Today, we are pleased to present a guest contribution written by Valerio Ercolani, from the Directorate General for Economics, Statistics and Research at the Bank of Italy. The views expressed in this note represent that of the author and not necessarily reflect those of Bank of Italy.

Last months saw an unprecedented rise in the US saving rate. Most of the accumulated saving was undoubtedly generated by the social distancing and lockdown measures imposed by the government, however part of it could also have been driven by precautionary motives due to grim labor prospects. Some back-of-the-envelope calculations show that the dynamics of the unemployment rate alone can induce a large increase in (precautionary) saving for this year, which can raise the prospect of a new saving glut.

The economic effects of the Covid-19 outbreak have become manifest in official data. In the US, GDP contracted by 5% in Q1, at an annual rate, with service consumption contributing the most to the contraction. During the last three months, employment fell by roughly 19 million units with unemployment rate settling at almost 14% in May. In fact, the latter can be underestimated, for example, Hamilton (2020) shows, through some back-of-the-envelope calculations, that the actual unemployment rate could be even larger, around 20%. In April, the saving rate rose 20.3 percentage points (to 33%), the largest monthly increase since the 1960s (Figure 1). In May, it retreated to a still high 22.3%. Personal saving skyrocketed itself: it almost tripled from February, moving from roughly 1400 to 4100 billions of dollars.

Figure 1: Saving rate and unemployment rate (monthly frequency, US). Source: Bureau of Economic Analysis and Bureau of Labor Services

Most of the accumulated saving was undoubtedly generated by the social distancing and lockdown measures imposed by the government during the emergency, which significantly limited the households’ ability to spend. However, part of it could also have been driven by precautionary motives, related to both medical and economic concerns, and even after the economy fully reopens and the effect of the lockdown on the savings rate wanes, those other factors may well continue to be an important driver. On the health front, the virus may not disappear until a vaccine is developed, produced and distributed on a large-scale; indeed, recent studies show that the pandemic could last between 18 and 24 months, because the virus won’t be halted until about 60-70% of the population is immune (Barry et al., 2020). If this is the case, fear of contracting the Covid-19 may continue to drive people’s behavior even in the post-emergency phase. Hence, demand for physical-proximity services will be subdued and, in some extreme cases, may not resume completely (e.g., in tourism-related or in transport sectors). At the same time, the supply side will likely remain an issue: in order to curb the probability of new outbreaks, several US States, including Florida, Texas, New York and California, are imposing on businesses some forms of social distancing that, while looser than during the emergency, will inevitably constrain their ability to produce and deliver goods and services. All that increases the likelihood of business defaults or resizing, amplifying the risk of income losses or layoffs and hence stimulating households’ precautionary saving (see, among others, Hugget, 1993).

Indeed, labor prospects are grim. Barrero et al. (2020) look at Covid-19 as a reallocation “labor” shock and conclude that almost half of the recent layoffs will be permanent. This is consistent with the results by Coibion et al. (2020) which reports that expectations about the level of unemployment have increased not only for the short term, but also over a medium horizon (three to five years).

In particular, unemployment rate has been shown to be a crucial driver of precautionary saving, on both theoretical and empirical grounds; see, for example, Carroll et al. (2012) and Mody et al. (2012). The latter shows, for a panel of 27 advanced countries (including US), that a 1 percentage point (pp) increase in unemployment rate – taken as a proxy for labor income uncertainty – raises the saving rate by roughly 0.3 pp. The estimated regression controls for a number of other determinants of saving, including the one-year ahead disposable income to capture for “first-moment effects” stemming from changes in economic activity.

While we know that a careful analysis on precautionary saving would need household-level data (see, e.g., Attanasio and Weber, 1993), we use the above mentioned result to perform some back-of-the-envelope calculations about the effects of unemployment rate on the average US saving rate in 2020, due to precautionary motives (Table 1). We present three scenarios, based on alternative assumptions about the increase in US unemployment during the year. The most conservative one is an increase from the 3.7% observed in 2019 to 10%, in line with the projections of IMF (2020). The worst-case scenario assumes an increase to 20%, more in line with the projections in Petrosky-Nadeu et al. (2020). We take an intermediate of 15% as our benchmark. Using the coefficient estimated by Mody et al. (2012), i.e. 0.3, in the intermediate scenario the saving rate would increase by 3.4 pp, which would represent the largest yearly surge since the 1960’s. Indeed, so far, the largest increase occurred in 1970 and amounted to 1.9 pp.

Applying the same back-of-the-envelope calculation, the observed rise in unemployment during the Great recession period (3.5 pp between 2008 and 2009) would have predicted an increase of the saving rate in 2009 of 1 pp; although other factors have clearly been at play, this prediction is strikingly close to the observed increase (1.1 pp).

Table 1: Precautionary saving and unemployment rate in US. Notes: the table reports the back-of-the-envelope calculations using Mody et al. (2012) and Muellbauer (2020).

The flip side of a saving expansion is a fall in consumption. Muellbauer (2020) estimates an augmented-consumption function for US, finding that one pp increase of unemployment in a quarter generates a fall in consumption of roughly 0.5%, in the same quarter. Back-of-the-envelope calculations, which convert quarterly changes into yearly ones, show that that the intermediate scenario in Table 1 is associated to a fall in consumption of roughly 3%. The implied increase in the saving rate of such a fall in consumption, holding income constant, is roughly 2.9 pp, very close to the figure obtained from Mody et al. (2012).

All in all, the proposed figures depict only part of the story because saving is driven by a number of other factors such as income, wealth and interest rate dynamics (see, among others, Dynan, 2008), on top of the continuation of the social distancing measures. Therefore, a new saving glut could emerge globally to the extent that the saving dynamics observed and anticipated for the US is not an unicum. Indeed, saving rates in the major European countries rose sharply last March as well as bank deposits (Arnold and Storbeck, 2020). As conjectured by Blanchard (2020) and Goy (2020) such a striking surge in (precautionary) saving could be long-lasting, potentially reinforcing some already existing tendencies such as persistently low levels of inflation and of the natural rate.

This post written by Valerio Ercolani.