In this note we address a question that has arisen repeatedly following the December FOMC meeting: is a positive productivity shock hawkish or dovish for a central bank? We examine this issue through the lens of the Federal Reserve’s FRB/US model and preview the main conclusion upfront. Within this framework, a productivity shock tends to be dovish for the Fed. That said, the result is not immediate.
Simulating the productivity shocks
To clarify the underlying mechanics, we simulate two distinct types of productivity shocks. The first is a level shock to productivity: we permanently increase the level of productivity in the model by 1 percent. In FRB-US, productivity is captured by a multifactor productivity (MFP) trend. The Impulse Response Functions (IRFs from “PING” program in FRB-US) of this experiment are shown in Figure 1. The output gap initially declines on impact before turning positive over time, while inflation falls, implying a disinflationary effect. The impact on inflation is intuitive; the behavior of the output gap is less so. In the model, a level shock raises actual output, but potential output increases immediately and proportionally with productivity—and by more than actual output. As a result, the output gap narrows on impact, not because output falls, but because potential output rises more sharply.
Figure 1. IRFs to a shock to the level of productivity.
While theoretically coherent, this type of shock is arguably less realistic in practice. Productivity shocks—particularly those related to technological innovations such as artificial intelligence—are unlikely to be adopted instantaneously by all sectors of the economy. Even if the technology is available, diffusion and effective implementation typically occur gradually, often over several years.
For this reason, we consider a second, possibly more realistic scenario: a shock to the growth rate of productivity. In this case, productivity growth is shocked once and then remains persistently higher than in the baseline. The results, shown in Figure 2, are more consistent with the current context. The inflation response is broadly similar to that in the level-shock scenario, remaining disinflationary at least for a few years. The response of the output gap, however, is fundamentally different. Output rises, as expected, and potential output also increases, but much more gradually than in the level-shock case. As a result, the output gap turns positive almost immediately.
Figure 2. IRFs to a shock to the growth rate of productivity
The labor market implications differ accordingly. Under the level shock, unemployment rises temporarily—indeed for several quarters—reflecting the initial compression of the output gap. By contrast, under the growth-rate shock, unemployment declines, consistent with a stronger cyclical expansion.
Implications for monetary policy
Assuming a shock to the level of productivity leads to unambiguously dovish implications for monetary policy: such a shock results in an immediate increase in the unemployment rate and a decline in core inflation.
By contrast, a shock to the growth rate of productivity tends to reduce both the unemployment rate and inflation. The net effect depends on model-specific calibrations—particularly the size and persistence of the shock, as well as the slope of the Phillips curve and related structural parameters.
Productivity Shock with Reduced Labor Intensity
The productivity shocks discussed so far—and the results we have presented—refer to pure productivity shocks. That said, the artificial intelligence (AI) shock currently unfolding may not only raise productivity but also reduce labor intensity relative to capital. This raises a natural question: how can this additional feature be incorporated into the productivity shocks described above?
To address this issue, it is necessary to delve into the mechanics of the model. Conceptually, there are two main ways to simulate a reduction in labor intensity relative to capital. The first would be to intervene directly in the production function itself. We do not pursue this route, as it would effectively require rewriting parts of the model. The second approach—which we adopt instead—is to reduce labor intensity by lowering the trend growth of employment. (A third possibility would be to reduce the natural rate of unemployment, but we do not follow this path either, as it is substantially more complex and less transparent.)
For readers interested in the technical details, we provide the relevant equations and a brief explanation at the linked reference, including the equations governing potential business output, potential employment, and the desired level of the civilian labor force here.
Under this scenario (which we circulated recently – see here), we jointly shock productivity (positive shock) and population growth (negative shock). As explained, the latter has a direct effect on trend employment and therefore reduces labor intensity within the FRB/US production framework. The results of this exercise are shown in figure 3 where we plot the current FRB-US baseline (red lines) and the simulated scenario (gray lines).
Figure 3. Current FRB-US baseline (red lines) and simulated scenario (gray lines).
Note: the red lines show the current model baseline, while the gray lines show the constructed scenario. Real GDP growth and core PCE price inflation are shown in YoY terms.
The implications are straightforward. The combination of these two shocks is clearly dovish for the Fed. On the one hand, the productivity shock is disinflationary. On the other hand, the negative population shock limits the rise in unemployment by reducing labor supply, keeping the unemployment rate relatively contained. As a result, the net effect on the federal funds rate is dovish. Inflation declines, reinforcing the case for easier policy, while the labor market does not generate upward pressure on unemployment. At the same time, the reduction in population growth introduces a negative demand effect on overall economic growth within the model. Taken together, the net effect of these interacting forces is unambiguously dovish. In this scenario, the implied path of the federal funds rate is lower and more accommodative than in the baseline that excludes this combination of shocks.
Listening to Chair Powell’s most recent press conference, it is clear that views within the FOMC remain widely dispersed. Nevertheless, it appears that some Committee members—and possibly Chair Powell himself, or even the Board staff—may have precisely this combination of shocks in mind.