Transcript mukherjee_S1000_16_pres

Why IMF Stabilization Programs Fail to Prevent Currency Crises in
Some Financially Distressed Countries, But Not Others?
Bumba Mukherjee
Assistant Professor
Dept. of Political Science and Dept of. Economics & Econometrics
University of Notre Dame
Visiting Associate Research Scholar
Princeton University
[email protected]
Do IMF stabilization programs increase (decrease) the likelihood of a currency crisis?
 Article I IMF: Promote Stability of Exchange Rates and Currency Markets
 IMF Failures in the 1990s: Russia, Indonesia; “Abolish” IMF: for e.g. Stiglitz
 IMF’s record in preventing currency crisis (not crisis resolution) is “mixed”
Daily Brazilian Real/$, 20022003
100%
Volatility Clustering
80%
60%
IMF SBA loan
40%
%Change
20%
0%
-20%
2003
2002
-40%
Smoothed Transition Probability of Currency Crisis (S=1) in Thailand 1997-98
from Markov-Switching Model
1.0
IMF stabilization package
p11
II IMF SBA
1-p22
Prob.
(S=1)
0
Jan 97
july 97
feb 98
jan 99
531 Stabilization Programs 82 countries, 1974-2002: 60% Prevented Crisis, 40% Failed
What explains variation in Effect of IMF Programs on Currency Crises?

Scholars examine how IMF programs affect macroeconomic outcomes, e.g. growth (Barro,
Dreher, Vaubel, Stone, Vreeland), but not currency markets.

Political Scientists..…study how domestic institutions affect currency/financial markets
(Leblang; Bernhard; Freeman; Hays; Satyanath)…. how international institutions such as
WTO affect trade (Mansfield, Reinhardt; Goldstein,Tomz & Rivers; Gowa & Kim).

Answer: Impact of IMF programs on likelihood of currency crisis conditional on extent of
institutionalized state intervention in borrowing country’s financial sector.

Greater (lesser) the state’s role in the financial sector of the borrowing country’s financial
sector, the higher (lower) the likelihood that IMF loans under its stabilization programs will
lead to a currency crisis.
Model of Speculative Trading
 3 Players: Currency traders, IMF, Debtor Country (financial problems but not
fully blown currency crisis) that borrow IMF loans

Macroeconomic fundamentals of debtor country; s = signals about
 Currency Trader’s Payoff:
u( si , a )  
P ( )
 Traders’ start a speculative attack if
r( )h( | si )d  t
*
*

1 if    ; s  s
ai ( s )  
*
*

0 if    ; s  s
 IMF: Prevent speculative attack; provides m conditional on financial reforms

*
  , m   bm if   
 ( , m)  

 bm if    *
Debtor government that gets m implements reforms l, reform implementation
(l) affected by extent of formal state intervention in financial sector, i.e. v
arg max U G  (1  vl   )( ml  C )  (vl   ) l  l 2
l
Causal Story and Hypotheses

Nash equilibrium: l *  vC  m   (   m)
2(1  (   m)v)
l *
 0,
 comparative statics v
lim l *  0

  0, a (s)  1 as   
i
*
; s  s*
Causal Story….
 Higher state intervcntion in financial sector of debtor country…
 Greater political resistance to financial reforms suggested by IMF
 Ex ante commitment to implement reforms lack credibility &
 IMF loan/program engenders moral hazard under weak commitment
 Declining fundamentals = speculative attack =currency crisis

H1: IMF stabilization programs engender “moral hazard” in borrowing countries
with high state intervention in financial sector

H2: IMF stabilization programs increases likelihood of currency crises in borrowing
countries with high state intervention in financial sector
Sample and Statistical Model
 82 countries, 1974-2002
 2 Methodological Issues:
Non-random participation in IMF programs (selection)
spatial dependence in likelihood of currency crisis & participation in IMF program
 Spatial Autoregressive Error (SAE) Bivariate Probit model
y1*i   0  x1i1  u1i ; u1i    ciju1 j   1i (Selection)
j i


y1*i   0  x1i1  u1i , uy
i 
 ciju01 j x1i2 i (Selection
1i 2
1  u 2)i ;
*
j i
u2i    ciju2 j   2i (Outcome )
y   0  x2 i 1  u2i , u2i    ciju2 j   2i (Outcome)
*
2i
j i
y2i  1 (Currency crisis) iff y2*i  0 and y1*i  0
j i
y2i  1 (currency crisis) if y2*i  0 and y1*i  0

y1i  1(IMF program)
cij  C ( spatial weights); autoregressive parameters  and 
Weights in (1  C ) 1 and (1  C ) 1 given by geographic distance
Key Variables
 DV in outcome equation: Currency Crisis =1 if change in index of exchange
rate pressure exceeds mean plus 2 times the country specific std deviation.
 Index: weighted average of real exchange rate changes and % reserve losses
 IV in outcome equation interaction term: IMF Program x State Credit/GDP

IMF Program dummy for IMF loans provided for short-run financial
stabilization via SBAs, BSFF, CSF, SRF, CFF and EFF; Not PGRF and SAF

State Credit/GDP: Share of State Owned Credit GDP; Proxy for state
intervention in financial sector
 Several Controls in Selection and Outcome equation: M2/Reserves, Divided
Government, Terms of trade growth, external debt….
Outcome Equation of SAE Bivariate Probit Model: Select Variables
Global
Developing
Global
(without EFF)
IMF prog.
.073 (.082)
.031 (.040)
.050 (.046)
Credit/GDP
.098 (.077)
.065 (.092)
.023 (.0840
IMF prog x credit/GDP .122** (.036) .138** (.044)
.131** (.052)
M2/Reserves
.023** (.011) .025** (.012)
.037** (.018)
Democracy
-.027 (.023)
-.038 (.073)
-.036 (.032)
SAE parameter (γ)
.045* (.020)
.040* (.017)
.035* (.011)
Log likelihood
-214.36
-177.23
-182.78
N
2174
1889
2174
** (*) Indicates significance at 1% (5%)level; Substantive Effect of Interaction term: 18%
Conclusions & Future Research
 Effects of IMF programs on currency markets –especially the
likelihood of currency crisis – not direct
 Conditional on state’s role in financial sector
 Study more closely the details of the IMF programs, how
financial markets and domestic politics in debtor countries
respond to these programs
 Gather additional data as well
Selection Equation: Select Variables only
Global
Developing
Lag inflation
.098 (.077)
.065 (.092)
External debt/exports
.031 (.022)
.025 (.021)
REER valuation
.028 (.071)
.030 (.022)
SAE parameter (δ)
.021* (.012)
.032* (.020)
N
2174
1889