Lecture 4A: Empirical Literature on Banking Capital Shocks

Size: px

Start display at page:

Download "Lecture 4A: Empirical Literature on Banking Capital Shocks"

Philippa Lloyd
5 years ago
Views:

1 Lecture 4A: Empirical Literature on Banking Capital Shocks Zhiguo He University of Chicago Booth School of Business September 2017, Gerzensee

2 ntroduction Do shocks to bank capital matter for real economy? Often, we call these as bank credit supply shocks. Bank lending channel of the monetary policy Several conceptually important questions Can we separate credit supply (bank side) shocks from credit demand ( rm side supply shocks)? f credit supply shock matters, can rms substitute the bank credit by other forms of nancing? Bernanke (1983): f bank credit is really special, shocks to the banking sector a ect the level of economic activity and the nancial sector may amplify macroeconomic uctuations in output

3 First Generation: Aggregate Evidence (1) Kashyap et al. (1993): When tight monetary policy shrinks the size of the banking sector, it reduces the overall supply of credit to bank-dependent borrowers But, the reduction of supply of bank credit comes together with an increase in commercial paper issuance

4 First Generation: Aggregate Evidence (2) Kashyap and Stein (2000): Similar methodology, but focus on cross-sectional di erences of banks in response to policy shocks The e ect of monetary policy should be stronger for banks with weaker balance sheets Small banks, with less liquid balance sheets, etc Findings: impact of monetary policy on lending is larger for less liquid banks, and the e ect is largely attributable to the smaller banks Caveat: liquid/illiquid and large/small banks lend to di erent kinds of borrowers, whose demand may uctuate with the business cycle in a di erent way

5 Second Generation: Exogenous Variation Peek & Rosengren (2000) Results Look for shocks to banks credit supply that originate in a di erent geographical location Shock: Japanese banking crisis in the early 1990 s, arguably external to US credit markets mpact on geographically distinct commercial real estate credit markets with a large degree of Japanese bank penetration Lending by Japanese banks drops more than for domestic banks in markets with large Japanese bank presence Uses the crisis as an instrument to estimate the e ect of credit on commercial real estate activity Caveat: see Figure

6 Peek and Rosengren (2000)

7 Ashcraft (2005) (1) Variation of the same idea: bank failures that are not driven by local economic conditions Two incidents of healthy-failures: healthy subsidiaries of a multi-bank holding company (MBHC) failed following the failure of unhealthy lead banks "Cross-guarantee" provision the FRREA Act of 1989: allows the FDC to charge o any expected losses related to the failure of one subsidiary bank of a (MBHC) to the capital of a related subsidiary bank n 1992, lead banks of First City Bancorporation were declared insolvent (for loan losses to raqui banks made before the Gulf war). FDC expected losses of $500 million. 18 other subsidiaries held less than $300 million in primary capital, so they failed n July 1988, FDC chose not to renew assistance to the First RepublicBank Corporation. FDC hence take the subsidiaries of the MBHC as collateral, failing these non-lead banks

8 Ashcraft (2005) (2)

9 Ashcraft (2005) (3) Concludes that healthy bank failures have a signi cant and permanent e ect on real economic activity Main issue on causal interpretation: The co-existence of multiple non-healthy failures indicates that there were substantial negative local economic shocks in neighboring counties at the same time the healthy failures occurred Spillovers across county borders will lead to changes in economic activity that are not due to the healthy-bank failures n principle a good idea, by ideally you want a setting where nothing else is going on Same issue in Peek & Rosengren (2000): it is very di cult to argue that there is indeed nothing else going on

10 Third Generation: "Within" Estimators General idea: DiD that compares lending by a bank that receives a shock, relative to lending by a bank that did not, to the same rm Bertrand & Mullainathan (2003): compare plants in same location to control for local economic conditions By comparing two simultaneous outcomes for the same rm we get rid of rm speci c time varying shocks Shocks to rm demand for credit (e.g. due to changes in investment opportunities) are wiped out if in expectation they a ect the demand for credit across banks in the same way Requires data at the bank- rm-time level (indexed j, i, and t) Djt = 1 fbank j is a ectedg The following OLS regression gives you the e ect of the shock on credit supply Credit ijt = α ij + δ it + D jt + ε ijt

11 Khwaja & Mian (2008) (1) Shock: Following the nuclear tests in Pakistan in 1998, the government froze the exchange rate at which dollar denominated deposits in the Pakistan banking system could be withdrawn (default in dollar deposits) Massive withdrawals of dollar deposits

Khwaja & Mian (2008) (2) Data: Credit Registry Bureau of the State Bank of Pakistan, quarterly loan level information in Pakistan, July 1996 - March 2000 Classify banks in

12 Khwaja & Mian (2008) (2) Data: Credit Registry Bureau of the State Bank of Pakistan, quarterly loan level information in Pakistan, July March 2000 Classify banks in two groups Positive liquidity banks: banks that had above median growth in deposits after the (nuclear) shock Negative liquidity banks: those with below median deposit growth

13 Khawaja & Mian (2008) (3)

14 Khawaja & Mian (2008) (4) Loans from the bank experiencing a 1 percent larger decline in liquidity drops by an additional 0.6 percent, and the e ect can be entirely attributed to credit supply Caveats: "Within" approach only works for the intensive margin and for outcomes that are bank speci c (saying something about rm-level outcomes, such as defaults, sales or investments requires making additional assumptions) Di cult to draw conclusions on aggregate outcomes (same for any DiD estimation) Rely on the assumption that banks do not specialize Say banks with high dollar deposits specializes in funding importers/exporters, and massive change in real exchange rate collapses international trade This causes these specialized banks to shrink lending to these specialized rms

NBER WORKING PAPER SERIES TRACING THE IMPACT OF BANK LIQUIDITY SHOCKS: EVIDENCE FROM AN EMERGING MARKET. Atif Mian Asim Ijaz Khwaja

NBER WORKING PAPER SERIES TRACING THE IMPACT OF BANK LIQUIDITY SHOCKS: EVIDENCE FROM AN EMERGING MARKET Atif Mian Asim Ijaz Khwaja Working Paper 12612 http://www.nber.org/papers/w12612 NATIONAL BUREAU