RESEARCH The Initial Household Spending Response to COVID-19

Evidence from Credit Card Transactions

COVID-19 has rapidly transformed our nation. Following the declaration of a national emergency on March 13, 2020, the U.S. caseload exceeded 100,000 on March 29, and by April 6, 90 percent of the U.S. population was subject to “stay-at-home” orders. Within a matter of weeks, vacations and special events were cancelled, and routine trips to the store, workplace, and restaurants became hindered by both the virus and the policies designed to prevent its spread. 

These almost universal disruptions to normal activity have already had unprecedented consequences for the economy. The pandemic has shut down large sectors of the economy deemed “non-essential,” leaving millions of workers jobless. Social distancing restrictions have all but prohibited the consumption of certain goods and services. The government has responded with a massive recovery act to bolster income by funding stimulus checks, Unemployment Insurance (UI) supplements, and the Payroll Protection Program. 

In this report, we provide preliminary high-frequency evidence of the reaction of consumer spending to these events. We ask two main questions. First, how much has individual spending fallen, and how does this drop vary across households1? Second, can heterogeneity across households provide suggestive evidence about the spending decline caused by the nearly ubiquitous pandemic and policies intended to contain it versus the initial round of income losses during that period? With consumer spending accounting for roughly 70 percent of GDP, understanding the magnitude and causes of changes in consumption is critical to identifying policy interventions that could aid in accelerating an economic recovery.  This will be increasingly important as the pandemic and policy impacts interact with increasing job loss and additional policies to ameliorate the job loss, such as stimulus payments and UI.

To answer these questions, we use a dataset based on the universe of transactions made on Chase credit cards through April 11, 2020. We focus on a sample of 8 million families across all fifty states who have been active users of their credit cards since January 2018. For a subset of our analyses we pair these credit card data with checking account data through February 2020, which allow us to segment our population by income levels and industry of employment before the COVID shock3.

The key strengths of these data are a large sample size and the ability to track the spending patterns of specific households across time. We are thus able to provide detailed estimates of the spending drop and to analyze heterogeneity in this drop across household characteristics and across categories of spending. We decompose the spending drop into non-essential and essential spending, speaking to the pandemic-induced closure of many non-essential businesses. We also look at changes in spending across the pre-COVID income distribution. Finally, we stratify the sample by individuals’ industry of employment to test whether those employed in sectors with higher expected rates of job loss cut spending by a larger amount. 

Despite these strengths, our findings come with several important caveats that stem from the fact that, at the time of writing, we only observe the subset of spending that occurs on a household’s Chase credit cards through April 11. We do not observe spending using debit cards, cash, electronic payments, and non-Chase credit cards. Our estimates could be biased to the extent that there is substitution between these alternative channels and Chase credit cards coincident with our analysis period. We may be particularly concerned about this type of substitution at a time of acute economic disruption when households might turn to credit cards to smooth their consumption. In addition, the pandemic might accelerate the growth in card transactions as people avoid the risks associated with exchanging physical cash and because of the growth in online spend. This might cause us to understate the drop in spending.

Second, while our data include households across a wide cross-section of income levels and geographies, Chase credit card holders tend to be more affluent than the average U.S. household. As we show below, if higher-income families cut their spending to a greater extent, the sample frame could cause us to overstate the drop in spending. Thus, the net effect of these biases on our spend estimates is ambiguous.

Third, at the time of this release our preliminary data only cover the initial phase of the pandemic. Spending changes, and how these vary with household characteristics, may evolve over time particularly as income disruptions become more widespread.

In the future, we will be able to partially address these three limitations by looking at a longer time-period of data and examining checking account transactions to provide an integrated view of income and spending.

We have four main findings. First, we find that average weekly household credit card spending fell by 40 percent year-over-year by the end of March 2020, coinciding with a dramatic increase in COVID-19 cases, social distancing policies, and job losses. The magnitude of the spending drop is enormous; it is eight times larger than the spending drop typically observed among UI recipients in the first month after job loss. Second, spending cuts on non-essential goods and services account for nearly all of the total spending decline. Spending on essentials initially spiked 20 percent before falling back, while spending on non-essentials declined by 50 percent. Third, spending dropped substantially for households across the entire income distribution, with slightly larger drops for higher-income households driven by cuts in non-essential goods and services. Fourth, spending dropped dramatically for workers in all industries of employment. Similar drops occurred in industries with high and low rates of job loss as of April 2020.

In summary, we provide evidence suggesting that, as of the second week of April, the 40 percent drop in consumer spending appears to be driven to a greater extent by the pandemic and social distancing policies implemented across the country to prevent its spread and to a lesser extent by the initial round of income losses. However, as the pandemic unfolds, the balance of factors contributing to spending behavior could change dramatically. We will continue to track and disentangle these dynamics over time using administrative banking data.

 

Finding One: Average household credit card spending had fallen by 40 percent year-over-year by the end of March 2020.

Figure 1 plots the year-over-year percentage change in weekly credit card spending in 2020 and in 2019, and Figure 2 shows levels of average weekly credit card spending in 2020 and in 2019. 

Changes in spending follow a distinctive pattern — spend is stable through the beginning of March, then declines precipitously by 40 percent relative to 2019 from the second through fourth week of March. It then appears to stabilize at this lower level in the first two weeks of April.  The size of the spending drop is largely consistent with other estimates from similar administrative data sources during the same time frame.

Figure 1: Average weekly household credit card spending had fallen by 40 percent year-over-year by the end of March 2020.

Year-over-year percent change in credit card spending in 2020

Source: JPMorgan Chase Institute
Line graph showing the year-over-year percent change in credit card spending per household in 2019 and 2020 from January 4 to April 11. Spend is stable through the beginning of March, then declines precipitously by 40 percent relative to 2019 from the second to fourth week of March. It then stabilizes at a lower level in the first two weeks of April.

Figure 2: Average weekly credit card spending per household was more than $300 lower in April 2020 compared to April 2019. 

Average weekly credit card spending per household ($)

  • 2019
  • 2020
Source: JPMorgan Chase Institute
Line graph showing levels of average weekly credit card spending in 2020 and 2019 from January 4 to April 11. Average weekly credit card spending was more than $300 lower in April 2020 than in April 2019.

The timing of the spending drop mirrors the spread of the virus and staggered national implementation of government social distancing orders. A national emergency was declared on March 13, 2020.  Over the following three weeks, the number of states with stay-at-home orders increased from zero to forty-five, and then also remained stable (see Figure 3). The prevalence of COVID-19 also increased dramatically with over 300,000 cases and 5,000 COVID-related deaths in the U.S. by the month’s end.

At the same time, the drop in spending also closely tracks the pattern of initial job losses. UI claims began spiking in the third week of March, with more than 20 million UI claims filed by April 11. This raises the question as to how much of the 40 percent drop in credit card spending is due to the pandemic itself, the social distancing policies, or income losses.

Importantly, while we know from UI claims that jobs have been lost, it is unlikely that the income supports extended by the government in response to COVID-19 would have been received by the end of our time series—the second week of April. The median time between job loss and the first UI benefit receipt is roughly five weeks, which would mean that many of the first 3 million people to file for UI during the third week of March—the initial surge in UI claims—may not have started receiving their UI benefits until late April. In addition, families did not start receiving stimulus checks until the third week of April. Thus, to the extent that income losses are playing a role, they would likely not yet have been offset by policy interventions to mitigate those losses.

Nonetheless, it is still useful to calibrate the size of the spending drop relative to what we have observed among those who lose a job involuntarily during normal times. We have previously used these data to measure the spending drop around job loss among UI recipients, and observed an initial credit card spending drop of roughly 5 percent (Ganong and Noel 2019). In other words, the spending drop in March 2020 is roughly eight times larger than the average household credit card spending drop in the first month of unemployment for UI recipients in normal times. This puts into perspective how dramatic the spending drop is and suggests that the pandemic and policies aimed at preventing its spread are contributing substantially to the drop in spending. We explore this possibility further in Findings 3 and 4.

Figure 3: UI claims, social distancing policies, and COVID-19 cases all increased dramatically during late March and early April.  

Number of Unemployment Insurance (UI) claims and prevalence of social distancing policies and COVID-19 cases

  • Number of new Unemployment Insurance (UI) claims (Source: The Department of Labor)
  • Number of states with stay-at-home orders in place (Source: Pew Stateline)
  • Number of COVID-19 cases in the U.S. (Source: The New York Times)
Source: JPMorgan Chase Institute
Line graph showing the number of COVID-19 cases in the United States, the number of new Unemployment Insurance claims, and the number of states with stay-at-home orders between January 4 and April 11, 2020. In the three weeks after March 13, the number of states with stay-at-home orders increased from 0 to 45 and then remained stable. The prevalence of COVID-19 also increased with over 300,000 cases by the end of March. UI claims began spiking in March with more than 20 million claims filed by April 11.

Finding Two: Spending on essentials initially spiked 20 percent before falling to below pre-pandemic levels, while spending on non-essentials declined by 50 percent and accounted for nearly all of the total spending decline.

While Finding 1 shows a sharp drop in aggregate spending, there is reason to think that specific spending categories would be differentially impacted. Many non-essential businesses, like bars and salons, were closed by state and local governments. Similarly, stay-at-home orders limited the ability of individuals to travel. Beyond the mechanical effect of social distancing regulations, individuals may also have independently curtailed spend in certain categories to avoid risk of infection or as a response to income loss.

We begin by disaggregating total spending into essential and non-essential categories, as commonly defined in state “stay-at-home” orders. Figures 4 and 5 show a dramatic difference in the path of essential and non-essential spending. Essential spending spiked in early March, up almost 20 percent by the second week. It then fell back down, stabilizing at a decline of around 20 percent by early April. In contrast, spending on non-essential categories fell sharply throughout March before stabilizing down 51 percent in early April4.

Figure 4: Spending in non-essential categories dropped by roughly 50 percent year-over-year compared to 20 percent for essential categories.

Year-over-year percent change in credit card spending by spend category

  • Essential
  • Non-essential
Line graph showing year-over-year percent change in credit card spending for essential and non-essential types. Essential spending spiked in early March up almost 20 percent by the second week before falling back down, stabilizing at a decline of around 15% by early April. Spending in non-essential categories dropped by roughly 50%. Note: We use state social distancing orders that restricted non-essential goods and services to categorize spend. “Essential” categories include fuel, transit, cash, drug stores, discount stores, auto repair, groceries, telecom, utilities, insurance, and healthcare. “Non-essential” includes department stores, other retail, restaurants, entertainment, retail durables, home improvement, professional and personal services, and miscellaneous. Although flights, hotels, and rental cars are sometimes categorized as “essential” and not technically closed, we include them in the “non-essential” group because they are affected by stay-at-home restrictions on non-essential travel.
Source: JPMorgan Chase Institute

Figure 5: Average weekly household spending on non-essential categories dropped by roughly $200 

Average total credit card spending per customer by spend category ($)

  • Essential
  • Non-essential
Line graph showing average household weekly credit card spending on essential and non-essential types. Average weekly household spending on non-essential categories dropped by roughly $200. Note: We use state social distancing orders that restricted non-essential goods and services to categorize spend. “Essential” categories include fuel, transit, cash, drug stores, discount stores, auto repair, groceries, telecom, utilities, insurance, and healthcare. “Non-essential” includes department stores, other retail, restaurants, entertainment, retail durables, home improvement, professional and personal services, and miscellaneous. Although flights, hotels, and rental cars are sometimes categorized as “essential” and not technically closed, we include them in the “non-essential” group because they are affected by stay-at-home restrictions on non-essential travel.
Source: JPMorgan Chase Institute

Given the fact that households were ordered to stay at home except to make essential trips in most states, one might ask why households were still spending roughly $250 a week on non-essential categories in April. First, there is variation in the degree of closures across geographies and in what is deemed non-essential in each place. Second, our spending categories do not map perfectly to each specific non-essential category. Third, households may be able to switch some non-essential services from in-person to remote — for example from movie theatre entertainment to online streaming or from in-restaurant dining to take-out.

Figure 4 shows the percentage change in spending within each category, but how much did each category then contribute to the aggregate drop in spend? This requires understanding what share of aggregate spending went towards essential and non-essential categories at baseline5.  These shares are shown, before and during the pandemic, in Table 1. Multiplying the baseline shares by their relative percentage drops, we find that non-essential spending accounts for 84 percent of the aggregate decline, and essential spending accounts for 16 percent.

Table 1: The drop in non-essential spending accounted for 84 percent of the aggregate drop in spend.

 

Essential

Non-Essential

 

Share of spending

Year-over-year percent change

Share of spending

Year-over-year percent change

April 2019

33%

 

67%

 

April 2020

45%

-20%

55%

-51%

Contribution to Aggregate Drop in Spend*

16%

 

84%

 

* Percent contribution to aggregate drop in spend is calculate as: (% Drop in Category A)*(Baseline Share of Category A)/(% Drop in Aggregate).

To further illustrate the divergence in spending patterns across essential and non-essential categories, we show the year-over-year change in spending at grocery stores, drug stores, and restaurants. Figure 6 shows that spending spiked dramatically on groceries and remained elevated relative to baseline. Spending on drugstores also increased initially, before declining slightly by the end of March and early April. In contrast, spending on restaurants fell by about 70 percent. 

Figure 6: Year-over-year percent change in spending at grocery stores and drug stores surged initially, while spending at restaurants dropped by 70 percent.

Year-over-year percent change in credit card spending by spend category

  • Groceries
  • Drug Stores
  • Restaurants
Source: JPMorgan Chase Institute
Line graph showing year-over-year percent change in credit card spending on groceries, drug stores, and restaurants. Spending spiked on groceries and also spiked for drug stores, though less dramatically than for groceries. Both declined slightly by the end of March and early April. Spending fell by about 70%.

Finding Three: Spending dropped substantially for households across the entire income distribution, with slightly larger drops for higher-income households.

We next explore whether spending reductions (both in aggregate and by category) varied across the pre-pandemic income distribution. We stratify our sample into income quartiles based on total labor inflows in 2019. For context, those in the bottom quartile make less than $39,000 in take-home labor income per year, while those in the top quartile earn more than $92,0006.

Figure 7 plots the year-over-year change in spending for each quartile, both in percentage and dollar terms. The top income quartile reduces spend by about 46 percent, or $400, by the second week of April, while the bottom quartile reduces spend by 38 percent, or $150. The difference in the spending drop between income quartiles is starker in dollar terms than percentages, since high-income households have a higher baseline level of spending7.

The sharp decline in spending across the entire income distribution may be surprising. Recent research suggests that lower-income households work in jobs that are harder to perform at home, require higher physical proximity, and therefore may be more impacted by distancing restrictions (Mongey, Pilossoph, and Weinberg 2020). Perhaps as a result, recent evidence from administrative ADP data shows that job losses were four times higher for workers in the bottom income quintile than in the top income quintile, with a staggering 35 percent employment decline for the lowest-income workers (Cajner et al 2020). In response to greater income losses, we might have expected lower-income workers to have cut their spending by more. If anything, we find the reverse—higher-income households cut their spending by slightly more.

Figure 7: Year-over-year reductions in aggregate spending are slightly larger for households in the upper portion of the income distribution.

.

Year-over-year percent change in credit card spending by income quartile

  • Quartile 1 (lowest)
  • Quartile 2
  • Quartile 3
  • Quartile 4 (highest)
Note: Income quartiles are defined as follows: Quartile 1: less than $39,200; Quartile 2: $39,200 - $58,900; Quartile 3: $58,900- $91,800; Quartile 4: greater than $91,800.
Source: JPMorgan Chase Institute

Average total credit card spending per household by income quartile ($)

  • Quartile 1 (lowest)
  • Quartile 2
  • Quartile 3
  • Quartile 4 (highest)
Line graphs showing year-over-year percent change in credit card spending by income quartile and average weekly credit card spending by income quartile. The top income quartile reduced spend by about 46%, or $400 by the second week of April. The bottom quartile reduced spend by 38% or $150. The difference in the spend drop is starker in dollar terms, since high-income households have a higher baseline level of spending. Note: Income quartiles are defined as follows: Quartile 1: less than $39,200; Quartile 2: $39,200 - $58,900; Quartile 3: $58,900- $91,800; Quartile 4: greater than $91,800.
Source: JPMorgan Chase Institute

One potential reason that high-income households cut total spending slightly more than low-income households could be that non-essential categories represent a larger share of spending for high-income households — 70 percent of spending in April 2019 for households in the top income quartile compared to 61 percent for those in the bottom income quartile. Additionally, higher-income families exhibited a slightly larger drop in non-essential spending, while we see little divergence across the income distribution in essential spending (Figure 8). Thus, the reduction in non-essential spending accounted for a slightly larger share of the total spending decline for high- versus low-income households (88 percent compared to 81 percent, Figure 9).

Figure 8: Year-over-year changes in essential spending were consistent across the income spectrum, while higher-income households cut non-essential spending slightly more than lower- income households. 

Year-over-year percent change in essential spending by income quartile

  • Quartile 1 (lowest)
  • Quartile 2
  • Quartile 3
  • Quartile 4 (highest)
Note: Income quartiles are defined as follows: Quartile 1: less than $39,200; Quartile 2: $39,200 - $58,900; Quartile 3: $58,900- $91,800; Quartile 4: greater than $91,800.

We use state social distancing orders that restricted non-essential goods and services to categorize spend. “Essential” categories include fuel, transit, cash, drug stores, discount stores, auto repair, groceries, telecom, utilities, insurance, and healthcare. “Non-essential” includes department stores, other retail, restaurants, entertainment, retail durables, home improvement, professional and personal services, and miscellaneous. Although flights, hotels, and rental cars are sometimes categorized as “essential” and not technically closed, we include them in the “non-essential” group because they are affected by stay-at-home restrictions on non-essential travel.
Source: JPMorgan Chase Institute

Year-over-year percent change in non-essential spending by income quartile

  • Quartile 1 (lowest)
  • Quartile 2
  • Quartile 3
  • Quartile 4 (highest)
Line graphs showing year-over-year percent change in essential spending by income quartile and year-over-year percent change in non-essential spending by income quartile. Year-over-year changes in essential spending were consistent across the income quartile. Note: Income quartiles are defined as follows: Quartile 1: less than $39,200; Quartile 2: $39,200 - $58,900; Quartile 3: $58,900- $91,800; Quartile 4: greater than $91,800.

We use state social distancing orders that restricted non-essential goods and services to categorize spend. “Essential” categories include fuel, transit, cash, drug stores, discount stores, auto repair, groceries, telecom, utilities, insurance, and healthcare. “Non-essential” includes department stores, other retail, restaurants, entertainment, retail durables, home improvement, professional and personal services, and miscellaneous. Although flights, hotels, and rental cars are sometimes categorized as “essential” and not technically closed, we include them in the “non-essential” group because they are affected by stay-at-home restrictions on non-essential travel.
Source: JPMorgan Chase Institute

Figure 9: The drop in non-essential spending accounted for a slightly larger share of the drop in total spending among higher-income households

Share of drop in total spending by income quartile

  • Non-essential
  • Essential
Bar graph showing the drop in essential and non-essential spending as shares of the drop in total spending by income quartile. Among high-income households, the drop in non-essential spending contributed a much greater share to the drop in total spending (88%) compared to low-income households (81%). Note: Income quartiles are defined as follows: Quartile 1: less than $39,200; Quartile 2: $39,200 - $58,900; Quartile 3: $58,900- $91,800; Quartile 4: greater than $91,800.

We use state social distancing orders that restricted non-essential goods and services to categorize spend. “Essential” categories include fuel, transit, cash, drug stores, discount stores, auto repair, groceries, telecom, utilities, insurance, and healthcare. “Non-essential” includes department stores, other retail, restaurants, entertainment, retail durables, home improvement, professional and personal services, and miscellaneous. Although flights, hotels, and rental cars are sometimes categorized as “essential” and not technically closed, we include them in the “non-essential” group because they are affected by stay-at-home restrictions on non-essential travel.
Source: JPMorgan Chase Institute

Finding Four: Spending dropped dramatically for workers in all industries of employment.

Findings 1 and 2 show that drops in spend are especially pronounced in non-essential categories, and mirror the timing of emergency declarations, the implementation of social distancing policies, and the prevalence of the disease.  Finding 3 shows that the spending drops were dramatic across the income distribution, even though income losses may have been more concentrated among lower wage earners unable to perform their duties from home. This suggests that the pandemic largely contributed to the reduction in spending.

Here we further examine whether income losses could also be playing a central role: as individuals working in affected sectors lose their jobs or see hours reduced, they may additionally cut down on spending. Indeed, Figure 3 shows that UI claims started spiking in the same week that commerce would have been restricted by state stay-at-home orders.

We speak to this hypothesis by splitting the sample by industry of employment and comparing spending across industries that may have been differentially impacted by earnings losses. We examine a subset of our credit card sample who also have a Chase checking account and infer their industry of employment based on the payer associated with their payroll income received in February 2020. However, we observe the payer associated with their payroll income for only 24 percent of households, and most of these payers tend to be large employers.

Figure 10 plots spending changes by industry of employment for each industry where we have significant sample size. We aggregate to industries at the two-digit NAICS code. The one exception is retail, which we break out into grocery stores, drug stores, and discount stores—generally considered essential businesses and kept open under social distancing policies—and clothing and department stores, which were generally deemed non-essential businesses and where layoffs have been greater (Cajner et al 2020).

We find that spending declined dramatically across all industries of employment. Workers in professional services, manufacturing, healthcare, education, and finance all cut spending similarly. For the most part, differences in spending declines between industries are not statistically significant. This may be surprising given initial evidence of large differences across industries in hours reductions (Bartik et al. 2020) and employment (Cajner et al 2020). Even government workers, who have experienced some of the lowest employment losses since the beginning of the pandemic, cut spending by about 35 percent. This is only a few percentage points lower than the 40 percent spending cut for all other workers.

Perhaps the most direct test of the income channel is to compare retail workers employed by different types of retail stores. Workers employed in grocery, drugstore, and discount stores cut spending by 35 percent, only a few percentage points less than the 41 percent cut in spending observed among workers employed by clothing and department stores, who might have experienced larger drops in earnings. Similarly, when we disaggregate the spending behavior into essential and non-essential spending, we see comparable spending drops across households with individuals who work in government and retail sectors (Figure 11). 

 

Figure 10: Spending dropped dramatically for workers in all industries of employment.

Year-over-year percent change in credit card spending by industry of employment

  • Government
  • Grocery, Drugstore, Discount Stores
  • Health Care
  • Finance and Insurance
  • Manufacturing
  • Education
  • Clothing and Department Stores
  • Professional
Line graph showing year-over-year percent change in credit card spending by industry of employment. Workers in professional services, manufacturing, healthcare, education, and finance all cut spending similarly—by about 40%. Government workers also cut spending to a similar degree, about 35% despite experiencing the lowest employment losses since the beginning of the pandemic. Note: We use state social distancing orders that restricted non-essential goods and services to categorize spend. “Essential” categories include fuel, transit, cash, drug stores, discount stores, auto repair, groceries, telecom, utilities, insurance, and healthcare. “Non-essential” includes department stores, other retail, restaurants, entertainment, retail durables, home improvement, professional and personal services, and miscellaneous. Although flights, hotels, and rental cars are sometimes categorized as “essential” and not technically closed, we include them in the “non-essential” group because they are affected by stay-at-home restrictions on non-essential travel.
Source: JPMorgan Chase Institute

Figure 11. Similar drops in spending on essentials and non-essentials occurred in industries with high and low rates of job loss.

Year-over-year percent change in essential spending by industry of employment

  • Grocery, Drugstore, Discount Stores
  • Government
  • Clothing and Department Stores
Source: JPMorgan Chase Institute

Year-over-year percent change in non-essential spending by industry of employment

  • Grocery, Drugstore, Discount Stores
  • Government
  • Clothing and Department Stores
Line graphs showing the year-over-year percent change in credit card spending for essentials and non-essential by three industries of employment: Grocery, Drugstore, and Discount Stores, Government, and Clothing and Department Stores. Workers across all three industries had comparable drops in spending across essential and non-essentials. Note: We use state social distancing orders that restricted non-essential goods and services to categorize spend. “Essential” categories include fuel, transit, cash, drug stores, discount stores, auto repair, groceries, telecom, utilities, insurance, and healthcare. “Non-essential” includes department stores, other retail, restaurants, entertainment, retail durables, home improvement, professional and personal services, and miscellaneous. Although flights, hotels, and rental cars are sometimes categorized as “essential” and not technically closed, we include them in the “non-essential” group because they are affected by stay-at-home restrictions on non-essential travel.
Source: JPMorgan Chase Institute

Assuming income losses do vary systematically across sectors, one potential interpretation is that the income channel accounts for only a small share of the initial spending decline through mid-April. This may not be surprising given the magnitude of the spending decline. As mentioned previously, we document that average household spending fell almost 40 percent, while the typical unemployed worker receiving UI only cuts credit card spending around 5 percent in normal times (Ganong and Noel 2019). 

However, there are at least four reasons for caution in concluding that the income channel is playing a small role even in this early phase of the pandemic.  First, industry of employment may be a poor proxy for job loss in our sample. To the extent that we can ascertain industry of employment primarily for employees of large firms, we may not be capturing the income losses for employees of small businesses. Second, job losses may not yet have translated into income losses within our time frame. The peak of the UI claims occurred the penultimate week of our sample. Households who have lost jobs may still be receiving their last paychecks. We may expect to see larger income-related spending declines over time due to both past and future job losses8.

Third, current conditions of the pandemic make comparing the magnitude of the spending response in April 2020 to that of UI recipients during normal times highly uncertain. In normal times only one in four unemployed households receives UI, and spending declines may be greater for those who do not receive such benefits. Presently, due to the CARES act, UI benefits are much more generous in level and duration, available to many more workers, and may coincide with stimulus checks. Thus, current income supports might buffer against income-related spending declines to a greater extent. On the other hand, the economic situation is highly uncertain, and the labor market is rapidly weakening, which could cause the unemployed to cut spending to a greater extent.

Finally, we analyze spending solely on the universe of Chase credit cards, which may not fully capture the spending response to income loss due to both sample selection and measurement error. Since Chase credit card holders tend to be more affluent than the average U.S. household, we may be missing those households who might cut spending the most due to income declines. In addition, impacted individuals may turn to credit cards to finance their spending and to avoid the risk of infection posed by other means of transacting9. In the future we can test the limitations of the credit card sample by studying checking account transactions and spending on debit cards.

Conclusion

In summary, we provide two pieces of evidence suggesting that, as of the second week of April, the 40 percent drop in consumer spending appears to be driven to a greater extent by the pandemic and social distancing policies implemented across the country to prevent its spread and to a lesser extent by the initial round of income losses. First, the 40 percent drop in spending was observed across the income distribution and regardless of industry of employment. Second, the drop in spending was most dramatic at merchants which provide non-essential goods and services

However, we analyze only the initial, short-run reaction of spending to the pandemic. The balance of factors contributing to spending behavior could change dramatically as the pandemic unfolds. If the virus and economic disruptions remain widespread even after social distancing restrictions are lifted, or if income supports, such as UI and stimulus payments, provide only temporary relief, consumer spending may not return to baseline levels. In future work we will continue to track the path of consumer spending and evaluate the extent and impact of income disruptions by extending and complementing our current view of credit card spending with checking account transactions. 


Acknowledgements

First and foremost, we thank Tanya Sonthalia and Therese Bonomo for their outstanding analytical contributions to the report. We are additionally grateful to Samantha Anderson, Maxwell Liebeskind, Robert McDowall, Shantanu Banerjee, Melissa Obrien, Erica Deadman, Sruthi Rao, Anna Garnitz, Chris Knouss, and Preeti Vaidya for their support and contributions along the way.

We are also thankful for the invaluable constructive feedback we received from external experts, including Jesse Edgerton, Michael Feroli, Daniel Silver, Joseph Lupton, Joseph Vavra and Arlene Wong. We are deeply grateful for their generosity of time and insight.

This effort would not have been possible without the diligent and ongoing support of our partners from the JPMorgan Chase Consumer and Community Bank and Corporate Technology teams of data experts, including, but not limited to Brian Maddox, Kyung Cho-Miller, Michael Aguilar, Albert Raymond, Breann Zickafoose, Scott Dodds, Jay Mathuria, Roma Patel, Andrew Goldberg, Derek Jean-Baptiste, Anthony Ruiz, Suresh Devarar, Ravi Tummalapenta, Jeff Hamroff, Senthilkumar Gurusamy, and Melissa Goldman. The project, which encompasses far more than the report itself, also received indispensable support from our Internal partners in the JPMorgan Chase Institute team, including Elizabeth Ellis, Alyssa Flaschner, Carolyn Gorman, Sarah Kuehl, Carla Ricks, Gena Stern, Parita Shah, Haley Dorgan, and Tremayne Smith.

Finally, we would like to acknowledge Jamie Dimon, CEO of JPMorgan Chase & Co., for his vision and leadership in establishing the Institute and enabling the ongoing research agenda. Along with support from across the firm—notably from Peter Scher, Max Neukirchen, Joyce Chang, Marianne Lake, Jennifer Piepszak, Lori Beer, Derek Waldron, and Judy Miller—the Institute has had the resources and support to pioneer a new approach to contribute to global economic analysis and insight.