Will anchored inflation expectations actually anchor inflation?

The most compelling argument for a benign inflation outlook is that long-run inflation expectations remain anchored about where the Federal Reserve wants them to be. This argument was also one of the principal reasons that so many forecasters predicted low inflation in 2021—a forecast that ended up being wildly at variance with the resulting inflation. Is there reason to believe that this argument will work any better in 2022?

In a PIIE Policy Brief, David Reifschneider and David Wilcox offer one of the most compelling versions of the anchored long-run inflation expectations argument. They point out that in the 25 years before the pandemic, inflation shocks were barely persistent and that inflation always soon returned to around the Federal Reserve's 2 percent inflation target. Their central estimate is that the Fed's de facto inflation metric—the core personal consumption expenditures price index—will smoothly fall to the Fed's 2 percent target over the next two years or so.

Will the statistical relationships of the 25 years leading up to the pandemic reassert themselves in 2022? They may. But it would not be my central case: Short-run inflation expectations and wage- and price-setting behavior indicate a degree of inflation inertia that is closer to what was experienced in the 1960s–1980s than the low inflation of recent decades. Moreover, counting on inflation to fall could lead to policy errors that could actually prevent it from happening.

Which Inflation Expectations Matter?

The fundamental problem for any forecast of inflation is uncertainty about how to model expected inflation—a much more important source of large variations in inflation across countries and across time than differences in slack. Unfortunately, the data do not clearly answer the question of how to model expected inflation. Moreover, even if it were possible to model it, there is no reason to think that those stable relationships hold in the very different environment of today.

Three factors determine inflation: expected inflation, an increase (or decrease) relative to that level based on labor market conditions, and an error term that captures everything else, such as supply shocks and taste shifts (the error term may be correlated with itself over time):

Inflation = Expected inflation – θ * (unemployment – natural rate of unemployment) + error term.

The economics underlying this relationship is that the unemployment rate can stay below the natural rate either by "tricking" workers through prices that are higher than they expected in setting their wages or by running ahead of the built-in wage and price increases (Friedman 1968 and Phelps 1967).

Estimates of θ vary but have recently been in the range of 0–0.3, depending on the model and the measure of inflation—a relatively flat Phillips curve. These estimates mean that if the natural rate of unemployment is 4 percent, then even an exceedingly tight labor market with an unemployment rate of 1 percent would not add more than 0.9 percentage point to the inflation rate. If expected inflation is anchored at 2.0 percent, then even this almost impossibly tight labor market would not raise the total inflation rate above 3 percent.

Inflation rates vary widely over time, for reasons largely unrelated to slack. In the United States, the inflation rate has been 5 percent or higher on a sustained basis in the past; in other countries, like Argentina and Brazil, it has been 50 percent, 500 percent, or even higher. These wide variations in inflation rates over time are best understood as reflecting differences in expected inflation.

The original Phillips curve was a special case of the general Phillips curve but one that sets expected inflation at a constant rate (say, 2 percent). This assumption worked reasonably well over the 25 years before the pandemic. Under this model, inflation will come down sharply and cannot be much above 2 percent. Moreover, bringing inflation down further would require permanently higher unemployment.

The original Phillips curve breaks down if policymakers try to take advantage of the relatively modest unemployment effect, because sustained higher actual inflation becomes higher expected inflation. This observation has given rise to an alternative approach that explicitly models the expected inflation term as a combination of the actual inflation experienced and longer-term inflation expectations, which Reifschneider and Wilcox take from the Federal Reserve Bank of Philadelphia's Survey of Professional Forecasters (SPF). This approach worked well in recent decades and is grounded in their observation that inflation very quickly returns to its mean value. The SPF long-term forecasts vary very little, because its forecasters appear to be expecting the Federal Reserve to be able to maintain inflation at 2 percent. As long as forecasters expect the Federal Reserve to hit its target over the next decade, the Reifschneider and Wilcox forecast will not be very different from the original Phillips curve with fixed expectations.

Using the long-run SPF forecast for inflation as a key input into a forecast of short-run inflation raises two issues. The first is whether the long-term forecast from a group of economic forecasters is the right way to think about inflation over the next several years. In a way, the Reifschneider and Wilcox forecast is less the prediction of an economic model and more the restatement of what another group of forecasters think. Ricardo Reis (2021) questions whether economic forecasters in a number of countries have consistently missed structural breaks in inflation. Second, the relevant time horizon for expectations of inflation is not clear. Short- and medium-run inflation expectations by financial markets, businesses, consumers, and professional forecasters rose sharply over the last year, even as longer-term expectations remained anchored (figure 1). Short-term expectations, especially by consumers, were not good predictors in the period of stable inflation, but there is reason to take them more seriously now that inflation is more unstable than it has been in 40 years.

The second issue, which I return to in the last section of this blog, is that the Reifschneider-Wilcox Phillips curve worked very well in a period in which Federal Reserve interest rate setting hewed closely to a version of a Taylor rule that sets rates based on deviations of unemployment and inflation from their desired values. The Fed is now very, very far from what such a rule would say. Do historical estimates based on sticking closer to that rule still apply? The fact that long-term inflation expectations are anchored suggests that maybe they do. But then again, maybe they do not.

Will the Economy Experience Wage-Price Persistence or a Wage-Price Spiral?

Inflation expectations are critical in both price setting and wage setting. But price setting also matters for wage setting and vice versa. The link is not clear in data from the period of stable inflation, but it is probably a mistake to extrapolate from that period to the very unusual one we are in today. Instead, it is better to rely on a combination of economic theory and a wide range of recent data and behavior.

Let us first consider the pass-through or catch-up from prices to wages. It happened more quickly in an era with more powerful labor unions and indexed contracts. But the importance of these institutional arrangements in price–wage pass-through has often been exaggerated. At their peak in the 1960s and 1970s, unions covered only about a quarter of the economy. Much of the weakening of the rapid price–wage linkage was caused by the shift to low and predictably stable inflation. Moreover, the forces of supply and demand are by themselves sufficient to create a strong link between prices and wages.

Consider a simple supply and demand framework in which labor demand and labor supply are graphed as a function of nominal wages. If prices go up, a business can sell its products at a higher price. For any given nominal wage, it will want to hire more workers (as the real wage has fallen), shifting the demand curve out. Labor supply for any given nominal wage would also move in response to changes in the real wage. The result is a higher nominal wage that exactly offsets the increase in prices. Nominal wages rise without any company being aware of the full picture; individual firms know only that they want to hire more workers, that they are willing to pay more for them, and that they will have to do so or lose workers to competitors who are paying more. (The same logic applies in a richer set of models of the labor market involving institutional features like bargaining.)

Faster price growth thus almost certainly leads to faster wage growth. Similarly, faster wage growth increases the cost of production and is passed on to prices. This relationship can be seen in figure 2, which shows that growth in unit labor costs (growth in labor costs minus productivity growth) closely tracked price growth over the last 70 years. How quickly these two pass-throughs happen, whether they are dampened or spiral over time, and whether wages or prices rise by more are open questions.

The magnitudes of wage increases are currently very far from a 2 percent inflation rate. Over the last two years, productivity rose at an annual rate of 2.3 percent, only slightly above trend, while compensation per hour rose at 7.0 percent per year, far above the trend rate (figure 3). As a result, unit labor costs are up 4.7 percent per year, a rate that is consistent with a similar rate of inflation if the labor share is unchanged. (Note that compositional changes boost both productivity and compensation, as lower-productivity/compensation workers drop out of the data because their employment has not fully recovered. These composition issues matter less for unit labor costs, because they cancel out when the two are divided.

Moreover, nominal wage growth has been very strong recently, although there was a slowdown in February (figure 4). Given that wages can be sticky and are adjusted only periodically, it is likely that much of the big price increases will show up in wages going forward. It is likely that nominal wage growth over the next year or two will be at least 5.5 percent, given the combination of catch-up for past price increases, staggered wage setting, and tight labor markets. A reasonable forecast is that annual productivity growth will be about 1.5 percent. It could be lower, however, as some of the above-trend productivity has been temporarily boosted by people working more intensely because of labor shortages, a trend that will either fade away or need to be matched by higher pay. If this scenario plays out, it would be consistent with price inflation of about 4 percent.

There is, of course, considerable uncertainty. The most likely benign inflationary scenario is that the rapid wage increases of 2021 were a one-time adjustment and that the pace of wage growth will slow. Goldman Sachs analysis shows that employers are expecting to increase wages by only 4 percent in 2022. The most desirable, but less likely, inflationary scenario is that we are entering an era of faster productivity growth, fueled by developments like working from home. Almost as desirable, but implausible, would be a persistent and large disconnect between real wages and productivity growth that increases the labor share of income over time as business markup over costs is compressed. Any of these scenarios is possible, but there is a large gap between the plausible case (at least 5.5 percent wage growth and 1.5 percent productivity growth) and what would be needed for 2 or even 3 percent inflation.

Why Else Might Inflation Be Higher or Lower than Expected?

This note focuses on two of the biggest issues in inflation: expected inflation and the wage–price relationship. Other developments could bring inflation down. They include supply chain unsnarling, a shift from goods to services (assuming the supply of services can respond elastically without big price increases), a return of workers to the labor force (assuming the supply they produce exceeds the demand they generate), and a waning of the pandemic. Demand could fall as a result of the fading of fiscal support, a drop in the stock market, and a rise in interest rates. All of these factors can be understood as the error term in the Phillips curve going from a large positive value to zero. In fact, in some scenarios—such as a glut of goods that leads to a temporary fall in goods prices—the error term would be negative in 2022. If it is, inflation in 2023 could be higher than in 2022.

Other factors could go the other way. The Russian invasion of Ukraine is likely to add several tenths to the inflation rate in the coming months, and the course of the war and the economic responses could have long-lasting effects on the level of energy and other prices. The ebbing of the pandemic and increases in employment could unlock further increases in demand and a return of prices for COVID-19-sensitive services to or above their pre-pandemic levels. Shelter costs, which have been rising (figure 5), could drive up service price inflation in 2022. Labor markets are considerably tighter than they were a year ago, especially when measured by quits (figure 6), which seem to predict inflation better than some other measures. Finally, inflationary expectations are higher than they were a year ago.

What Are the Implications for Monetary Policy?

The debate over the trajectory of inflation has mostly not been determinative of the question of what the short-run path of interest rates should be. The reason is that under almost any inflation forecast, the interest rate paths under discussion are much lower than what almost any monetary policy rule would recommend. Any monetary policy rule that is based solely on current conditions would call for the federal funds rates to already be well above the neutral rate, plausibly around 7 percent or possibly even higher. Inertial rules that penalize policy for moving too quickly would not call for an immediate jump, but they would have rates above neutral by the end of the year, even with a relatively benign outlook for inflation.

Figure 7 presents three possible trajectories: (a) the Survey of Professional Forecasters' inflation forecast and the original Taylor formulation, (b) a dovish version of the balanced rule (which places twice the weight on employment deviations and assumes a natural rate of 3.5 percent and a real neutral interest rate of 0 percent), and (c) an inertial version of the balanced dovish rule. The appendix provides the parameters for all three monetary policy rules.

I am uncertain about inflation, but conditional on an inflation forecast (and mine would be higher than that of Reifschneider and Wilcox), I am even less certain about monetary policy. Some tightening has already been built into long-run rates, although to the degree that inflation expectations have risen, this tightening may be less than meets the eye. Overly rapid rate increases risk market overreactions and other uncertainty. So it may well be that the Fed is on a reasonable course, raising rates once every meeting.

It is, however, increasingly unlikely that the Fed will be able to stop at a neutral interest rate. It should make clear that it will continue its steady increase in interest rates as long as inflation remains substantially elevated and the job market is not worsening. This type of soft, data-dependent forward guidance would help tighten financial conditions today, raising long-term rates and preventing the inversion of the yield curve. It would also reduce the chances of communication errors creating a disconnect between market expectations and the actual course of Fed policy. It is impossible to say how far the interest rate would rise under this soft, data-dependent forward guidance, but it could go well above 3 percent before it starts to come back down again as inflation comes under control.

Inflation coming under control need not be strictly defined as 2 percent inflation or 2 percent average inflation. Stabilizing inflation at 3 percent would itself be an accomplishment. If inflation does settle there, it would be very painful to bring it down much more: With a flat Phillips curve, doing so would likely require a recession. The ideal outcome could be inflation settling at 3 percent—and the target resetting in the Fed's next framework review—a measure that could improve macroeconomic stability by giving the Fed more scope to cut nominal interest rates to combat future recessions.

For now, however, even achieving inflation as low as 3 percent could be challenging. And it would be premature for the Fed to start talking about new targets when it desperately needs to keep inflation expectations—especially short-run expectations, which could become embedded into nominal wage and price growth—as low as possible.

Appendix

The three monetary policy rules shown in figure 7 follow the general form:

where i_t is the federal funds rate, π_t is year-on-year personal consumption expenditures (PCE) inflation, u_t is the unemployment rate, and the other parameters are as specified in table A.1 for each of the three rules.

Table A.1 Taylor rule parameters
Parameter	Taylor rule	Dovish Taylor rule	Inertial dovish Taylor rule
π*, target inflation rate (%)	2.0	2.0	2.0
r*, natural real federal funds rate (%)	0.5	0.0	0.0
1/b, inverse Okun's coefficient	-1.4	-1.4	-1.4
u*, natural/long-run rate of unemployment (%)	4.0	3.5	3.5
ρ, interest rate inertia/smoothing parameter	0.0	0.0	0.8
α, weight on inflation gap	0.5	0.5	0.5
β, weight on output gap	0.5	1.0	1.0
Source: Author's calculations based on data from Federal Reserve Bank of Cleveland.