Transcript rosendorff_vreeland_S200_2_pres

Democracy and
Data Dissemination:
The Effect of Political
Regime on Transparency
B. Peter Rosendorff, NYU
James R. Vreeland, Yale
IPES, Princeton, November 2006
1
Is transparency “incentive
compatible”?

Generally
 Under
what conditions do rulers act in the best
interests of the ruled?

Representatives may divert resources
may act as disciplining devices –
retrospective voting
 Elections


Consumers/voters demand transparency
When will policymakers supply transparency?
 Is
transparency incentive compatible?
2
Key insights

Policymakers have an incentive to be
more transparent when
 Survival
in office depends on electoral
accountability

(democratic polities)
 Election
outcomes are uncertain
3
Policy Instruments

Monetary base


Direct control
Inflation


Costly to society
Redistributive effects


Towards borrowers from lenders
Tax on money holdings


Revenues accrue to the government
Seignorage

"The amount of real purchasing power that government can extract
from the public by printing money." -- Cukierman 1992
4
Intuition: Incentive Compatible
Transparency

Monetary policy is made under uncertainty

Policymakers are uncertain about the aggregate state of the economy at
election time
 Electorate can’t distinguish between (exogenous) aggregate economic
shocks
 And the executive's excessive inflation tax (endogenous).

For example

If voters observe high real income in any period



If voters observe low incomes




Voters infer aggregate conditions were good, and executive has not been
excessively inflationary
Voters inclined to reelect the incumbent
Voters infer aggregate state was poor and/or
The executive extracted/inflated too much
Voters inclined to evict the incumbent
Possibility of “unfair eviction”

Extraction was modest, but exogenous conditions conspired against the
executive.
5
Accountability

Autocrats
 Executives
are less accountable to the electorate
 Will extract more and worry little about eviction

Democratic Executives
 Sensitive
to the will of the voters
 Will moderate extraction, but still under bad enough
shocks be kicked out of office

Democrats are more likely to be subject to unfair
eviction.
6
Transparency

Transparency in policymaking



Voters can better (but still imperfectly) separate aggregate
shocks from policy
Voters punish extractive behavior, not low incomes
Smart policymaker

One that is subject to unfair dismissal – democratic executive



The autocrat – does not fear unfair dismissal


May choose to relinquish opportunities for rent extraction
In exchange for eliminating the risk associated with being unfairly
dismissed
Will not be willing to relinquish rent extracting opportunities
Hence those politicians sensitive to the electorate adopt
transparent modes of policymaking

In order to enhance the possibility of remaining in office.
7
Model




Large number of identical consumers/voters
Infinite horizon
Voters choose savings and reelection rule
Govt chooses inflation tax

Inflation redistributes and is a public bad.
4.
Non-Transparency, in each period
Executive
 Inflation
Nature
 Shock
Voters
 Savings
Voters
 Reappoint?

Payoffs:

1.
2.
3.


t
t
st
ut  U

t2 
u  t   u1  mt 1  s    u2  mt s 1   t  
Voters:

2
2

Government
t 
v  t   mt s t 
2
8
Non-transparency



Will of the voters: ex ante probability of voters
recommending reelection
  t ,U   Pr u  t   U 
Regime Type: The ex ante probability of keeping
office
  t ,U ,      t ,U   1   
σ in [0,1] : the degree to which the sentiments of
the voters are binding on the executive
σ
=1 : pure democracy
 σ =0 : pure autocracy
9
Executive’s Reaction Function
π


U
u
10
Transparency



Voters do not directly observe π
The policymaker announces ex ante a policy 
Credible source (WB, IMF, independent agency)
 Announce

if
 
Sequence
 Announcement
and Policy
 Savings
 Signal
 Election

Nash
 ,  ,  s, 
t
11
Transparency

Voters condition their reelection rule on the
announcement
 As before, s  s  t 
   and  t  
 If
reelect,
 evict otherwise


Voters are weakly better off
12
Proposition

NE

s      / 2
1
If   1  *
1     1
*
*2
   s   / 2


then  *     and   1;
 
otherwise  *  ,    0,   and   0.

Transparency is preferred by both players
when the polity is sufficiently democratic
13
The Parameter Space
σ
Transparency
Preferred by policymaker
Non-transparency
preferred
    
1
δ
Sufficiently democratic executives prefer transparency.
14
Evidence

Transparency – take advantage of a hindrance

Missing data – World Development Indicators (World Bank)
 Inflation
 Unemployment



GDP growth
Infant Mortality Rates
Regime

Przeworski et al (2000) - Dichotomous measure

Key government offices (executive and legislature) are filled through
contested elections
15
Reporting of Data by Regime
Figure 1: Reporting of data by regime
Percentage of observations with data
1
0.8
0.6
0.4
0.2
0
Inflation
Unemployment
Grow th
Infant Mortality Rate
Regim e
Dictatorship
Democracy
16
But Democracies are…
Richer (control for GDP/capita)
 Recent (use hazard model)

We also control for:
 IMF participation
 Country fixed effects
 Regional effects
17
Controls: GDPPC and IMF
Table 2: The pooled effect of democracy on transparency
controlling for level of economic development and IMF participation
Inflation
Unemployment
Growth
Infant Mortality Rates
The effect of
Democracy
2.02**
1.65**
1.97**
0.24**
(Robust std error)
(0.15)
(0.11)
(0.24)
(0.09)
The effect of
GDP/capita
0.0001**
0.0002**
0.00002
0.00001
(Robust std error)
(0.00002)
(0.00001)
(0.00002)
(0.00001)
The effect of IMF
participation
0.82**
0.66**
1.41**
0.23**
(Robust std error)
(0.09)
(0.11)
(0.16)
(0.09)
Constant
0.31**
-2.46**
1.67**
-2.06**
(Robust std error)
(0.07)
(0.12)
(0.09)
(0.08)
# of obs.
5,042
2,935
5,042
5,128
Log pseudolikelihood
-1932.65
-1428.22
-1114.83
-2084.83
18
Controls and Country Fixed effects
Conditional Logit
Table 3: The effect of democracy on transparency
controlling for economic development, IMF participation, and country specific effects
Inflation
Unemployment
Growth
Infant Mortality Rates
The effect of
Democracy
4.93**
1.96**
1.59*
0.60**
(Robust std error)
(0.87)
(0.64)
(0.94)
(0.10)
The effect of
GDP/capita
0.0014**
0.0005**
0.0010*
0.0001**
(Robust std error)
(0.00035)
(0.00016)
(0.00051)
(0.00001)
The effect of IMF
participation
2.73**
0.96**
2.45**
0.16*
(Robust std error)
(0.43)
(0.25)
(0.49)
(0.08)
# of ctries
65
73
34
169
# of obs.
1,806
1,306
915
5,101
Log pseudolikelihood
-466.41
-477.21
-301.03
-1793.52
19
Regional
Effects,
Controls.
Basic
Logit
Table 4: The effect of democracy on transparency controlling for economic development, IMF
participation, and regional effects
Inflation
Unemployment
Growth
Infant Mortality Rates
The effect of
Democracy
1.87**
1.45**
1.57**
0.36**
(Robust std error)
(0.16)
(0.13)
(0.30)
(0.11)
The effect of
GDP/capita
0.0001**
0.0001**
0.00003
0.00002**
(Robust std error)
0.00002
0.00002
0.00002
0.00001
The effect of IMF
participation
0.87**
0.62**
1.40**
0.19**
(Robust std error)
(0.09)
(0.13)
(0.17)
(0.09)
Africa
-0.88**
-4.83**
-2.82**
0.16
(Robust std error)
(0.26)
(0.44)
(0.72)
(0.34)
South Asia
-2.86**
0.09
(Robust std error)
(0.47)
(0.39)
East Asia
-1.49**
0.21
(Robust std error)
(0.32)
(0.45)
S. E. Asia
1.20**
-1.52**
0.17
(Robust std error)
(0.38)
(0.43)
(0.38)
Oceania
-3.64**
(Robust std error)
(0.58)
-3.31**
(0.79)
Middle East
-0.50*
-2.88**
-2.87**
0.10
(Robust std error)
(0.28)
(0.43)
(0.72)
(0.36)
Latin America
0.33
-1.55**
-1.31*
0.02
(Robust std error)
(0.30)
(0.43)
(0.78)
(0.34)
Caribbean
-1.18**
-3.71**
-0.71
-0.19
(Robust std error)
(0.30)
(0.44)
(1.01)
(0.36)
East Europe
-1.37**
-1.57**
-3.96**
0.38
(Robust std error)
(0.30)
(0.44)
(0.73)
(0.36)
Industrial ctries
-0.58
-2.22**
-0.25
(Robust std error)
(0.53)
(0.81)
(0.36)
Constant
1.06**
0.93**
4.29**
-2.21**
(Robust std error)
(0.25)
(0.41)
(0.72)
(0.33)
# of obs.
4,013
2,935
4,762
5,128
Log pseudolikelihood
-1775.38
-1083.57
-1025.76
-2078.72
20
Table 5: The effect of democracy on transparency controlling for duration dependence
Regional
Effects,
Controls.
Duration
Dependence
Inflation
Unemployment
Grow th
Infant Mortality Rates
The effect of
Democracy
1.30**
0.65**
0.62**
-0.14
(Robust std error)
(0.27)
(0.30)
(0.23)
(0.18)
The effect of
GDP/capita
-0.00003
0.00004*
-0.00003
-0.0001*
(Robust std error)
(0.00006)
(0.00002)
(0.00003)
(0.00004)
The effect of IMF
participation
0.55**
0.42*
0.44**
-0.29**
(Robust std error)
(0.17)
(0.24)
(0.19)
(0.14)
Africa
-3.19**
-2.51**
-1.54**
-2.53**
(Robust std error)
(0.25)
(1.08)
(0.38)
(0.85)
South Asia
-2.12**
-1.48
0.05
(Robust std error)
(0.52)
(1.18)
(0.10)
East Asia
-3.18**
0.36*
0.07
(Robust std error)
(0.38)
(0.21)
(0.11)
S. E. Asia
-1.49**
-0.81
-0.09
0.07
(Robust std error)
(0.30)
(1.15)
(0.21)
(0.07)
Oceania
-2.78*
-16.10**
-14.76**
(Robust std error)
(1.56)
(1.06)
(1.30)
Middle East
-2.38**
-1.46
-1.43**
0.24**
(Robust std error)
(0.36)
(1.12)
(0.45)
(0.09)
Latin America
-2.20**
-0.29
-0.55**
0.43**
(Robust std error)
(0.41)
(1.08)
(0.27)
(0.13)
Caribbean
-3.31**
-1.59
-0.44
-3.25**
(Robust std error)
(0.65)
(1.09)
(0.29)
(0.95)
East Europe
-2.26**
-0.34
-1.85**
(Robust std error)
(0.39)
(1.07)
(0.52)
Industrial ctries
-1.25**
0.57
-0.47
0.65**
(Robust std error)
(0.39)
(1.14)
(0.39)
(0.23)
Constant
0.20
-2.23**
-0.33
0.08*
(Robust std error)
(0.24)
(1.08)
(0.22)
(0.04)
Duration dependence
(ln p)
-0.03
0.11
-0.05
0.30**
(Robust std error)
(0.07)
(0.08)
(0.06)
(0.15)
97
# of countries:
144
160
113
# eventually reporting:
126
103
110
95
# of obs.
998
1,431
457
270
Log pseudolikelihood
-123.30
-162.24
-126.86
-107.12
Note: Coefficients not hazard ratios are reported.
21
Conclusions

Methodological Implications
 Definition


Minimalist definition of democracy actually covers more than
just elections
Implies transparency too.
 Missing


Data
Highly correlated with regime type
Researchers interested in the consequences of regime type
cannot simply use the available data


of democracy
Must recognize the bias that emerges.
Modest attempts to separate out the oft conflated notions
of democracy, accountability and transparency.
22