One area where competencies are, at the moment, almost completely centralized at the European level is financial regulation. This post explains how explicit and implicit government guarantees for bank debt incentivize banks to take on excessive risk. The vast amount of financial regulations produced by governments over the decades constitutes the hapless attempt to deal with the perverse incentives created by those government guarantees.
Brexit gives the UK the opportunity to embark on a new approach in this field. In the 19th century banks funded themselves with 40% to 50% equity. (1) In 2007, the US banking industry’s equity to asset ratio was 3.8 percent. For the 10 largest banks it was only 2.8 percent. (2) What had happened?
The reason for the drastic decline in capital ratios during the 20 th century is simple: since the great the depression bank debt has been explicitly (such as in the US) or implicitly (such as in the UK where bank deposit insurance was introduced not until 1979) guaranteed by the government.
If creditors of a bank will be repaid regardless of the condition of the bank, they stop worrying about the condition of the bank. In particular, they will give the bank a discount on the cost of the funds it borrows and will not demand compensation for a lower capital ratio (i.e. a higher probability of bankruptcy).
In short: the government guarantee is a massive subsidy of debt financing. And minimizing equity to asset ratios maximizes the value of this subsidy to the banks’ shareholders.
A razor-thin capital ratio creates perverse incentives for a bank’s investment decisions. Consider the following example:
A bank is funded by 3 billion of capital and 97 billion of debt. The bank’s management can choose between two alternative strategies:
- Strategy A: The bank invests the 100 billion in relatively secure government and company bonds. After one period the return will be either 110 billion or 100 billion (with a probability of 50%, respectively).
- Strategy B: The bank invests the 100 billion in US subprime mortgages. After one period the return will be either 65 billion or 125 billion (with a probability of 50%, respectively). In order to keep things simple, we normalize the interest rate on debt to 0%.
Then the different stakeholders’ payoffs for the two possible strategies are given as follows:
Payoffs for strategy A (in billions):
|return if lucky||97||113||110|
|return if unlucky||97||3||100|
Payoffs for strategy B (in billions):
|return if lucky||97||28||125|
|return if unlucky||97||0||65|
Note that strategy A is the efficient strategy because it maximizes the expected return from lending. However, by following an inefficiently risky business model (strategy B) the bank can increase the expected profits for its shareholders. If we assume that the bank’s management acts in the interest of the shareholders, it will choose strategy B.
Given (explicitly or implicitly) guaranteed bank debt, there are basically four options available to try and reduce the incentives for banks to take on excessive risk (3):
- Minimum capital requirements
- Directly regulating the banks’ lending policies
- Limiting competition in the financial industry so that banks have a large franchise value, which they are reluctant to put at risk
- Private regulation
In the UK, there was no government regulation of banking until 1979. Instead, the behavior of banks was subject to tight private regulation. This private regulation of banking was then substituted by government regulation in the 1980s.
I do not want to write a lengthy discussion on the question of which alternative is the least costly in dealing with the incentive problems arising from the implicit subsidy by the taxpayer. There are good reasons to believe an incremental, decentralized and evolutionary system of market-based regulation to be superior to centrally designed government regulation. (4)
But even if this is the case, private regulation arising as a response to the incentive problems resulting from explicit and/or implicit government guarantees is still costly. Indeed, the evolved system of private regulation in the UK banking industry was giving the appearance of a restrictive cartel. If my analysis is correct, this “cartel” served a useful social function, namely to deal with the incentive problems created by the implicit government guarantee. Nevertheless, it also involved costs.
At the root of the problem are the taxpayer guarantees. If the UK government abolished deposit insurance (which is, of course, not insurance in the true sense of the word but simply a government guarantee) and also announced that there would be no bailing out of creditors any more in the future, would this eliminate the perverse incentives in the financial industry?
Without being accompanied by certain changes in the institutional environment (that I will discuss below) such an announcement would simply not be credible and hence not effective: there are, at the moment, banks that are (potentially) “too important to fail”. Hence, the government has a strong incentive to renege on its promise and rescue such banks in the event of a crisis.
Anticipating this too-important-to-fail problem facing the government, the creditors of such banks will be content with relatively low risk-premiums. Again, the capital structure irrelevance principle (5) will not hold and the banks will still be able to reduce their weighted average cost of capital by increasing their leverage to astronomical levels.
Without first solving the too-important-to-fail problem, the government cannot credibly commit to not bail out failed banks. But is there a solution to this problem at all?
Turmoil in the financial industry can affect the economy as a whole, if the turmoil causes disturbances to the flow of credit in the economy. Such disturbances can be the result of either a liquidity crisis or a solvency crisis.
A bank run, e.g., constitutes a sudden surge in liquidity demand. A surge in liquidity demand can potentially force a bank or several banks out of business, which again can cause disturbances to the flow of credit in the economy.
However, a solution to the problem of liquidity crises has already been found more than 100 years ago by Walter Bagehot (6): liquidity provision by the central bank. As long as there is no failure of the interbank market, this liquidity provision takes the form of ordinary monetary policy. Otherwise, the central bank can provide additional liquidity by lending to individual banks on collateral. (By lending only on collateral, the central bank can avoid having to make a judgement on the solvency of a borrower. )
The Bank of England has followed Bagehot’s advice. Therefore, Britain has had an effective lender of last resort since 1866. Since then, sudden surges in liquidity demand have not resulted in significant disturbances to the economy any more. (8)
The Bank of England has the tools at hand to ward off liquidity crises but, under the current monetary policy regime, the central bank’s effectiveness is limited in the case of a solvency crisis (such as the one that started in 2007/08).
In a solvency crisis (i.e. in a situation where the banking system is undercapitalized), simply providing banks with liquidity is not enough (at least as long as nonbank credit is no perfect substitute for bank credit).
If a bank is undercapitalized, i.e., if its capital to asset ratio is too small relative to market and/or regulatory standards, it may well ration its credit supply. (9) Note that “credit rationing” means declining to lend at any interest rate, not merely a simple leftward shift of the loan supply (holding constant the safe real interest rate). A bank engaging in credit rationing will simply not increase its balance sheet any further.
Under the assumption of perfect substitutability of bank and nonbank credit, even credit rationing or bankruptcies across many banks in the economy would not have any adverse effect on the economy: a decline in bank credit would be exactly offset by an increase in nonbank credit without being associated with an increase in the real interest rate.
In reality, bank and nonbank credit are not perfect substitutes. Bank-dependent borrowers such as some small businesses, whose traditional lender is engaging in credit rationing or has gone out of business, may be cut off from lending or forced to borrow from other banks.
Compared to the traditional lender, the cost of the credit obtained from the new bank is likely to be higher at any given safe real interest rate. The reason for this is that the new lender has not had the opportunity to acquire the same amount of information on the borrower as the old lender. Hence, a higher risk premium can be expected.
If bank-dependent borrowers have to pay higher interest rates or are completely cut off from lending, investment demand falls at any given safe real interest rate. If the central bank does not reduce the safe interest rate by expanding the money supply, this will lead to a fall in aggregate demand and therefore a deviation of output from its natural level. (10)
Unless the safe nominal interest rate is at (or too close to) zero, the adverse effect of credit rationing and bank failures on aggregate demand can be completely offset by conventional monetary policy.
The more banks fail or engage in credit rationing, the more sharply the central bank needs to cut the safe nominal interest rate in order to keep aggregate demand stable. If the negative demand shock resulting from credit rationing and bank failures is sufficiently strong, the central bank will not be able to fully offset this shock by cutting rates: the nominal safe interest rate cannot be (substantially) lower than zero.
The fact that the central bank’s power to offset negative aggregate demand shocks by cutting the safe nominal interest rate is limited (namely by the zero lower bound) is the main argument put forward to rescue failed banks with taxpayers’ money.
The failure of a sufficiently large bank (or even several banks) may well cause a massive negative shock to aggregate demand that cannot be offset by conventional monetary policy (i.e. by cutting the nominal safe interest rate). The economic cost associated with such an outcome (especially the cost in the form of a potentially significant rise in unemployment) may well be deemed too high to be politically acceptable.
However, while there is obviously a limit to reductions in nominal interest rates, there is no limit to the extent to which the central bank can increase the money supply. That was Milton Friedman’s point when he complained about the failure of US monetary policy in the 1930s.
Of course, it is not enough to just increase the money supply. In order for the expansion of the money supply to increase aggregate demand, markets have to believe the increase of the money supply will be permanent. In short: the central bank has to commit to temporarily higher inflation (NGDP growth) in the future. (11)
A fixed inflation target cuts off the Friedmanite money expansion route to boosting aggregate demand at the zero lower bound. If the central bank does not commit to deviate from its inflation target, market participants know that the central bank will collect the newly printed money again as soon as the economy is not subject to the aggregate demand shock any more. Hence, the effect of the monetary expansion on spending will be negligible. (12)
Bank failures do not only constitute shocks to aggregate demand but possibly also to aggregate supply. (13)
Even if bank failures and/or credit rationing do not reduce total spending on investment (due to expansionary measures taken by the central bank), investment by bank-dependent borrowers will decrease in favor of investment by bank-independent firms. This distortion in the allocation of investment spending can impact the productive capacity of the economy.
A large firm with access to the credit market may be induced to carry out more investment than usual, while a small firm without access to the capital market may not be able to fund an investment project with a relatively higher expected return (compared to the marginal investment project carried out by the large firm).
Aggregate supply shocks cannot be offset by monetary policy. However, as long as the central bank stabilizes aggregate demand, the negative consequences of the supply shock will be limited. In particular, they will not pertain to the rate of unemployment.
A supply shock resulting from bank failure(s) always reduces the standard of living. But it only leads to higher unemployment, if the central bank allows aggregate demand to drop alongside aggregate supply.
When nominal spending falls, there is less money to pay employees. If nominal wages were fully flexible, they would adjust downwards. The number of jobs would stay the same. In reality, nominal wages are very sticky, so instead the number of jobs adjusts downwards.
If the central bank stabilizes aggregate demand (i.e. nominal spending), real wages will (as a result of price rises) fully adjust to the diminished productive capacity of the economy. Unemployment will stay the same. Only the composition of NGDP growth will temporarily change (more inflation, less real growth). This may still be unpleasant but compared to higher unemployment the economic and political cost are negligible.
Endowing the central bank with the mandate to target some growth rate of NGDP, say, 5% would mean that the central bank automatically stabilizes aggregate demand (which is the same as nominal spending which again is the same as NGDP).
To summarize, explicit and implicit government guarantees for bank debt constitute a subsidy of debt financing and provide an incentive for banks to minimize equity to asset ratios. Razor-thin capital ratios again incentivize banks to take on excessive risk. At some point excessively risky investment behavior will produce heavy losses and lead to a solvency crisis. Government guarantees for bank debt do not only stand at the beginning of this causal chain – they also stand in the way of a solution to the solvency crisis.
Undercapitalized banks do not necessarily need to engage in credit rationing. The much simpler solution would be to issue more capital. But the taxpayer subsidy for debt has made capital artificially expensive relative to debt. Hence, banks are reluctant to issue more capital and the solvency crisis drags on. (14)
Removing explicit government guarantees for bank debt is easy. Doing the same for implicit government guarantees is far more difficult. Without solving the too- important-to-fail problem, the government cannot credibly commit to not bail out failed banks.
There are three arguments put forward against allowing large banks to fail:
- a run on other banks may occur (the liquidity crisis argument);
- the effect on aggregate demand may be too big to be offsettable by the central bank (the aggregate demand shock argument);
- possible negative effects on aggregate supply (the aggregate supply shock argument).
The central bank has always been able to offset any surge in liquidity demand. The central bank would also be able to offset any aggregate demand shock, if it were allowed to commit to a more expansionary monetary policy in the future when conventional monetary policy today is constrained by the zero lower bound.
Additionally, if the central bank operated under a Nominal GDP targeting regime, a possible supply shock resulting from bank failure(s) would not increase the rate of unemployment. (15)
This means that the too-important-to-fail problem can be solved; the government can credibly commit to not bail out failed banks. All that is necessary is to endow the central bank with an appropriate mandate such as the following:
Target a growth rate of NGDP of x%. If conventional monetary policy is not enough to meet this target, commit to higher future NGDP growth in order to meet the NGDP growth target today.
As soon as the central bank is endowed with the appropriate mandate, the whole body of government regulation pertaining to the banking industry can be scrapped and regulatory responsibility can be given back to the markets. (16) Undoubtedly, financial regulation can serve a useful purpose in many cases. This is precisely why in such cases it will be provided by the market.
(1) Admati, A. and Hellwig, M., 2013. The Bankers’ New Clothes –What’s Wrong with Banking and What to Do about It. Princeton: Princeton University Press, p. 178.
(2) Hoenig, T.M., 2013. Basel III Capital: A Well-Intended Illusion.
(3) These options are, of course, not mutually exclusive.
(4) See, e.g., Arthur, T. and Booth, P., 2010. Does Britain Need a Financial Regulator? – Statutory Regulation, Private Regulation and Financial Markets. London: Institute of Economic Affairs.
(5) Modigliani, F. and Miller, M.H, 1958. “The Cost of Capital, Corporation Finance and the Theory of Investment.” The American Economic Review, 48(3), pp. 261-97.
(6) Bagehot, W., 1873. Lombard Street: A Description of the Money Market, Wiley & Sons, (reprinted 1999).
(7) Hawtrey, R., 1932. The Art of Central Banking. London: Longman, Green and Co., pp. 126-127.
(8) Two questions may have occurred to the reader: Why is it that the US government introduced deposit insurance in 1933 given that it had the Fed (which was founded 20 years earlier)? And why did the UK government introduce deposit insurance in 1979 (which was then expanded in 2007)? In the case of the US the government simply responded to the abject failure of the central bank to do its job. (For an account of the complete inattention and incompetence exhibited by the Fed during that time see, e.g., Friedman, M. and Schwartz, A.J., 1963. A Monetary History of the United States, 1867-1960. Princeton: Princeton University Press.) In the UK the introduction of deposit insurance in 1979 simply constituted the implementation of a 1977 EEC Directive “on the coordination of the laws, regulations and administrative provisions relating to the taking up and pursuit of the business of credit institutions”.
(9) This does not necessarily have to be the case. The bank could also issue more capital.
(10) See, e.g., Bernanke, B.S. and Blinder, A.S., 1988. “Credit, Money, and Aggregate Demand.” American Economic Review, 78(2), pp. 435-39.
(11) See, e.g., Eggertsson, G. and Woodford, M., 2003. “The Zero Bound on Interest Rates and Optimal Monetary Policy.” Brookings Papers on Economic Activity, 34(1), pp. 212-19.
(12) This is why Quantitative Easing (QE) had only a negligible effect on aggregate demand in those countries where it was attempted during the Great Recession.
(13) Blinder, A.S., 1987. “Credit Rationing and Effective Supply Failures.” Economic Journal, 97, pp. 327- 52.
(14) This quote from Mervyn King puts it aptly: “Of all the many ways of organising banking, the worst is the one we have today” (King, M., 2010. Banking: From Bagehot to Basel, and Back Again, p. 18).
(15) NGDP targeting also has other advantages over inflation targeting. For example, under an NGDP targeting regime, the central bank would automatically offset aggregate demand shocks – without needing detailed information about the output gap.
(16) This essay has not explicitly dealt with financial industries other than banking. Government regulation of the financial sector is, of course, not restricted to the banking industry. However, banking constitutes the strongest case for government regulation (and has been the focus of regulatory efforts by the government). Therefore, the recommendation of this essay (namely to take away the regulatory responsibility from the state and to give it to the markets instead) pertains to the rest of the financial sector as well.
The Freeconomist blogs at belowpotential.com