As the US unemployment rate continues to drift down to levels not seen in decades, many observers point to relatively low wage and price inflation as evidence that the Phillips curve is dead. Yet inflation is behaving exactly as the Phillips curve—which shows the inverse relationship between the inflation rate and the unemployment rate—would predict. The decline in the US unemployment rate is too recent and too small to have caused any significant rise in inflation to date. Inflation is likely to pick up over the course of this year and next, albeit with a considerable degree of uncertainty. This post updates and expands on my post of November 17, 2017.
The Natural Rate of Unemployment
Nowadays, economists tend to express the Phillips curve as a relationship between deviations of unemployment from a "natural rate"—that is, the rate that still remains at full employment—and increases or decreases in inflation.1 This is sometimes referred to as the "accelerationist" Phillips curve. When unemployment is below the natural rate, inflation tends to rise (prices accelerate), and when unemployment is above the natural rate, inflation tends to fall.
Changes in demographics and labor market institutions can move the natural rate over time. Figure 1 displays the natural rate of unemployment estimated by the Congressional Budget Office (CBO). The gradual rise in the 1960s and 1970s probably reflects the large influx of young people into the labor market, and the gradual decline since 1980 probably reflects the maturing of that cohort. The natural rate is far less volatile than the actual rate of unemployment. Over nearly 70 years, the natural rate is contained within a band only 1.6 percentage points wide, whereas the unemployment rate fluctuated within a range more than 8 percentage points wide. For 2018, the natural rate is estimated to be 4.6 percent.
The Effect of Downward Wage and Price Rigidity
The unemployment rate fell below the natural rate in the second quarter of last year after being above the natural rate for several years (figure 1). The May 2018 unemployment rate of 3.8 percent (not shown) is 0.8 percentage points below the CBO natural rate. This is too small and too recent of a gap to have had much effect on inflation yet.
The real puzzle is why the massive excess of unemployment between 2009 and 2016 did not push inflation down very far. The best answer seems to be that firms and workers are highly reluctant to accept cuts in prices and wages. This phenomenon is also at work in Europe and Japan, where comparable or larger periods of excess unemployment did not push inflation significantly below zero.
Figure 1 US unemployment rate and natural rate, 1949Q1–2018Q1
Source: Haver Analytics.
Phillips Curves with Low and High Inflation
Downward wage and price rigidity matters only when overall inflation is very low. Figure 2 plots Phillips curves when the previous year's inflation rate was below 3 percent (on the left) and when it was above 3 percent (on the right). In each plot, the vertical axis displays the four-quarter core inflation rate minus its average value over the previous three years. The horizontal axis is the CBO natural rate minus the unemployment rate; thus, a high unemployment rate results in a negative employment gap.
The right panel of figure 2 displays a symmetric Phillips curve in which negative employment gaps pull down inflation and positive employment gaps push up inflation. The left panel, in contrast, displays an asymmetric Phillips curve. When inflation is already very low, even large negative employment gaps have little effect, but there is some evidence that positive gaps push inflation up. If the asymmetry is driven by downward wage and price rigidity, there is no reason to doubt that a significant positive employment gap will eventually push inflation up.
Figure 2 US Phillips curves with low and high inflation, 1963Q1–2018Q1
Note: Change in inflation is the four-quarter percent change in personal consumption expenditures (PCE) prices (excluding food and energy) minus the average percent change over the previous three years. The employment gap is the Congressional Budget Office estimate of the natural rate of unemployment minus the actual rate of unemployment as a percent of the labor force. The employment gap is lagged by four quarters. The split between left and right panels is based on four-quarter inflation as of four quarters earlier.
Sources: Haver Analytics and author's calculations.
The latest observation (2018Q1) of the change in inflation and (lagged) employment gap is labeled in the left panel of figure 2. Recent inflation performance is exactly what would be predicted by the accelerationist Phillips curve. The May 2018 employment gap is 0.8 percentage point. If the 2018Q2 employment gap turns out to be 0.8 percentage point, the figure suggests that inflation may rise 0.5 to 1 percentage point over the next four quarters, but outcomes both higher and lower than that are possible. Based on the current three-year average rate of core inflation of 1.6 percent, an employment gap of 0.8 percentage point is likely to lead to a core inflation rate of around 2¼ percent in 2019Q2, but outcomes between 1¾ percent and 3 percent cannot be ruled out.
1. Alternatively, some models are expressed in terms of deviations of inflation from where it had been expected to be. There are a host of issues in measuring such expectations.
The data underlying this analysis can be viewed here [zip].