One of the key channels through which the ongoing global health pandemic affects economic conditions is through household expectations.1 Thus, a key component to evaluating the economic impact of the COVID-19 pandemic, and the measures implemented to contain it, is an assessment of how these have affected household beliefs and expectations. This is doubly important in that the impact of the pandemic on beliefs and expectations could also have scarring effects (Kozlowski et al. 2020).
In this column, I directly assess in what way the pandemic has affected household expectations using the latest consumer survey data on European households. I use the methodology in Ambrocio (2020) to document how the pandemic has affected household expectations along three dimensions: sentiment, disagreement, and uncertainty. Sentiment reflects average views on the expected state of the economy and the households' own financial situations (a first moment of beliefs), disagreement reflects how views in the cross-section of households differ from each other, and, finally, uncertainty reflects the fraction of the population who are uncertain about future changes in the economy and their own (financial) situation.2
Evolution of household sentiment and uncertainty in past crises episodes
Historically, household sentiment tends to fall during recessions while household uncertainty peaked in past crises episodes such as the Global Crisis, the sovereign debt crisis, and Brexit. Figure 1 plots the evolution of these three indices for the euro area over the last two decades.3
Figure 1 Household expectation indices for the Euro area
Notes: CSI is a measure reflecting average household views on the general economic situation and households' financial situation over the next 12 months. DIS is a measure reflecting the dispersion or disagreement of views in the cross-section of households. HUN measures household uncertainty, specifically the proportion of households who respond with "Don't Know" when asked about their views on the general economic situation and their own financial situation. Shaded areas reflect Euro Area Business Cycle Network peak-trough periods.
Household expectations during the COVID-19 pandemic
The COVID-19 pandemic has led to a universal and significant decline in household sentiment over the last two months. Figure 2 plots the development of household expectations across European countries over the last four months.4 In relative terms, the drop in sentiment parallels the global crisis episode for the euro area, although the recent decline has occurred at a much more rapid pace. This drop in sentiment is accompanied by an increase in disagreement across households for many countries. This may indicate that the pandemic has hit households differentially and thus households' views about the future has further diverged. On the other hand, and unlike in previous crisis episodes, household uncertainty was relatively unchanged. This may be due to the rapid evolution of the situation which has enabled households to form (pessimistic) opinions in between the survey dates.
Figure 2 Recent developments in household expectations across Europe
Notes: the figures plot developments in the expectation indices for the 27 European Union countries excluding Lithuania, Latvia, and Malta, the Euro area (solid black), and the United Kingdom over the last four months. The indices have each been rescaled such that the value for January of 2020 is 100. CSI is a measure reflecting average household views on the general economic situation and households' financial situation over the next 12 months. DIS is a measure reflecting the dispersion or disagreement of views in the cross-section of households. HUN measures household uncertainty, specifically the proportion of households who respond with "Don't Know" when asked about their views on the general economic situation and their own financial situation.
The impact of the pandemic and containment measures on sentiment
How do these shifts in household expectations correlate with the severity of the pandemic and the stringency of containment measures implemented to counter it? To answer this question I collect data on confirmed cases and deaths due to COVID-19, as well as a measure of the stringency of containment measures from Hale et al. (2020) and the economic stimulus index of Elgin et al. (2020).5 It should be noted that these are estimated figures and could be subject to measurement error. Nevertheless, the number of confirmed cases and deaths would be the same information that households were exposed to when forming their views.
Table 1 reports correlations between the changes in household sentiment, disagreement, and uncertainty from the January to April surveys of this year against the number of confirmed cases and deaths (per 100,000 population), as well as the index of the stringency of containment measures as of the end of March and the economic stimulus index across European countries.6
Table 1 Correlations between household expectations and COVID-19 across Europe
Notes: the table reports correlation coefficients between the change in the household expectation indices for 24 European Union countries (excluding Lithuania, Latvia, and Malta) and including the United Kingdom over the last four months against the number of confirmed COVID-19 deaths and cases (per 100,000 population), the stringency index of containment measures by Hale et al. (2020) as of end of March 2020, and the Elgin et al. (2020) economic stimulus index (CESI). CSI is a measure reflecting average household views on the general economic situation and households' financial situation over the next 12 months. DIS is a measure reflecting the dispersion or disagreement of views in the cross-section of households. HUN measures household uncertainty, specifically the proportion of households who respond with "Don't Know" when asked about their views on the general economic situations their financial situation.
The stringency of containment measures appears to be most correlated with changes in household expectations. In particular, countries with more stringent containment measures are associated with larger drops in consumer sentiment and more disagreement across households. Paradoxically, more confirmed deaths are correlated with smaller drops in sentiment. Further, more confirmed cases and deaths are also correlated with milder increases in disagreement. On the other hand, disagreement is increasing in the stringency of containment measures and the intensity of economic stimulus which, in turn, suggests that these policy responses may have affected households differently. Finally, household uncertainty is only weakly positively correlated with the number of deaths.
In Figure 3, I plot the change in household sentiment against the number of deaths and the stringency of containment measures for several European countries. I also report the coefficients from a regression of the change in household sentiment on all of the pandemic-related measures. The negative relationship between household sentiment and the stringency of containment measures is supported by the regression results. Moreover, the puzzling positive relationship between deaths and sentiment remain after controlling for the stringency of the containment measures. This positive relationship may indicate that policy responses to the pandemic are not adequately covered by the stringency and stimulus indices. More cases and deaths could also indicate stronger responses from the government in other areas and thus a smaller drop in sentiment.
Figure 3 Household sentiment against COVID-related deaths and containment
Notes: the figures plots the change in the household sentiment for 24 European Union countries (excluding Lithuania, Latvia, and Malta) and including the United Kingdom over the last four months against the number of confirmed COVID-19 deaths (per 100,000 population) and the stringency index of containment measures by Hale et al. (2020) as of end of March 2020. CSI is a measure reflecting average household views on the general economic situation and households' financial situation over the next 12 months. Standard errors of the regression coefficients are reported in parentheses below the point estimates.
The relative importance of the stringency of containment measures in explaining the change in household expectations parallels the results in Baker et al. (2020) who find that US states with stronger shelter-in-place policies feature sharper decreases in household spending.
Concluding Remarks
In sum, the COVID-19 pandemic has largely affected household expectations through a drop in sentiment, a first moment of beliefs. Furthermore, stringent containment measures correlate positively with larger drops in sentiment. On the other hand, household uncertainty has been relatively less affected by the pandemic. These results suggest that the expectations channel of the transmission of the health pandemic to economic conditions may largely operate through an increase in household pessimism and less so in terms of increases in household uncertainty. Nevertheless, uncertainty may have risen due to COVID-19 in other areas. For instance, Baker et al. (2020) document an increase in uncertainty for businesses, financial markets, and in the news media in the US.
What are the implications? Results from vector-autoregressions in Ambrocio (2020) suggest that sentiment shocks are like demand shocks in which drops in sentiment lead to higher unemployment and lower inflation. On the other hand, disagreement and uncertainty shocks tend to lead to a delayed increase in unemployment and higher inflation. The observed decline in sentiment, and increase in household disagreement, indeed suggest that unemployment or economic activity may fall through a household expectations channel, while the overall effects on inflation would depend on country specifics.
Author's note: The views and opinion presented in this column are the author's and do not necessarily reflect those of the Bank of Finland.
References
Ambrocio, G (2020), “Inflationary household uncertainty shocks”, Bank of Finland Research Discussion Papers, May.
Baker, S, N Bloom, S Davis and S Terry (2020), “COVID-induced economic uncertainty”, NBER Working Paper No. 26983.
Baker, S, R A Farrokhnia, S Meyer, M Pagel and C Yannelis (2020), “How does household spending respond to an epidemic? Consumption during the 2020 Covid-19 Pandemic”, NBER Working Paper No. 26949.
Baldwin, R and B Weder di Mauro (2020a), Economics in the Time of Covid-19, CEPR Press: VoxEU.org eBook.
Baldwin, R and B Weder di Mauro (2020b), Mitigating the COVID Economic Crisis: Act Fast and Do Whatever it Takes, CEPR Press: VoxEU.org eBook.
Elgin, C, G Basbug and A Yalaman (2020), “Economic policy responses to a pandemic: Developing the COVID-19 economic stimulus index”, Covid Economics 3. CEPR Press.
Hale, T, S Webster, A Petherick, T Phillips and B Kira (2020), “Oxford COVID-19 Government Response Tracker”, Blavatnik School of Government.
Kozlowski, J, L Veldkamp and V Venkateswaran (2020), “Scarring body and mind: The long-term belief-scarring effects of Covid-19”, Covid Economics 8. CEPR Press.
Endnotes
1 See Baldwin and di Mauro (2020a, 2020b) for a broad overview.
2 Specifically, respondents' answers with regard to their expectations on the change in employment, general economic situation, their expected ability to save, and expected change in their financial situation (all over the next 12 months) are quantified to a range from 1 (Much better/more) to -1 (Much worse/less) and responses of "Don't know" are counted separately. The CSI is the average quantified response across questions, disagreement is the cross-sectional standard deviation of the responses, and uncertainty is the fraction of respondents choosing "Don't know." The indices use the set of questions consistent with the definition of the official Consumer Confidence index prior to January 2019 such that all questions employed in the indices are forward-looking.
3 The indices have been standardized such that 100 equals the historical mean over the period January 2000-December 2019 and a 10 point change reflects a one standard deviation change (calculated over the same period).
4 The March survey was largely conducted prior to the implementation of containment measures across most countries. The April survey was conducted under unusual (web-assisted) circumstances for some countries due to the containment measures in place. The countries included in the analyses are the 27 European Union member countries excluding Lithuania, Latvia, and Malta, and including the United Kingdom. The updated dataset on household expectations indices can be found at https://sites.google.com/site/ambrociogpg/research .
5 The number of confirmed cases and deaths are consistent with the European Center for Disease Prevention and Control figures. The economic stimulus index by Elgin et al. (2020) standardizes economic policy responses across countries which cover fiscal, monetary, and exchange rate measures.
6 The change from January to March is used for Italy given the available survey data.
