The Role of Industrial Composition in Driving the Frequency of Price Change 
This paper assesses the degree to which a shift in the industrial composition in the United States from manufacturing toward services has reduced the frequency of price change and thus flattened the slope of the Phillips curve and affected the associated inflation dynamics.
One of the most crucial parameters in monetary economics is the frequency of price change. If the frequency of price change is low, then nominal shocks have real effects—because prices cannot adjust quickly enough, restoring equilibrium in the economy requires real changes. Thus, with a lower frequency of price change, the economy demonstrates greater monetary non-neutrality. This also implies that when the frequency of price change is low, achieving a given reduction in inflation requires a larger increase in unemployment.
In this way, the frequency of price change is a key determinant of the slope of the Phillips curve, which depicts the inverse relationship between inflation and the unemployment rate. The negative relationship was most clearly observed in the post–World War II period. However, in recent decades and particularly following the Great Recession and before the onset of the COVID-19 pandemic, the Phillips Curve was difficult to detect. The flattening of the Phillips curve in the United States could be explained by a reduction in the frequency of price change, which in turn could be explained by the shift in the economy's industrial composition.
The flattening of the Phillips curve has coincided with a shift in the focus of the US economy from primary and secondary industries (industries that extract raw materials and those that process them, respectively) toward tertiary industries (industries that provide services). Because products in the tertiary industries have lower frequencies of price change compared with products in the primary and secondary industries, it is reasonable to expect that as the tertiary-industry share of the economy increases over time, there would be a concurrent downward shift in the distribution of the frequency of price change and that this shift would flatten the Phillips curve.
Key Findings
- Changes in industrial composition in the United States have led to large declines across the distribution of the frequency of price change over the 1947–2019 period. The median monthly frequency fell from 9.2 percent in 1947 to 6.9 percent in 2019. The mean fell from 24.2 percent to 15.1 percent.
- These declines have led to a 28.5 percent flattening of the slope of the Phillips Curve, from –0.26 in 1947 to –0.18 in 2019.
- Despite a flattening of the Phillips curve, large or persistent expansionary demand shocks—such as the one that accompanied the US economy's recovery from the COVID-19 pandemic—still can cause a sharp rise in inflation.
- Small positive demand shocks hitting the economy for six consecutive months in 2019 would have resulted in 8.7 percent inflation compared with 9.5 percent in 1947.
Implications
A larger service-industry share of the economy could account for the flattening of the Phillips Curve. Because inflation is less responsive to real changes in the economy when the Phillips curve is flatter, industrial composition shifts could, therefore, explain some of the missing deflation during the Great Recession and the COVID-19 recession. It could also account for the missing inflation during the recovery from the Great Recession.
A flatter Phillips curve implies that monetary policymakers face a greater trade-off in stabilizing inflation versus employment. The long-term structural changes that have left the Phillips curve more than one-quarter flatter than it was seven decades ago are unlikely to revert in the short run. Therefore, monetary policymakers must account for the flattening in their policy decisions.
Abstract
We analyze the impact of shifts in the industrial composition of the economy on the distribution of the frequency of price change and its consequences for the slope of the Phillips curve for the United States. By combining product-level microdata on the frequency of price change with data on industry shares from 1947 through 2019, we document that shifts in industrial composition led to a gradual reduction in the median monthly frequency of price change from 9.2 percent in 1947 to 6.9 percent in 2019. Other percentiles of the distribution of the frequency of price change show similar reductions. These declines were broadly driven by a shift in the industrial composition of the economy from primary and secondary industries toward service industries. In a calibrated multisector general equilibrium menu cost model, we find that this effect flattened the Phillips curve by 28.5 percent from 1947 to 2019. However, despite a flatter Phillips curve, persistent shocks to aggregate demand still can cause significant inflation.