HOUSEHOLD DEBT AND BUSINESS CYCLES WORLDWIDE

Size: px

Start display at page:

Download "HOUSEHOLD DEBT AND BUSINESS CYCLES WORLDWIDE"

Egbert Baldwin
5 years ago
Views:

1 DISCUSSION OF: HOUSEHOLD DEBT AND BUSINESS CYCLES WORLDWIDE BY MIAN, SUFI AND VERNER Emi Nakamura Columbia University December 2015 Nakamura Inflation Expectations December / 24

2 Could a credit boom sow seeds of destruction for the macroeconomy?

3 COMPLETE MARKETS BENCHMARK Households borrow to smooth consumption High borrowing presages high growth (and low current income) Aguiar and Gopinath (JPE, 2007) High expected growth leads to CA deficits Nakamura Inflation Expectations December / 24

4 INCOMPLETE MARKETS Households are borrowing constrained High borrowing when borrowing limits expand Endogenous: High output implies greater collateral e.g., Midrigan and Philippon (2011) Exogenous: Shocks to credit supply e.g. Eggertsson & Krugman (2012); Korinek & Simsek (2015) This paper! Nakamura Inflation Expectations December / 24

5 MAIN EMPIRICAL RESULT (1) (2) VARIABLES Growth GDP Growth GDP Lag HHD Growth *** *** (0.0707) (0.0706) Lag NFD Growth (0.0287) Constant 9.810*** 9.956*** (0.305) (0.305) Observations R-squared Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Nakamura Inflation Expectations December / 24

6 GDP GROWTH (T TO T+3) VS. HHD GROWTH (T-4 TO T) Nakamura Inflation Expectations December / 24

7 PRECURSOR: JORDA, SCHULARICK AND TAYLOR (2011) Nakamura Inflation Expectations December / 24

8 PRECURSOR: JORDA, SCHULARICK AND TAYLOR (2014) Nakamura Inflation Expectations December / 24

9 MY COMMENTS Robustness Endogenous vs. exogenous credit Extending the sample Nakamura Inflation Expectations December / 24

10 ROBUSTNESS (1) (2) (3) VARIABLES Growth GDP Growth GDP Growth GDP Lag HHD Growth *** *** *** (0.0707) (0.0785) (0.0812) Trend *** (0.0547) Constant 9.810*** 377.7*** 7.058*** (0.305) (109.0) (1.226) Year FE No No Yes Observations R-squared Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Nakamura Inflation Expectations December / 24

11 CORRELATION COMES FROM LOW GROWTH PERIODS (1) (2) VARIABLES Baseline >2% Growth over 3 Years Lag HHD Growth *** (0.0707) (0.0546) Trend ** (0.0565) Constant 9.810*** 290.3** (0.305) (112.4) Observations R-squared Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Nakamura Inflation Expectations December / 24

12 GDP GROWTH (T TO T+3) VS. HHD GROWTH (T-4 TO T) Nakamura Inflation Expectations December / 24

13 ONLY > 2% GROWTH OVER 3 YEARS Nakamura Inflation Expectations December / 24

14 ENDOGENOUS VS. EXOGENOUS CREDIT Mian et. al argue a credit boom is destructive But what if credit expands with Y...And Y is mean-reverting Implies: Expansion in credit (and Y) today associated with future Y decline But there is nothing bad about the credit growth Can this explain the data? Nakamura Inflation Expectations December / 24

15 NOT REALLY... (1) (2) VARIABLES GDP Growth GDP Growth Lag HHD Growth *** *** (0.0707) (0.0645) Lag GDP Growth (0.127) Constant 9.810*** 10.13*** (0.305) (1.326) Observations R-squared Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Nakamura Inflation Expectations December / 24

16 DO EXOGENOUS CREDIT SHOCKS EXPLAIN DATA? Eggertsson & Krugman (2012); Korinek & Simsek (2015): Credit expansion leads to boom Subsequent credit contraction leads to bust Would expect effect of lagged credit on output to occur via effect on current credit Do we see this in the data?...not quite Nakamura Inflation Expectations December / 24

17 CONTROLLING FOR CURRENT CREDIT GROWTH (1) (2) (3) VARIABLES GDP Growth GDP Growth GDP Growth Lag HHD Growth *** *** (0.0707) (0.0716) Curr. HHD Growth 0.140** (0.0669) (0.0786) Constant 9.810*** 8.124*** 9.385*** (0.305) (0.293) (0.414) Observations R-squared Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Nakamura Inflation Expectations December / 24

18 CONTEMPORANEOUS OUTPUT VS. CREDIT GROWTH Nakamura Inflation Expectations December / 24

19 FIRST VS. SECOND DERIVATIVES Lagged credit growth appears to have a negative effect on future growth even controlling for current credit growth Negative effect of credit boom not just a consequence of future credit bust Nakamura Inflation Expectations December / 24

20 DYNAMICS OF DEBT VS. OUTPUT (1) (2) VARIABLES Output Dynamics Debt Dynamics Lag HHD Growth *** 0.116* (0.0707) (0.0644) Constant 9.810*** 4.025*** (0.305) (0.278) Observations R-squared Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Nakamura Inflation Expectations December / 24

21 FIRST VS. SECOND DERIVATIVES Lagged credit growth appears to have a negative effect on future growth even controlling for current credit growth Negative effect of credit boom not just a consequence of future credit bust Credit keeps rising after positive credit shock (albeit at a slower rate) Nevertheless, output contracts Need a story where slowdown in credit growth causes recession (Kermani, 2012) Nakamura Inflation Expectations December / 24

22 EXPANDING THE SAMPLE PERIOD Relatively short sample period for many countries Median country <20 yrs data Limitation: Household debt data But...Mian et al. also consider IV for household debt Instrument: Spread on 10-yr bonds vs. US Much longer time series available Nakamura Inflation Expectations December / 24

23 REGRESSIONS USING SOVEREIGN SPREADS VARIABLES (1) (2) (3) IV OLS >=1965 Baseline Baseline Lag HHD Growth *** (0.240) 10-Yr Bond Spread vs. US 0.626*** (0.180) (0.179) Trend *** (0.0302) Constant 10.91*** *** 408.3*** (0.820) (0.0195) (60.57) Observations R-squared Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Nakamura Inflation Expectations December / 24

24 RESULTS USING LONGER SAMPLE PERIOD IV analysis suggests increase in spreads leads to lower growth due to contraction in credit Reduced form: Higher spreads yield lower future growth Can run reduced form on larger sample back to 1965 Effect much weaker Different effect of household debt historically? Nakamura Inflation Expectations December / 24

25 CONCLUSIONS Robustness Key role of big recessions Endogenous vs. exogenous credit Controlling for current credit doesn t kill effect of lagged credit It should in simple models Extending the sample Yield spread effects go away for older data Nakamura Inflation Expectations December / 24

Household Debt and Business Cycles Worldwide

Household Debt and Business Cycles Worldwide Atif Mian, Amir Sufi, and Emil Verner Princeton University, University of Chicago and Princeton University June 2016 Motivation: Household Debt and Output Growth