1. Stigma in Financial Markets NORGES BANK Government Intervention and Moral Hazard in the Sector Financial September 3 2010 Olivier Armantier, Eric Ghysels, Asani Sarkar, and Jeffrey Shrader The views in this paper are those of the authors and not necessarily those of the Federal Reserve Bank of New York or the Federal Reserve System.

2. 2 “In August 2007,... banks were reluctant to rely on discount window credit to address their funding needs. The banks’ concern was that their recourse to the discount window, if it became known, might lead market participants to infer weakness — the so- called stigma problem.” -- Chairman Ben Bernanke (2009)

3. 3 Background  DW essential mechanism by which Fed supplies liquidity  Crisis period: Lender of last resort to mitigate systemic liquidity shortage when interbank market is dysfunctional  Normal times: Provide discretionary funding for solvent banks that are temporarily illiquid  Monetary policy: DW rate puts ceiling on interbank market rates  Focus on primary credit  Eligibility: All banks in good standing as determined by its Reserve Bank  Use supervisory ratings and data on bank capital  Collaterized loans  Only overnight loans (until 08/2007)

4. 4 Effectiveness of DW Lending  Historically, DW effectiveness has been called into question due to the perception of stigma  (e.g. Continental Illinois during crisis that began in 1982)  DW structure revised 2003 to become more like a standing facility  Prior to 2003: – Rates below market – Discretionary + administratively burdensome  After 2003: – Penalty rate (initially 100bp above Target FF rate) – No questions asked

5. 5 DW Effectiveness During Recent Crisis  August 2007: Fed strongly encouraged banks to borrow from DW  Cut penalty rate from 100 to 50bp  Extended term of DW loans from overnight to 30 days  Send messages to encourage DW borrowing  Banks did not borrow from DW  At least one DW borrower identified in media (Deutsche Bank)  DW stigma was perceived to be a problem  Fed introduced TAF in part to mitigate stigma

6. 6 DW and TAF Borrowing During Crisis

7. 7 DW and TAF Participation During Crisis

8. 8 Lack of Rigorous Evidence of DW Stigma  Anecdotal: DW Stigma often mentioned in the popular press  Suggestive evidence for pre-2007 DW mechanism:  Peristiani (1998): banks in the 1980’s may have become reluctant to borrow from Fed  Furfine (2001, 2003): banks are willing to borrow Fed Funds at rates above the DW rate  Fed funds and DW borrowings are not directly comparable  Ennis and Weinberg (2009) propose a theoretical model in which stigma emerges endogenously

9. 9 DW Stigma: Theory  DW stigma from adverse selection in loan markets. Conditions:  DW visit observed or inferred  DW visit more informative than private market borrowing  We are agnostic regarding the source of stigma  Ennis and Weinberg (2009):  Banks sell assets of unobservable quality to pay back IB loans  DW borrowing may be negative signal of asset quality  Banks borrow at IB rate > DW rate  Philippon and Skreta (2010)  Asset quality inferred from participation in government programs outside funding options cost of government programs  Bad banks opt out to signal good quality and get better rates

10. 10 Questions we Address  Part 1: Existence  Is there DW stigma?  How does incidence of DW stigma vary with bank characteristics and market conditions?  Part 2: Magnitude  What is the magnitude of the effective DW stigma premium (lower bound on the DW stigma premium)?  How did effective stigma premium change during crisis? Part 3: Market Effects  Effect of DW and TAF visits on banks’ interbank borrowing rates

11. 11 Contributions  Effectiveness of central bank liquidity supply during crisis  Stigma creates uncertainty as to the price of liquidity  Variation across banks and over time: Response of banks to DW policy becomes difficult to estimate  Liquidity mechanism design  Auction mechanism versus standing facility  Consequences of transparency  Market price effects of disclosure

12. 12 Results  Existence of stigma:  Strong evidence of existence of DW stigma  Several banks regularly bid in a way consistent with stigma  Probability of bidding above DW rate is higher:  For banks anticipating greater funding needs  When market funding conditions worsen  Size of Stigma:  At least 37bp at height of crisis (at least 150 bp after Lehman)  Asset price effect of DW and TAF visits:  Higher borrowing rates on the day of DW visit but not TAF

13. 13 What is DW Stigma?  Stigma exists if banks willing to pay more to avoid borrowing from the DW  Two possible forms of Stigma, wrt:  Market participants: – Market participants interpret DW borrowing as poor financial health  The Fed: – May trigger regulatory action (Camel rating, access to primary credit) – May limit ability to access DW in the future  We mostly focus on stigma wrt market

14. 14 Methodology: Basic Assumptions  Basic idea: compare banks’ TAF bids with DW rate  TAF and DW borrowing:  Similar eligibility criteria  Same collateral requirement  If rates equal, banks should be indifferent or favor DW borrowing -DW loans can be prepaid -DW open every day

15. 15 TAF and DW are Close Substitutes Term Auction Facility (TAF)Primary Credit Facility CollateralSame as DW for 28 day loans, additional overcollateralization of 25% required on 84 day loans Same broad set of collateral allowed as 28- day TAF auctions Eligible BanksPrimary credit eligible banks in good standing with enough collateral to make minimum TAF bid All banks with reserve account and high supervisory rating Minimum bid or loan amount $5 millionNone FrequencyGenerally once every two weeksAny time during normal business hours Loan TermGenerally 28 days or 84 daysOvernight through 90 days, renewable by borrower (up to 30 days before March 17, 2008) Maximum bid or loan amount 10 percent of total auction size or up to available collateral (whichever is smaller) Up to available collateral PrepaymentNot possibleFixed RateDetermined through competitive bidding at auctions Spread over FF target (FF+50 bp until 16 Mar 2008, FF+25 bp after) Similarities Differences

16. 16 Methodology: Existence of Stigma  Assumption: Bank has a maximum willingness to pay (MWTP) for borrowing from Fed  Depends on funding needs and outside funding options  Dominated strategy for bank to bid at TAF above its MWTP  Absent DW stigma, rational bank should never bid at TAF above its outside option (i.e. the prevailing DW rate)  If a bank’s TAF bid > DW rate, then this is evidence of stigma

17. 17 TAF bidding in the absence of stigma

18. 18 Rising share of banks bid above DW rate

19. 19 Banks bidding above DW rate, did so regularly

20. 20 Banks bidding above DW rate, did so regularly The majority of banks that bid above the DW rate did so regularly

21. 21 Summary: Existence of Stigma  Strong evidence of existence of DW stigma  Several banks regularly bid in a way consistent with stigma

22. 22 Determinants of bidding above DW rate  Probit model with bank specific random effects  Probability of bidding above DW rate is higher when:  Banks anticipate greater funding needs (by increasing collateral pledged before TAF and having more collateral)  Market funding conditions worsen (i.e. volatility of fed funds borrowing rate increases; LIBOR-OIS spread is higher)  Banks are smaller  Banks bid above DW rate previously

23. 23 DW stigma premium  DW stigma rate= highest rate banks is willing to pay to avoid DW  DW stigma premium=DW stigma rate – DW rate  Before TAF, bank visits DW when DW stigma rate < MWTP  DW stigma rate is a latent variable  Observable proxies can only provide lower bound

24. 24 Effective DW stigma premium  For a bank bidding above DW rate, Effective DW stigma premium = Highest TAF bid - DW rate  Banks are allowed two bids per auction: take the max of these bids  Only defined when banks TAF bid > DW rate  Limitations  Provides a lower bound on the bank’s DW stigma premium  Determinants of effective DW stigma premium need not be the same as that of the DW stigma premium  DW stigma premium may be constant but the effective premium could vary over time

25. 25 Estimate of Effective DW Stigma Premium Full SampleSummer 2008Lehman Mean effective DW stigma40.836.9142.7 Note: Summer 2008 is all auctions between Mar 24, 2008 and Sep 8, 2008.

26. 26 Economic Cost of Bidding Above DW rate A. Potential Cost Full SampleSummer 2008Lehman Total per Auction (millions USD)17.815.9164.4 Average per Bank per Auction (millions USD)0.430.262.05 Bid Above Cost/Interest Paid12.3%12.4%46.5% B. Realized Cost Full SampleSummer 2008Lehman Total Cost per Auction (millions USD)6.75.574.7 Average per Bank per Auction (millions USD)0.250.102.41 Bid Above Cost/Interest Paid9.1%5.6%40.0% Note: Summer 2008 is all auctions between Mar 24, 2008 and Sep 8, 2008. Maximum or potential cost = (TAF bid - DW rate)*Amount bid Actual or Realized cost= (TAF stop out rate - DW rate)*Amount received

27. 27 Summary: Effective DW Stigma Premium  Mean (median) stigma at least 26 (37) bp when TAF stop out rates were consistently > DW rate  Effective DW stigma premium varies across banks  Larger for small banks and banks with more collateral

28. 28 Market Response to DW and TAF Visit  Is visiting the DW and TAF associated with a subsequent decline in a bank’s interbank borrowing rate?  Expect lower impact after TAF visits if lower stigma  DW visit is not disclosed  Perhaps can be inferred from interbank borrowing activity or balance sheet changes  Timing of DW visits and interbank borrowing:  DW borrowing throughout the day  IB borrowing throughout the day, spike at day-end  Expect to see IB effect on same day or day after DW visit

29. 29 Fed Funds Borrow Rate after TAF Visit: Probit Analysis

30. 30 Fed Funds Borrow Rate: Matched Sample Analysis

31. 31 Conclusion  Banks preferred to pay premium at TAF to avoid borrowing at DW  Consistent with DW borrowing stigma  Banks paid a premium of at least 37 bp on average to avoid the DW during the height of the crisis  Incidence of stigma varied with bank and market funding conditions  Banks’ interbank borrowing rates are higher after DW visit but not after TAF visit

32. 32 TAF bidding in the presence of stigma