Since the U.S. economy is now below our 2 percent inflation target, and since unemployment is in the vicinity of full employment, it is sometimes argued that the link between unemployment and inflation must have been broken. I don't believe that. Rather the link has never been very strong, but it exists, and we may well at present be seeing the first stirrings of an increase in the inflation rate--something that we would like to happen.

[The Phillips Curve] is essentially a theory that fits reasonably, but certainly not perfect, explaining the inflation process. And it's a theory that says first that inflation expectations play a key role in determining inflation. Second, that various supply shocks, such as movements in the price of oil or commodities or import prices, also play an important role. And third, that the degree of slack in the labor market or the degree of more generally of pressure on resources in the economy as a whole exert an influence on inflation as well.

And that theory underlines the kind of statement that I've made, that if inflation remains -- inflation expectations remain well anchored and the transitory influence of energy prices and the dollar fade over time, that in a tightening labor market with higher resource utilization, I expect inflation to move back up to 2 percent. It is consistent with that Philips curve theory.

Regardless of its source, the presence of sluggish nominal adjustment brings to the fore three key elements driving wage and price dynamics: inflation expectations, supply shocks, and resource utilization...The tendency of some prices to adjust very quickly to changing circumstances in conjunction with sluggish adjustment in other prices and wages implies that large, sharp price movements, such as a change in the price of oil, lead to relative price distortions throughout the economy; these distortions imply that relative price shocks have important implications for the functioning of the economy.

An efficient monetary policy {following an oil shock} should attempt to facilitate the needed economic adjustments so as to minimize distortions to economic efficiency on the path to achieving, over time, its dual objectives of price stability and maximum employment.⁹

In particular, an appropriate monetary policy following a jump in the price of oil will allow, on a temporary basis, both some increase in unemployment and some increase in price inflation.  By pursuing actions that balance the deleterious effects of oil prices on both employment and inflation over the near term, policymakers are, in essence, attempting to find their preferred point on the activity/inflation variance-tradeoff curve introduced by John Taylor 30 years ago.  Such policy actions promote the efficient adjustment of relative prices: Since real wages need to fall and both prices and wages adjust slowly, the efficient adjustment of relative prices will tend to include a bit of additional price inflation and a bit of additional unemployment for a time, leading to increases in real wages that are temporarily below the trend established by productivity gains.

Empirical work on inflation, including much of the classic work on Phillips curves, has generally treated changes in commodity prices as an exogenous influence on the inflation process, driven by market-specific factors such as weather conditions or geopolitical developments.  By contrast, some analysts emphasize the endogeneity of commodity prices to broad macroeconomic and monetary developments such as expected growth, expected inflation, interest rates, and currency movements.  Of course, in reality, commodity prices are influenced by both market-specific and aggregate factors...

I have only mentioned a few of the issues raised by commodity price behavior for inflation and monetary policy.  Here are a few other questions that researchers could usefully address:  First, how should monetary policy deal with increases in commodity prices that are not only large but potentially persistent?  Second, does the link between global growth and commodity prices imply a role for global slack, along with domestic slack, in the Phillips curve?  Finally, what information about the broader economy is contained in commodity prices?  For example, what signal should we take from recent changes in commodity prices about the strength of global demand or about expectations of future growth and inflation?

Over the past decade, formal work in the modeling of inflation has treated marginal cost, particularly the marginal cost of labor, as central to the determination of inflation.²  However, the empirical evidence for this linkage is less definitive than we would like.³  This mixed evidence is one reason that much Phillips curve analysis has centered on price-price equations with no explicit role for wages.⁴

Problems in the measurement of labor costs may help explain the absence of a clearer empirical relationship between labor costs and prices.  Compensation per hour in the nonfarm business sector, a commonly used measure of labor cost, displays substantial volatility from quarter to quarter and year to year, is often revised significantly, and includes compensation that is largely unrelated to marginal costs--for example, exercises (as opposed to grants) of stock options.  These and other problems carry through to the published estimates of labor's share in the nonfarm business sector--the proxy for real marginal cost that is typically used in empirical work.  A second commonly used measure of aggregate hourly labor compensation, the employment cost index, has its own set of drawbacks as a measure of marginal cost.  Indeed, these two compensation measures not infrequently generate conflicting signals of trends in labor costs and thus differing implications for inflation.

I do not expect the recent elevated inflation rates to persist.  In my view, the adverse dynamics of the financial markets and the economy have presented the greater threat to economic welfare in the United States.  But the recent information on prices underlines the need to continue to monitor the inflation situation very carefully.

...

I expect the run-up in headline inflation to be reversed and core inflation to edge lower over the next few years.  This projection assumes that energy and other commodity prices will level out, as suggested by the futures markets.  Moreover, greater slack in the economy should reduce pressure on prices and wages.  Despite high resource utilization over the past couple of years and periods of elevated headline inflation, labor cost increases have remained quite moderate, and inflation expectations remain reasonably well anchored.

Recent research at the Federal Reserve Bank of St. Louis suggests that such movements along a short-run Phillips curve or transitory shifts up and down in that curve only account for a relatively minor portion of the observed inflation in the United States since the mid 1950s. The dominant factor in U.S. inflation history over the past 50 years has been changes in inflation expectations, or semi-permanent shifts up and down in the short-run Phillips curve. When it comes to the forces behind U.S. inflation, expectations trump the gap.(6) While some observers might be startled by this conclusion, reflection on the broad outline of our economic history should allay any apprehensions. In the 1960s and 1970s, successive business cycle peaks had both higher inflation and higher unemployment rates, explained by increases in inflation expectations. After the recession of 1990-91, both inflation and unemployment trended down for the remainder of the decade. In the textbook paradigm, such patterns can only be produced by shifts in the short-run Phillips curve generated by changes in inflation expectations. Indeed, direct evidence on inflation expectations suggests that expectations did trend gradually down over the 1990s.

The conclusion that expectations trump the gap in generating inflation is extremely important for monetary policy. It implies that low and stable inflation will only be observed when the private sector’s expectations of inflation are solidly entrenched at a low level.

The Phillips curve is a core component of every realistic macroeconomic model. It plays a critical role in policy determination, because its characteristics importantly influence the short- and long-run tradeoffs that central banks face as they strive to achieve price stability and, in the Federal Reserve’s case, maximum sustainable employment—our second, congressionally mandated goal. I will argue that behavioral economics can enhance our understanding of the Phillips curve, and this is important for two reasons: First, better models of the inflation process help improve our forecasts and clarify limitations on what monetary policy can do. Second, the theoretical underpinnings of the Phillips curve are important in understanding what central banks should do. In other words, beyond determining the “constraints” governing what is feasible, models underpinning the Phillips curve have implications for the way in which central banks should interpret their price stability mandate and for assessing the welfare costs of fluctuations in output and inflation.

 Interestingly, however, the system approach does not seem to forecast price inflation as well as single-equation Phillips curve models do. This weaker performance appears to reflect, at least in part, the shortcomings of the available data on labor compensation. The two principal quarterly indicators of aggregate hourly compensation are the employment cost index (ECI) and nonfarm compensation per hour (CPH). Both are imperfect measures of the labor costs relevant to pricing decisions. For example, the ECI's fixed employment and occupation weights may not reflect changes in the labor market, and the ECI excludes stock options and similar forms of payment. CPH is volatile, perhaps in part because it measures stock options at exercise rather than when granted, and it is subject to substantial revisions. Moreover, these two hourly compensation measures often give contradictory signals.

The history of the Phillips curve has three distinct phases: the Phillips curve as a stable menu of policy options; the Phillips curve as a short-run relationship that depends crucially on people’s expectations; and the Phillips curve as one piece of a larger model that describes the complicated interactions of the decisions made by diverse participants in the economy.  While this last phase may sound impractically complex, we believe it offers a clear understanding of macroeconomic behavior and a useful way to frame current policy debates.

Indeed, many economists criticize the Phillips curve, with some even declaring it dead.⁶ 

In other words, the evidence suggests that the Phillips curve has flattened.

The finding that inflation is less responsive to the unemployment gap, if taken at face value, suggests that fluctuations in resource utilization will have smaller implications for inflation than used to be the case. From the point of view of policymakers, this development is a two-edged sword: On the plus side, it implies that an overheating economy will tend to generate a smaller increase in inflation. On the negative side, however, a flatter Phillips curve also implies that a given increase in inflation will be more costly to wring out of the system. We can quantify this latter consideration using the so-called sacrifice ratio--the number of years that unemployment has to be 1.0 percentage point greater than its natural rate to reduce the inflation rate 1.0 percentage point. Averaging estimates obtained from a comprehensive battery of equation specifications suggests that the sacrifice ratio may be 40 percent larger--that is, it may be 40 percent more costly to reduce inflation than it was two decades ago.

Another apparent change in the inflation process has been a reduction in the correlation between inflation and unemployment (Atkeson and Ohanian, 2001; Roberts 2006). Now, this relationship was always loose, as most of the historical variation in inflation has reflected influences aside from movements in unemployment or other measures of resource utilization. Still, in the 1960s and 1970s, a reasonably strong empirical relationship between inflation and unemployment could be found for the United States, with inflation tending to rise in periods when unemployment was low and vice-versa. Starting in the 1980s, however, this correlation began to weaken noticeably. In fact, some researchers now find no relationship at all, whereas others tend to find one that is of reduced economic importance.

In the 1960s, political pressure for low interest rates combined forces with a growing consensus among economists and policymakers that moderate inflation is an acceptable way to boost employment and economic growth. Monetary policymaking was viewed as simply a matter of selecting from among a menu of inflation and unemployment options. Choose a little more inflation and unemployment would fall, according to this theory. Accept somewhat higher unemployment, on the other hand, and inflation would be a bit lower.

The infamous Phillips curve made policymaking seem beguilingly simple. Based on this theory, several influential economists argued that the menu of inflation-unemployment options offered by the Phillips curve could be improved upon if policymakers were willing to discard their old-fashioned attraction to price stability. Forego price stability, these economists argued, and the labor market would operate more efficiently, employment would rise and the economy would grow faster.