The crisis has left deep scars, which will affect both supply and demand for many years to come.
The Covid-19 pandemic in the US has led to volumes of initial claims for unemployment and unemployment rates not seen since the Great Depression period, pushing the economy into a recession. As policymakers map out potential recovery paths, much of the debate tends to focus on short-run and medium-run implications. Can we hope for a ‘V-shaped’ rebound, at least once vaccines have been widely distributed, or will it take a long time between economic decline and subsequent recovery akin to a ‘U-shaped’ rebound or worse (Baldwin and di Mauro 2020, An and Loungani 2020)? What has received less attention are the potential long-run implications. History tells us that economic crises like the current one can alter consumer behaviour in the long-run – beyond the effects captured by standard economic variables such as current employment and employment prospect, current income, and wealth. The most famous example might be the Great Depression and the lingering “mood of pessimism that for a long time affected markets” (Friedman and Schwartz 1963). But even looking back at the Great Recession of 2008, we have observed that the crisis had long-term effects on consumer demand. Consumption was slow to return to pre-crisis levels, not only in absolute levels, but also relative to the growth of income, net worth, and employment (Petev et al. 2011, De Nardi et al. 2012, Fatás and Summers 2016). If standard life-cycle consumption channels, such as time-varying financial constraints, as well as explanations building on loss of worker skills or low private investment (Hansen 1939, Blanchard and Summers 1986, Delong and Summers 2012, Summers 2014) fail, what, then, explains the lasting effects of economic crisis on consumer demand?
Estimating scarring in consumption
In a recent paper (Malmendier and Shen 2021), we propose that past economic crises leave lasting scars on those who have lived through them, with important implications for our understanding of the long-run consequences of recessions. We operationalise the notion of scarring effects building on the macro-finance literature on the long-lasting effects of stock-market and inflation experiences (Malmendier and Nagel 2011, 2015). What this prior literature has shown is that individual investors tend to assign significantly more weight to realisations of economic conditions that have occurred during individuals’ lifetimes than other realisations or learned information.
To test the long-term effects of past economic conditions in our context of consumer expenditures, we focus on personal exposure to unemployment as the source of scarring effects. Coibion et al. (2015) have singled out unemployment as the most spending-relevant variable. Here we ask whether even unemployment experiences far in the past leave a detectable impact on consumer choices. We consider both the macro experiences of living through spells of high local and national unemployment, and personal unemployment experiences (controlling for the macro environment). The latter most clearly distinguishes our approach and the proposed scarring effects from existing learning models (e.g. Kozlowski et al. 2020).1 The measures are constructed to not only account for all experiences individuals have accumulated during their lifetimes, but also allow for experience effects to decay over time, for example as memory fades or structural change renders old experiences less relevant. To ensure that we distinguish scarring effects from known earnings implications of job loss (Jacobson et al. 1993, Couch and Placzek 2010), we control for earnings in the recent past and exclude experiences from the recent past from our measures.
We estimate the effect of prior personal and macro unemployment experience on consumption using the Panel Study of Income Dynamics (PSID).2 Our estimation model controls for all known determinants of expenditures: wealth (first-order and second-order logarithm of liquid and illiquid wealth), income (first-order and second-order logarithm of income and lagged income), age dummies, household characteristics (current employment status, family size, gender, years of education, marital status, and race), as well as time dummies, state dummies, and household dummies.3,4 Our identification relies on three margins of variation: people differ in their prior exposure to unemployment depending on their (1) cohort and (2) location at each given point in time, and these cross-sectional differences evolve over (3) time. Figure 1 illustrates the sources of identifying variation through a simple example of the relation between experience shocks and consumption from the Great Recession.
Figure 1 An example of experience shocks from the Great Recession
Notes: Two individuals (A and B) have the same age (born in 1948) but live in different states (Pennsylvania and Alabama) during the 2007-2013 period. A third person (C) lives in the same state as B (Alabama) but differs in age (born in 1975). The red (dark) bars depict the 2007 and 2013 unemployment experiences of person A and the red (dark) line the corresponding change of total consumption per member of A’s family. Similarly, the blue (medium dark) bars and line show person B’s unemployment experiences and consumption and the green (light) bars and line person C’s unemployment experiences and consumption. A enters the crisis period with a higher exposure to unemployment than B (5.81% versus 5.70%). But, since unemployment rates were lower in Pennsylvania than in Alabama during the crisis, A’s lifetime experience worsens less over the course of the financial crises and becomes relatively more favourable than that of B by 2013 (6.06% versus 6.11%). Person C has even lower macroeconomic unemployment experiences before the crisis period than B (5.46%), but, being the younger person, C is more affected by the crisis which leads to a reversal of the lifetime unemployment experience between the old and the young by the end of the crisis (6.11% versus 6.20%). The increase in unemployment experiences of Person A, B, and C by 0.25%, 0.41%, and 0.74%, respectively, were accompanied by decreases in consumption in the same relative ordering, by 7%, 13%, and 21%, respectively. All consumption expenditures are measured in 2013 dollars, adjusted using PCE. Person A’s ID in the PSID is 45249; person B’s ID in the PSID is 53472; person C’s ID in the PSID is 54014.
Source: Malmendier and Shen (2021).
Four main results
We find a robust effect of both personal unemployment experiences and exposure to (national and local) unemployment rates on consumption expenditures, controlling for financial constraints, income, wealth, and demographics. A one standard-deviation increase in personal unemployment experiences is associated with a 0.92%–1% ($344–$370) decline in total annual consumption spending, and a one standard-deviation increase in the macro-level measure with a 1.65%–1.90% ($615–$708) decrease.5,6
What channel drives the effect of past experiences on consumption? Using the Michigan Survey of Consumers (MSC) from 1953 to 2018, we find that individuals who have lived through worse times consider their own financial future to be less rosy and times to be generally bad for spending on durables, controlling for all historical data, current unemployment, and other macro conditions. The findings suggest that economic conditions individuals have experienced in the past have a lingering effect on their beliefs about the future, giving rising to experience-based learning.7
We were concerned that consumers’ reduction in consumption and their pessimism could reflect (unobserved) determinants of households’ future income that are correlated with past unemployment experiences. To test this alternative explanation, we relate the same measure of lifetime unemployment experiences to actual future income, up to five PSID waves (ten years) in the future. We fail to identify any robust relation. In other words, while past experiences of unemployment exert a strong influence on beliefs about the future and on consumption expenditures, actual future income does not explain these adjustments.
Future wealth build-up
If consumers become more frugal in their spending after negative past experiences, even though they do not earn a reduced income, we would expect their savings and ultimately their wealth to increase. We confirm this prediction in the data. Using a horizon of three to six PSID waves (six to 12 years into the future), we find that a one standard-deviation increase in lifetime exposure to personal (macroeconomic) unemployment leads to additional precautionary savings and resulting wealth build-up of about 1.6% or $3,500 (1.5% or $3,100) ten years later.
Heterogeneity across cohorts
A key implication of the experience-effect hypothesis is that younger cohorts react more strongly to a shock than older cohorts since that shock makes up a larger fraction of their life histories so far. As a result, the cross-sectional differences vary over time as households accumulate different histories of experiences. Consistent with this conjecture, the time-series of household expenditures in Figure 2 (expressed as deviations from the cross-sectional monthly means from the Nielsen data) reveals that the spending of younger cohorts is more volatile in general and was significantly more negatively affected by the Great Recession than those of other age groups.8 More broadly, this result implies that past exposure to unemployment shocks gives rise to generational differences in consumption patterns.
Figure 2 Monthly consumption expenditure by age group
Notes: Six-month moving averages of monthly consumption expenditures of young (below 40), mid- aged (between 40 and 60), and old individuals (above 60) in the Nielsen Homescan Panel, expressed as deviations from the cross-sectional mean expenditure in the respective month and deflated using the personal consumption expenditure (PCE) price index of the US Bureau of Economic Analysis (BEA). Observations are weighted with Nielsen sample weights.
Source: Malmendier and Shen (2021).
Experience effects could constitute a novel micro-foundation underlying fluctuation in aggregate demand and long-run effects of macroeconomic shocks. In Figure 3, we provide suggestive evidence on the aggregate implications of experience effects. We relate an aggregate measure of lifetime experiences in the US population to a measure of aggregate consumption expenditure in the US from 1965 to 2018. As shown, there exists a negative relationship between the two measures: times of higher aggregate unemployment experience coincide with times of lower aggregate consumer spending. The strong negative correlation pattern not only adds credibility to our micro-level estimates but also suggests the possibility that personally experienced labour market conditions may be a significant granular source of aggregate fluctuations.
Figure 3 Aggregate unemployment experience and consumer spending
Notes: The figure relates an aggregate measure of lifetime experiences in the US population to a measure of aggregate consumption expenditure in the US from 1965 to 2018. Aggregate unemployment experience is calculated as a weighted average of national unemployment experience, with the weights being US population by age (restricted to age 25 to 75) from the Census. Aggregate consumer spending is measured as real personal consumption expenditure (PCE) from the US Bureau of Economic Analysis (BEA) normalised by real gross domestic product (GDP), detrended by removing a linear time trend from the series.
Source: Malmendier and Shen (2021).
An, Z and P Loungani (2020), “Forecasting Recoveries is Difficult: Evidence from Past Recessions”, VoxEU.org, 26 April.
Baldwin, R and B Weder di Mauro (2020), “Economics in the Time of COVID-19: A New eBook”, VoxEU.org, 6 March.
Blanchard, O (2012), “Sustaining a Global Recovery”, Finance and Development 46(3): 9–12.
Blanchard, O and L Summers (1986), “Hysteresis and the European Unemployment Problem”, NBER Macroeconomic Annual 1: 15–90.
Coibion, O, Y Gorodnichenko and G H Hong (2015), “The Cyclicality of Sales, Regular and Effective Prices: Business Cycle and Policy Implications”, American Economic Review 105(3): 993–1029.
Couch, K A and D W Placzek (2010), “Earnings losses of displaced workers revisited”, American Economic Review 100(1): 572–89.
Delong, B and L Summers (2012), “Fiscal Policy in a Depressed Economy”, Brookings Papers on Economic Activity 44 (1): 233–297.
De Nardi, M, E French and D Benson (2012), “Consumption and the Great Recession”, Federal Reserve Bank of Chicago Economic Perspectives 1: 1–16.
Fatás, A and L Summers (2016), “Hysteresis and Fiscal Policy during the Global Crisis”, VoxEU.org, 12 October.
Friedman, M and A J Schwartz (1963), A Monetary History of the United States, 1867-1960, Princeton University Press.
Hansen, A H (1939), “Economic Progress and Declining Population Growth”, The American Economic Review 29(1): 1–15.
Jacobson, L S, R J LaLonde and D G Sullivan (1993), “Earnings losses of displaced workers”, The American Economic Review 83(4): 685–709.
Kozlowski, J, L Veldkamp and V Venkateswaran (2020), “The tail that wags the economy: Beliefs and persistent stagnation”, Journal of Political Economy 128(8): 2839–2879.
Low, H, C Meghir and L Pistaferri (2010), “Wage Risk and Employment Risk over the Life Cycle”, American Economic Review 100(4): 1432–67.
Malmendier, U and S Nagel (2011), “Depression Babies: Do Macroeconomic Experiences Affect Risk-Taking?”, Quarterly Journal of Economics 126: 373–416.
Malmendier, U and S Nagel (2015), “Learning from Inflation Experiences”, Quarterly Journal of Economics 131(1): 53–87.
Malmendier, U and L S Shen (2021), “Scarred Consumption”, Working paper.
Petev, I, L Pistaferri and I Saporta-Eksten (2011), “An Analysis of Trends, Perceptions, and Distributional Effects in Consumption”, In B W David, B Grusky and C Wimer (eds.), The Great Recession, Handbook of the Economics of Finance, pp. 161–195, Russell Sage Foundation.
Summers, L H (2014), “US Economic Prospects: Secular Stagnation, Hysteresis, and the Zero Lower Bound”, Business Economics 49(2): 65–73.
1 We are the first to contrast effects of personal experiences from exposure to macro conditions, which had been the focus of the prior literature.
2 The PSID has the advantage of containing rich information on household wealth and long time-series coverage that allows us to construct lifetime experience measures for each household. We also replicate the analysis using the Nielsen Homescan Data, which contains details about the products that households purchase at the Universal Product Code (UPC) level for each shopping trip and allows us to control more finely for time (year-month) effects, and the Consumer Expenditure Survey (CEX), which contains a more comprehensive list of product categories, and thereby more extensively sheds light on the impact of unemployment experience on total, durable, and non-durable consumption. The estimated effects across all three datasets are not only consistent but in fact very similar in economic magnitude.
3 To the best of our knowledge, our analysis is the first to identify experience effects solely from time variation within household. The inclusion of household fixed-effects fully controls for any unspecified time-invariant household characteristics and implies that we identify scarring effects solely from time variation in the within-household co-movement of consumption and unemployment histories. We have also estimated a model with only cohort fixed effects.
4 In the paper, we conduct a broad range of robustness checks and replications using variations in wealth, income, and liquidity controls, including four additional variants of wealth controls (third-order and fourth-order liquid and illiquid wealth, decile dummies of liquid and illiquid wealth, separate controls for housing and other wealth, and controls for positive wealth and debt) and four additional variants of the income controls (third-order and fourth-order income and lagged income, quintile dummies of income and lagged income, decile dummies of income and lagged income, and five separate dummies for two-percentile steps in the bottom and in the top 10% of income and lagged income). Furthermore, we address the concern about measurement error in income by incorporating estimates of the extent of measurement error in income in our regressions and assessing whether they affect our coefficients of interest.
5 We find consistent results of similar magnitude in the Nielsen and CEX data. Furthermore, we find that households who have lived through worse employment conditions lower the quality of their consumption, as they are more likely to use coupons, purchase lower-end products, and allocate more expenditures toward sale items.
6 In the paper, we further distinguish the proposed scarring mechanism from an even broader array of life-cycle consumption factors using a stochastic life-cycle model from Low et al. (2010). We show that the negative relationship between past downturns and consumption cannot arise from financial constraints, income scarring (the notion that job loss reduces income flows because of lower match quality in future jobs), or unemployment scarring (the notion that unemployment, once experienced, makes individuals inherently less employable).
7 The evidence on experience-based learning (beliefs channel) does not rule out that experience-based taste changes (preference channel) are also at work. Nevertheless, we find that the effects do not operate through the channel of habit formation.
8 In the paper, we directly test for the implication that the young would lower their consumption expenditure to a greater degree than older cohorts during economic busts and, vice-versa, increase it more during booms and find supporting evidence.