The Zero Lower Bound was Irrelevant

Many economics writers, including Ben Bernanke, Neil Irwin, and Justin Wolfers, worry that the Fed will not be able to combat the next recession. Current interest rates, the sad story goes, are already close to zero. Since a downturn will push the economy to the zero lower bound (ZLB), the Fed will not be able to lower rates further, thereby prolonging the recession.

Of course, for such a story to make sense, the ZLB must be a fundamental constraint that inhibits monetary policy. In a new NBER working paper, Davide Debortoli, Jordi Galí, and Luca Gambetti consider whether the ZLB was actually the problem during the last recession. They say the ZLB was irrelevant. The authors come to this conclusion by studying two types of evidence: measures of macro volatility, and the response of macro variables to aggregate shocks through a vector autoregression.

One may worry that the economy becomes more volatile at the ZLB because the Fed is unable to reduce the federal funds rate, its usual stabilizing policy. However, Debortoli, Galí, and Gambetti find little evidence that volatility increased when the economy hit the ZLB, a period ranging from January 2009 to December 2015, as shown in the table below. The left two columns show the relative variance (a measure of volatility) of the ZLB period compared to a baseline period when the economy was not at the zero lower bound. They exclude the Great Recession in the middle column.

If anything, volatility dropped at the ZLB. For example, GDP was 0.92 times as volatile when at the ZLB compared to when not at the ZLB. For inflation measures, such as CPI and PCE, the volatility was cut in half. Not only was inflation low during this period, as has been well acknowledged, but it was also steady. For all the complaints about how the Fed has failed to hit its inflation target, it did a good job at one of its mandates: price stability.

As the authors show through simulations, the empirical finding that volatility can go down at the ZLB directly contradicts the predictions of a baseline New Keynesian model. (Galí is one of the world’s most famous proponents of the New Keynesian model.) It is consistent, the authors show, with a New Keynesian model in which the central bank follows a “shadow” interest rate rule. Such a model captures the effects of unconventional monetary policy and involves some reinterpretation.

In addition to documenting volatility, the authors consider how the economy responded to different aggregate shocks, depending on whether it was at the ZLB or not. Since any particular shock only hits the real economy once, the authors use a standard model called a vector autoregression to act as if the same shock hit multiple times. The authors use the empirical model to estimate how the economy evolved over time, accounting for differences caused by the ZLB. Then the modeler can use the estimated model to ask questions like, What if the demand shock we experienced in August 2007 (non-ZLB) actually took place in November 2012 (ZLB)? If the response is different, then we have reason to say that the ZLB had an effect on the economy.

The authors find that, on average, the economy seemed to respond similarly whether or not the ZLB was binding. Similar shocks appear to generate similar responses. The results are plotted in the figure below. The columns represent the type of shock induced (technology, demand, monetary policy, or supply), the rows are the model responses (output, inflation, nominal interest rates, real interest rates), and the plotted lines are “average non-ZLB economy” (blue) and “average ZLB economy” (red below). The plots are almost overlapping.

That tells us that the economy appears to react the same way, regardless of whether the ZLB is binding. This matches the volatility story.

Although the empirical evidence that the authors find is strong that the ZLB did not impact the macroeconomy, they argue that we should not downplay the significance of the Great Recession. Instead, they argue,

no special role should be attributed to the ZLB constraint as (an) explanation for the depth and persistence of the recession. Instead, the size, persistence and financial nature of the shocks experienced by the U.S. economy (before the start of the ZLB episode) are instead more likely explanations for the severity of the downturn, as had been the case for many other financial crises experienced by different countries in the past, and which did not generally involve a binding ZLB constraint.

While the ZLB may not have affected the economy, it did have an important effect on monetary policy. Since the Fed could not use its main policy instrument (interest rates) at the ZLB, it relied on unconventional monetary policies, such as quantitative easing and forward guidance. But those policies, other issues aside, did fill the policy void left by the ZLB. As the authors write, “We interpret the previous results as suggesting that unconventional monetary policies may have been highly effective in steering the long rate as desired during the ZLB period.”

Looking ahead to the next recession, Debortoli, Galí, and Gambetti’s careful analysis gives hope. If the binding zero lower bound did not prolong the last recession, it may not prolong the next.