Macroeconomics Then and Now

Edward M. Gramlich
Member, Board of Governors of the Federal Reserve System

I am happy to be back in New Haven, the place where I learned much of the economics, particularly macroeconomics, that I still use today. I was a student here in the early 60s, receiving my degree in 1965. My career has been inside and outside of academia. My first job was as a research economist at the Federal Reserve Board -- followed by stints at the Office of Economic Opportunity, the Brookings Institution, Congressional Budget Office, and a long spell at the University of Michigan. Two years ago I was asked if I wanted to come back to Washington.  I said yes, and now I am a governor of the Federal Reserve Board, completing a professional cycle that began back in 1965.

Before getting into my remarks, I would like to say a word about Yale, which has been a leader in the economics profession for many decades now.  I am proud to share this panel with some of the many distinguished graduates of the program, and also proud to have been taught by the Yale faculty. In my case my main teachers were James Tobin, who is here today, and Arthur Okun, who unfortunately is not.  Since I have followed both of their works carefully since leaving Yale, it is hard to remember exactly what I learned as a graduate and since. A lot of nice things have been said about Jim, and I heartily endorse all of them -- he has been a lifelong hero.  But I also wish that Art were still here, especially to see our present non-inflationary environment -- he really cared about that.  I also benefited profoundly from our chair Bill Brainard, who hardly looks old enough to have been my teacher.

In line with Bill’s instructions, I will engage in some then and now analysis.  Here’s what I thought then about some matter, here’s what I think now, here’s why views have changed, if indeed they have changed. Topically, I will stick to macroeconomics, and particularly the macroeconomics practiced at the Fed.

Assignment of Macroeconomic Roles

Then, as now, we defined macroeconomic policies in terms of monetary and fiscal policies. While international economics was a lively field in those days, for the most part macroeconomics was taught, and thought of, in terms of closed economies. To the extent that we worried about exchange rates at all, we generally considered them fixed, as of course they were in the Bretton Woods era.

In that type of closed economy, either monetary or fiscal policy was in general capable of bringing about full employment. Beginning from an underemployment equilibrium, monetary expansion would generate full employment output with relatively low interest rates; fiscal expansion with relatively high interest rates. If one were worried about defending the exchange rate, fiscal expansion would be preferred. If one were worried about long term growth for an undersaving economy, monetary expansion would be preferred, down to the interest rate dictated by Ned Phelps’ golden rule of capital accumulation. Now there’s a notion that brings back memories!

The analysis has not changed that much today, but the environment has. One important change is the breakdown of the Bretton Woods system of fixed exchange rates. Another is increasing globalization. Nowadays macroeconomics is usually taught, and thought about, from an open economy perspective, where the openness accentuates the possibility that large-scale international capital flows will even out real interest rates around the world. If exchange rates are fixed, monetary policy is necessarily devoted to keeping them so, which is to say keeping interest rates at the world level. This leaves fiscal policy as the only effective stabilization policy tool.

But when exchange rates are flexible, as they have been for the U.S. since the early 70s, fiscal impacts are crowded out by the trade balance, which makes monetary policy the only policy that is effective for stabilization purposes. Fiscal policy remains influential, but its importance involves longer run questions of national saving. In principle fiscal policy could then be guided by the golden rule conditions for growth, except that in the open economy there is no diminishing marginal product of capital, and hence no definable golden rule. A country can still raise its living standards in the long run by saving more, but in the open economy I know of no objective way to define optimality.

Consider now the United States, essentially an open economy with flexible exchange rates and low national saving. The standard policy prescription would be tight fiscal policy to try to raise the national saving rate; and flexible monetary policy, to stabilize output near the preferred level. Indeed, this is exactly the mix that has been followed for most of the past decade, in both the Bush and Clinton Administrations. While the tightness of fiscal policy has at times been politically difficult for both Administrations, from a macroeconomic standpoint the policy mix seems to be working. National saving is still low, but it is at least rising and will rise further if the country can continue to run budget surpluses.

Inflation and Unemployment

The other part of the story involves stabilization strategy, particularly for monetary policy. How exactly is the Fed supposed to stabilize output near its long run sustainable level?

There have been all levels of changes in conditions and thinking in the past three decades, far too many to mention. There was a worldwide acceleration of inflation until about the mid-80s, followed by worldwide deceleration. Now many countries are even talking about deflation and liquidity traps, ideas I thought I would never hear discussed in my professional life. On the theoretical side, the no-expectations Phillip Curve tradeoff was supplanted by an adaptive expectations vanishing tradeoff and then later by a rational expectations model in which only surprises affect real output.

But let me stick to questions of how the Fed should operate. A first issue is the matter of targeting – should the Fed conduct policy in terms of monetary quantities or interest rates? In the old days this was an open question – my new (and then) Fed colleague Bill Poole showed how the answer depended on the likely errors in predicting final demand (IS errors) as compared to the likely errors in predicting money demand (LM errors). Nowadays there has been a fairly decisive shift in thinking – generally LM errors have been far larger than IS errors and almost all monetary analysts have shifted over to thinking about policy directly in terms of interest rates.

How then should interest rates be set? John Taylor has worked out a neat little rule of thumb that allows for countercyclical policies around a long run anchor. It sets desired short term interest rates at some equilibrium real level, with deviations upward if inflation exceeds its target value and downward if unemployment exceeds its target value. Taylor determines his response coefficients from properties of econometric models – working in both automatic and optimal discretionary responses.

It works out nicely, but one does have to know the target values of inflation and unemployment to conduct policy properly. When Taylor himself, blessed with the wisdom of hindsight, goes back through history, he faults the Fed of the 1960s for raising interest rates too little in response to incipient inflation. He does that on the assumption that the target unemployment rate, or NAIRU, is about 6 percent, which is what we all thought when Taylor wrote his article a few years ago. But back in the 60s the general feeling was that NAIRU was about 4 percent, and if one goes back and looks at Fed policy under the assumption that NAIRU was 4 percent, it looks like the Fed at that time was doing a reasonable job. The problem, in a word, was with the then conventional wisdom about NAIRU.

By the same token, one can bring Taylor’s analysis up to date, and again find fault with the current Fed for raising interest rates too little in response to strong growth in aggregate demand and associated low unemployment. If NAIRU is indeed 6 percent now, this criticism would be appropriate. But if NAIRU has slipped back down towards 4 percent again, maybe the current Fed is not doing such a bad job. The upshot is that in appraising policy now, as then, there is still this huge uncertainty about the target unemployment rate.

What’s a poor central banker to do when the econometrics are unclear and the analysts are fundamentally undecided about such a crucial variable? It seems to me that the proper strategy is to go to a version of the Taylor Rule that avoids making a decision about NAIRU. Suppose we use what I’ll call a change version of the rule. The Fed might not know the precise target values for inflation and unemployment, but it would know if inflation and unemployment were in a desirable band. If they were, the Fed might just work to preserve a good thing and keep inflation and unemployment in that band. At least in times when supply shocks are minimal, this policy in turn involves keeping the forecast growth of aggregate demand roughly equal to the forecast growth of aggregate supply (which I think can be predicted), and thereby preventing both inflation and unemployment from worsening.

A variant of this rule, known as inflation-targeting, just involves working against forecast inflation. Without going into details, there could be difficulties with strict inflation-targeting in supply shock periods, and also if inflation forecasting is itself dependent on knowing the value of NAIRU.

Whatever the variant, these new rules may in fact come close to what the old time central bankers were trying to do, before all this fanciness about relative shocks, target values, response coefficients, and all the rest. Indeed, the new rules are suspiciously close to William McChesney Martin’s remark about removing the punch bowl when the party heats up too much. That may not be much to show after all these years of research and analysis, but the procedure might even work.