Two Histories of the Same Economy: What the GFC and 2018–19 Reveal

This chapter puts the competing gap measures to work in two contrasting U.S. episodes: the Global Financial Crisis of 2008–2009 and the expansion of 2018–2019. The Federal Reserve paper uses these periods to show that the same broad observations about output, unemployment, inflation, and wages can support different estimates of the underlying star. The value of the comparison is not to declare one historical decomposition infallible, but to see exactly how a model’s benchmark changes the judgment attached to a low or high unemployment rate.

2008–2009: a collapse that does not fit one story

The Global Financial Crisis is the easier case to imagine as a cyclical demand shock. Output fell sharply, unemployment rose, and the economy operated far below the path many people had expected before the crisis. A conventional trend-cycle model can describe this as a large negative cyclical gap: the supply-driven potential remains on its prior path while the observed economy moves below it. The unemployment rate then looks high because demand is temporarily too weak to employ the available labor force.

The Federal Reserve paper does not deny that cyclical forces were important. Its point is that the real-side collapse and the nominal response need to be explained together. In the paper’s account, the fall in price and wage inflation during 2008–2009 was relatively small compared with the historic collapse in real activity. That combination is awkward for a model in which a large cyclical gap has a strong, stable relationship with inflation and wages. If the entire real decline is assigned to demand, the model may need a very persistent cycle with a comparatively weak Phillips-curve response, or it may need other shocks to offset the predicted nominal movement.

The scarring specification has another option. It assigns an important role to adverse supply shocks and allows the history-dependent capacity to respond to the past cyclical conditions. In that interpretation, the economy’s productive and labor-market capacity moved in a way that absorbed part of the real weakness. The remaining cyclical gap was shallower than in the benchmark model.

That conclusion can sound like a claim that the crisis was not a demand collapse. It is not. The distinction is between the source of the initial disturbance and the path of capacity afterward. A demand-driven shock can have persistent supply-side consequences through investment, job-finding, worker attachment, or the organization of production. The model is asking how much of the later observed path should be attributed to a continuing cycle and how much to a capacity that has been altered by the cycle’s history.

The numerical contrast is striking. Around 2008–2009, the scarring specification estimates the natural rate of unemployment at about 7 percent, compared with about 5 percent in the benchmark specification. Those numbers are not readings from a labor-market survey. They are model-specific estimates of the hidden reference point. With a higher estimated natural rate, the observed unemployment rate is judged to be less far above sustainable unemployment than it would be under the lower benchmark. The output and unemployment gaps become shallower, even though the observed downturn remains severe.

This is a different kind of statement from saying that the crisis was mild. The people and businesses affected by the collapse experienced a substantial shock. The model-specific claim concerns the counterfactual path used to measure the gap. If capacity itself was damaged or displaced, the distance from the new capacity can be smaller than the distance from the pre-crisis trend.