Transcript brooks_mosley_F230_1_pres

Risk, Uncertainty, and Autonomy:
Financial Market Constraints in
Developing Countries
Sarah Brooks & Layna Mosley
Ohio State University & University of North Carolina, Chapel Hill
Prepared for the 1st meeting of the International Political Economy Society,
Princeton University, Nov. 17-18, 2006
Political Risk

Exposure to the possibility that political events will
adversely affect the profitability of an investment (Bailey
& Chang 1995: 542)

Especially problematic for developing countries


Less transparent to foreign investors
Uncertainty / incomplete information are central to the
problem of political risk
Previous Research

Elections are a key source of political risk:

Uncertain outcomes

Incentives for opportunistic behavior, policy change

Associated with turbulence in currency, bond markets (Martinez
& Santiso 2003; Reinhart 2002)

Both the outcome of the election and the possibility
of policy discontinuity are considered sources of
uncertainty (Jensen & Schmith 2005)

Uncertainty is said to be largely resolved following the
election (Bernhard & Leblang 2002)
Political Risk v. Political Uncertainty


Difference between risk and uncertainty largely
elided
Knight (1957):



Risk can be quantified, measured
Uncertainty is unmeasurable
Difference between risk and uncertainty has to do
with the degree of knowledge about a situation

Important implications for market behavior, government
autonomy
Risk and Uncertainty in International
Bond Markets
Sources of risk and uncertainty in sovereign debt
markets:
1.
Default risk


2.
3.
Ability to pay
Willingness to pay
Currency risk (including inflation)
Policy Stability (long-run value of asset)
Risk and Uncertainty in International
Bond Markets

Information derived from institutional, ideological
factors, and past policy behavior



Challengers to the incumbent often lack a past policy
record
Elections thus increase uncertainty, even if outcome is
predictable
Information frictions: little incentive to become
fully informed (Calvo & Mendoza 2000)
Risk and Uncertainty in International
Bond Markets

Investors can and do acquire information to reduce
uncertainty about candidates, new incumbents

Partisan cues

Assimilate candidates into a group of similar situations
for which probability of default can be calculated

Left/Right partisan archetypes

Right more committed to fiscal probity, willing to honor debt
commitments

Left more likely to inflate economy, default
Risk and Uncertainty in International
Bond Markets

Since 1990s, these archetypes have been less valid


Left governments adopting orthodox liberal policies

“Nixon in China”: e.g., Menem, Fujimori

Not universal: e.g., Kirschner, Morales, Chávez
Partisan cues – for the Left – thus are ambiguous

Elections do not fully resolve uncertainty about Left
candidates’ willingness to honor sovereign debt
commitments
Observable Implications
1.
Bond markets should respond more negatively to
Left candidates and new presidents than to those
on the Right, all else being equal.
2.
Time in office (opportunity to gain information
about type) should reduce market premium for
Left governments, but should not change
assessments for the Right, all else being equal.
Empirical Analysis: Dependent Variable

Sovereign risk is captured through spreads:

Difference between yield on government bond and
benchmark ‘risk-free’ U.S. Treasury bond

Higher spreads indicate negative market sentiment about a
country’s economic and political risk (Eichengreen & Mody 1998)

J.P. Morgan’s EMBI Global Index



Foreign currency denominated debt instruments: isolates
default risk; no currency risk
33 ‘emerging market’ countries
1993-2004
Empirical Analysis: Independent Variables



Partisanship of Executive
Years in Office
Partisanship * Years in Office
Empirical Analysis: Controls

Political Controls




Creditworthiness Indicators





Divided Government
Democracy
Presidential Election Year
Inflation (ln)
External balance
Debt/GDP Ratio
Debt Service/Export Ratio
Macroeconomic Controls



GDP (ln)
Growth (% per capita)
Trade exposure
Empirical Model
ΔS = β0St-1 + β1Partisanship + β2YrsOffice +
β3Partisan* YrsOffice + β4 Divided Govt +
β5Democracy + β6Election Year +
β[Creditworthiness] + β[Macroeconomy] + β7Time



Random effects cross-sectional, time-series estimation
Temporal dependence: AR-1
Cross-sectional correlation: PCSE
Political Variables
Left Executive
Years in Office
Left Executive * Years Office
Majority
Presidential Election Year
142.372**
205.658**
218.025***
(66.655)
(84.432)
(87.070)
-4.935
-3.354
-4.430
(3.811)
(3.764)
(3.484)
-
-13.630*
-15.705*
(7.955)
(8.766)
-130.655*
-134.859*
-133.283*
(72.792)
(71.981)
(72.852)
-
-
97.428
(90.374)
Democracy
-8.006*
-8.827*
-10.610***
(4.698)
(4.621)
(4.128)
Creditworthiness
External Balance
Inflation (Ln)
Debt: GDP
GDP Growth
Debt Service: Exports
4.825
4.659
5.249
(5.677)
(5.688)
(5.889)
-11.847
-15.003
-17.801
(19.354)
(19.469)
(18.991)
919.580***
970.53***
971.356***
(272.468)
(280.069)
(279.721)
-6.437***
-6.321***
-6.505***
(1.974)
(1.994)
(1.871)
-2.894
-3.086
-3.072
(1.946)
(2.064)
(2.002)
Macroeconomy
GDP (ln)
15.482
21.305
23.124
(33.243)
(33.450)
(33.248)
-2.177**
-2.312**
-2.275**
(1.1133)
(1.130)
(1.097)
-7.740
-6.680
-5.915
(16.178)
(16.241)
(16.071)
-85.046
-239.949
-288.483
(855.098)
(862.632)
(857.938)
rho
0.02
0.029
0.041
Observations (N)
193
193
193
0.316
0.32
0.328
Trade
Year
Constant
R-squared
Case Study: Brazil

2002 Presidential election

Front-runner: Luiz Inácio Lula da Silva (“Lula”)






Left-wing Worker’s Party
Earlier rhetoric of debt restructuring
Recent Argentine default
Risk of policy discontinuity
Previous elections followed by costly devaluation
Market response:


Wall Street firms paid close attention to domestic polls
Spreads jumped sharply
Case Study: Brazil 2002
Brazil and Emerging Market Indices, 2001-2003
2500
2000
1500
1000
500
Brazil
Composite
00
3
6/
1/
2
00
3
4/
1/
2
00
3
1/
2
2/
/2
00
2
12
/1
/2
00
2
10
/1
00
2
8/
1/
2
00
2
6/
1/
2
00
2
4/
1/
2
00
2
1/
2
2/
/2
00
1
12
/1
10
/1
/2
00
1
0
Case Study: Brazil 2002
Market response unlikely to be due solely to
changes in fundamentals or risk aversion

1.
Debt, although rising, was considered manageable (Martinez &
Santiso 2003: 371)
2.
Brazil spreads diverged from EMBI Global Composite
Index

Composite reflects broader risk appetite for emerging market debt
Case Study: Brazil 2002
Brazil-Composite
Inflation
6
Ja
n0
5
Ja
n0
4
Ja
n0
3
Ja
n0
2
Ja
n0
1
Ja
n0
0
Ja
n0
9
% annual change in CPI
20
18
16
14
12
10
8
6
4
2
0
Ja
n9
8
1400
1200
1000
800
600
400
200
0
-200
Ja
n9
EMBI Index
Index Differentials and Inflation
Entropy
Brazil minus Composite
10/28/2002
10/14/2002
9/30/2002
9/16/2002
9/2/2002
8/19/2002
8/5/2002
7/22/2002
7/8/2002
6/24/2002
6/10/2002
5/27/2002
5/13/2002
4/29/2002
4/15/2002
Case Study: Brazil 2002
Brazil, 2002 Spreads
1400
1.05
1200
1
1000
0.95
800
600
0.9
400
0.85
200
0.8
0
0.75
Lula vs.Closest Rival
Brazil minus Composite
10/28/2002
10/14/2002
9/30/2002
9/16/2002
9/2/2002
8/19/2002
8/5/2002
7/22/2002
7/8/2002
6/24/2002
6/10/2002
5/27/2002
5/13/2002
4/29/2002
4/15/2002
Case Study: Brazil 2002
Brazil, 2002 Spreads
1400
35
1200
30
1000
25
800
20
600
15
400
10
200
5
0
0
Case Study: Brazil 2002

Efforts during the campaign to win confidence

All three candidates signed a Letter of Intent with IMF


Funds to be dispersed after the election
Lula’s “Letter to the Brazilian People”

Pledged to respect existing agreements with international
institutions, companies

Laid out reform agenda
Case Study: Brazil

Lula in office: Confidence-enhancing measures






Maintained, deepened policies of previous government
Wall Street executive Henrique Meireles named President
of the Central Bank
Increased the autonomy of the Central Bank
Reforms of social security, tax systems
Primary surplus above IMF target
Market response:


Sharp decline in spreads
Currency and domestic stock market recovered
/2
0
5/ 01
1/
20
9/ 01
1/
20
1/ 01
1/
20
5/ 02
1/
20
9/ 02
1/
20
1/ 02
1/
20
5/ 03
1/
20
9/ 03
1/
20
1/ 03
1/
20
5/ 04
1/
20
9/ 04
1/
20
1/ 04
1/
20
5/ 05
1/
20
9/ 05
1/
20
1/ 05
1/
20
5/ 06
1/
20
9/ 06
1/
20
06
1/
1
Case Study: Brazil 2002-2006
EMBI Indices, Blended Spreads, 2001-2006
1400
1200
1000
800
600
400
200
0
-200
Brazil minus Composite
Case Study: Brazil 2006

2006 Election Campaign







Lula once again front-runner
Favorable global conditions
Net government debt lower, dollar exposure diminished
Strong currency (appreciated 53% since Lula took office)
Low growth, string of high-profile scandals
Scandals led to resignation of nearly all cabinet ministers,
top leaders of Lula’s Workers’ Party
Market response was sanguine
Case Study: Brazil 2006
Brazil and Emerging Market Indices, 2005-2006
500
450
400
350
300
250
200
150
100
50
11
/3
/2
00
5
1/
3/
20
06
3/
3/
20
06
5/
3/
20
06
7/
3/
20
06
9/
3/
20
06
11
/3
/2
00
6
3/
20
05
9/
3/
20
05
7/
3/
20
05
5/
3/
20
05
3/
1/
3/
20
05
0
Brazil
Composite
Case Study: Brazil 2006
Public Opinion Polls and Spreads,
2005-2006
120
35
30
100
25
80
20
15
60
10
40
5
0
20
-5
14
/2
00
6
2/
14
/2
00
3/
6
14
/2
00
6
4/
14
/2
00
6
5/
14
/2
00
6
6/
14
/2
00
6
7/
14
/2
00
6
8/
14
/2
00
6
9/
14
/2
00
10
6
/1
4/
20
06
1/
4/
2
12
/1
4/
2
11
/1
00
5
-10
00
5
0
Lula minus Closest Rival (Alckmin/Serra)
Brazil minus Composite
Conclusions

Market learning extends beyond the government
formation period


Especially for developing countries, left governments.
Partisanship matters



Left governments, all else equal, represent higher risk of
default.
Left partisan signal is more ambiguous: thus Left
governments also generate greater uncertainty.
Policy uncertainty, however, is resolved as Left executives
govern, so that investment risk declines as time in office
increases.