September 24, 2007
Beware the moral hazard fundamentalists
By Lawrence Summers
Central to every policy discussion in response to a financial crisis or the prospect of a crisis is the concept of moral hazard. Unfortunately, there is great confusion in many quarters about the circumstances when moral hazard is, and is not, a problem. The world has at least as much to fear from a moral hazard fundamentalism that precludes actions that would enhance confidence and stability as it does from moral hazard itself.
The term "moral hazard" originally comes from the area of insurance. It refers to the prospect that insurance will distort behaviour, for example when holders of fire insurance take less precaution with respect to avoiding fire or when holders of health insurance use more healthcare than they would if they were not insured.
In the financial arena the spectre of moral hazard is invoked to oppose policies that reduce the losses of financial institutions that have made bad decisions. In particular, it is used to caution against creating an expectation that there will be future "bail-outs".
Moral hazard forms the basis for criticism of a wide range of measures including, among others; large International Monetary Fund loans to countries experiencing financial panics; public sector actions to facilitate co-ordination of creditors, as in the famous 1998 case of the New York Fed and Long Term Capital Management; lender of last resort activities by central banks through their discount window; aggressive cuts in interest rates following collapses in asset prices; and the extension of government guarantees or quasi-guarantees to liabilities of financial institutions, as in deposit insurance or the US government’s support for the credit of mortgage lenders Fannie Mae and Freddie Mac.
Moral hazard fundamentalists misunderstand the insurance analogy, fail to recognise the special features of public actions to maintain confidence in the financial sector and conflate what are in fact quite different policy issues. As a consequence, their proposed policies, if followed, would reduce the efficiency of the financial sector in normal times, exacerbate financial crises and increase economic instability. They are wrong in three crucial respects. First, granting for the moment the relevance of the insurance analogy, as the economist Michael Mussa has pointed out, the prospect that people may smoke in bed is not usually taken as an argument against the existence of fire departments. Moreover, if there is “contagion” as fires can spread from one building to the next, the argument for not leaving things to the free market is greatly strengthened. In the presence of contagion there is every reason to expect that individual institutions will under-insure because they will not feel obliged to take account of the benefits their insurance will have for others. Second, the insurance analogy fails to take account of what is a key aspect of the financial context - moral hazard and confidence are opposite sides of the same coin. Financial institutions can fail because they become insolvent, as misguided lending or borrowing causes their liabilities to exceed their assets. But solvent institutions can also fail because of illiquidity simply because creditors rush to withdraw their funds and assets cannot be liquidated fast enough. In this latter case the availability of external support averts needless panic and contagion. More subtly, but no less important, the knowledge that efforts will be made to stand behind solvent institutions facing runs reduces the capital institutions have to hold, encourages investment in productive but illiquid projects and reduces the risk of contagion. Third, in the insurance template used in thinking about moral hazard, the insurer pays more out because of the behavioural changes induced by insurance, such as when the failure to install fire extinguishers makes fires more costly. Something parallel happens when the government guarantees a financial institution’s liabilities. But much of what financial authorities do in response to crises does not impose any costs on taxpayers and may actually make them better off. In the much criticised LTCM case no taxpayer money, except perhaps the cost of a lunch, was spent. A competent lender of last resort - in Bagehot’s sense of one who lends freely at a penalty rate against good collateral - actually turns a profit, as the IMF did in its response to the financial crises of the 1990s. Monetary policies that prevent deflation of the kind that cost Japan a decade of growth in the 1990s are another example of how a policy can respond to stress without imposing costs on taxpayers or the economy. Where does all of this leave policy? It certainly suggests that moral hazard is not always a negative with respect to policy responses to financial stress. In particular, the idea put forward by some that a central bank should act only once it is clear that financial problems have become serious enough to threaten a breakdown of the financial system or a sharp downturn in economic activity cannot be right. Instead, these considerations suggest that prudent central banks will make judgments during financial crises not on the basis of “avoiding moral hazard” but rather by asking themselves three questions. First, are there substantial contagion effects? Second, is the problem a liquidity problem where a contribution to stability can be provided with high probability or does it involve problems of solvency? Third, is it reasonable to expect that the action in question will not impose costs on taxpayers? If the answers to all three questions are affirmative, there is a strong case for public action. The writer is the Charles W. Eliot professor at Harvard University
Gerhard Illing: Larry Summers is right in one sense: during a financial crisis, prudent central banks have no other choice than to take public action. But he is awfully wrong to claim that acting this way is without cost. It may not impose costs on taxpayers, but it is bound to impose serious costs on financial stability in the long run. That is exactly what happened after the LTCM case. Rescuing reckless lenders is sowing the seeds for worse turmoil in the future.
Mervyn King stated the key point very precisely: “The provision of large liquidity facilities penalises those financial institutions that sat out the dance, encourages herd behaviour and increases the intensity of future crises.”
The tragedy is that, if a central banker tries to make such an obvious point transparent during a crisis, he is sure to lose the game. Martin Wolf called it a game of chicken. Unfortunately, it is much worse. All know right from the beginning that the central bank will have no other option than to play chicken. So better to not start playing at all.
There is a serious problem of dynamic inconsistency: ex post, there is no other choice than to ignore these crazy moral hazard fundamentalists. Financial markets, however, would be much more efficient if central banks could commit ex ante to the type of policy Mervyn King tried to implement. Liquidity provision needs to be priced efficiently. Otherwise you invite speculators to a free ride. But there is no free lunch in economics.
It was a humiliating but predictable experience to see the credibility of the Bank of England being smashed by a Northern rock. It is stunning that people like Summers or Barro do not recognise the consistency problem in this context. Clever financial market participants notice the commitment issue right from the start; they base their investment strategy on that knowledge. Reliance on central banks coming to rescue as lender of last resort is bound to discourage prudent behaviour. It is straightforward to show that this can result in serious distortions. For clever people, it should be obvious that you have to address the moral hazard issue in some way. I agree that not lowering interest rates during a crisis is not a sensible recipe. So the incentive problem needs to be addressed in some other way - by stricter regulation or alternative instruments. It is simple Tinbergen logic that you need more instruments to target several objectives.
Gerhard Illing is professor of economics at the university of Munich
Posted by: Gerhard Illing | September 25th, 2007 at 11:38 am | Report this commentWillem Buiter: Larry’s contribution contains a nugget of sense about liquidity, but this is buried deep under several layers of dross about moral hazard – a term I consider unhelpful. Its use encourages getting sidetracked into a didactic, essentialist argument about whether the bail-outs and other official financial support operations under discussion are indeed creating moral hazard in the strict, insurance-technical sense of the word. What we should be talking about is bad incentives producing bad - inefficient and inequitable - outcomes.
Providing liquidity to support markets. Liquidity is a key property of assets. It refers to the ability to sell the asset at short notice and at low transaction cost at a price close to its fundamental or fair value (fundamental or fair value is what you would pay for the asset if it could be bought and sold instantaneously and at zero transaction cost, that is, if ownership could be transferred costlessly and instantaneously). Liquidity is distinct from maturity or duration. Securities can have long remaining maturities or duration, yet be highly liquid because of the existence of deep, well-functioning secondary markets. Market liquidity is about trust and confidence. When normally liquid markets dry up, only the central bank can provide the public good of trust that restores liquidity swiftly and at little or no private or social cost. So it should be done.
More formally, correcting or mitigating market failure need not distort private incentives; injecting liquidity into a market that has become illiquid need not create moral hazard by distort private incentives for appropriate risk management in the future. Markets, that is, mechanisms for matching willing buyers and sellers at a price acceptable to both, are, in the case of assets like securities (or any store of value that can be resold in the future), subject to an inherent network externality: the likelihood of my being willing to buy a security at a price close to its fundamental or fair value is a an increasing function of the likelihood I attach to my being able to find a willing buyer for that security in the future at a price close to its future fundamental or fair value. When I believe that (1) I may have to sell the security in the future (possibly unexpectedly) and that (2) the future probability of finding a buyer is high, I am likely to buy now. If there are a lot of market participants with similar beliefs, the market today will be liquid. If there are a lot of market participants today with pessimistic beliefs about finding a future buyer at a price close to future fair value, the market today will be illiquid. Such a market will have at least two kinds of equilibria. One has self-fulfilling optimistic beliefs about future liquidity. Such a market will be liquid today. The other has self-fulfilling pessimistic ideas about future liquidity. Such a market will be illiquid today.
When the bad (illiquid) equilibrium prevails, one way to move to the good (liquid) equilibrium is for an agency whose liabilities have unquestioned perfect liquidity to inject liquidity into that market. In doing so it supports the market for the illiquid security. It does not bail out individual private businesses, that is, it does not act as a Lender of Last Resort (LOLR). The action will help the private businesses that hold the illiquid securities, but this assistance efficient: it corrects a distortion. The intervention renders liquid those securities that, because of fundamentally arbitrary albeit self-fulfilling beliefs, have become illiquid. The agency acts as a Market Maker of Last Resort (MMLR). The central bank is the natural agency to ‘liquidify’ (or should that be ‘liquefy’?) normally liquid markets that have become illiquid. That is because it is the source of ultimate, unquestioned, costless and instantaneous liquidity – the monetary liabilities of the central bank: commercial bank reserves with the central bank and currency.
Unlike the Fed and the ECB, the Bank of England does not appear to understand the nature of market liquidity and what could cause it to disappear and reappear. Instead of thinking of liquidity as a public good, it thinks of it as a private good that should be managed by individual financial institutions the same way they manage default risk or price risk.
Indeed, liquidity can be managed privately. Commercial banks could hold as assets only things that are highly liquid, like reserves with the central bank and government securities for which the secondary markets are normally deep and orderly (Treasury bills, gilts etc.). This would eliminate liquidity risk. However, such highly liquid asset portfolios would be socially inefficient (as well as unprofitable). We want our intermediaries to intermediate in support of long-term commitments by households and non-financial corporations. Some of the most productive assets are inherently illiquid. Someone has to hold them. If it can only be the originator of the illiquid asset (say a private entrepreneur investing in plant and equipment) the productive efficiency of the economy would be gravely impaired. Confidence that when some key financial market becomes illiquid, the central bank will support that market, by acting as MMLR (or buyer of last resort), is essential if our economy is to optimise its ability to generate productive but illiquid assets.
The Bank of England, until it changed its mind last week and decided to intervene in the three-month repo market against illiquid collateral (mortgages), appeared to believe that any market operations by the Bank at longer than zero maturity (overnight), represented a bail-out of all potential or would-be sellers of the illiquid collateral. That is a nonsense. It may be that some banks and other financial institutions indeed had too few liquid assets on their books, even for orderly market conditions. In that case, charging a premium over the Bank’s marginal cost of funds (Bank Rate) on the Bank’s lending in the three-month repo market makes sense. The Bank has decided to do so, setting the rate it charges for access to the Standing Lending Facility (the Bank’s discount window, 100 basis points above Bank Rate) as the floor for the rate it will charge on its three-month repos. It should also value the illiquid collateral according to its fair value rather than its face value, and impose other constraints to safeguard the interests of the tax payer. Finally, it should impose an appropriate haircut (discount) on the (conservatively estimated) fair value of the collateral. If all that is done, market liquidity support (overnight or at a 1, 3, 6, 12 or 24 month horizon) is not a reward for past reckless lending or borrowing. It is correcting a distortion - mitigating market failure.
Bailing out undeserving private financial businesses. Larry’s rather blanket support for bailing out distressed financial businesses (as distinct from supporting markets) is quite unconvincing. Arguments by analogy are cute but prove nothing. No, smoking in bed is not an argument against have a fire department. It is, however, an argument for having a clause in the homeowners’ insurance contract stating that no valid claim exists if the house burns down because one of the occupants was smoking in bed.
Contagion (in the sense of irrational herd behaviour) is as frequently mentioned (and modelled in neat academic papers) as it is uncommon in practice. When many private institutions or many countries are being dragged down by a common tidal wave, it tends to be because they have the same flawed fundamentals, not because of contagion. Contagion is an argument for deposit insurance, if the contagion takes the form of panicky depositors. It takes the form of market support (MMLR) action rather than support for individual financial businesses (LOLR) action if the contagion affects the liquidity of the markets for other financial instruments. State entities, including the central bank, the deposit insurance agency and the Treasury should support markets and other social mechanisms with clear public good properties, like the payment, settlement and clearing systems. Individual private businesses should be directly supported only if this is necessary for the safeguarding of some socially valuable ‘institution’ (in the proper sense of the word institution, as opposed to its use in financial ‘institution’, where it simply means ‘business’).
I cannot think of a single financial institution that is too big to fail, in the sense that it would damage some systemically important social institution. If deposit insurance is deemed important, whether because deposits are deemed an important part of the payment mechanism or because of distributional, social or political concerns, let’s guarantee deposits, but allow the institutions issuing them to fail. In the UK, Northern Rock was both granted an uncapped and open-ended Liquidity Support Facility (credit line) with the Bank of England and an unlimited guarantee for its existing depositors (and most other unsecured creditors, except for the holders of subordinated debt!). You might be able to make a case for either one of these support interventions, but not for both.
To hold out the disgraceful bail-out of LTCM as an example of how to act in a crisis is extraordinary. Indeed no public money was involved. But the Fed (through the Federal Reserve of New York) put is reputation at risk, and in my view damaged it severely, by enabling and facilitating this shoddy arrangement - offering its ‘good offices’.
As a result of the bail-out of LTCM, there was never any serious effort to address the potential conflicts of interest arising from simultaneously financing hedge funds, investing in them, and making money executing trades for them, as many investment banks did with Long-Term Capital Management. The results are there today for all to see. Things were even worse because apart from the institutional potential conflict of interest that is present whenever a party is both a shareholder in and a creditor to a business, the bail-out created a serious corporate governance problem because executives of one of the financial institutions that funded the bail out had themselves invested $22m in LTCM on their personal accounts. Using shareholder resources for a bail-out of a company to which you have personal exposure is unethical, even where it is legal. The bail-out of LTCM smacks of crony capitalism of the worst kind. The involvement of the Fed smacks of regulatory capture.
It is clear from Larry’s record at the World Bank (1991-1993) and at the US Treasury (from 1993 till 1995 as Under Secretary of the Treasury for International Affairs, from 1995 till 1999 as Deputy Secretary of the Treasury and from 1999 till 2001 as Secretary of the Treasury), that he has never seen a potential bail-out he did not like: the United States support programme for Mexico in the wake of its 1994-1995 financial crisis, the international response to the Asian financial crisis of 1997 and the 1998 Russian crisis and the Fed’s response to the 1998 LTCM crisis. I recognise the upside of bail-outs for those who arrange them: they look like movers and shakers, making and shaping events. It’s heroic, in an industry where heroism can be rarely displayed. But in all of the examples mentioned above, the bail-out did more harm than good.
Finally, Larry needs to add at least two other questions to his list of three - (1) Are there substantial contagion effects?; (2) is there a liquidity or a solvency problem?; (3) will there be costs to the taxpayer? - central banks ought to ask themselves during financial crises. These are:
(4) Will this action (Lender of Last Resort bail-out of individual private financial businesses, Market Maker of Last Resort liquidity injections into the markets) have a material impact on the likelihood and severity of future financial crises?”
Posted by: Willem H. Buiter | September 26th, 2007 at 2:57 pm | Report this comment(5) Will this action produce any net social benefit?