« Arbitrary Coherence | Main | The Federal Reserve Expands Direct Borrowing »

Sunday, March 16, 2008

Alan Greenspan: Models of Risk

The short version of this argument by Alan Greenspan is: models of risk are not perfect, they can miss inflating bubbles, so don't expect the Fed to know when a bubbling is inflating or to predict when a bubble might burst. And even if the Fed could identify bubbles, it couldn't stop them since a bubble "will not collapse until the speculative fever breaks on its own." The subtext is: don't blame me for missing the bubble that created the current mess, even if I did recognize it I couldn't have stopped it, and don't use the current crisis as an excuse to limit "market flexibility and open competition" through regulatory responses:

We will never have a perfect model of risk, by Alan Greenspan, Commentary, Financial Times: The current financial crisis in the US is likely to be judged ... as the most wrenching since the end of the second world war. It will end eventually when home prices stabilise... Although inventories of vacant single-family homes ... have recently peaked, until liquidation of these inventories proceeds in earnest, the level at which home prices will stabilise remains problematic. ...

Home prices have been receding rapidly under the weight of ... inventory overhang. ... The level of home prices will probably stabilise as soon as the rate of inventory liquidation reaches its maximum... That point, however, is still an indeterminate number of months in the future.

The crisis will leave many casualties. Particularly hard hit will be much of today’s financial risk-valuation system, significant parts of which failed under stress. ...

The problems, at least in the early stages of this crisis, were most pronounced among banks whose regulatory oversight has been elaborate for years. To be sure, the systems of setting bank capital requirements ... will be overhauled substantially...  Also being questioned, tangentially, are the mathematically elegant economic forecasting models that once again have been unable to anticipate a financial crisis or the onset of recession.

Credit market systems and their degree of leverage and liquidity are rooted in trust in the solvency of counterparties. That trust was badly shaken on August 9 2007 when BNP Paribas revealed large unanticipated losses on US subprime securities. Risk management systems – and the models at their core – were supposed to guard against outsized losses. How did we go so wrong?

The essential problem is that our models – both risk models and econometric models – as complex as they have become, are still too simple to capture the full array of governing variables that drive global economic reality. ...

I do not say that the current systems of risk management or econometric forecasting are not in large measure soundly rooted in the real world. The exploration of the benefits of diversification in risk-management models is unquestionably sound and the use of an elaborate macroeconometric model does enforce forecasting discipline. It requires, for example, that saving equal investment, that the marginal propensity to consume be positive, and that inventories be non-negative. These restraints, among others, eliminated most of the distressing inconsistencies of the unsophisticated forecasting world of a half century ago.

But these models do not fully capture ... the innate human responses that result in swings between euphoria and fear that repeat themselves generation after generation with little evidence of a learning curve. Asset-price bubbles build and burst today as they have since the early 18th century, when modern competitive markets evolved. To be sure, we tend to label such behavioural responses as non-rational. But forecasters’ concerns should be not whether human response is rational or irrational, only that it is observable and systematic.

This, to me, is the large missing “explanatory variable” in both risk-management and macroeconometric models. Current practice is to introduce notions of “animal spirits”, as John Maynard Keynes put it, through “add factors”. That is, we arbitrarily change the outcome of our model’s equations. Add-factoring, however, is an implicit recognition that models ... are structurally deficient...

We will never be able to anticipate all discontinuities in financial markets. Discontinuities are, of necessity, a surprise. Anticipated events are arbitraged away. But if, as I strongly suspect, periods of euphoria are very difficult to suppress as they build, they will not collapse until the speculative fever breaks on its own. ...

In the current crisis, as in past crises, we can learn much, and policy in the future will be informed by these lessons. But we cannot hope to anticipate the specifics of future crises with any degree of confidence. Thus it is important, indeed crucial, that any reforms in, and adjustments to, the structure of markets and regulation not inhibit our most reliable and effective safeguards against cumulative economic failure: market flexibility and open competition.

You have just been appointed as a member of the Federal Reserve Board and you are now in charge of the Bubble Management Division. Will you be able to identify the next bubble? How will you do it? Remember, when some prices shoot up rapidly it is an equilibrium response to changes in supply or demand that signal resources to flow into the industry. High profits for a period of time - profits that look excessive when compared to other industries - are required to signal people and resources to enter the industry. If we shut down all large changes in prices, and prevent large profits, then we will shut down this signal and resources will fail to move efficiently in response to variations in market conditions.

But we don't want resources to respond to false signals and that is the problem with bubbles, resources flow to the wrong places because the price signal is distorted. So the problem for a policymaker is to separate the price changes that are properly directing the flow of resources from those that aren't, and the size or rapidity of price changes is not enough to make this determination, the two types of price changes are difficult to sort out.

So, since it's your job to do this, can you? Or is there a chance you might shut down an industry that appears to be an inflating bubble when in fact it is an equilibrium response to changing conditions? It's generally possible to tell both sets of stories, to explain how it could be an equilibrium response, and to explain that it could be a bubble, so what degree of certainty do you require before intervening and shutting down price movements, thereby taking profits away from participants in the market? It's easy to see bubbles looking back, it's not so easy to see them as they are developing. I'm not saying we can't do it, that we'll never recognize a bubble as it is inflating, just that it's not as easy as it might seem.

Stepping away from bubbles for a moment and talking more generally about prices sending false signals, the point behind monetary policy in New Keynesian models of sluggish wage and price adjustment is to shut down misleading price signals. The flow of resources in the economy is dictated by what we call relative prices, the price of good A divided by the price of good B (e.g., the number of apples you can get for an orange), and in the presence of inflation and prices that do not adjust quickly these relative prices can be pushed away from their long-run, equilibrium levels. Inflation requires all prices to move, and local conditions require prices to move differentially, but the presence of price rigidities prevents these adjustments causing relative prices to move away from their desired values. This sends false signals to both input and output markets, resources go to the wrong places, the wrong goods are produced, and this makes us worse off.

The policy solution is to make sure relative prices do not deviate from their optimal values. How do you do that? There are two types of prices, flexible prices (e.g. the price of gasoline and asparagus change rapidly in response to current conditions), and not-so-flexible prices which stay fixed or move slowly over time even when current conditions are changing (e.g. prices printed in catalogues mailed to households). If you are going to have any chance of keeping these ratios in line, it is the flexible prices that will have to do most of the adjusting.

For example, suppose that good A has inflexible prices, but good B has perfectly flexible prices. Then so long as participants in the market for good B are fully aware of and can predict the price of good A, the price of B can be adjusted so that relative prices are correct. That is the basis of monetary policy in the New Keynesian model, make the prices that are inflexible as stable and predictable as possible so that the flexible prices can adjust optimally and prevent false signals from misdirecting resource flows.

How do you make these prices stable and predictable? You first create a price index - call it core CPI or core inflation if it is a percentage change - that contains only the inflexible prices and leaves out the flexible prices. You then target those core prices using monetary policy so they are as stable and as predictable as you can make them, and allow the flexible prices which are not part of the index to adjust in response to changing conditions and maintain the optimal relative prices.

Notice something important. If you use overall CPI instead of core CPI, i.e. if you use the price index that includes both flexible and inflexible prices, then you would be targeting and limiting movement in the very prices - the flexible ones - you need to adjust to keep relative prices in line and prevent false signals from sending resources to the wrong places. That's why you don't want the flexible prices to be part of the price index the Fed is monitoring.

I've wandered from the topic of bubbles a bit, so back to the main topic. As the newly appointed Head of Bubble Management, where do you think the next bubble will appear, how will you detect and verify that it is a bubble, and how will you deflate it without harming other sectors of the economy in the process?

Update: Paul Krugman comments on Greenspan's remarks.

    Posted by on Sunday, March 16, 2008 at 02:26 PM in Economics, Financial System, Monetary Policy | Permalink  TrackBack (0)  Comments (31)


    TrackBack URL for this entry:

    Listed below are links to weblogs that reference Alan Greenspan: Models of Risk:


    Feed You can follow this conversation by subscribing to the comment feed for this post.