Basel 2 has several fundamental weaknesses. First, it relies on banks own risk estimates. This is not incentive-compatible. The review was prompted largely by the observation that banks in large financial centers were increasingly circumventing the rules by regulatory arbitrage. The conclusion that the Committee came to at the very beginning of the process was that differences between regulatory and economic capital caused regulatory arbitrage, and the way to eliminate regulatory arbitrage was to make the regulatory rules converge on economic capital, that is, to make the rules 'risk-sensitive'. Given the flood of regulatory arbitrage, this was the obvious conclusion to draw, but it was wrong, and the review has run into severe difficulties as a result. Because they have limited liability, banks shift risk to others but keep the rewards, and so benefit too much from risk. They also suffer from internal agency problems that induce their traders and credit officers to seek risk. Banks may manage their risks perfectly for shareholders but pose too much risk to others. Important properties - such as risk, correlations, liquidity - of the very system that the Basel Committee is trying to protect are not exogenous but defined by the collective behavior of banks and other financial institutions. If banks manage risk in the same way, shocks, news and changes of opinion have greater effect.
Daníelsson, Shin and Zigrand report show that adding a value at risk constraint affects the demand for assets, and hence prices and price distributions. Their simulations suggest that general use of VaR lowers asset prices, increases their volatility, and increases the amplitude and duration of asset price response to large shocks. The same authors compare the fallacy of using for policy purposes models that assume exogeneity of risk to the Lucas Critique: the risks derive from banks' behaviour and are not invariant to changes of regime.
Basel 2 framework, in which several approaches to credit and operational risk are available, replaces one form of regulatory arbitrage with another. The IRB approach 13 generates higher capital requirements than the standardised approach on lower-quality assets, but lower requirements on higher-quality assets. Banks on the IRB approach will tend to acquire all the high-quality assets and banks on the standardised approach all the low-quality assets - the assets for which the standardised approach undercharges.
Banking groups will be tempted to 'cherry-pick' between the two regimes, but even if they do not, the banking system will do it automatically. Banks on the IRB approach may branch into jurisdictions where only the standardised approach is offered, or vice versa. Adverse selection is seen as a price worth paying to achieve a framework that can be applied by different banks and that contains incentives to improve risk management. But if the IRB approach is not a better form of risk management from a social perspective, these costs may not be worth bearing.
The second problem is that supervision may fail to achieve its objectives for any one of a large number of reasons. Supervision has two advantages over formal regulation: supervisors use a broad range of information, notably of a soft, subjective nature; and they generally try to encourage improvements in risk management, which is for many risks a more efficient form of insurance than capital. However, in reality supervisors may not have the skill, the power or the incentives to supervise effectively. A supervisory regime such as Basel 2 requires powerful yet benevolent supervisors, the stock of which is limited. Supervision, on the other hand, requires bureaucrats to exercise discretion, which may be used for good or ill. The effectiveness of supervision therefore depends on the incentives of supervisors, and of others such as bankers, politicians and auditors.
These incentives are affected by formal constraints, such as the law, but also by informal constraints of culture and norm, which vary across countries and persist over time. It is easy to build incentive mechanisms that fail to achieve regulatory objectives, and difficult to build mechanisms that succeed. Organizational structures give even the most public spirited supervisor reasons to misbehave. The power of supervision also depends on convention and norm, rather than on formal law; the tools of persuasion are subtle, and persuasion is difficult to monitor. Supervision is intrinsically interpersonal, and human relationships can distort decisions. Furthermore, supervision is a complex activity in which the supervisor acquires and process a great deal of information and makes decisions under uncertainty. Rationality is bounded. Moreover, even the richest supervisory agencies possess scant information about the relationship, if any, between supervisors' actions and bankers' behaviour. Whether supervisory actions are effective or entirely superstitious is not known. In summary, the new regime is founded on a largely untested belief that supervision works. It is not robust to supervisory failure. Thirdly, a shift towards banks' internal risk measures and on supervision implies a shift from rules to standards. This has hugely important implications barely considered by the Committee. Standards work better in some jurisdictions than others. High-level principles and qualitative standards delegate to bureaucrats the task of giving content to the law. Not all legal and political systems can easily accommodate this delegated administrative approach. Indeed, even in the US, the courts' willingness to accept delegation of powers to administrators is a modern phenomenon, and even now, as in the UK, administrators are subject to procedural constraints in the exercise of their powers.
Fourthly, the primary purpose of the international capital adequacy aregime is to protect against international competition in laxity as said by Kapstein in 1989, and the new regime will fail to achieve its purpose. Regulators impose externalities on each other because banks may branch across borders under home country regulation, affecting competitive conditions in other countries. Regulators respond strategically to other regulators' laxity by increasing their own. The 1988 Accord appears to have reversed the long-term trend towards lower capitalisation. Unfortunately, the race to the bottom has already begun again, for two reasons: the shift from rules to standards reduces observability, and there are more areas of national discretion. Standards are not contractible. Another regulator's standard can only be observed by observing both the formulation of the standard and every decision and the information used to give content to the standard in every case. Low regulatory and supervisory standards are therefore easy to disguise, as the IMF has observed. In the new world, peer pressure will have less disciplinary effect than in more easily observable, rule-based regimes such as the 1988 Accord. And markets have neither the information nor the incentives to punish those inadequately supervised. The number of areas in which national regulators are to choose between different options has jumped sharply. National discretion is useful only when Basel Committee members cannot agree on a single approach, and so by a revealed preference argument, they can be expected to use that discretion differently. National discretion is equivalent to a hole in the international regime. When it comes to Pillar 2, there is little desire for coordination. Institutional arrangements and attitudes to supervision differ widely across countries, and there is little common understanding of the role and purposes of supervision.
The final problem is excessive faith in disclosure requirements. The modern rationale for bank regulation is based on asymmetries of information and externalities. Banks know more than their depositors and regulators, and less than their borrowers. Disclosure requirements are that they reduce the asymmetries. Some disclosure requirements are almost certainly beneficial. There is a strong case for improving disclosure standards in most countries. Weak accounting and disclosure standards undermine the effectiveness of all three pillars. However, that since the production and processing of information is costly, the optimal disclosure requirement is finite. Furthermore, disclosure cannot correct for all the market failures. Banks are large, and impose extra costs when they fail. These costs are borne by creditors deemed worthy of protection (depositors), and by the economy in general. This last aspect is often neglected by those promoting disclosure. It is implied that those to whom bankers shift risk will require compensation if they are adequately informed. However, if all participants in the economy suffer indirectly from a bank failure, transactions costs may rule out such an outcome even under full information. Nordic countries suffered a banking crisis in the early 1990s in spite of a high level of disclosure.
Disclosure requirements, therefore, cannot by themselves correct for all externalities. Moreover, some asymmetries of information are inevitable. Banks exist as agents for depositors in lending to firms that cannot provide credible information on their financial health. As Eichengreen (1999) says, "it is unavoidable that borrowers should know more than lenders about how they plan to use borrowed funds. This reality is a key reason why banks exist in market economies. Bank fragility is inevitable." Disclosure may not have the desired disciplinary effect. A supply of accurate and timely information is necessary, but not by itself sufficient to discipline bankers There are several steps in the process of market discipline, and all may fail.
First, market participants must change behaviour in response to new information. Lamfalussy (2000) reports that bank lenders ignored relevant information in the public domain, notably BIS statistics showing a build-up of external debt and a shortening of its maturity, for some time before the Asian crisis hit. (The 1996 BIS Annual Report noted that Thailand had become the biggest debtor in the world.) They must acquire, process and act on information. This is costly - and may be subject to increasing marginal costs - and so participants will not act hyper-rationally but with 'bounded rationality'.
Participants must also have the incentive to acquire and process information, and so they must believe themselves to be uninsured; such a belief would be irrational in many countries. Furthermore, particularly among those with only a small stake in the bank's survival there are externalities to monitoring and incentives to free ride. Indeed, one reason why not all credit is allocated through the market is that banks do not free ride in the monitoring of their own private loans; and one rationale for regulation is that the regulator acts as a monitor on behalf of depositors.
Secondly, the market price of bank liabilities must reliably reflect participants' judgements of the 'fundamental' risks of the bank, rather than estimates of other people's judgements. Market efficiency requires that there be traders unconstrained by liquidity constraints or risk-aversion willing to trade against noise traders. Even when the market disciplines, it does not necessarily do so beneficially. Markets are driven by expectations about average opinion.Market shifts are associated with increasing agreement among participants. Homogeneity can be the enemy of stability and of liquidity. Anything that aids coordination can induce instability, and information can do that.
Market discipline is also procyclical. It has a deflationary impact in bad times. If there are some in the market who trade on short-term information, perhaps constrained by short-term liquidity, market discipline will be volatile. Informed traders should then trade not only on long-run fundamentals, but on their expectations of noise traders' behaviour. In good times, the market disciplines those, like PDFM, unwilling to acquire inflated shares in companies that have never produced a profit. Regulators are well aware of the problem of overreaction when it comes to suggesting that their own assessments of banks (CAMELS ratings, individual capital requirements) might be published. In Europe, the latest draft directive (European Commission Services, 2002) Article 128 states that individual capital requirements above the minimum shall not be published. The IMF/World Bank assessments of countries financial sectors are not generally published, for the same reason.
Thirdly, for bank managers to respond to market discipline by reducing risk, their personal welfare must fall with the price of the bank's liabilities. This requires strong corporate governance standards. In the new regime, market participants will be required to understand each bank's risk measurement system and, in comparing banks' risks, somehow correct for differences in reporting systems. They will also be expected to use the soft information required in Pillar 3 to correct for supervisors' different standards. The Basel 2 disclosure requirements embody an optimistic view of the benefits of greater disclosure.
To summarise, in its reliance on market discipline and on internal risk estimates, the new framework relies too much on the absence of market failure. This is intellectually incoherent. It is market failure that justifies regulation. And in its reliance on supervisory judgements, the framework is not robust to government failure.
