SERIES: Brookings Policy Brief Series | # 4 of 186 « Previous | Next »

Low Inflation or No Inflation: Should the Federal Reserve Pursue Complete Price Stability?

Although the Fed's performance has hardly ever been better, with both inflation and unemployment at low levels, some politicians and economists want the Fed to go further and to pursue zero inflation as its primary goal. Economists have argued that the costs of such a policy would be temporary and small while the long-term gains would be great. We reexamine these costs and find that previous studies have seriously understated them. The costs of maintaining zero inflation would be a permanent reduction in gross domestic product of 1 to 3 percent and a permanent drop in employment by the same amount. Complete price stability should not be the Fed's goal.


In recent hearings on Capitol Hill, Senator Daniel Patrick Moynihan (D.-N.Y.) hailed Alan Greenspan as "a national treasure." Such acclaim is unprecedented for a Federal Reserve chairman and the institution he represents. Throughout most of the postwar period, the Fed has drawn fire from one side or another. It was blamed for frequent recessions in the 1950s, high inflation in the 1970s, and high interest rates in the 1980s.

Both by historical precedent and through legislation passed in the 1970s, the Fed's responsibility for stabilizing the U.S. economy has encompassed goals for both employment and inflation. In the past, the Fed has come under attack when one goal conflicted with the other. The widespread appreciation the Fed currently enjoys has come as the rate of consumer price inflation has stabilized at a 30-year low of less than 3 percent, with the economy in its fifth year of expansion and unemployment at less than 5.5 percent.

Although the Fed's performance has hardly ever been better, many policymakers and economists want it to go even further and to pursue zero inflation as its primary goal. Senator Connie Mack of Florida has introduced the Economic Growth and Price Stability Act, which would amend the Federal Reserve Act. It would replace the old instruction that the Fed should "promote effectively the goals of maximum employment, stable prices, and moderate long-term interest rates," with the single instruction that it should "promote price stability." The cosponsors of this legislation included nearly every member of the Senate Republican leadership, including former Majority Leader Bob Dole. The same bill was introduced in the House of Representatives by Jim Saxton (R.-N.J.).

Several studies have been done on the impact of going to zero inflation. Nearly all suggest that the costs would only be transitional. In addition, it has been argued that inflation causes costly distortions in saving and investment, because investment income is taxed on the basis of its nominal rather than inflation-adjusted or real value. These distortions are a permanent cost of even low inflation and could be avoided if the Fed achieved zero inflation. So, some argue, the benefits of achieving zero inflation exceed the temporary costs of getting there.

Whether or not the Mack bill passes, the Fed will certainly have to consider whether or not it still wants to pursue lower inflation. We have examined the costs of maintaining a zero inflation rate and find that contrary to previous work, the costs of zero inflation are likely to be large and permanent: a continuing loss of 1 to 3 percent of GDP a year, with correspondingly higher unemployment rates. Therefore, zero inflation would involve large real costs to the American economy.

The reason that zero inflation creates such large costs to the economy is that firms are reluctant to cut wages. In both good times and bad, some firms and industries do better than others. Wages need to adjust to accommodate these differences in economic fortunes. In times of moderate inflation and productivity growth, relative wages can easily adjust. The unlucky firms can raise the wages they pay by less than the average, while the lucky firms can give above-average increases. However, if productivity growth is low (as it has been since the early 1970s in the United States) and there is no inflation, firms that need to cut their relative wages can do so only by cutting the money wages of their employees. Because they do not want to do this, they keep relative wages too high and employment too low. Spillovers cause effects on the economy as a whole to be greater than the employment effects in the affected firms.

Evidence on the Frequency of Wage Cuts

Employers almost never cut their employees' wages because they fear that doing so would cause serious morale and staff retention problems. Studies of popular sentiment suggest why. Most people consider it unfair for a firm to cut wages, except in extreme circumstances. On the other hand, most do not consider it unfair if a firm fails to raise wages in the face of high inflation.

Downward money-wage rigidity used to be a core tenet of macro economics. But the validity of this assumption is now doubted by many macro economists. A series of recent studies argues that money wages are almost as flexible downward as upward. We have reviewed a wide range of data on this question, and we reject these findings. Downward wage rigidity is indeed an important feature of the economy. Studies of general wage increases in manufacturing, union contracts, employer surveys, and our own phone survey of workers allows us to directly examine whether wage cuts are frequent. These data show that wage changes vary across firms, yet few employees receive wage cuts even when inflation is low. Many receive wage increases and many no wage change at all, but the distribution is abruptly truncated at zero. For example, in 1962, when inflation was about 1 percent, 53 percent of production workers in nonunion manufacturing firms received general wage increases, and the average wage change was a 3.2 percent increase. However, though 47 percent of workers received no general increase in that year, less than one-tenth of one percent of workers were employed by firms that made general wage cuts. Firms are extremely reluctant to cut workers' wages.

So why do some other studies claim that wage cuts are frequent? All share the same significant flaw: They do not look directly at wage changes. Instead, they compute these changes from wages reported by workers in surveys taken a year apart. The problem is that survey data of this type is rife with error. People often cannot remember (or simply don't bother to accurately report) their wages to survey takers. With less than half the respondents reporting their wages accurately, "wage changes" computed in this manner are more likely to result from reporting errors than actual wage changes. A recent study by John Shea of the University of Maryland matched a number of people in one of these survey studies with their union contracts. He found that though 21 percent of the wage changes computed for these survey respondents showed declines, only 1.3 percent actually had wage declines in their respective union contracts. Using direct evidence on survey response error, we have shown that the typical errors from panel surveys are easily large enough to produce the appearance of frequent wage cuts, even when the true distribution of wage changes has no such cuts.

Macroeconomic Implications of Downward Rigidity

If employers cannot cut wages, what does this mean for the economy, and particularly for inflation targets? To answer this question, we developed a simulated economy with thousands of firms, each subject to random demand and supply shocks that affected its desired level of employment and wages. We then simulated the behavior of that economy at high, moderate, low, and zero rates of inflation. In this simulation model, unemployment rises at low rates of inflation. There are costs to pursuing low inflation, and these costs are as permanent as the gains of maintaining zero inflation. The effects are permanent because, in the turbulence of the economy, there are always some firms that would want to cut their workers' real wages, and nominal wage rigidity makes this impossible when inflation is low. For the aggregate economy, the consequence of real wages that are too high is employment that is too low. This real cost is not only permanent but also much larger than any reasonable estimate of the gains created by going to zero inflation.

We conducted thousands of simulation experiments to explore the sensitivity of our results and to determine whether there were plausible parameter values that would produce only small effects from nominal rigidity. Our best estimate of the cost of lowering inflation from 3 percent to zero is an increase in unemployment of between 1 and 3 percentage points. Only a few extreme assumptions yielded effects below this range.

One may ask whether there is any direct evidence that the economy behaves like the simulation. To answer this question, we developed a simplified version of the simulation model using U.S. postwar economic data. When fitted to the data, this model did marginally better at predicting the rate of inflation at any level of unemployment than the standard model. More should not be expected, because for most of this period inflation has been above the range where downward nominal rigidity would play a major role.

As a strong test of the usefulness of our model, we attempted an ambitious exercise. The behavior of prices during the Great Depression has always defied explanation through conventional models which assume that only one level of unemployment (the so-called natural rate) is consistent with constant inflation. Unemployment was always above any reasonable estimate of the natural rate, so standard theory predicts accelerating deflation for the entire decade of the 1930s. In fact, there was deflation for the first few depression years. But then a year of significant inflation was followed by a period of low inflation, more deflation, and then inflation again.

We took our model, which was estimated using postwar data, and back-cast the price behavior of the Great Depression. Figure 1 shows the behavior of the standard natural-rate model and our model. The standard model goes wildly off track, whereas our model (which embodies the effects of nominal rigidity) tracks the Great Depression with uncanny accuracy. Both models predict deflation in the early 1930s. However, in the mid-to late 1930s the standard model predicts continuing deflation. Our model predicts that the effects of nominal rigidity finally catch up with the economy and create positive and varying inflation rates, despite high unem-ployment.

Encouraged by these results, we used our estimated model of inflation and unemployment to see what would happen if the Fed were to attempt to move the U.S. economy from a hypothetical 6 percent inflation and 6 percent unemployment to either 3 percent or zero percent inflation. The results are shown in figure 2. With a target of 3 percent inflation, unemployment settles to where it has been since mid-1994 between 5.5 and 6 percent. However, if the Fed were to shoot for zero inflation, the initial costs would be much higher and the long-term unemployment rate more than 2 percentage points higher.


Zero inflation is far from costless, even in the long run. The fortunes of firms continually change, and inflation greases the economy's wheels by allowing these firms to slowly escape from paying real wages that are too high without actually cutting the wages they pay. This adjustment mechanism allows the economy to avoid a large employment cost. At very low rates of inflation and productivity growth, such adjustments are short circuited, and employment suffers.

Though it might be argued that zero inflation over many years would lessen workers' resistance to wage cuts, the interview studies we cite make this seem unlikely. Workers' resistance to nominal wage cuts is tied to their fundamental feelings about fairness and their suspicions of employer motives. The experience of the Great Depression is instructive. After considerable deflation in the early 1930s, resistance to nominal wage cuts apparently stiffened in the mid-to late 1930s. Legal and institutional changes supporting wage rigidity were put in place. A decade of high unemployment and stable prices left nominal rigidity an even more important feature of the economy than before.

Plausible estimates of the benefits of zero inflation are certainly less than the unemployment costs of zero inflation we have documented. A low, steady rate of inflation is a reasonable target for the Fed. We cannot say precisely what low rate of inflation best serves the American people, but we are confident it is not zero.

SERIES: Brookings Policy Brief Series | # 4