Comments on…Using Price Information as an Instrument of Market Discipline in Regulating Bank Risk Andrew Powell Inter American Development Bank and Universidad Torcuato Di Tella Prepared for World Bank Conference on Bank Regulation and Corporate Finance, Oct 26/27 2006
Why did the organizers ask me (1)? • Perhaps due to development of BASICin Argentina: 1996-1998? • Market and Traditional Supervision Complementary in 2 Ways: • Information Sets Different • Discipline • Bonds (Sub) Debt • Auditing • Supervision • Information • Credit Rating
Complementarities in Discipline • Regulators can force banks to disclose information to the market: • Information (Transparency), I of BASIC • Auditing rules A of BASIC • If markets react, supervisors will have to, less forbearance • Bonds in Basic • Credit Rating in BASIC • Markets rarely forbear!
Argentine Experience • Idea Summarized in “Basic Banking Supervision” mimeo Central Bank. • Experience Summarized (critically) in Calomiris and Powell (2000) in Mishkin ed. Prudential Supervision, What Works and What Doesn’t. • So I’m on the record…
Why did the organizers ask me (2)? • Or a Recent Paper with Arturo Galindo and Ana Maria Lobuguerrero (2004) • Panel-VAR tests on whether depositors punish risky banks across many countries in Latin America • AND whether banks respond, reducing risk • 3 Equations Quantity (Deposits), Price (Interest Rate) and Risk (Capital Ratio) • We find that a) depositors respond to risk AND that b) banks respond to depositors • Results summarized in UNLOCKING CREDIT: The Quest for Deep and Stable Bank Lending (IDB, J.Hopkins, 2005)
What do the authors do here? • A theoretical analysis of what we might expect if regulators condition their bank auditing decisions on market information • Its a very elegant model and authors obtain a set of quite fascinating and actually quite intuitive results
Result Number One • The “benchmark model” (with no stock price) has a mixed strategy equilibrium • The bank invests some portion (mixed strategy?) in the risky asset and the regulator audits with some probability • Using market information may increase the investment in the risky asset. • OR may drive the bank’s investment in the risky asset to zero!
The Model Considers Certain Aspects of Market Discipline • The model has markets anticipating the action of the regulator and endogenizing the incentive to collect information. • But there is no potential for forbearance and hence no role for the market to attempt to resolve this “supervisory failure”
Introducing Forbearanceinto the Model? • Assumption 1: always optimal to close a bank with risky assets when regulator signal is sA= -1. • Suppose this is not the case, as closing the bank has some non-pecuniary costs for the Super, but if left open the bank might recover – forbearance! • Now there may be room for the market to ensure (or increase the probability) that the supervisor will act. Does this enhance welfare?
“Timely Intervention” • The introduction motivates the model discussing timely intervention to avoid the collapse of a financial institution. • But is closing the bank in the model timely intervention? Timely intervention motivated by the concern of gambling for resurrection • But there is no real gambling in the model • As the bank gets closer to default, the incentive to invest in risky-asset rise probably does rise, but rebalancing is not allowed. • Market discipline may be particularly important at this point – Galindo et al Panel-VAR results. • But this is with a debt-like instrument
The Main Result • If the risky asset is very attractive relative to the safe one then, if the regulator conditions audit policy on the stock price, the bank may wish to invest more in the risky asset. • This startling result is actually intuitive, but again isn’t this why its best to use debt rather than equity? • Might argue it’s the same when you only have 2 states, but when “relative attractiveness” is compared then the upside is what is important! If it was a debt contract, would this result follow?
Sub-Debt “Trigger Rule” • In the model, the regulator always closes the bank on her bad signal and keeps it open on her good one. • In fact the strict-form sub-debt “trigger rule” proposal would: • Close the bank when the stock (debt) price low • Keep it open when high • No need to audit! Would this make any sense in the model? Would it make the perverse result even more likely (with equity/debt) ?
Argentina’s “Sub Debt Rule”:An Aside • Argentina’s so called sub-debt rule focused on the primary market not the secondary one. • Akin to ensuring that there is a stock (debt) price, but not conditioning closure on what the stock (debt) price turned out to be, with a specific rule. • The model here has a single investor but suppose there are inside and outside investors, a transparent price for all to see may have other advantages, another model?
Re-Audit Rules? • In the model the regulator audits more when the stock price is low, less when it is high • But when the stock price is low, the audit may give a positive signal, should this situation trigger a re-audit? • And vice-versa: when the stock-price is high auditing may give a negative signal. • Would re-auditing reduce the cost of closing (keeping open) a solvent (insolvent) bank? • On a disagreement, one might let the bank stay open another period but re-audit with a high probability. • What is the optimal n-period (re-) audit strategy?
Final Comment • A feature of financial markets is that they are essentially continuous. • But the model has 3 discrete periods and there is really only one market signal. • Distance to default measures (again as discussed in the introduction) consider first and second moments - tough to compute in this setting. • The set-up does not do justice to the richness of available market information
The Bottom LineTheory versus Policy • An elegant model, very thought-provoking, as perhaps economic theory should be. • But do the (perverse) results stem from a rather special model? • I’m holding onto my prior beliefs!