The Price of Political Opposition: Evidence from Venezuelas Maisanta

1
The Price of Political OppositionEvidence from
Venezuelas Maisanta

• Chang-Tai Hsieh, U.C. Berkeley and NBER
• Edward Miguel, U.C. Berkeley and NBER
• Daniel Ortega, IESA
• Francisco Rodríguez, Wesleyan University

2
Political Polarization and the Economy

• How do political polarization and conflict affect
  economic performance?
• -- Patronage and clientelism are salient issues
• Do individuals who join the opposition pay a
  price?
• What are the economic impacts for firms,
  individuals, and society as a whole?
• -- Existing work on the economic impacts of
  political conflict, instability is based on
  cross-country regressions
• Our case Hugo Chávezs Venezuela

3
Venezuelas Maisanta List

• Unique data on individual political actions all
  registered voters (12 million) signatures on
  recall petitions
• -- This provides a measure of real-world
  political behavior for the whole population, not
  just elites
• -- The data became widely available within
  Venezuela
• A preview of our main findings
• -- Pro-opposition firms pay higher taxes (33)
  and get less access to foreign exchange (-49)
• -- These and other policy distortions led to a
  5 decline in aggregate manufacturing
  productivity (lower bound)
• -- Lower earnings for pro-opposition individuals

4
Valuing political connections, preferences

• Several studies estimate the economic
  benefits/cost of political connections and
  preferences
• -- Fisman (2001) study the value of links to
  Suharto in Indonesia, Khwaja and Mian (2006)
  estimate the rents to politicians in terms of
  securing bank loans in Pakistan
• -- Li et al (2007), Morduch and Sicular (2000)
  estimate returns to communist party membership in
  China
• -- Dunning and Stokes (2007) on social programs
• Most studies that study the economic impacts of
  political instability use cross-country
  regressions
• -- Alesina et al (1996) the average effect of a
  coup is
• -0.6 to -1.4 of aggregate output

5
Presentation outline

• Introduction
• Background on Venezuelan political economy
• The Maisanta database
• A simple model of petition signing
• Empirical results
• Data and matching
• Firm analysis
• Deriving aggregate TFP impacts
• Labor market analysis
• Discussion and future work

6
Hugo Chávezs Venezuela

• Venezuela has strong democratic traditions since
  the 1950s, and was spared the coups and violence
  that swept most of Latin America in the 1970s and
  1980s (Karl 1997, Corrales 2002)
• -- Venezuelas oil abundance is a defining
  characteristic
• -- Per worker GDP declined 32 between 1978-1998
• Hugo Chávez, a former army officer, won December
  1998 presidential elections with 56 of the vote
• -- Conventional wisdom Chávez stoked the
  resentment of the poor and is despised by the
  business elite

7
Hugo Chávezs Venezuela

• Chávez quickly moved to consolidate power in a
  new constitution, elections, extensive
  institutional reforms
• -- Recall referendum new to the 1999
  constitution
• A failed coup attempt in April 2002 contributed
  to increased political polarization
• -- Opposition mass demonstrations, National
  Strike (12/02-1/03), attempts to recall Chávez in
  2002-2004
• -- Chávezs popularity fell sharply throughout
  2002-2003 as the economy slumped

8
Timing of the signatures, recall referendum

• Three waves of recall efforts in 2002-2003
• (1) November 2002 1.57 million signatures for a
  referendum calling for Chávezs resignation
• -- Invalidated by the Supreme Court February
  2003 due to claims about forged signatures
• (2) August 2003 2.79 million signatures
  submitted for a Recall Referendum against Chávez
• -- Invalidated by the new National Electoral
  Council, despite being above the 20 threshold,
  since signatures were collected before the middle
  of Chavezs term in office (the constitutional
  requirement)

9
Timing of the signatures, recall referendum

• (3) December 2003 3.48 million signatures
  submitted to recall Chávez, in officially
  supervised signing booths
• -- Pro-government groups also submitted over 2
  million signatures to recall opposition leaders
  from Congress
• -- The National Electoral Council rejects 34 of
  opposition signatures, to be re-validated May
  2004
• In the meantime, Chávezs popularity rises in
  2004, with high oil prices and expanding social
  programs
• Recall Referendum finally held in August 2004
• -- 59 of voters oppose the recall, Chávez
  survives

10
Political lists posted on the internet

• January 2003 Pro-Chávez legislator Luis Tascón
  claims many signatures for the first petition
  were forged and posts the list of signers on his
  webpage
• -- Tascóns List was updated with later waves of
  petition signatures, through August 2004

11
Political lists posted on the internet

• January 2003 Pro-Chávez legislator Luis Tascón
  claims many signatures for the first petition
  were forged and posts the list of signers on his
  webpage
• -- Tascóns List was updated with later waves of
  petition signatures, through August 2004
• The Maisanta database (July 2004) contains all
  petition signers (pro-opposition and
  pro-government)
• -- Widely distributed to government electoral
  battle units, government offices, sold on the
  street in Caracas
• -- Quickly became well-known and politically
  salient
• -- 12.3 million registered voters, 77 of voting
  age adults

12
(No Transcript)
13
The Maisanta database interface
14
The Maisanta database interface
15
Uses of the petition information

• Immediate accusations the information was used to
  discriminate against firms, employees and job
  seekers
• Whoever signs against Chávez their name will be
  there, registered for history, because theyll
  have to put down their first name, their last
  name, their signature, their identity card
  number, and their fingerprint.
• -- Hugo Chávez, televised address, Oct. 17,
  2003
• Many anecdotes of public sector workers fired for
  signing
• -- Media accounts of the government using tax
  audits and fines against pro-opposition firms
• -- Also claims private pro-opposition firms
  required employees to bring proof of signing
  against Chavez

16
Uses of the petition information

• There are still places that use Tascon's List to
  determine who gets a job and who doesn't.
• -- Hugo Chávez, televised address, April 15,
  2005
• The database remains widely held and available
• -- Even beyond Maisanta, political affiliations
  are increasingly salient due to rising political
  polarization, and this is important for the
  interpretation of our results

17
A simple model of petition signing decisions

• Many factors could affect individuals signing
  decision
• -- The time costs of signing
• -- A taste for expressing ones political
  preferences
• -- Expected punishment from the government, and
  rewards from the opposition
• Political preferences are in turn a function of
  how people expect to fare under Chavez versus the
  opposition
• In the case of a secret ballot, expected
  punishments / rewards are not a factor (e.g., for
  the first petition signing round in 2002), but
  they are important for later petition rounds
  where people knew their names would be posted

18
A simple model of petition signing decisions

• We focus on the differences in post-2003 outcomes
  for signers and non-signers
• -- Selection bias is a concern if signers and
  non-signers expect to have different income
  trends if Chavez wins. In fact, these differences
  could be driving signing choices
• -- We attempt to address omitted variable bias
  using rich panel datasets on firms and
  individuals, with geographic, sectoral,
  demographic time trends
• Can we interpret these differences as the
  willingness to pay for dissident political
  expression?
• -- Only under the (empirically incorrect)
  assumption that everyone fully expected Chavez to
  win the referendum

19
Firm datasets

• Industrial Survey panel data for 1996 to 2004
• -- Approximately 2,300 plants (private, public)
  per round
• -- Census of plants with at least 100 employees,
  representative sample of plants with 5-99
  employees
• -- Data on 927 firms for 2001, 2002, and 2003
  rounds
• -- Revenue, sales, employment, taxes, profits

20
Economic outcomes in a polarized society

• Firm board members IDs are public information
  and were then matched to Maisanta
• -- We matched data on 453 private manufacturing
  firms in 95 municipalities (including the largest
  cities)
• -- Contains 34 of private national
  manufacturing output
• -- The Industrial Directory allows us to
  identify most firms
• The private sector is dominated by the Opposition
• -- Among our firms, 56 of board members signed
  against Chávez, only 4 signed against the
  Opposition
• -- Meanwhile the public sector was increasingly
  dominated by Chávez and his allies

21
(No Transcript)
22
Results Firm survey estimates

• Difference-in-differences estimation with firm
  and year fixed effects, and firm sector time
  trends as controls
• -- 50 3-digit industrial sectors (e.g., 6
  categories for textiles/apparel alone, 5 for food
  processing)
• -- We effectively estimate impacts of signing on
  outcomes within these narrow industrial sectors
• Pro-opposition firms profits fell substantially
  relative to other firms post-2003 (Table 2). Why?
• Pro-opposition firms paid roughly 33 more in
  taxes
• -- On average, 76,340 more in taxes per year
• -- Pro-opposition firms also got significantly
  less foreign exchange (-49) from the government
• -- Opposite effects for pro-government firms

23
(No Transcript)
24
(No Transcript)
25
TFP Impacts of factor misallocation

• Hsieh and Klenow (2007) quantify the effect of
  factor misallocation on manufacturing TFP, in the
  presence of firm-specific distortions taxes
  (?Y,i), costs of capital (?K,i)
• Marginal revenue products of capital, labor are
  higher for firms facing larger firm specific
  distortions ?Y,i and ?K,i
• -- A production function allows us to convert
  observed average products of capital, labor into
  marginal products
• -- Greater marginal revenue product dispersion
  across firms is indicative of greater
  firm-specific distortions, and the resulting
  misallocation leads to lower aggregate TFP

26
TFP Impacts of factor misallocation

• Firm marginal revenue products diverge
  substantially from sector averages over time in
  Venezuela, and these changes are strongly
  correlated with political preferences
• -- Pro-government firms have sharply falling
  average products of labor and capital, marginal
  revenue products, meaning they are becoming less
  efficient
• -- Pro-opposition firms show small increases
  they are relative productive but face constraints
  to expansion (e.g., higher taxes or less access
  to foreign exchange)

27
(No Transcript)
28
(No Transcript)
29
(No Transcript)
30
TFP Impacts of factor misallocation

• We compute the increase in marginal revenue
  product dispersion correlated with firms
  politics (Table 3)
• The effect of this factor misallocation is a drop
  of 5 in aggregate total factor productivity
  (TFP) in 2003-2004
• -- Including the steady-state endogenous
  response of capital investment increases output
  effects to -7.5
• -- This ignores any sector-wide distortions, or
  broader inefficiencies from Chavezs economic
  policies
• -- Alesina et al (1996) find a small -1 effect
  of a coup

31
Household Datasets

• Household Survey, biannually for 1997-II to
  2006-I
• -- Approximately 55,000 households per round in
  a rotating panel, households are retained for six
  semesters
• -- Panel of earnings, employment status and
  sector, as well as individual education,
  household size
• 20 signed Anti-Chávez, 8 signed Pro-Chávez

32
Matching Maisanta and household data

• Maisanta identifies individuals voting center,
  which can be placed in a particular locality
  (parroquia)
• -- Locality information, exact date of birth and
  gender, uniquely identifies 45 of individuals in
  Maisanta
• -- Another 19 are in DOB-gender-parroquia cells
  where all individuals share a political
  preference
• HHS data matched to Maisanta using these
  variables
• -- HHS data on 87,100 individuals in Maisanta,
  296,087 individual-semester observations
• -- The matched, unmatched similar (Appendix
  Table 1)
• -- Re-weight observations by 1/Locality match
  probability

33
Results Descriptive statistics on signers

• Pro-government and pro-opposition signers have
  similar earnings at baseline, pro-opposition
  signers have more education (Table 4)
• -- Pro-government signers more likely to live in
  Caracas

34
(No Transcript)
35
Labor markets in a polarized society

• Petition signing allows people to signal their
  political type
• -- Labor demand employer discrimination, a
  taste for workers with similar political views
• -- Labor supply employees may also have a taste
  for working with like-minded people
• Regardless of the exact cause, exogenous job
  displacement could destroy job match surplus and
  have negative social costs (Mortensen and
  Pissarides 1998)
• -- Loss of firm-specific human capital, worker
  job search

36
Results Labor market effects

• Difference-in-differences estimation of effects
  of Maisanta information on labor market outcomes
• -- Include individual fixed effects, year fixed
  effects
• -- Time-varying omitted variables correlated
  with politics are the key identification concern.
  Include individual characteristics (female, year
  of birth, years of schooling, locality)
  interacted with time trends
• Negative earnings effects for both groups of
  signers starting in 2003, the year the
  information was released, but larger and
  statistically significant effects for
  pro-opposition supporters a 3.9 drop in mean
  earnings (Table 5)

37
(No Transcript)
38
(No Transcript)
39
Results Household survey data estimates

• There is considerable churning in the labor
  market public sector employment decreases for
  pro-opposition signers, and private sector
  employment decreases for pro-government signers
  (Table 6)
• -- Effects at roughly 4-6 of pre-Maisanta
  employment
• -- Stronger effects for males no differential
  effects by respondent education
• It is challenging to estimate the aggregate
  welfare consequences of this increased job
  switching, without job match surplus estimates
  for Latin America
• (Ideas?)

40
(No Transcript)
41
Discussion

• Political polarization and favoritism can have
  substantial negative effects on aggregate
  economic performance, in a setting with weak
  institutional checks and balances
• -- Echoes of other cases of Latin American
  populism and political patronage (Peron in
  Argentina, Garcia in Peru)
• -- Pro-opposition firms were punished,
  reducing aggregate manufacturing productivity by
  at least 5
• -- Individual Venezuelans who signed the
  anti-Chavez petition also had worse labor market
  outcomes
• These findings provide a partial explanation for
  the stability of dictatorships or
  pseudo-democracies when the price of political
  opposition is high
• -- A rationale for preference falsification
  (Kuran 1995)

42
An example

• Ms. Rocío San Miguel worked for 13 years as a
  contract worker for the Venezuela National
  Borders Council
• -- Fired on March 12, 2004. Her boss "How could
  it have occurred to you to sign against the guy
  who pays you?
• Three other co-workers who had signed also fired
• One decided not to validate his signature (in the
  reparos) and the lay-off letter was withdrawn
• Ms. San Miguel taped phone conversations where
  her boss stated she was fired for signing the
  recall petition
• -- Case at Inter-American Human Rights Commission

43
(No Transcript)
44
(No Transcript)