The COVID-19 pandemic will hit the global economy hard. How hard exactly is still uncertain and the subject of an ongoing debate (Baldwin and Weder di Mauro 2020, Barro et al. 2020, or Fornaro and Wolf 2020). In this column we summarise our ongoing work on the topic (Dietrich et al. 2020). This work documents how household expectations about the economic impact of COVID-19 have evolved over time. Our evidence is based on the first wave of a survey of US households that we started on 10 March 2020. In what follows, we highlight key aspects of our survey results. Based on these, we derive a scenario for the short-run economic impact implied by the survey expectations. The two main observations are:
- Although the pandemic had been raging in China for months, the survey nevertheless shows a very sharp revision of household expectations within just a few days. On 10 March, US households in the survey anticipated a negligible effect of COVID-19 on US GDP over the next 12 months. By 20 March, the average expected 12-month output loss amounted to 6.8%.
- Uncertainty about these costs is very large, both for individuals and in the cross-section. The standard deviation of the point forecast across respondents is 16.5 percentage points (ranging between +1% and -15% of GDP over the next twelve months).
Survey evidence illustrates arrival of both news and uncertainty
Over the period from 10 March to 20 March we asked Qualtrics Panels to survey 1,100 respondents that are representative of the US population. The main questions of the survey are modelled after the Survey of Consumer Expectations of the New York Fed, focusing on expected GDP losses instead of household income. Among other things, we elicit individuals’ entire forecast distribution as well as their point estimates. We first ask respondents whether they expect the overall effect to be positive or negative. Then we ask them to provide a quantitative answer. Specifically, we ask:
- What do you expect the overall impact of the coronavirus to be over the next 12 months? Please give your best guess.
- I expect the overall economic impact of the coronavirus to be [positive/negative] _____ percent of GDP.
In order to be conservative, we drop a third of the observations with extreme responses.
Figure 1 Expected GDP change due to COVID-19 over 12-month period
Panel A: Average responses over time
Panel B: Distribution of responses
Notes: Panel A reports the average expected GDP change across respondents over time. Whiskers indicate one standard deviation. Panel B presents kernel estimates of the distributions across respondents on 10 March 2020 (blue) and 13-20 March 2020 (red).
In Panel A of Figure 1 we show how the average expected output loss evolves over time. Our observations start on 10 March. At this point the expected output loss is basically zero. Recall that on that day the number of infections in the US just exceeded 1,000 cases. The next observation is three days later (initially we did not solicit responses on each day). At this point, the expected loss is 5.8% on average. In the period up to 20 March, we observe a further adjustment of the expected economic costs of the COVID-19 pandemic. The latest observation is from 20 March. At this point the expected output loss caused by COVID-19 amounts to about 6.8% on average. The largest downward shift so far took place between 10 March and 13 March. It seems news about a significant economic fallout arrived at the representative US household during this period.
In Panel B of Figure 2 we show kernel estimates of the expected output changes. In particular, we compare two distributions. The blue line reflects the responses on 10 March. The red line reflects the responses since then. We observe that not only the centre of the distribution shifted to the left, but the left tail also thickened notably. Some of the very adverse realisations which were considered impossible by the respondents on 10 March were no longer ruled out. This increase in dispersion indicates that uncertainty about the economic fallout from COVID-19 has increased. At the level of individual forecast distributions, we observe a similar level of uncertainty. We present more evidence on this and other details on the survey in our paper (Dietrich et al. 2020).
Short-run economic impact implications for monetary policy
The survey results show a sharp drop in expected GDP and heightened uncertainty around that prediction. In the paper, we discuss some implications of COVID-19 for the conduct of monetary policy. Under normal circumstances, if monetary policy were not constrained by the zero lower bound, offsetting the COVID-19 shock would call for a very large cut in interest rates.
Monetary policy analysis first necessitates classification of the shock as supply or demand. The answer is: “it is complicated”. To us, COVID-19 is first and foremost a supply shock. It will depend on the response of fiscal and monetary policy if the demand side propagates the shock. There can be little doubt, however, that the shock and its side-effects reduce the productive capacity of the world economy and the US. Still, both the news and the uncertainty component mean that this supply shock induces an (endogenous) contraction of private-sector demand. Intuitively, in the face of bad news and increased uncertainty, households and firms increase their savings and reduce expenditures.
In a world without (pricing) frictions, the real interest rate would decline strongly in such situations in order to incentivize households to save less and spend more. This would mitigate the short-run impact of the shock. The adjustment of the natural rate of interest that would obtain under such circumstances is an important benchmark for monetary policy. According to New Keynesian monetary theory, if monetary policy could set the policy rate in sync with the natural rate of interest, it would stabilize the economy at its (temporarily lower than pre-COVID-19) potential output level. And it would maintain price stability.
Figure 2 Response of natural rate of interest to shift in expectations
Notes: This figure presents the response of the natural rate of interest to a shift in expectations due to COVID-19 in mid-March 2020. Computation is based on survey responses and an asset-pricing equation. The horizonal axis measures household risk aversion, while the vertical axis measures response of the natural rate in percentage points.
In our paper, we argue that the quantitative effect of the COVID-19 shock on the natural rate of interest might be large. As a first pass, we obtain this result using a standard asset-pricing equation which ties the natural rate of interest to the expected change in GDP while accounting for uncertainty about future output realisations. We use this equation to evaluate the effect of the expected GDP loss as manifest in the distribution in the red line of Panel B in Figure 1 above. In doing so, we account for both the average output loss and the uncertainty about the output loss. The effect of this uncertainty on the natural rate of interest depends on the degree of household risk aversion. In Figure 2 we show results for various degrees of risk aversion, measured along the horizontal axis. For a value of risk aversion of one, the drop in the natural rate amounts to 870 basis points. A value of one in terms of risk aversion is frequently used in macroeconomic models, but the range of empirical estimates is very wide. However, this does not matter for our main point: The natural rate of interest drops sharply in response to the COVID-19 shock. Ignoring uncertainty for a moment and accounting for the fact that households expect output to drop by 6.1% over the course of the year, the natural rate falls by 613 basis points if risk aversion is one. It falls somewhat less for smaller values of risk aversion and more strongly so if the value is larger. Uncertainty adds about a further one-third to the aforementioned effect.
Recalling the 2007-2009 Global Crisis puts these magnitudes into perspective. During the 2007-2009 crisis, the Fed lowered short-term interest rates by about 500 basis points, starting in the second half of 2007. Rates were cut to zero by the end of 2008. Clearly today’s environment is different. Interest rates were already very low before the COVID-19 outbreak. For this reason, it will likely be harder for central banks to accommodate the drop in the natural rate. The Fed cumulatively lowered interest rates on 3 March and 15 March by 150 basis points in two unscheduled FOMC meetings.
Markets did not calm down afterward. Our survey evidence and the theory-based considerations above provide a clear narrative to explain this reaction. In this reading, the Fed’s rate cuts helped, but they overlapped with the arrival of bad news and a notable rise in uncertainty.
In our paper we show further effects of the COVID-19 outbreak by focusing on the short-run dynamic path of the economy. We explicitly account for the news and the uncertainty aspects and consider alternative assumptions regarding monetary policy. We find that if monetary policy is unconstrained, then it can mitigate the adverse economic short-run effects of the COVID-19 pandemic, simply by tracking the natural rate. If it fails to do so, for instance because it is constrained by the zero lower bound, then there is a substantial additional output loss in the short run. This is in addition to the adverse effects on potential output which materialize quite independently of monetary policy.
Authors' note: The views expressed here are solely those of the authors and do not necessarily reflect the views of the Federal Reserve Bank of Cleveland or the Federal Reserve System.
Baldwin, R and B Weder di Mauro (2020), Economics in the Time of COVID-19, a VoxEU.org eBook, CEPR Press.
Barro, B, J Ursua and J Weng (2020), “Coronavirus meets the Great Influenza Pandemic”, VoxEU.org, 20 March.
Dietrich, A, K Kuester, G. Müller and R Schoenle (2020), “News and uncertainty about COVID -19: Survey Evidence and short-run economic impact”, mimeo.
Fornaro, L and M Wolf (2020), “Coronavirus and Macroeconomic Policy”, VoxEU.org, 10 March.