Everyone agrees that the housing and mortgage bubbles of the first decade of this century turned out to be incredibly costly. There is no need to review the data. Many observe the enormous costs and conclude that there just must be a way, or we must invent a way, to prevent a recurrence of a bubble in the housing market or any other significant market. So, I want to go back to basics. Just because something is incredibly costly does not necessarily mean that there are solutions that will leave us better off. I intend my argument to be general but actual policy might well depend on the market involved. I am thinking primarily about a monetary policy device of some sort. But the argument holds also for a fiscal device such as a tax that might go on and off depending on efforts to control the bubble.
From work in the 1950s and 1960s — Tinbergen, Theil, Brainard and others — we settled certain things. We need at least as many policy instruments as goals. If we are going to control a bubble by using an aggregate monetary policy instrument — ordinarily, the federal funds rate — we will have to accept unpleasant trade‐offs because tighter policy may push the economy below full employment. So, let us use the economist’s favorite can‐opener device — assume there exists policy lever Z that has no unwanted side effects. There just has to be a Z out there somewhere.
We also want our policy levers to have predictable effects. A policy lever Z that would control a targeted asset price only within a range of plus or minus 20 percent would not be a good lever. Well, as a long‐time economist, I know how to solve that problem, too — let there be policy lever ZZ that controls the target asset price with no error.
I will talk about regulation later, but beyond that hope, everyone knows that there is no ZZ. Many reach the same point as I have in this analysis but say that we must be prepared to use some policy instrument anyway because failure to do so has been so ghastly. This reaction is altogether wrong. Assume we really could find instrument ZZ, which would be better than what we actually have. From work in the 1970s — Lucas, Sargent, Taylor and others — we know that we cannot analyze the application of instrument ZZ without analyzing asset pricing and decisions that depend on asset prices.
How is the market going to trade an asset knowing that the government has ZZ in operation? Asset prices are not exogenous. How will trading in other assets be affected? How will the market make investment decisions that depend on asset prices knowing that ZZ is in operation? We know that we cannot understand how a policy is going to work without analyzing private behavior based on expectations of what the policymaker is going to do. I do not know of even the simplest toy model illustrating how private agents would behave in the market for houses, or equities, or any other asset in the face of policymakers wielding instrument ZZ. If we do not understand how the market would work with an ideal instrument ZZ, how can we possibly know what will happen using the federal funds rate or a variable tax or some other instrument?
Actually, my claim that we do not have an example of a closed model is a bit of an overstatement, as we do know something about how the foreign exchange market works with government intervention. In an open‐economy macro model, we can understand government intervention defined in terms of international capital flows. But there is no accepted model of government policy in the foreign exchange market of the sort we actually observe, such as the G7 intervention in the yen market on March 18 of this year. That makes it impossible to model private expectations of government behavior. In contrast, a Taylor‐type rule of monetary policy allows a macroeconomic model to be closed with a defined monetary policy.
We also have experience in a related problem, that of employing a variable investment tax credit for economic stabilization. The interaction of policymakers’ uncertainty and of market expectations created an unsatisfactory outcome in the 1960s. Adjustments in the ITC were almost perfectly mistimed from a stabilization perspective. We gave up on the idea of active use of the ITC for very good reason.
But I want to drive this argument further. Would we really want a government agency setting an asset price? Asset pricing is one of the most fundamental characteristics of a market economy. I can almost hear the objection to this conclusion: “You have taken logic too far. The government should influence when appropriate but not set asset prices.” This fuzzy argument ignores what we learned from control theory and the rational expectations revolution. And I can also hear Milton Friedman’s delightful and gleeful reply, which I heard on a number of occasions. Milton would put it this way: “As far as possible is the only place you can take logic.” The bottom line is that once the government makes an effort to influence an asset price on a systematic basis we really do not know what may happen.
To avoid misunderstanding, of course the central bank should take account of the behavior of asset prices in setting monetary policy. Asset prices often contain valuable information as the central bank pursues policy to maintain aggregate price stability and does what it can to cushion disturbances to employment and output. Using information in asset prices is completely different from targeting asset prices.
The Mirage of Better Regulation
An asset price bubble creates a problem when the asset is held in highly leveraged accounts, such as bank portfolios. The dot‐com bust, which took the NASDAQ average down by 70 percent, did not yield a financial crisis because the stocks were mostly held in unleveraged portfolios. A much smaller decline in house prices did create a crisis because homeowners had too much mortgage debt and highly leveraged commercial and investment banks held that debt in highly leveraged portfolios.
I am skeptical that tighter safety and soundness regulation can solve the problem of banking instability. Right now, everyone is cautious. But caution will fade. After all, the most recent crisis occurred with lots of regulators around and less than two decades after the S&Ls were shut down.
Regulation is a political animal. The old saw about the making of sausage and of legislation is true. I don’t like that process. But, worse than that, the process assures that regulation will not at the end of the day be effective. Over time, lobbyists wielding campaign contributions and the promise of votes will persuade powerful congressmen to insert obscure provisions in obscure legislation. Dodd‐Frank leaves us worse off than before because it institutionalizes too‐big‐to‐fail and, potentially, adds large nonbanks to protected list.
To avoid misunderstanding, of course the central bank should take account of the behavior of asset prices in setting monetary policy. Asset prices often contain valuable information as the central bank pursues policy to maintain aggregate price stability and does what it can to cushion disturbances to employment and output. Using information in asset prices is completely different from targeting asset prices.
The so‐called Volcker Rule institutionalizes a permanent game over the permissible activities of banks. Banks with lobbying power will work endlessly to carve out special provisions. To me, the logic of the Volcker Rule is flawed from the beginning. Risk is very hard to measure, but it would seem almost impossible for regulators to take account of the covariance of returns across different activities. Banning banks from certain activities may actually increase risk in a bank portfolio.
Structural Reforms
Are we doomed, then, to suffer asset price bubbles from time to time and their damage to the general economy through banking instability? To a degree, perhaps, but not to the degree of the recent disaster. Three reforms would help: fair‐value accounting, higher capital requirements and elimination of the deductibility of interest on all tax returns. These reforms do not raise control‐theoretic issues. Fair‐value accounting is an issue like standards for weights and measures, though obviously much more complicated to implement.
Nevertheless, implementation is not impossible — we have been doing it with mutual funds for over a half century. Within a few minutes of the market close every day, I can find the net asset value of my mutual fund shares on the Internet. By “fair‐value accounting,” I mean the use in accounting statements of market values for actively traded instruments — sometimes called “marked‐to‐market” — and best estimates of value when market values are not available.
Current accounting practice under Generally Accepted Accounting Principles — GAAP — is not much better than allowing a grocery store to sell breakfast cereal by the box without disclosing weight. GAAP is equivalent to being very careful about when a cereal box was manufactured without insisting that the weight of the cereal in the box be disclosed. Under GAAP, banks can hold securities in a portfolio labeled “held‐to‐maturity” and report them at prices that can be substantially different from observed market prices. The standard banker argument is that fair‐value accounting would make earnings too volatile. The argument is specious; original‐cost accounting does not make earnings less volatile but simply hides the volatility that exists.
Examples abound. Firms just do not want to recognize losses. The Wall Street Journal reported just a few days ago, on March 28, that the SEC required Berkshire Hathaway, a firm revered for its high standards, to restate some asset values to reflect market values. Berkshire’s chief financial officer complained that current stock prices did not reflect the true value of the shares. That might be, but the place to make the argument is in the letter to shareholders and not in a falsely stated balance sheet. A well‐capitalized firm can ride out periods of depressed asset prices. That is what capital is for.
Besides resistance to recognizing losses, I have another hypothesis as to what is going on. Consider an example from the current situation in which the cost of overnight money to a large bank — one too big to fail — is the current federal funds rate of about 15 basis points. Suppose the bank borrows in the funds market and invests the proceeds in 5‑year treasuries at 2.25 percent. The bank plans to keep rolling over its fed funds borrowing for 5 years. Using original‐cost accounting, this position will yield positive carry as long as the funds rate remains below 2.25 percent. However, using fair‐value accounting this position might quickly lose money if market expectations about the future fed funds rate rise and the 5‑year bond drops below par. Using original‐cost accounting, employees getting bonuses from reported profits will continue to get their bonuses even if the position quickly creates a capital loss. The loss will not show up until the fed funds rate exceeds 2.25 percent.
Original‐cost accounting does not change the ultimate size of a gain or loss but does change its timing. Fair‐value accounting would do more than report financial information more accurately — it would change management incentives. Knowing that a bad bet on the direction of the 5‑year T‑bond might yield an immediate reported loss for the bank, the incentive to put on the trade would change. Fair‐value accounting is a powerful incentive for good risk management and an inducement to hold adequate capital.
In fact, fair‐value accounting may be the most powerful incentive for sound risk management we can find. The S&L disaster would never have happened if S&Ls had been required to use fair‐value accounting. The whole world would have observed the deteriorating financial position of these firms in the mid to late 1960s. Moreover, a decade of federal government policies to prop up the industry would probably have been untenable politically with the losses out there for all to see. Thus, the policies pursued by both the S&Ls and the federal government would have been different. The incentive effects of fair‐value accounting do not garner enough attention.
As for the recent crisis, I suspect that the potential for deterioration in the fair value of a subprime mortgage portfolio would have made banks more cautious from the outset. I realize that putting a fair‐value estimate on the complicated CDOs backed by subprime mortgages might have been difficult, but that problem arose anyway and would have been faced earlier. Fair‐value accounting would have forced the Countrywides of the world to slow down. The transparency that more accurate accounting can bring should not be underestimated as a way to address the bubble problem.
Stiffer capital requirements for banks also deserve attention. The Dodd‐Frank legislation did not solve the problem; capital requirements for the largest banks remain too low. The current situation is reminiscent of Fanny and Freddie before they were conserved in September 2008. Eventually, a big bank will get into trouble. The competitive pressures on banks will encourage one or more to accept risks that the banks and we do not understand today. The most powerful incentive effects would come from a system in which banks were required to issue long‐term subordinated notes. That idea has been thoroughly vetted in the literature. Capital requirements are a non‐intrusive regulation, unlike regulations that substitute examiner judgment for management judgment with regard to asset quality, adequacy of liquidity, compensation practices and a host of other management responsibilities.
Another powerful structural reform would be to reduce or eliminate the deductibility of interest on all tax returns. Public finance experts have recommended such a reform for years. Given the instability that flows from excessive leverage, it makes no sense to provide a tax incentive for debt. Interestingly, the Wall Street Journal reported just a few days ago, on April 4, that a congressional study of this issue is under way.
In terms of reform, fair‐value accounting ranks above higher capital requirements. The reason is simple. Capital is the difference between assets and liabilities and therefore cannot be measured accurately without measuring assets accurately.
Concluding Comment
The objections to macroeconomic policies to deal with bubbles are decisive: the case against even trying is clear. However, microeconomic policies to change incentives and improve bank strategy are available and should be implemented.
A banking crisis of the magnitude of the recent one is not inevitable. We know how to construct a more robust banking system. Unfortunately, it seems that large banks have won the debate for now. They operate under federal protection, or think they do, as did Fannie and Freddie and the S&Ls before them. Bubbles can occur, run on or pop, reflecting the market’s efforts to price assets appropriately and the enthusiasms of an innovative market economy. Without banking instability, bubbles are interesting but do not pose great risks to the macroeconomy.
Thank you. I would love to have some questions and comments.
