University of Pennsylvania & NBER. This paper reflects only the authors views, and not those of the IMF

Transcription

1 An Anatomy of Credit Booms and their Demise Enrique G. Mendoza University of Pennsylvania & NBER Marco E. Terrones IMF This paper reflects only the authors views, and not those of the IMF Motivation and objectives Conventional wisdom holds that bad times often follow credit booms, but yet until recently we had 1. Limited evidence on CBs, and their macro/micro linkages. Lacked robust method for identifying & measuring credit booms Mendoza & Terrones () proposed a new thresholds method to study credit booms, but with a sample ending in we missed all the fun!!! In this paper: In this paper: 1. Review/upgrade our methodology to identify/measure CBs. Examine macro dynamics around CB episodes 3. Conduct frequency analysis of CBs, crises and triggers. Examine linkages between CBs and firm & bank indicators 1

2 Main findings 1. Using 19-1 data for 1 ICs & EMs, we find 35 credit booms in each group (1 around recent crisis),.% overall frequency. Booms synchronized globally, centered around big events 3. Three striking similarities across EMs and ICs: a) Similar duration and magnitude (normalized by s.d. of credit) b) Banking crises, currency crises or Sudden Stops often follow credit booms, and with similar frequencies in EMs and ICs c) Credit booms follow financial reforms, surges in capital inflows and TFP, and are far more common w. managed ex. rates. CBs feature clear cyclical pattern in production, absorption, current account, capital flows, real ex. rates (in EMs), asset prices and firm- and bank-level financial indicators Mendoza-Terrones Thresholds Method Country i is in a credit boom if: l it, ( l i) =1.5 boom threshold, set to std. Normal Pr( l )=.5 it, / ( li) 1.5 l it deviation from HP trend in log real credit per capita ( l i ) standard deviation of l it. Key differences with Gourinchas et al.1 1. Country-specific lower bound ( ( l i ) ). Standard HP detrending, not expanding trend 3. Credit measured as real credit per capita, not share of GDP Significant quantitative implications: robust to credit measure, unbiased credit trends, macro/crises linkages

3 Comparing methods: The case of Chile Credit booms: seven year event windows (Cross-country means and medians of cyclical component of credit) 35 Industrial Countries 35 Emerging Economies

4 Normalized credit boom episodes (relative to s.d. of credit in each country) INDUS TRIAL C O UNTRIES BEL_1979 GRC_7 USA_7 JPN_199 Normalized credit boom episodes (relative to s.d. of credit in each country) EMERGING MARKET EC O NOMIES 1/ AUS_1973 AUS_19 IRL_7 BEL_199 IRL_1979 ESP_7 JPN_197 AUT_1979 BEL_7 CHE_7 SWE_7 NOR_197 USA_19 AUT_197 DEU_ ITA_199 GRC_197 DEU_197 ITA_1973 FRA_199 NOR_7 CHE_199 DNK_197 PRT_ GBR_199 SWE_199 NZL_197 NLD_1979 FIN_199 GBR_1973 PRT_ RUS_7 PAK_19 IND_199 POL_ 1/ Ongoing credit booms are shown in green. URY_19 PER_197 THA_197 BRA_199 CHL_19 NGA_19 EST_7 TUR_1997 KOR_199 TUR_197 SVN_7 ECU_1997 ROM_199 HKG_1997 CRI_1979 PHL_1997 URY_ NGA_ IDN_1997 SGP_193 EGY_191 HKG_1 COL_199 PHL_193 CIV_1977 THA_1997 ZAF_7 VEN_7 MYS_1997 ISR_197 MEX_199

5 CBs are synchronized around big events 1 1 IND EMEs Global Financial Crisis / Ongoing credit booms are shown in green. Duration statistics

6 Macro credit cycles Seven-year event windows centered at CB peaks Clear credit cycle pattern similar across EMs and ICs 1. Y, C, I, YN, KI (and RER & G in EMs) rise above trend in upswing, move below trend in downswing. CAY shifts from deficit to surplus 3. Normalized by standard deviations fluctuations are similar (except YN and RER larger in EMs). Minor changes in inflation and in ICs little change in RER & G 5. Equity prices rise 5-3%, housing prices rise 1-15% 15% followed by price crashes in downswing 5% of CBs feature booms in Y & C in EMs and ICs (%/3% are I booms and 31%/% are YN booms in ICs/EMs) Output credit cycles (Cross-country means and medians of cyclical component of GDP) Industrial Countries Output (Y) Emerging Economies Output (Y)

7 Credit cycles in absorption (Cross-country means and medians of cyclical components) CB & Output/Absorption Booms Industrial Countries Emerging Economies Output.9. Non-tradable output.31. Consumption.9. Investment..3 7

8 Credit cycles in YN & RER (Cross-country means and medians of cyclical components) Industrial Countries Emerging Economies Non-tradables Output (YN) Non-tradables Output (YN) Real Exchange Rate (RER) Real Exchange Rate (RER) Credit cycles in the current account Credit cycles in the current account (Cross-country means and medians of cyclical component of CAY)

9 Credit cycles in net capital inflows (Cross-country means and medians of cyclical component) Industrial countries Emerging markets Credit cycles in asset prices (Cross-country means and medians of cyclical component) Industrial countries Emerging Markets 9

10 Credit cycles and inflation (Cross-country means and medians of cyclical component) Credit cycles: regional features (Cross-country medians of cyclical component) 1. Industrial Countries: G7 vs Nordic Countries t= G7 1.1 Real credit Nordic 19. G7. Output (Y ) Nordic 3.79 G7.11 Non-tradable Output (YN ) Nordic.533 G7.3 Consumption ( C ) Nordic 3.3 G7 9.9 Investment t(i ) Nordic 1.5 G7-3.9 Real Exchange Rate (RER ) Nordic 3. G Current Account-GDP ratio (CAY ) Nordic

11 Credit cycles: regional features (Cross-country medians of cyclical component). Emerging Economies: Latin America (LA) vs Asia vs Transition t= LA Real credit Asia.175 Transition 3.1 LA 7.79 Output (Y ) Asia 5.77 Transition 7.1 LA 1.31 Non-tradable Output (YN ) Asia.37 Transition.3 LA.51 Consumption ( C ) Asia.3 Transition.31 LA.97 Investment (I ) Asia 1. Transition.71 LA 9.3 Real Exchange Rate (RER ) Asia.75 Transition.5 LA -.99 Current Account-GDP ratio (CAY ) Asia -.33 Transition Financial crises after credit boom peaks (frequency analysis).3.5 Advanced economies Emerging markets. All Banking Crisis Currency Crises Sudden Stops 11

12 Credit booms and potential triggers (frequency analysis).5.5 Advanced economies. Emerging markets.35 All Large capital inflows Significant productivity gains Large financial sector changes Logit analysis of CB triggers Explanatory Variables (1) () (3) () Lagged Net Capital Inflows.35***.319*** (percent of GDP, five-year average of yearly changes) [.17] [.11] Lagged Financial Reform.15.1 (five-year average of yearly changes) [.] [.7] Lagged Total Factor Productivity Growth.19*. (five-year average) [.] [.] Lagged Total Factor Productivity Growth x Advanced.39** Country Dummy [.173] Advanced Country Dummy.113 [.95] Constant -3.*** -3.*** -3.73*** -3.77*** [.11] [.1] [.1] [.] Memorandum Number of Observations Log Likelihood AUC

13 Credit booms and exchange rate regimes (frequency analysis)..7 Advanced economies Emerging markets All. Managed Dirty Float Floating Mixed Micro credit cycles: corporate financial indicators (firm-level medians averaged across countries) 9 7 Total Debt to Book Value of Equity (LBV) Total Debt to Market Value of Equity (LMV) Industrial Countries 3 Industrial Countries 3 Emerging Markets 1 Emerging Markets... Tobin's Q (Q1) 1 1 Effective Interest Rate (EIR) Emerging Markets. 1. Industrial Countries Industrial Countries 1.. Emerging Markets.. 13

14 EM corporations: Tradables v. Nontradables Total Debt to Book Value of Equity (LBV) Tradable Total Debt to Market Value of Equity (LMV) RER (Right Axis) Non- Tradable RER (Right Axis) Tradable Non- Tradable Current Liabilities to Sales (LWK) RER (Right Axis) Non-Tradable Tradable Rajan-Zingales Index of External Financing (RZ) Non- Tradable Tradable RER (Right Axis) Bank-level indicators 1 Asset Quality 5 Credit Operations 1 NPL 55 LAR Profitability 15 Capital Adequacy CAR 1 ROA

15 Conclusions M-T method to study CBs in a sample of 1 countries for 19-1 period yields six main findings CBs in ICs and 35 in EMs, aggregate frequency:.%. Similar size, duration and macro fluctuations, w. regional diffs. 3. CBs associated with surges in capital inflows, large TFP gains, fin. reforms, and managed exchange rates. CBs followed by financial/currency/ss crises 5. CBs aligned with fluctuations in firm-level fin. indicators. CBs aligned with changes in banking fragility These are stylized facts that need to be explained (both upswing and downswing of credit booms matter) Monitoring tool for tracking CBs & macro/micro finance linkages 15