Introduction Literature Methodology Empirical Finding Robustness - PDF document

Mohammad Karimi (University of Ottawa) May 2012  Introduction  Literature  Methodology  Empirical Finding  Robustness Tests  Conclusion 1

18 16 14 12 10 8 6 4 2 0 Mexico 1994 Korea 1997 Chile 1981 Malaysia 1997 Finland 1990 Tailand 1997 Indonesia 1997 Argen�na 2002 Source: Krugman, Paul, 2010    First-generation models: inconsistent macroeconomic policies and fixed exchange  Second-generation models: multiple equilibria and self-fulfilling attacks  Third-generation models: interconnection between exchange rate and financial fragility and financial institiutions  Contagion 2

 Structural models  Panel data models (discrete-variable techniques)  Crisis definition - Exchange market pressure index - Conventional  signaling methods  Contagion definition: - Interdependence, or; - Pure contagion  Transmission channels: common shocks, trade linkages, financial linkages, political linkages, macroeconomic similarities  Panel data models, Eichengreen et al. (1996) - assumption of independently distributed equation errors across countries introduce a substantial bias, Pesaran and Pick (2007). 3

 Duration analysis - Semi-parametric continuous models, - Semi-parametric discrete models, - Parametric models  several robustness checks, running the models on two different crisis episodes sets identified based on monthly and quarterly data Crisis episodes identified by extreme value theory Table 1. Descriptive statistics of spells monthly Quarterly Total number of spells 266 190 Number of right censored spells 21 21 Mean of duration of spells 10.3 14.8 Median of duration of spells 5 8 Shortest completed spell 2 1 Longest completed spell 55 70 4

Figure 1. Empirical hazard of spells .25 .2 Empirical Hazard .15 .1 .05 0 0 5 10 15 20 Duration monthlytype quarterly Table 2. Cox proportional hazard estimation results (monthly-type spells) Contemporaneous Lagged Variable Model (I) Model (II) Model (III) Model (IV) Unemployment volatility 0.05* 0.05** 0.06** 0.07** Inflation volatility -1.89 -0.85 Size of economy 0.96** 0.83** 0.83* Whole period GDP growth 0.02* 0.01 Previous crises GDP growth rate 0.03 0.02 -0.08 -0.05 Inflation rate 0.30*** 0.25** 0.26*** 0.21* Unemployment rate -0.01 -0.02 0 0 Share price index growth -0.03*** 0 0 0 Real effective exchange rate 0 0 0 0.02** Money growth 0.02 -0.02 0.01 0.02 Openness degree -0.12 0.05*** -0.31 0.01 Current account / GDP 0 0.00* -0.05 0 Capital account / GDP 0.29 0 -0.56 0 Financial account / GDP 0.10* 0.00* -0.12** 0.00* Budget deficit / GDP 0.02 0 0.01 0 Trade linkages 0.14** 0.12* 0.15** 0.19** Financial linkages -0.03 0 -0.01 -0.02 Macroeconomic similarities 0.03 0.06 -0.02 -0.06 Log likelihood -107.93 -87.45 -106.85 -82.49 5

Table 3. Cox proportional hazard estimation results (monthly-type spells) Contemporaneous Lagged Variable Model (I) Model (II) Model (III) Model (IV) Unemployment volatility 0.02 0.03 0.04* 0.02 Inflation volatility -0.57 -1.42 -0.32 Size of economy 0.27 0.53 0.38 Whole period GDP growth 0.01 0.01 Previous crises -0.41* -0.38 -0.36 GDP growth rate 0.02 0.05 0.12 0.08 Inflation rate 0.22*** 0.21** 0.17* 0.20* Unemployment rate 0.04** 0.03 0.08** 0.04* Share price index growth -0.02* 0 0.02* -0.01 Real effective exchange rate 0 0 0.02*** 0.02*** Money growth -0.03 -0.03** 0.02 0.02 Openness degree 0.18 0.02 0.17 0.04* Current account / GDP 0 0 0.03 0 Capital account / GDP 0 0 -0.12** 0 Financial account / GDP -0.02 0 0 -0.02 Budget deficit / GDP 0 0 0.01 0 Trade linkages 0.19* 0.19** 0.18** 0.19* Financial linkages 0 0 0 0 Macroeconomic similarities 0 0 0.02 -0.07 Log likelihood -118.92 -106.24 -115.62 -96.79 Estimated hazard by Cox continuous model (monthly-type spells) .25 .2 .15 .1 .05 0 0 5 10 15 20 Duration Empirical_Hazard Cox_Hazard_Model2 Cox_Hazard_Model4 6

Estimated hazard by Cox continuous model (quarterly-type spells) .2 .15 .1 .05 0 0 5 10 15 20 Duration Empirical_Hazard Cox_Hazard_Model_2 Cox_Hazard_Model_4 Goodness of fit for Model 2 (monthly-type) 4 3 2 1 0 0 1 2 3 4 Cox-Snell residual H1 Cox-Snell residual 7

Goodness of fit for Model 4 (monthly-type) 4 3 2 1 0 0 1 2 3 4 Cox-Snell residual H Cox-Snell residual Goodness of fit for Model 4 (quarterly-type) 4 3 2 1 0 0 1 2 3 4 Cox-Snell residual H Cox-Snell residual 8

Table 4. Treatment of ties: Efron (a) versus the partial calculation (b) estimation results Monthly-type Quarterly-type Variable Model (II) (a) Model (II) (b) Model (IV) (a) Model (IV) (b) Unemployment volatility 0.05 0.05 0.02 0.02 Inflation volatility Size of economy 0.39 0.38 Whole period GDP growth 0.02 0.02 Previous crises -0.38 -0.36 GDP growth rate 0.02 0.02 0.08 0.08 Inflation rate 0.22 0.25 0.2 0.2 Unemployment rate -0.02 -0.02 0.04 0.04 Share price index growth 0 0 -0.01 -0.01 Real effective exchange rate 0.01 0 0.02 0.02 Money growth -0.02 -0.02 0.02 0.02 Openness degree 0.04 0.05 0.02 0.04 Current account / GDP 0 0 0 0 Capital account / GDP 0 0 0 0 Financial account / GDP 0 0 0.02 -0.02 Budget deficit / GDP 0 0 0 0 Trade linkages 0.1 0.12 0.2 0.19 Financial linkages -0.01 0 0 0 Macroeconomic similarities 0.07 0.06 -0.07 -0.07 Table 5. Semi-parametric discrete hazard (a) and Gompertz parametric hazard (b) estimation results Monthly-type Quarterly-type Variable Model (II) (a) Model (II) (b) Model (IV) (a) Model (IV) (b) Unemployment volatility 0.01 0.03* 0.02 0.04 Inflation volatility Size of economy -0.17 0.38 Whole period GDP growth 0 0.02* Previous crises -0.07 -0.22 GDP growth rate -0.16 -0.05 -0.04 0.09 Inflation rate 0.32** 0.27* 0.25 0.29** Unemployment rate -0.06 0 0.02 0.07 Share price index growth -0.01 0 -0.03 -0.01 Real effective exchange rate 0 0.01 0.02* 0.03** Money growth 0.01 0.01 0.01 0 Openness degree 0.7*** 0.06*** 0.02 0.05* Current account / GDP 0 0.00* 0 0 Capital account / GDP 0 0 0 0 Financial account / GDP 0.00* 0.00** 0 -0.05 Budget deficit / GDP 0 0 0 0 Trade linkages 0.26*** 0.19*** 0.21** 0.22 Financial linkages -0.01 0.01 0.07* 0 Macroeconomic similarities 0.01 0.07 0.01 0.01 Log likelihood -153.8 -57.41 -140.36 -56.06 9

Estimated hazard by Discrete semi-parametric model (monthly-type spells) .25 .2 .15 .1 .05 0 0 5 10 15 20 Duration Empirical_Hazard Discrete_Hazard_Model2 Discrete_Hazard_Model4 Estimated hazard by Discrete semi-parametric model (quarterly-type spells) .2 .15 .1 .05 0 0 5 10 15 20 Duration Empirical_Hazard Discrete_Hazard_Model_2 Discrete_Hazard_Model_4 10

Estimated hazard by Gompetrz model (monthly-type spells) .25 .2 .15 .1 .05 0 0 5 10 15 20 Duration Gompertz_Hazard_Model2 Empirical_Hazard Gompertz_Hazard_Model4 Estimated hazard by Gompetrz model (quarterly-type spells) .2 .15 .1 .05 0 0 5 10 15 20 Duration Empirical_Hazard Gompertz_Hazard_Model_2 Gompertz_Hazard_Model_4 11

 probability of currency crisis rises with undesired changes in job markets as well as macroeconomic fundamentals  Estimation results for trade linkages were constantly significant 12