The Turkish Currency Crisis -A Balance Sheet Effect

The Turkish Currency Crisis
-A Balance Sheet Effect Framework• Place of the Turkish Crisis within the
currency crisis framework
• Introduce a third generation model based on
balance sheet effects of devaluations
• Empirically test the model in Turkey’s case
• Learnings for the exchange rate policy in
emerging economies
1
Perspective on Currency Crisis Models Literature
• First generation models – Irresponsible government
policies
• Second generation models – multiple equilibria
• Third generation models
– Excess borrowing
– Bank run models
– Balance sheet effects
2
First Generation Models
• Krugman (1979), Flood&Garber (1984), starting from
commodity price fixing models
• The crisis is a consequence of the government’s pursuit of
leverage leading to foreign reserves depletion
• Through backward induction, the timing of the attack is the
moment when the shadow price exceeds the peg parity
• The crisis is the only outcome possible, given
government’s policies
3
Second Generation Models
• Obstfeld (1994), as first generation models failed to
explain the EMS crisis
• The peg is abandoned by a rational government unwilling
to sustain it, although it might have been able to keep it
• Continuous assessment of the cost of maintaining the peg
vs. the cost of removing the peg
• Expectations of the peg being abandoned in the future
increase the cost of defending the peg, leading to
MULTIPLE EQUILIBRIA
4
The Need for a New Generation of
Crisis Models
• The SE Asia crisis has revealed the need for a new
framework for looking at currency crisis
• Neither first, nor second generation models provide a
rationale for the fall in output after the crisis has occurred
• The central role the financial system played in the crisis,
leading to the concept of “Twin Crises”
5
Third Generation Models
• Excessive lending caused by implicit government
guarantees (Krugman 1998, Corsetti, Presenti, Roubini)
– The governement assesses the costs of making good/defaulting on
its guarantees, a la second generation models
• Sachs and Radelet – model of financial fragility
– The currency crisis is, in fact, an international banking crisis
• Balance Sheet Effects Model by Krugman 1999
6
The Balance Sheet Effect
Model of Currency Crisis
7
The Classical Mundell Flemming
Framework
• (1) y=d(y,i) + NX(eP*/P,y)
• (2) M/P=L(y,I)
• (3) i=i*
8
The Balance Sheet Effect of Domestic
Currency Devaluation
• Firms have debt denominated in hard currency while their
revenues are denominated in local currency
• A domestic currency real depreciation will thus deteriorate
the firm’s balance sheet
• High net worth is essential in obtaining financing because
of asymmetrical information
• Investment projects are assessed based on their hard
currency-returns
9
Effects on Output of a Domestic Currency
Depreciation Induced by the Balance Sheet Effect
• Contractionary effect in the middle of e
ranges
• The balance sheet effect fades for extreme e
values (both favorable and unfavorable)
• The effect on output does not depend on the
maturity of the foreign currency
denominated debt
10
The Crisis Mechanism
• An expected real devaluation translates into
an expected fall in output
• An expected fall in output makes domestic
assets unattractive and leads to a flight of
funds
• The flight of funds fulfills the expectations
of devaluation
11
Mundell Flemming Framework with Balance
Sheet Effect
• (1’) y = d(y,i,eP*/P)+NX(ep*/P,y)
• (2’) M(e)/P=L(y,i) with M decreasing in e –
Central Bank’s fear of floating
• Where d(y,i,eP*/P) is a decreasing function
of eP*/P
12
Macroeconomic Developments in
Turkey in the Pre-Crisis Period
• External debt of private sector increased more than 10 times since
1987 and more than 3 times vs. the 1994 crisis level, reaching 12% of
GDP
• M3/M1 ratio tripled vs. 1994 – development of the financial sector
• Stock market index drop of 50% between April and September 2000
• A banking crisis in November 2000
• Central Bank foreign reserves increased 12 times vs. 1987 and three
times vs. 1994 crisis level
• Domestic credit by the central bank eliminated end 1999 (IMF
stabilisation agreement) until November 2000 (failing banks bail-out)
• Continuous real appreciation of the Lira with the exception of the 1994
crisis
13
Empirical Testing of the Balance
Sheet Effect in the Case of Turkey
14
The Variables
• Investment – as measured by the gross capital formation
• HCPI= (CPIt/CPI0)/(et/e0) - hard currency price index
• economic significance = the degree to which domestic
firms are able to price-up for the depreciation of their
national currency - “Moral Dollarisation”
• Q1,Q2,Q3 – quarterly dummies
• C - free term
15
Estimation Output
Dependent Variable: DINVEST
Method: Least Squares
Date: 07/02/01 Time: 22:44
Sample(adjusted): 1987:2 2000:4
Included observations: 55 after adjusting endpoints
White Heteroskedasticity-Consistent Standard Errors & Covariance
Variable
Coefficient
Std. Error
t-Statistic
Prob.
C
-0.476791
0.143641
-3.319324
0.0017
DHCPI
2.950792
0.789090
3.739486
0.0005
Q1
-1.087536
0.225829
-4.815756
0.0000
Q2
1.874560
0.253766
7.386952
0.0000
Q3
1.208776
0.162309
7.447368
0.0000
R-squared
0.831850
Mean dependent var
0.096607
Adjusted R-squared
0.818398
S.D. dependent var
1.228359
S.E. of regression
0.523462
Akaike info criterion
1.629803
Sum squared resid
13.70062
Schwarz criterion
1.812288
F-statistic
61.83852
Log likelihood
Durbin-Watson stat
-39.81958
2.022949
Prob(F-statistic)
0.000000
16
Interpretation
Estimation Command:
=====================
LS(H) DINVEST C DHCPI Q1 Q2 Q3
Estimation Equation:
=====================
DINVEST = C(1) + C(2)*DHCPI + C(3)*Q1 + C(4)*Q2 + C(5)*Q3
Substituted Coefficients:
=====================
DINVEST = -0.4767909993 + 2.950791862*DHCPI - 1.087535506*Q1 + 1.874559665*Q2 + 1.208775582*Q3
• The variation in investment is positively correlated with the variation
of HCPI
• As real e is the inverse if HCPI, real e movement will be negativelly
correlated with investment, as suggested by the model
17
Conclusions
• The Balance sheet effect is validated
empirically in the case of Turkey
• We can expect a fall in investment induced
by the February devaluation
18
Lessons for other countries
• An expansionary policy of real
depreciation will work only if the external
financing of the private sector is not
important
19