
centralized ratings-based (CRB) approach, banks would rate their clients,
but the regulator would determine the rating scale and the way in which
the banks’ ratings map into default probabilities. Using a centralized scale
would facilitate comparison across banks and greatly ease the monitor-
ing of banks’ ratings. Those requirements would also be easier to monitor,
x E C O N O M I A , Spring 2005
8/17/2019 Andres Velasco Economia
12/257
since the regulator would determine how banks’ ratings feed into capital
requirements.
The hard part of the approach is deciding what kinds of standards the
regulator should apply, since what works in rich countries may not work
in emerging economies. Basel II’s IRB approach suggests a formula for cal-
culating a bank’s capital requirement as a function of three basic variables:
default probability, exposure at default, and loss given default. A regulator
might then ask a bank to hold provisions and capital to cover a specified
percentage of the distribution of losses to ensure the continued solvency
of the bank except in highly extraordinary circumstances. The calibration of
the Basel II IRB formula uses a value at risk of 99.9 percent with a hori-zon of one year—that is, a bank is only expected to use up its capital in
one year with a probability of 0.1 percent, or once every 1,000 years.
Majnoni and Powell employ a bootstrapping technique to calculate loss
distribution functions for Argentina, Brazil, and Mexico, using data on loan
performance from public credit registries. They then use these functions to
estimate the size of expected and unexpected losses of an average-sized
bank with a loan portfolio randomly drawn from the universe of loans
within the financial system. Their results show that these three countrieshave significantly higher default probabilities than Group of Ten (G10)
countries. As a result, both current practice under Basel I and the sug-
gested standards under Basel II may be inadequate. To achieve a 99 percent
level of protection, capital requirements would need to be close to 15 per-
cent, which is significantly higher than the 8 percent level recommended
in Basel I. Even higher levels would be required to achieve 99.9 percent
protection, as intended in Basel II. They also find that Basel II’s IRB
approach would result in levels of 90–95 percent protection rather than the99.9 percent goal. This is not surprising, since the IRB was calibrated for
the safer economies of G10 countries.
If bank regulation needs modernizing in Latin America, public trans-
port does too. The spectacle of streets packed with old buses spewing black
smoke is all too common in many cities of the region, from Mexico City
to Quito and from São Paulo to Santiago. Poor public transport induces
more private cars to enter the streets, worsening congestion and pollution.
If you think that this is a textbook case of the state not doing the job of
providing public services, think again. Bus systems are private in many
cities in Latin America, and that does not seem to solve the problem. As
Juan Carlos Echeverry, Ana María Ibáñez, Andrés Moya, and Luis Carlos
Andrés Velasco xi
8/17/2019 Andres Velasco Economia
13/257
Hillón document in their paper in this volume, the market for urban buses
is ripe with market failures: unclear definition of property rights on the
curbside and on the road; cartelization that results in fares set above the
competitive equilibrium levels; misalignment between the incentives of bus
drivers and owners, in a typical principal-agent conflict; and congestion and
pollution externalities. In many developing countries, these market fail-
ures are exacerbated by weak regulation and enforcement. The result often
is too many buses each carrying too few passengers in unsafe conditions,
clogging the streets and soiling the air as they move (or fail to move) along.
One city in Latin America to have tackled the problem head-on is
Bogotá, Colombia. Its so-called TransMilenio system is now being imitatedin Quito and Santiago, among others, as well as several cities in Colom-
bia. Echeverry, Ibáñez, Moya, and Hillón explain the logic behind the new
system and analyze is effects. The key elements of the new system are as
follows: (i) a hybrid public-private system, with concession contracts for
private service providers; (ii) competition “for the road” (rather than “on
the road”) in a tendering process, with fare-setting based on long-term
investment recovery; (iii) remuneration based on kilometers traveled rather
than passengers transported, so as to prevent drivers from fighting over pas-sengers on the street; (iv) separation between the transportation service and
the fare collection process; and (v) exclusive road and curb-side service in
metro-like stations.
Congestion, pollution, traffic accidents, travel times, and waiting times
all fell dramatically along the corridors where TransMilenio was first put to
work. The system was initially hailed as the solution of Bogotá’s serious
transport problems. Not all results were unambiguously positive, however,
as the paper makes clear. Increased ridership resulted in jammed busesand rising waiting times. Moreover, the full system covering the entire
city is not expected to be operating until 2015. This gradual transition did
not help: older buses were displaced to secondary streets, where traffic
and pollution increased.
A cost-benefit analysis of the system as is, with approximately 25 per-
cent of the routes in operation, reveals welfare gains for users of the new
routes, but an overall negative effect stemming primarily from increases
in travel time for passengers using the traditional transport system. Since
congestion costs are highly nonlinear, the welfare losses from heightened
congestion in unserved corridors more than offset the benefits from Trans-
Milenio, even though those benefits are sizeable. The authors conclude by
xii E C O N O M I A , Spring 2005
8/17/2019 Andres Velasco Economia
14/257
arguing that the adoption of a new public transport system must be cou-
pled with improved regulation of all other public transport providers, so as
to avoid the problem that arose in Bogotá.
What happens to workers’ wages and employment prospects once they
are displaced from their current jobs—for instance, by trade reform? If they
are likely to be re-hired quickly at comparable wages, then no policy
response is called for; but if some wage losses are large and lasting, then
targeted help for displaced workers may be called for. David Kaplan,
Gabriel Martínez, and Raymond Robertson study the issue for the case of
Mexico, using an administrative data set that allows them to follow indi-
vidual workers over a period of thirty-two quarters in four regions that varysignificantly in labor market conditions. They focus on the differences in
institutions, inequality, and labor market conditions that may explain dif-
ferences in wage behavior after displacement.
One striking result is the heterogeneity of worker experiences, which
range from large wage losses to many instances of gains after displacement.
This is consistent with earlier results for other countries, but it cannot be
attributed to differences in institutions (rates of unionization) or inequal-
ity, which are quite similar across Mexico. Rather, Kaplan, Martínez, andRobertson argue that labor market conditions, which vary quite a bit across
time and regions within Mexico, explain the heterogeneity of experiences.
In good times and in the most economically active regions, postdisplace-
ment wages are generally higher than they were in the previous jobs. How-
ever, workers who are fired during times of high unemployment and in
less economically active regions experience lasting effects on wages. If any
public assistance is to be disbursed, Kaplan, Martínez, and Robertson
argue, it should go to these workers.All papers but one included in this issue were presented at the panel
meeting held in San José, Costa Rica, in October 2004. The local hosts, and
particularly Juan Rafael Vargas, provided much help. As usual, associate
editors of Economía, members of the 2004 panel, and outside discussants
and referees have done an outstanding job. Thanks is due to them all.
Andrés Velasco xiii
8/17/2019 Andres Velasco Economia
15/257
8/17/2019 Andres Velasco Economia
16/257
The Electoral Consequences of the
Washington Consensus
No country in Latin America escaped the dictums of the Washington
Consensus. From Brazil under left-leaning Fernando Henrique Car-
doso to Mexico under ultra-orthodox economist Ernesto Zedillo
and Peru under Alberto Fujimori’s yoke, macroeconomic imbalances were
brought under control, barriers to international trade were lifted, and state-
owned enterprises were privatized. Whether this one-size-fits-all prescrip-
tion was imposed from outside or adopted at will by the governments elected
on the promise of improving the lot of their peoples may be a matter of debate. But all sides seem to agree on one point: the results did not meet
the expectations created both by outsiders and by those in power.
Up to the mid-1980s only two countries in Latin America had adopted
a package of policies similar to what became to be known as the Wash-
ington Consensus at the turn of the decade. Those two were undemocratic
Chile and impoverished Bolivia, by then among the most politically and
economically unstable countries, if not in the world, then certainly in Latin
America. Extreme cases, extreme policies: that was a common interpreta-tion of the two experiences. Less common was the expectation that those
policies were about to be adopted by virtually every Latin American coun-
try in the next few years, both those in which democracy had been the rule
1
E D U A R D O L O R A
M A U R I C I O O L I V E R A
Lora and Olivera are with the Inter-American Development Bank.
We are grateful for the valuable research assistance of Carlos Andrés Gómez. We also
thank Benito Arrunada, Mauricio Cárdenas, Stephen Kay, Ugo Panizza, Andres Rodríguez-
Clare, Mariano Tommasi, Jessica Wallack, and seminar participants at ISNIE-University
Pompeu Fabra, LACEA-PEG, Econnet-IADB, and the Economia panel meeting for com-ments and suggestions. We are especially grateful to Rafael Di Tella, Sebastian Galiani, and
Ernesto Dal Bó for their detailed and very useful comments and suggestions. We would like
also to thank Sebastian Saiegh for allowing us to use his data on political coalitions.
8/17/2019 Andres Velasco Economia
17/257
for decades, like Colombia, Costa Rica, and Venezuela, and those where
the third wave of democratization was just arriving, such as Argentina,
Brazil, and Uruguay.
The years of high expectations, both about democratization and about
Washington Consensus–type policies, are over. Latin Americans are still
convinced democrats, but enthusiasm has waned. Three out of every four
Latin Americans see democracy as the best form of government—or rather,
as the least bad, since 68 percent think that democracy is not function-
ing well in their countries. Latin Americans are even more sceptical about
the benefits of promarket economic policies. Only one out of four Latin
Americans considers privatization to have been beneficial for his or her coun-
try and barely 16 percent think that the market economy is doing a good job.1
Malaise is getting the upper hand in a number of places. Electricity and
water privatizations were blocked in Arequipa (Peru) and Cochabamba
(Bolivia), following violent clashes between vociferous opponents and the
police. An ambitious project to attract foreign direct investment to
Bolivia’s gas sector was derailed by the Indian communities. While these
events may be dismissed as isolated expressions of popular feeling, a new
crop of presidents from Néstor Kirchner in Argentina to Lucio Gutiérrez
in Ecuador and Tabaré Vásquez in Uruguay has won clear majorities inpopular elections after campaigning against the excesses of market-
oriented policies.
In an attempt to establish whether this malaise is justified or not, econ-
omists have devoted substantial effort to assessing the economic and social
consequences of the Washington Consensus policies. The dominant view
seems to be that they have had positive effects on economic growth and
income levels, though there is intense debate over the size of those effects,
over whether they are transient or permanent, and over the importance of
each of the components of the Washington Consensus. The dominant viewalso holds that the effects have been muted by lack of regulatory and insti-
tutional support for the liberalization efforts, though the specific forms of
regulation and institutions necessary for that purpose are far from clear.
Even more intense is the debate over the social and distributional effects of
fiscal stabilization and promarket reforms, which are the two main pillars of
the Washington Consensus.2
2 E C O N O M I A , Spring 2005
1. Opinion data come from the 2003 issue of Latinobarómetro, a public opinion surveyconducted by the Corporación Latinobarómetro, Santiago, Chile.
2. These debates are surveyed in Lora and Panizza (2002); Kuczynski and Williamson
(2003); and Lora, Panizza and Quispe-Agnoli (2004).
8/17/2019 Andres Velasco Economia
18/257
However, the future of these policies will depend not so much on their
efficacy but on whether they receive the support of the electorate. On this,
the state of knowledge is much more scant and fragmentary, as will be seen
below. This paper attempts to help fill that vacuum by evaluating through
econometric methods the electoral consequences of the Washington Con-
sensus. Although our approach is backward looking, it sheds considerable
light on the future. Our study shows that the electorate cares not only about
the outcomes of the policies (maybe about only some outcomes and not
others), but also about the policies themselves, irrespective of whether they
produce good or bad (observable) outcomes. In addition, the electorate
seems to care about whether the policies adopted by a government are
in line with the ideology of the incumbent’s party and with preelectoral
promises. Furthermore, in presidential regimes voters cast separate votes
for the executive and the legislature, and outcomes and policies affect each
vote differently. The presidential vote is more volatile and more suscepti-
ble to economic outcomes and policies, but votes for the legislature are not
completely immune: policies in which the legislature clearly plays a role,
such as privatizations, tend to have electoral consequences. These results
provide a nuanced landscape for the future of Washington Consensus poli-
cies, where neither bold backslashes nor aggressive promarket reformsshould be expected in the future. Not only is the time of high expectations
over; perhaps the time for deep reforms is also past.
In the next section of the paper we present a short survey of the literature
assessing the electoral consequences of the Washington Consensus policies
and derive our empirical hypotheses. On that basis, we then discuss the the-
oretical and econometric approaches that support the empirical analysis.
In subsequent sections we describe the data, present the econometric find-
ings, and discuss our conclusions.A note on terminology is in order before proceeding. “Neo-liberal,”
“market-oriented,” “orthodox,” and a variety of other labels have been
attached to the set of economic policies in vogue since the early nineties in
Latin America and elsewhere. We use these terms interchangeably, but not
loosely: for the sake of clarity and brevity, this paper deals with the ten poli-
cies summarized in the classic article by Williamson that made the term
“Washington Consensus” famous.3 We assume that all those labels refer to
that same set of policies (as detailed below in the section titled “Data”).
Eduardo Lora and Mauricio Olivera 3
3. Williamson (1990a).
8/17/2019 Andres Velasco Economia
19/257
Review of the Empirical Literature and Some Testable Hypotheses
The most straightforward view of the response of the electorate to eco-nomic policies is based on the “economic voting” argument: people base
their electoral decisions on cost-benefit calculations. If the policies bring
net benefits to them, they cast their votes to support the government, or the
party, administering those policies; if the policies bring losses to them,
they lend their support to the candidate, or the party, opposing them. Eco-
nomic voting is usually assumed to be retrospective: voters observe past
performance and assume that past trends will persist into the future if the
government or the party remains in power. If those trends are deemed
acceptable, given a set of standards or expectations voters decide to reelect
the incumbent, or his party if the option of reelection does not exist. There-
fore, in retrospective economic voting policies play no direct role, since
voters decide entirely on the base of past outcomes.4
Considerable evidence from advanced industrial democracies supports the
view that past economic performance influences people’s voting decisions
and their support for governments.5 An important empirical finding from this
literature is that voters base their decisions on aggregate (or “sociotropic”)
economic outcomes such as growth, inflation, and unemployment, ratherthan on individual (or “pocketbook”) outcomes. Most of the empirical liter-
ature on developed countries comes from single-country analyses, based
either on time-series electoral outcomes or public opinion polls. The eco-
nomic voting hypothesis is more robust for public opinion polls than for
actual electoral outcomes.6 Empirical studies of electoral behavior in the
United States using state-level data lend support to the simple economic
voting hypothesis, in the sense that voters are able to evaluate their state’s
economic performance relative to that of the national economy. Further-more, they (irrationally) reward state governors for economic fluctuations
that are unrelated to gubernatorial actions, which implies that they have lim-
ited ability to filter aggregate economic information.7 The ability of voters
4 E C O N O M I A , Spring 2005
4. Stokes (2001b, pp. 1–18) provides a concise review and discussion of the theoretical
underpinnings of retrospective economic voting.
5. Based on the seminal work by Downs (1957); among the initial papers on economic
voting in the United States are Kramer (1971); Meltzer and Vellrath (1975); and Arcelus and
Meltzer (1975).6. Lewis-Beck (1988) is a salient example of the early empirical literature based on opin-
ion polls in European countries. For a review of this literature, see Stokes (2001b, pp. 2–8).
7. Wolfers (2002).
8/17/2019 Andres Velasco Economia
20/257
to gather and update information is a central issue in the theoretical and
empirical literature on economic voting.8 Although some evidence points
out to the presence of prospective as well as retrospective behavior, uncer-
tainty about the workings of the economy and the relatively high cost of
gathering and processing the information necessary to forecast outcomes
is consistent with the importance of retrospective voting in the empirical
findings.9
Empirical support for the economic voting hypothesis in Latin America
has been uncovered by Karen Remmer, Michael Coppedge, Kenneth Roberts
and Erik Wibbels, and Susan Stokes.10 A concise summary of these findings
is presented in table 1. Based on data for twenty-one competitive elections
between 1982 and 1990, Remmer has found that conditions of economic
crisis undermine support for incumbents and provoke high levels of elec-
toral volatility.11 The magnitude of the electoral change is found to be asso-
ciated with the depth of the crisis during the campaign period, with variations
in exchange rates, GDP, and inflation highly correlated with various
indicators of electoral outcomes. Her results also suggest that the effect
of economic conditions on electoral instability are mediated by the struc-
ture of the party system (insulating two-party systems from the volatility
experienced by more fragmented systems). However, as Stokes pointsout, these results are anomalous given the predictions of normal economic
voting, as she “finds that incumbent parties suffered larger losses at the
polls when inflation went down (significant) and when GDP rose (not
significant).”12
In a subsequent paper, Remmer presents new estimates on the influence
of inflation and growth on the incumbent vote in presidential elections.13
Her new database covers forty-nine elections for seven countries between
1983 and 1999. Her results indicate that after controlling for the advantageof incumbency as well as major differences in the structure of party sys-
tems, electoral outcomes are strongly influenced, in the direction expected,
by macroeconomic performance in the year before the election. That is, infla-
tion is found to be negatively correlated with electoral support, whereas
Eduardo Lora and Mauricio Olivera 5
8. For a review of this debate, see Duch and Stevenson (2004); and Keech (1995).
9. On prospective behavior, see, for instance, Lewis-Beck (1988).
10. Remmer (1991, 2003); Coppedge (2001); Roberts and Wibbels (1999); Stokes (2001b).11. Remmer (1991).
12. Stokes (2001b, p. 27).
13. Remmer (2003).
8/17/2019 Andres Velasco Economia
21/257
growth is positively correlated with it. Furthermore, inflation is significant
in all the regressions presented, while growth is more significant for the
elections held in the 1990s than for those in the 1980s, indicating that the
sensitivity of the electorate to economic performance has increased rather
than waned over time.Coppedge’s empirical work focuses on the impact of changes in inflation
on legislative vote shares. His dependent variable consists of 132 changes in
legislative vote shares for major parties in eleven countries from 1978 to
1995. His only indicator of economic performance is the change in (the
log of) inflation from the last year of the previous government to the last
year of the current government. By interacting this variable with appropri-
ate dummies, Coppedge finds that changes (whether increases or decreases)
in inflation affect electoral support for the incumbents’ parties in theexpected way, while only increases in inflation improve the vote share of
the opposition parties. However, these results apply only to parties “with a
fluid base,” that is, parties that do not count on a strong party identification.
When there is such identification, voters are reluctant to question their
party identification on the basis of macroeconomic outcomes.
Roberts and Wibbels consider economic voting as a possible explana-
tion of electoral volatility in Latin America. Their database includes fifty-
eight congressional elections and forty-three presidential elections in sixteen
Latin American countries during the 1980s and 1990s. Their results show
that economic performance has an effect on electoral stability. Economic
growth stabilizes partisan support in legislative elections, whereas sharp
6 E C O N O M I A , Spring 2005
T A B L E 1 . Summary of Empirical Findings on Economic Voting in Latin America
Election type
(number of Estimation
Study Dependent variable countries) Period method Main results
Remmer (1991)
Remmer (2003)
Roberts and
Wibbels (1999)
Coppedge (2001)
Stokes (2001b)
Source: Authors’ calculations.a. Not significant.
Electoral volatility
Vote shares
Electoral volatility
Vote shares
Probability of a
security-oriented
candidate beingelected
Inflation −; GDPgrowth +a
Inflation −; GDPgrowth +
Inflation −a; GDPgrowth +
Inflation
Inflation −; GDPgrowth +
Presidential (12)
Presidential (8)
Legislative and
presidential (16)
Legislative (11)
Presidential (15)
1982–90
1983–99
1980–97
1978–95
1982–95
Pooled OLS
Pooled OLS
Pooled OLS
Pooled OLS
Probit
8/17/2019 Andres Velasco Economia
22/257
changes in the rate of inflation from one administration to the next, whether
positive or negative, produce the opposite effect. Short-term inflation influ-
ences support for incumbent presidents, but growth changes have only a
weak effect on the vote for incumbents, “which suggests that voters are
more inclined to hold them directly accountable for monetary stability
than economic growth.” Although electoral volatility is influenced by eco-
nomic performance, it is also related to the institutional characteristics of
political regimes and party systems, and to the structure and organization
of class cleavages.14
In her study of “neoliberalism by surprise,” Stokes uses data from
twenty-three elections in the 1980s and 1990s in order to assess how the
electorate judges incumbents who, having campaigned for stability-oriented
or protectionist policies, once in office switch to market-oriented ones. She
finds that for both, “switchers” and “non-switchers,” economic growth
and inflation affect their vote share in the expected ways. Furthermore,
voters are more sensitive to economic outcomes in the case of “switchers,”
although this result is not statistically significant (more on these results
below).15
These empirical studies taken together lend support to the retrospective
economic voting argument in both presidential and legislative elections.They make clear that voting decisions are also influenced by political, insti-
tutional, and structural factors and that some of these factors may influence
the severity with which voters judge economic outcomes. Therefore, based
on these studies, two testable propositions are derived:
1. Electoral support for the incumbent’s party is higher, the better the
aggregate economic outcomes during his or her administration.
2. The sensitivity of electoral support to economic outcomes depends on
the institutional characteristics of the political regime and the party system.As mentioned, in normal economic voting only past outcomes influence
people’s views. However, as in all six of the Stokes case studies on mar-
ket reforms in new democracies, people sometimes react to economic
deterioration by supporting the government more strongly; and conversely,
they sometimes respond to economic improvements with pessimism and
opposition.16 Normal economic voting is not the only pattern, especially in
the process of deep economic reform. If there are good reasons to believe
Eduardo Lora and Mauricio Olivera 7
14. Roberts and Wibbels (1999), quote from p. 584.
15. Stokes (2001a).
16. Stokes (2001a).
8/17/2019 Andres Velasco Economia
23/257
that past circumstances are not good indicators of the future, information
other than past economic outcomes may influence people’s electoral deci-
sions. For instance, voters may recognize that past circumstances were
affected by factors beyond the government’s control and exonerate the
incumbent from the responsibility for past declines in their welfare. Vot-
ers may then forecast their future welfare as a function of government pol-
icy, rather than as an extrapolation of the past. This sounds simpler than
it is, of course, because future government policies are unknown and
because the relationship between policies and outcomes is diffuse. Peo-
ple’s expectations of future policies may be formed on the basis of the
policies adopted or announced by the incumbent or on the basis of his
party’s ideology. These policy expectations may then be translated into
expected outcomes through a set of beliefs and hypotheses about their pos-
sible consequences.
It is often implicitly assumed that people’s (average) beliefs conform
to the actual functioning of the real world. If that is so, assessing the
effects of economic policies would help explain voters’ electoral decisions.
Economists have devoted considerable effort to evaluating the impact of
Washington Consensus policies on economic growth, income distribu-
tion, employment levels, and a host of other variables.17
However, therehas been no comparable effort to examine whether these results are con-
sistent with how the electorate responds to those policies. The only study
on the subject, by Carlos Gervasoni, has found positive correlations
between several indicators of heterodox (that is, anti-neoliberal) policies
and losses in the vote shares of the parties of the incumbents who adopted
those policies.18 The variable with the largest and most significant effect is
money supply growth. Import protection indicators are also significant,
whereas fiscal deficit and the share of the state in GDP are not significant.These results suggest that Washington Consensus policies do not entail
electoral costs and may even produce electoral benefits, probably because
they bring positive economic effects. It is suggestive that the most signif-
icant policy variable is the money supply, because it is well known that
inflation is, ultimately, a monetary phenomenon, and as mentioned, empiri-
cal evidence suggests that inflation is a key economic outcome influencing
electoral decisions.
8 E C O N O M I A , Spring 2005
17. For surveys of the literature, see Inter-American Development Bank (2003, chap. 5);
Kuczynski and Williamson (2003); and Lora and Panizza (2002).
18. See Gervasoni (1997), citing a 1995 study.
8/17/2019 Andres Velasco Economia
24/257
However, it is a great leap of faith to assume that people’s beliefs con-
form to the actual consequences of policies. In mapping policies on out-
comes, ideology and leaders’ opinions may be more important for most
people than their limited understanding of how policies work their influence
through the social and economic structures to affect production, employ-
ment, or income distribution. Evidence on how those factors influence
electoral responses to economic policies is very scant. However, in-depth
case studies on Argentina and Venezuela by Javier Corrales clearly show
that the reaction of the electorate to the adoption of neoliberal economic
policies in the 1990s was mediated by the party structure and other insti-
tutional factors.19 The cohesion and tactics of the Partido Peronista help
explain the electorate’s support of the neoliberal reforms in Argentina in
the early 1990s, as well as their demise a decade later. Venezuela’s Acción
Democrática lacked that cohesion, and its reforms were soon rejected by
the electorate.
If voters care about policies and not only about past outcomes, the pol-
icy announcements of presidential candidates will be a key source of infor-
mation. However, campaign promises are often poor predictors of actual
policy: according to Stokes, of the thirty-three Latin American govern-
ments that adopted promarket reforms between 1982 and 1995, only abouthalf (seventeen) hinted during their campaigns that such reforms were
going to be implemented.20 This raises several empirical issues. First, do
policy announcements in fact influence electoral decisions? Empirical evi-
dence from the United States and other advanced industrialized economies
shows that they do: people seem to base their opinions in part on campaign
announcements, and voters punish ambiguous campaigns.21 Of course,
some promises may resonate more than others, depending on, among other
things, economic circumstances. For thirty-eight Latin American electionsin the 1980s and 1990s, Stokes finds that stability-oriented candidates (as
opposed to market-oriented ones) stand a better chance of being elected,
the lower the rates of GDP growth and inflation.22
A second empirical issue is whether deviating from campaign promises
carries electoral costs for the incumbent. Although deviations may in
principle be costly, they may produce a positive payoff if they signal the
Eduardo Lora and Mauricio Olivera 9
19. Corrales (2002).20. Stokes (2001a).
21. For a brief review of this topic, see Stokes (2001a, pp. 4–5).
22. Stokes (2001a, pp. 93–97).
8/17/2019 Andres Velasco Economia
25/257
incumbent’s commitment to achieving highly desirable economic out-
comes at the expense of more immediate partisan support.23 According to
Stokes, deviating from campaign promises does carry electoral costs,
although only weakly.24 However, since her estimates control for eco-
nomic outcomes, this result implies that policy switches may still have a
positive electoral payoff if the new policies bring substantial economic
improvement. Neoliberalism by surprise may still be a good political
strategy.25
A common theme in the literature on economic voting is the conditional
nature of voters’ responses to economic outcomes and policies. As men-
tioned, the severity of their judgment depends on their attachment to the
party in power, the structure of the party system, and other institutional
considerations. It also depends, although weakly, on whether the policies
adopted by the incumbent are in line with his campaign pronouncements.
An additional variation on this theme holds that the electorate is better pre-
pared to support untested policies, even if they may cause short-term duress
or if they run counter to established beliefs, when economic conditions have
deteriorated.26 However, once conditions improve or simply stabilize, tol-
erance subsides and support for further reforms wanes. Therefore, while
uncertainty is welcome at the outset of the reform process, certainty is thekey factor for its consolidation. Based on case studies of Peru and Argentina,
Kurt Weyland offers persuasive evidence that the public was supportive to
the reform process while there was a perception of acute economic crisis.27
Even though the reformers were reelected, support for their economic pro-
grams was already diminishing. Corrales endorses this view in his analysis
of the reform process in Argentina and Venezuela, although he acknowl-
edges that in the latter case support for reform was never very strong.28
Therefore, the literature on economic voting suggests that policies, notonly outcomes, may influence electoral decisions. As with outcomes, voters’
position with respect to policies may be mediated by a host of factors,
including ideological considerations, policy pronouncements during the
10 E C O N O M I A , Spring 2005
23. For a theoretical approach, see Cukierman and Tommasi (1998).
24. Stokes (2001a, p. 95).
25. Cukierman and Tommasi (1998); Navia and Velasco (2003).
26. This behavioral hypothesis is based on seminal work by Thaler and others (1997),
Kahneman and Tversky (1979), and Tversky and Kahneman (1991), who find that people aremore prone, even eager, to assume risks after experiencing losses.
27. Weyland (2002).
28. Corrales (2002).
8/17/2019 Andres Velasco Economia
26/257
electoral campaign, and the state of the economy at the time of elections.
This gives rise to the following additional testable propositions:
3. Electoral support for the incumbent’s party depends on the economic
policies adopted. Policies may carry electoral costs even when they deliver
good economic outcomes.
4. The electorate’s tolerance of unpopular policies depends on the ide-
ology of the incumbent’s party, his or her campaign statements, and the
initial state of the economy.
Empirical Approach
None of the empirical literature just reviewed offers a full-fledged theo-
retical model of electoral behavior, and we have no intention of providing
one. However, the series of hypotheses arising from that literature can be
organized in a simple framework such that the persistence of the vote for
the incumbent’s party is a function of a vector of economic outcomes and
a vector of policies (both relative to their past values):
where V t and V
t −1are the share of the vote for the incumbent’s party at the
end and the beginning, respectively, of its term in office; X t
and X t −1
are the economic outcomes at the time of each election; and Pt and P
t −1
are the policies at those two moments. A is the set of other parameters that
may influence the stability of the vote for the party in office, and ut is an
error term. β and γ are our parameters of interest. In this simple frame-
work, hypothesis 1 states that β is positive for economic outcomes that aredesirable, such as growth, or negative for undesirable ones, such as infla-tion or unemployment (and assumes that γ is zero, since it ignores theinfluence of policies). Hypothesis 2 postulates that β is a function of somefeatures of the political system, such as party fragmentation or the ideo-
logical polarization of the party system. The stronger these features, the
higher the electorate’s response to the economic outcomes. Hypothesis 3,
which postulates that the electorate cares about the choice of policies,
implies that γ is not zero but probably negative if the policies are marketoriented. Finally, hypothesis 4 states that some aspects of the political
and economic context when the incumbent’s party was initially elected
may affect the way the electorate judges the adoption of policies. This
V
V A
X
X
P
P u
t
t
t
t
t
t
t
− − −= ∗
∗
∗1 1 1
β γ
,
Eduardo Lora and Mauricio Olivera 11
8/17/2019 Andres Velasco Economia
27/257
hypothesis can be incorporated in our framework by assuming that γ is afunction of those factors. More specifically, γ will be smaller (in absolutevalue) when the policies adopted were those announced by the incumbent
during his election campaign, when they are in line with his party’s ideol-
ogy, or when the economy started from a situation of crisis. Although our
framework is general enough to test further hypotheses, due to sample size
limitations and for the sake of parsimony and tractability, we restrict its
application to the hypotheses identified in the literature review.
Our economic voting framework is relevant both for presidential and
for legislative elections. An important feature of presidential systems is the
separation of powers between the legislative and the executive, aimed at
imposing checks and balances in order to discipline parties and make themaccountable.29 Since checks and balances force the two powers to agree on
policies, voters should be expected to pass judgment on the performance
of the incumbent’s party in both branches on the basis of economic out-
comes and policy decisions. Of course, we should expect that the influence
of each policy on presidential vis-à-vis legislative elections will depend on
whether such policy is controlled exclusively by the executive or not. While
legislatures have very little influence on monetary, exchange rate, and tariff
policies in most Latin American countries, they do have a strong (even over-riding) influence on tax policies, privatization decisions, and the regulation
of financial, capital, and labor markets. As Brian Crisp and Gregg Johnson
show, contrary to widespread belief, Latin American legislatures make use
of their powers to influence the timing and depth of promarket reforms.30
And according to Roberts and Wibbels, the electorate holds each branch of
power more accountable for some outcomes than for others.31 When assess-
ing the role of the legislature in policy decisions in Latin America, it is impor-
tant to keep in mind that the incumbent’s party (or the coalition of parties
backing the incumbent) usually holds the majority in that body (see below).
To estimate the relevant parameters, the previous expression can be
written in logs as
where d log(V t ) corresponds to the change in (the log of) the share of votes
for the incumbent party between t, the time when its performance is eval-
d V F d X d pt t t t log log log log ,( ) = + ( ) + ∗ ( ) + ∗ ( ) +α ψ β γ ε
12 E C O N O M I A , Spring 2005
29. Persson, Roland, and Tabellini (1997).
30. Crisp and Johnson (2003).
31. Roberts and Wibbels (1999).
8/17/2019 Andres Velasco Economia
28/257
uated, and t − 1, when it was elected for office; d log( X t ) and d log(P
t ) are
the changes in (log measures of the) outcomes and policies, respectively;
εt is equivalent to log(u
t ); and
α + ψ log(F ) is equal to log( A), with
αas a
constant parameter and F as a set of political control variables.
We estimate separate models for presidential and legislative elections
with panel data for seventeen countries starting from the mid-1980s
described below. Potential problems of heteroscedasticity and endogene-
ity need to be addressed in this type of specification. The former may arise
from country or party heterogeneity and is dealt with by the use of White
robust standard errors. The endogeneity problem stems from potential omitted
variables, since differentiating countries solely by the economic and policy-
related variables included in sets X and P may not capture all the sources of heterogeneity.32 This is partly dealt with by the inclusion as controls of a set
of political variables (represented by F ). However, other country-related
factors might bias the estimations if they are correlated with the explanatory
variables. To take care of this problem, we run all the regressions with coun-
try fixed effects (although, admittedly, our sample size is too small to get
precise estimation of these effects).33 The fixed effects estimator is
where C is the set of country dummies.
Data and Sources
Table 2 presents the structure of our database, and table 3 shows correlations
between the more relevant variables. The database includes a total of sixty-
six presidential elections and eighty-one legislative elections in seventeen
d V F d X d p C t t t t t log log log log ,( ) = + ( ) + ∗ ( ) + ∗ ( ) + +α ψ β γ λ ε
Eduardo Lora and Mauricio Olivera 13
32. We assume that the two other sources of endogeneity—reverse causality and measure-
ment error—are not latent in our model. Reverse causality is not a concern, since voters
evaluate the incumbent’s behavior after policies and outcomes are known. Measurement error
problems may be present, depending on the actual process of expectations formation. However,
ample empirical evidence provides support for the hypotheses of retrospective voting, which
for our framework implies that expectations are formed on the basis of past outcomes only.
33. All the regressions were also run without fixed effects: while virtually all the conclu-
sions are the same, in these regressions, some of the explanatory variables (especially those
measuring promarket policies) show higher levels of significance. We have also run the re-gressions including a common time trend, or including five-year period fixed effects, with-
out any important divergence from the results presented below. Results are available upon
request from the authors.
8/17/2019 Andres Velasco Economia
29/257
T A B L E 2
.
S t r u c t u r e o f t h e D a t a S e t
N u m b e r o f e l e c t i o n s ( p e
r i o d )
N u m b e r o f p a r t i e s t h a t h e l d . . .
F r a g m e n t a t i o n a
C o u n t r y
P r e s i d e n t i a l
L e g
i s l a t i v e
P r e s i d e n c y
L a r g e s
t s h a r e i n t h e l e g i s l a t u r e
M e a n
M i n i m u m
M a x i m u m
P o l a r i z a t i o n i n d e x
A r g e n t i n a
3 ( 1 9 8 9 –
9 9 )
8 ( 1 9
8 5 –
9 9 )
2
3
2 . 7
7
2 . 3
0
3 . 0
6
0 . 2
3
B o l i v i a
4 ( 1 9 8 5 –
9 7 )
4 ( 1 9
8 5 –
9 7 )
3
3
4 . 0
6
3 . 4
2
5 . 0
8
0 . 5
2
B r a z i l
3 ( 1 9 8 9 –
9 8 )
4 ( 1 9
8 6 –
9 8 )
2
2
6 . 6
0
2 . 7
6
8 . 2
7
0 . 2
5
C h i l e b
3 ( 1 9 8 9 –
9 9 )
4 ( 1 9
8 9 –
2 0 0 1 )
2
2
4 . 9
0
4 . 8
4
4 . 9
9
0 . 1
6
C o l o m b i a
5 ( 1 9 8 6 –
0 2 )
5 ( 1 9
8 6 –
9 8 )
2
1
2 . 6
6
2 . 2
1
3 . 0
9
0 . 1
6
C o s t a R i c a
5 ( 1 9 8 6 –
0 2 )
5 ( 1 9
8 6 –
2 0 0 2 )
2
2
2 . 3
1
2 . 2
1
2 . 5
6
0 . 4
2
D o m i n i c a n R e
p u b l i c
5 ( 1 9 8 6 –
2 0 0 0 )
4 ( 1 9
8 6 –
2 0 0 0 )
3
3
2 . 4
8
2 . 1
8
2 . 8
8
0 . 5
5
E c u a d o r
4 ( 1 9 8 8 –
9 8 )
7 ( 1 9
8 6 –
9 8 )
4
2
6 . 0
5
4 . 2
9
7 . 5
6
0 . 3
6
E l S a l v a d o r
4 ( 1 9 8 4 –
9 9 )
6 ( 1 9
8 5 –
2 0 0 0 )
2
2
2 . 6
8
2 . 4
1
3 . 0
6
0 . 3
9
G u a t e m a l a
4 ( 1 9 8 5 –
9 9 )
5 ( 1 9
8 5 –
9 7 )
4
4
3 . 3
1
2 . 3
5
4 . 4
4
0 . 2
4
H o n d u r a s
5 ( 1 9 8 5 –
2 0 0 1 )
5 ( 1 9
8 5 –
2 0 0 1 )
2
2
2 . 1
8
2 . 0
0
2 . 5
8
0 . 4
2
M e x i c o
3 ( 1 9 8 8 –
2 0 0 0 )
6 ( 1 9
8 5 –
2 0 0 0 )
2
1
2 . 3
8
1 . 8
5
2 . 8
2
0 . 3
2
N i c a r a g u a
3 ( 1 9 9 0 –
2 0 0 1 )
3 ( 1 9
9 0 –
2 0 0 1 )
1
2
2 . 0
5
2 . 0
5
2 . 0
5
0 . 5
8
P e r u
4 ( 1 9 8 5 –
2 0 0 0 )
4 ( 1 9
8 5 –
2 0 0 0 )
3
4
3 . 8
0
2 . 5
0
5 . 8
3
0 . 5
1
P a r a g u a y
4 ( 1 9 8 9 –
2 0 0 3 )
4 ( 1 9
8 9 –
2 0 0 3 )
1
1
2 . 2
1
1 . 8
8
2 . 5
4
0 . 4
0
U r u g u a y
3 ( 1 9 8 4 –
9 9 )
3 ( 1 9
8 4 –
9 9 )
2
2
3 . 1
9
2 . 9
2
3 . 3
2
0 . 4
2
V e n e z u e l a
4 ( 1 9 8 8 –
2 0 0 0 )
4 ( 1 9
8 8 –
2 0 0 0 )
4
1
3 . 9
2
2 . 3
4
5 . 7
9
0 . 3
0
T o t a l o r a v e r a
g e
6 6
8 1
2 . 4
2 . 2
3 . 3
9
2 . 6
2
4 . 1
1
0 . 3
7
S o u r c e : P a y n e a n d o t h e r s ( 2 0 0 2 ) , c o m p l e m e n t e d w i t h t h e
P o l i t i c a l D a t a b a s e o f t h e A m e r i c a s ( O r g a n i z a t i o
n o f A m e r i c a n S t a t e s a n d G e o r g e t o w n U n i v e r s
i t y ) .
a .
E f f e c t i v e n u m b e r o f p a r t i e s i n t h e l e g i s l a t u r e .
b .
I n C h i l e , t h e e f f e c t i v e n u m b e r o f p a r t i e s d i f f e r s f r o m t
h e
n u m b e r o f c o a l i t i o n s ( C o n c e r t a c i ó n a n d A l i a n z a p o r C h i l e ) , w h i c h a r e c l o s e t o 2 i n e f f e c t i v e t e r m s a n d o f w h i c h o n l y C o n c e r t a c i ó n h a s h e l d t h
e p r e s i d e n c y .
8/17/2019 Andres Velasco Economia
30/257
T A B L E 3
.
C o r r e l a t i o n s
I n fl a t i o n ( l o s s
I n s t i t u t i o n a l
V o t e s
G r o w t h
o f p u r c h a s i n g
U n e m p l o y m e n t
G i n i i n d e x
M a c r o i n d e x
S t r u c t u r a
l i n d e x
i n d e x
( s h a r e )
F r a g m e n t a t i o n
P o l a r i z a t i o n
P r o m i s e s
I d e o l o g y
( l o g , c h a n g e )
p o w e r , c h a n g e )
( c h a n g e )
( c h a n g e )
( l o g , c h a n g e )
( l o g , c h a n g e )
( l o g , c h a n g e )
P r e s i d e n t i a l e l e c t i o n s
V o t e s ( s h a r e )
1 . 0
0
F r a g m e n t a t i o
n
− 0 . 3
2
1 . 0
0
P o l a r i z a t i o n
− 0 . 2
0
− 0 . 0
9
1 . 0
0
P r o m i s e s
− 0 . 1
6
− 0 . 0
5
− 0 . 3
2
1 . 0
0
I d e o l o g y
− 0 . 1
6
0 . 0
3
− 0 . 4
5
0 . 3
7
1 . 0
0
G r o w t h ( l o g , c h a n g e )
0 . 2
1
− 0 . 0
2
− 0 . 0
6
0 . 2
5
0 . 1
4
1 . 0
0
I n fl a t i o n ( l o s s
o f
p u r c h a s i n g
p o w e r , c h a n g e )
− 0 . 2
4
− 0 . 5
5
0 . 3
3
0 . 0
2
− 0 . 1
6
−
0 . 4
1
1 . 0
0
U n e m p l o y m e
n t ( c h a n g e )
− 0 . 0
2
0 . 1
3
− 0 . 4
6
− 0 . 1
6
0 . 2
8
−
0 . 2
7
− 0 . 4
8
1 . 0
0
G i n i i n d e x ( c h
a n g e )
− 0 . 2
2
− 0 . 1
6
− 0 . 1
9
0 . 2
1
0 . 0
9
0 . 2
3
0 . 1
Top Related