Indeed, considerable progress has been made in recent years, at the Board and elsewhere, in developing dynamic stochastic general equilibrium (DSGE) models detailed enough for policy application. These models have become increasingly useful for policy analysis and for the simulation of alternative scenarios. They are likely to play a more significant role in the forecasting process over time as well, though, like other formal methods, they are unlikely to displace expert judgment.

Although we use a variety of techniques for extracting information from asset prices, what we can learn has limits.  First, asset prices are tough to work with.  They change rapidly and are subject to short-run technical factors--swings in prices that are not related to fundamental and persistent shifts in supply and demand.  Second, and perhaps even more important, how asset prices embody risk and investors’ risk attitudes is complicated and varies over time.  We must use models to extract information on risk and risk preferences from prices, and because all models are simplifications of reality, we have to recognize that the results are only approximations of the underlying attitudes and circumstances and thus are subject to error.

It's true that the empirical evidence suggests that the link is looser, that there's less responsiveness of inflation to employment conditions than there perhaps may have been in past decades.

My own view is that we should take a very eclectic approach in thinking about inflation.

I look at the state of the economy. I try to assess whether demand is exceeding supply in some sense; whether the financial conditions are promoting growth in demand which is greater than the productive capacity of the economy.  But I also look at a wide variety of indicators, including commodity prices, including financial indicators like bond rates and inflation compensation.

I don't think we can rely on any single indicator, particularly one like the natural rate of unemployment concept. It's very difficult to know. Even if there is such a relationship, it's very difficult to assess in real time where that number might be.

And so we really have no alternative but to look at, you know, many indicators -- including {commodity prices} -- to try to assess where inflation's going.

From the Q&A session

Most central banks also strive to follow at least the spirit of Bayesian thinking by taking an eclectic approach to forecasting and to policy analysis.  To see this, consider the range of material that the staff supplies to the FOMC.  In the case of the economic projections contained in the briefing document we call the Greenbook, the staff consults a variety of indicators and models and then judgmentally pools this information to produce the baseline outlook.  The staff then supplements this analysis with various alternative scenarios intended to illustrate the primary risks to the outlook.  Although these scenarios are usually constructed using a single model (FRB/US), the simulations actually encompass a wider range of views about the nature of the economy.  For example, the simulations routinely consider alternative characterizations of such key aspects of the economy as the expectations formation process, wealth effects, and the sensitivity of inflation to changes in resource utilization and monetary policy.  Finally, the staff provides the FOMC with estimated confidence intervals for the forecast and produces studies addressing such questions as the optimal design of policy under different types of uncertainty. 

Despite these difficulties, the Federal Reserve will continue to monitor and analyze the behavior of money.  Although a heavy reliance on monetary aggregates as a guide to policy would seem to be unwise in the U.S. context, money growth may still contain important information about future economic developments.  Attention to money growth is thus sensible as part of the eclectic modeling and forecasting framework used by the U.S. central bank.

An often overlooked implication is that, all else equal, an increase in the growth rate of productivity will tend to put upward pressure on real interest rates.  But in fact we have not seen the predicted rise in real rates.  Of course, we do not live in the world of simple economic models so all other things are not equal.  In particular, I believe one reason is that sound economic policies have created a more stable economic environment, and with that has come low and stable inflation and an ongoing desire by foreigners to invest in the United States to reap higher returns associated with higher productivity growth than may be available in their economies.        

Given this reality, policymakers are well advised to follow two principles familiar to navigators throughout the ages:  First, determine your position frequently. Second, use as many guides or landmarks as are available In the context of monetary policy, these principles suggest that policymakers should monitor bond yields carefully in judging the current state of the economy--but only in tandem with the signals from other important financial variables; direct readings on spending, production, and prices; and a goodly helping of qualitative information. Ultimately, a robust approach to policymaking requires the use of multiple sources of information and multiple methods of analysis, combined with frequent reality checks. By not tying policy to a small set of forecast indicators, we may sacrifice some degree of simplicity, but we are less likely to be misled when a favored variable behaves in an unusual manner.

Although they are imprecise, simulations from the Federal Reserve Board staff's large-scale econometric model, which account for these effects, suggest that increases in spot and futures prices of energy from late 2003 to the present subtracted a 1/2 percentage point from real GDP growth in 2004 and more than 1 percentage point in 2005. The model suggests the subtraction this year will be about a 1/2 percentage point.

Rather, the Federal Reserve, together with all modern central banks, has found that the successful conduct of monetary policy requires painstaking examination of a broad range of economic and financial data, careful consideration of the implications of those data for the likely path of the economy and inflation, and prudent judgment regarding the effects of alternative courses of policy action on prospects for achieving our macroeconomic objectives. In that process, economic models can provide valuable guidance to policymakers, and over the years substantial progress has been made in developing formal models and forecasting techniques. But any model is by necessity a simplification of the real world, and sufficient data are seldom available to measure even the basic relationships with precision. Monetary policymakers must therefore strike a difficult balance--conducting rigorous analysis informed by sound economic theory and empirical methods while keeping an open mind about the many factors, including myriad global influences, at play in a dynamic modern economy like that of the United States.  Amid significant uncertainty, we must formulate a view of the most likely course of the economy under a given policy approach while giving due weight to the potential risks and associated costs to the economy should those judgments turn out to be wrong.

Economic modeling is as much art as science.  Economic policy makers face enormous uncertainty.  Economic models provide a set of useful tools to frame future outcomes; but as we were reminded repeatedly during our efforts to forecast the economy in 1974 and 1975, models can go off track in myriad ways...As hard as this can be to achieve, economic policy should take the long view.

We simulated FRB/US using the path for crude oil prices that futures market participants in December 2003 expected to prevail over the following three years. We also simulated the model with the revisions to futures prices that occurred subsequently over 2004 and through mid-September of this year. Based on a comparison of these simulations, we estimate that real GDP growth was held down 1/2 percentage point in 2004 and 1 percentage point this year relative to what it otherwise would have been. The drag on real GDP growth next year would be comparable to that in 2004. As higher energy prices are passed through to the prices of other goods and services, prices for core personal consumption expenditures (core PCE) are estimated by the model to have been boosted 1/4 percentage point last year and more than 1/2 percentage point in 2005. Given the lags in the inflation process, core PCE inflation rises a bit further relative to baseline next year.

Despite their historical importance for aggregate inflation, energy prices, for example are controlled for in only one of the structural [inflation] models discussed at this conference.  And this importance is not necessarily a concern of the past: Prices for oil and natural gas have soared since 2003, directly boosting the energy component of the consumer price index as well as raising the production costs, and ultimately to at least some degree the prices, of non-energy goods and services.  As a policymaker, I can assure you that any model of inflation that did not take account of these effects, and how they might or might not affect ongoing rates of inflation, would have been or little practical use to the FOMC over the past few years.

The economic world in which we function is best described by a structure whose parameters are continuously changing. The channels of monetary policy, consequently, are changing in tandem. An ongoing challenge for the Federal Reserve--indeed, for any central bank--is to operate in a way that does not depend on a fixed economic structure based on historically average coefficients...

Moreover, we recognize that the simple linear functions underlying most of our econometric structures may not hold outside the range in which adequate economic observations exist. For example, it is difficult to have much confidence in the ability of models fit to the data of the moderate inflations of the postwar period to accurately predict what the behavior of the economy would be in an environment of aggregate price deflation.

The failure of economic models based on history to anticipate the acceleration in productivity contributed to the recent persistent underprediction of economic growth and overprediction of inflation. Guiding policy by those models doubtless would have unduly inhibited what has been a remarkable run of economic prosperity.

I would remind you that in the 20 years prior to this recent episode, the Phillips curve based on NAIRU was probably the single most reliable component of any large-scale forecasting model. It was very useful in understanding the inflation episode over that entire period. Certainly, there is greater uncertainty today about where NAIRU is, but I would be very cautious about prematurely burying the concept.