The Impact of the COVID ‐ 19 Pandemic on the U.S. Economy: Evidence from the Stock Market

: The coronavirus crisis has damaged the U.S. economy. This paper uses the stock returns of 125 sectors to investigate its impact. It decomposes returns into components driven by sector ‐ specific factors and by macroeconomic factors. Idiosyncratic factors harmed industries such as airlines, aerospace, real estate, tourism, oil, brewers, retail apparel, and funerals. There are thus large swaths of the economy whose recovery depends not on the macroeconomic environment but on controlling the pandemic. Macroeconomic factors generated losses in industries such as production equipment, machinery, and electronic and electrical equipment. Thus, reviving capital goods spending requires not just an end to the pandemic but also a macroeconomic recovery.


Introduction
The COVID-19 pandemic has walloped the U.S. economy. Between January and July 2020, the unemployment rate rose from 3.6 percent to 10.1 percent, industrial production fell by 9 percent, and nonfarm employment fell by more than 12.5 million people.
The Federal Reserve and federal and state governments have fought the downturn and the pandemic. Beginning in March the Fed lowered the federal funds rate target by 150 basis points, provided forward guidance that interest rates would remain low, engaged in quantitative easing by buying Treasury and mortgage-backed securities, loaned to Treasury security primary dealers, backstopped money market funds, encouraged bank lending, and took other steps to maintain the flow of credit. 2 Congress passed several pieces of legislation in March, including the Coronavirus Aid, Relief, and Economic Security Act (CARES). CARES provides loans for small businesses to continue paying wages (Paycheck Protection Program), expands unemployment benefits, pays $1,200 per adult and $500 per child for individuals earning up to $75,000 (or $150,000 for taxpayers filing jointly), and channels funds to the health care system and to state and local governments. Several states and localities issued shelter-in-place (S-I-P) orders mandating that non-essential businesses close and that non-essential employees work from home.
What are the channels driving the economy's response to the pandemic and to the policy interventions? Chetty et al. (2020) employed daily data to examine how spending, revenues, employment, and other variables responded at the county and industry level. Since the fall in GDP between 2019Q4 and 2020Q1 was driven by a drop in personal consumption expenditures, they investigated consumer spending. They reported that more than half of the drop in spending in June 2020 relative to June 2019 came from the top income quartile and only five percent of the drop came from the bottom quartile. They found that three-fourths of the drop in spending between the pre-coronavirus period and the middle of April came from goods and services requiring close contact such as hotels, transportation, and restaurant meals. They also found that high-income households reduced spending at businesses producing non-tradables, causing these businesses to lay off low-income employees. CARES payments then stimulated spending by low-income individuals but did little to increase employment among the many laid off from jobs requiring close contact. The Paycheck Protection Program also did little to increase employment among these service workers. Chetty et al. concluded that stimulating aggregate demand and providing liquidity to businesses may not increase employment much when spending is constrained by health concerns. Goolsbee and Syverson (2020) investigated whether the U.S. coronavirus-related downturn arose from S-I-P policies or from people choosing to refrain from activities to avoid infections. They employed cell phone data on customer visits to 2.25 million businesses in 110 industries. Comparing visits to contiguous locations with different S-I-P policies, they reported that legal shutdown orders explained only 7 percentage points (ppt) of the 60 ppt drop in customer visits during the pandemic. They concluded that consumer actions to avoid infection rather than the lockdown policies were responsible for the lion's share of the drop in spending across exposed businesses. Eichenbaum et al. (2020aEichenbaum et al. ( , 2020bEichenbaum et al. ( , 2020c calibrated how pandemics affect aggregate demand and aggregate supply. They modeled the interactions between agents' economic decisions and the spread of the infection. They found that people avoiding infection risk caused labor supply and thus aggregate supply to fall and consumption and thus aggregate demand to drop. The simultaneous reduction in aggregate supply and aggregate demand in their model leads to a deep, persistent recession.
To investigate how the pandemic and the policy actions have affected the U.S. economy, this paper examines the response of sectoral stock prices. Finance theory indicates that stock prices equal the expected present value of future cash flows. They thus contain information about how investors expect firms and industries to be affected.
Previous research has investigated how the coronavirus crisis has affected asset prices. Ramelli and Wagner (2020) examined how COVID-19 affected U.S. stock returns. They used the capital asset pricing model and the Fama-French (1993, 2015 factors to adjust returns. They found that as news of the outbreak emerged between 20 January and 21 February 2020, adjusted returns on stocks of U.S. firms that traded with China fell more. Then as concern about the crisis exploded over the 24 February to 20 March 2020 period, they reported that adjusted returns on firms with less cash and higher leverage fell more. Pagano, Wagner, and Zechner (2020) found that stocks of firms that are resilient to social distancing perform better than non-resilient stocks. They employed Koren and Pető's (2020) affected share variable that measures the extent to which jobs can be done without close human contact. They classified firms as resilient if their affected share values are below the median and as not resilience if their affected share values are above the median. Over the 24 February -20 March 2020 period, they reported that capital asset pricing model-adjusted returns were 10 percent for the high resilience portfolio and -15 percent for the low resilience portfolio. The high resilience portfolio thus outperformed the low resilience portfolio by 25%. For returns adjusted using the French (1993, 2015) factors, they reported that the high resilience portfolio outperformed the low resilience portfolio by between 15% and 20%. Chan and Marsh (2020) compared the drop in the Dow Jones Industrial Average (DJIA) from 21 February 2020 to the end of March 2020 with drops during other market crashes and pandemics. They reported that the DJIA fell about as much over their sample as it did at a comparable point during the 1929 Great Depressions and much more than during previous pandemics such as the 1918 Spanish Flu epidemic. They attributed the large drop in 2020 to uncertainty about the technical characteristics of COVID-19, the effects of S-I-P policies on the economy, and the impact of the pandemic on global value chains. Gormsen and Koijen (2020) employed dividend futures to measure the economic impact of the crisis. As of 9 June 2020, their model predicted a 2 ppt fall in U.S. GDP growth forecasted over the next year relative to forecasts on 1 January 2020, a 9 ppt fall in U.S. dividends, a 3.1 ppt over the next year relative to forecasts on 1 January 2020, a 3.1 ppt drop in European GDP growth, and a 14 ppt drop in European dividends. They stated that their approach might understate the falls because the macroeconomic changes in 2020 are large relative to historical experience. This paper investigates how stock returns in 125 sectors have been affected during the pandemic. Black (1987, p. 113) observed that, "The sector-by-sector behavior of stocks is useful in predicting sector-by-sector changes in output, profits, or investment. When stocks in a given sector go up, more often than not that sector will show a rise in sales, earnings, and outlays for plant and equipment." This paper also decomposes cumulative returns into those portions driven by macroeconomic factors and by idiosyncratic factors. The results present evidence at a disaggregated level concerning how investors expect different parts of the U.S. economy to fare.
The next section presents the data and methodology. Section 3 contains the results.

Data and Methodology
This paper investigates stock returns for 125 sectors during the COVID-19 crisis. It focuses on the period beginning on 19 February 2020, when stock prices began falling, until 10 July 2020. The sectors investigated are listed in column (1) of Table 1 and the values on July 10 th of 1 dollar invested on February 19 th are presented in column (2). This paper investigates how both sector-specific factors and the macroeconomic environment affect returns. Over the sample period, the coronavirus pandemic was a key event driving these responses.
To capture how aggregate economic environment affects returns, several macroeconomic variables are used. The first is the return on the overall U.S. stock market, following a long tradition in finance of using the return on the market to control for economy-wide influences (see, e.g., Warner, 1980, 1985). The second is the return on the world stock market, capturing how changes in the world economy influence sectoral returns. The third is the nominal effective exchange rate, following a body of research that investigates sectors' exchange rate exposures (see, e.g., Dominguez and Tesar, 2006). The fourth is the price of oil, reflecting the large impact of oil prices across U.S. sectors (see, e.g., Thorbecke, 2019). The fifth is inflation, drawing on evidence that inflation is a state variable that matters for stock returns (see, e.g., Chen, Roll, and Ross, 1985). The other three variables are interest rates or interest rate spreads, building on the large literature indicating that these affect equity risk premia (see, e.g., Aït-Sahalia, Karaman, and Mancini, 2020). These last three variables are the change in the interest rate on three-month Treasury securities, the change in the spread between interest rates on tenyear and three-month Treasury security, and the change in the spread between interest rates on Moody's seasoned Baa corporate bonds and ten-year Treasury securities. The estimated equations take the form: ∆ , = 0 + 1 ∆ , , + 2 ∆ , , + 3 ∆ + 4 ∆ , + 5 ∆ + 6 ∆ ℎ , + 7 ∆ − ℎ , where the data are daily and ∆Ri,t is the change in the log of the stock price index for sector i, ∆Rm,US,t is the change in the log of the price index for the U.S. aggregate stock market, ∆Rm,World,t is the change in the log of the price index for the world stock market, ∆ert is the change in the Federal Reserve broad trade-weighted exchange rate index, ∆Poil,t is the change in the log of the spot price for West Texas Intermediate crude oil, ∆Inftt is the change in the five- year breakeven inflation rate calculated from TIPS, ∆ithree,t is the change in the interest rate on three-month Treasury securities, ∆iten-three,t is the change in the spread between interest rates on ten-year and three-month Treasury security, and ∆ibaa-ten,t is the change in the spread between interest rates on Moody's seasoned Baa corporate bonds and ten-year Treasury securities.
Equation (1) can be employed to divide returns into the part driven by macroeconomic variables and the part driven by sector-specific factors. The COVID-19 pandemic and the stimulus in response to it have affected the macroeconomic environment and also impacted sectors in various ways. The decomposition sheds light on when economy-wide and when sector-specific influences are driving performance.
As a robustness test the right-hand-side variables in equation (1) are replaced by the five Fama-French factors. French (1993, 2015) reported that five common factors are useful for explaining the cross section of stock returns. These factors are: 1) the return on the aggregate U.S. stock market index minus the return on one-month U.S. Treasury securities, 2) the average return on nine small capitalization stock portfolios minus the average return on nine large capitalization stock portfolios, 3) the average return on two high book value to market value portfolios minus the average return on two low book value to market value portfolios, 4) the average return on two robust operating profitability portfolios minus the average return on two weak operating profitability portfolios, and 5) the average return on two conservative investment portfolios minus the average return on two aggressive investment portfolios.
Regressing sectoral returns on these five factors provides an alternative way of decomposing returns into systematic and idiosyncratic portions. 4 Figure 1 plots aggregate U.S. stock prices from 1 January 2020 to 10 July 2020. After increasing between January 1 st and February 19 th , prices fell logarithmically by 42 percent between February 19 th and March 23 rd . They then increased by 37 percent between March 23 rd and July 10 th .
Smith and Badkar (2020) The focus in the next section is on sectoral performance over the 19 February 2020 -10 July 2020 sample period. The sample period is also divided into the 19 February -23 March sub-sample when news about the virus itself impacted markets and the 23 March -10 July subsample when news of government stimulus helped to revive markets. Table 1 presents the findings across sectors explained by the eight macroeconomic variables. The average adjusted R-squared in column (11) is 0.578. This is good for regressions explaining daily stock returns. The numbers in columns (2) through (10) indicate how much one dollar invested at the beginning of the period would be worth at the end of the period. Numbers below one indicate that one dollar invested at the beginning of the period would have lost value and numbers above one indicate that it would have gained value. Columns (2) -(4) concern investments over the 19 February -10 July period, columns (5) -(7) investments over the 19

Results
February -23 March period, and columns (8) -(10) investments over the 23 March -10 July period. For all three periods the leftmost column presents results for the overall stock return in a sector, the middle column for the portion of the return driven by macroeconomic factors, and the rightmost column for the portion driven by idiosyncratic factors.
According to column (2), the worst performing sector in terms of overall stock returns over the 19 February -10 July period is airlines. One dollar invested on 19 February would be worth only 38 cents on 10 July. The closely related aerospace sector that provides planes to the airline sector has also suffered, with a dollar investment falling to 50 cents by the end of the period. Columns (3) and (4) indicate that the lion's share of this fall has been driven by idiosyncratic rather than macroeconomic factors. The pandemic grounded flights and decimated these industries.  (3) and (4) indicate that two-thirds of the drop was due to idiosyncratic factors and one-third to the macro environment. Investments in two related sectors, retail REITs and mortgage REITs, also lost more than half of their value over this period. The crisis restricted visits to hotels and retail stores and jeopardized agents' ability to pay rents and mortgages.
The oil sector has also performed poorly. One dollar invested in oil equipment and services on 19 February fell to 42 cents, one dollar in crude oil production fell to 45 cents, and one dollar in oil refining and marketing and in pipelines fell to 55 cents. Columns (3) and (4) indicate that both macroeconomic factors (e.g., low oil prices) and idiosyncratic factors (e.g., S-I-P policies) roiled the industry. Brower (2020) reported that oil production in the U.S. fell by 30 percent during the crisis.
Tourism has been decimated. One dollar invested in casinos & gambling fell to 55 cents, one dollar in travel & tourism to 57 cents, and one dollar in hotels & motels to 58 cents.
Macroeconomic factors contributed to these losses but idiosyncratic factors contributed more.
Concern about infections and lockdowns have reduced cruise ship voyages, trips to crowded locations, and visits to hotels.
Recreational services such as fitness centers and also banks and consumer lending have suffered. One dollar invested in recreational services fell to 51 cents and one dollar invested in banks and consumer lending to 58 cents. Customers avoided fitness centers. Banks and consumer lending faced the danger that borrowers may be unable to repay loans. Noonan and Armstrong (2020) reported that U.S. banks set aside much more than expected for loan loss provisions in 2020Q2 and that Fed interest rate cuts also reduced interest rate margins and bank profitability. In addition credit card balances, a major profit source for banks, tumbled in 2020Q2 (Smith, 2020).
Other sectors that performed badly are brewers, apparel retailers, radio & TV broadcasting, and funerals. The returns on July 10 th to one dollar invested in these sectors on cents. In all of these cases macroeconomic factors were the primary driver of the losses, supported by idiosyncratic factors.
In columns (5) through (7), there are only five sectors where a one dollar investment on February 19 th was worth at least 85 cents on March 23 rd . These are nondurable household producers (e.g., Clorox), luxury items (e.g., gold watch makers), diversified retailers (e.g., Amazon), gold mining, and electronic entertainment (e.g., video game producers). In every case the macroeconomic environment led to losses and idiosyncratic factors offset some of these.
Columns (8) -(10) indicate that almost all sectors gained between March 23 rd and July 10 th . One of the few exceptions was brewers. As traffic to bars and restaurants fell, beer sales also tumbled. Funerals showed no gains and real estate sectors, banks, and airlines posted smaller gains then other sectors. In all of these cases the macroeconomic environment led to gains and idiosyncratic factors reduced these.
Columns (8) -(10) also indicate that, of the 13 best performing sectors over the March 23 rd -July 10 th period, more than half relate to the home and to home improvement. One dollar invested on March 23 rd would be worth $2.16 in the household furnishing sector, $1.86 in the renewable energy (including solar panel) sector, $1.76 in the recreational products (including home swimming pool) sector, $1.70 in the home improvement store sector, $1.69 in the electronic entertainment sector, $1.66 in the household appliance sector, and $1.63 in the home construction sector. These gains were driven first of all by the macroeconomic environment but also by idiosyncratic factors. As people sheltered at home, they invested in making their homes more comfortable and energy-efficient.
Other large gainers over the March 23 -July 10 period include transport services, recreational vehicles, and specialty retail. Transport services including logistics staged a partial comeback after being one of the deepest losers during the earlier sub-sample period.
Recreational vehicle demand rose since tourist activities involving close contact posed health risks. Specialty retail businesses such as Amazon continued to thrive as consumers sheltered at home.
As a robustness test, returns are regressed on the Fama-French (1993, 2015 Table 1 are robust to a very different choice of common factors. 6 The important implication of the findings in this section is that there are large swaths of the U.S. economy whose recovery depends not on the macroeconomic environment but instead on bringing the pandemic under control. These sectors include airlines, aerospace, real estate investment trusts, recreational services, brewers, apparel retailers, and funerals. On the other hand, many sectors that are important for capital investment such as production equipment, machinery, and electronic and electric equipment are dependent on the macroeconomy. A robust recovery is thus necessary to revive business investment.

Conclusion
The coronavirus pandemic is an exogenous shock. This paper uses sectoral stock price responses to trace out its effects on the U.S. economy. Stock prices are useful because they provide a measure of how investors expect shocks to impact future cash flows across sectors.
The paper also decomposes stock return changes into portions driven by sector-specific factors and portions driven by macroeconomic factors. Regressing returns on eight macroeconomic variables and on French's (1993, 2015)  and 10 July 2020. During the recovery period, seven of the 13 best performing sectors were related to the home and home improvement. As people sheltered at home, they spent on their homes. While this will make their homes more comfortable, the evidence reported here that the real estate sector has done so badly indicates that this spending might not yield a high return on investment (ROI).
Stimulus that raises spending but does not lead to a high ROI will produce short run gains. Chetty et al. (2020) reported that stimulating aggregate demand and providing liquidity to businesses may not increase employment much when spending is constrained by health concerns. Policymakers need to develop a new approach to promote sustainable recovery from the COVID-19-induced downturn. One step in this direction, as Barrero, Bloom, and Davis (2020) argued, is to speed the reallocation of labor and capital from sectors that are unlikely to recover to newly productive sectors.  (2) presents the stock market return as of 10 July 2020 from investing 1 dollar in the sector listed in column (1) on 19 February 2020. Column (3) presents the portion of returns in column (2) that can be attributed to the effect of eight macroeconomic factors on returns. These factors are 1) the return on the aggregate U.S. stock market index, 2) the return on the world stock market index, 3) the change in the Federal Reserve broad tradeweighted exchange rate index, 4) the change in the log of the spot price from West Texas Intermediate crude oil, 5) the change in the breakeven inflation rate calculated from U.S. Treasury inflation-protected securities, 6) the change in the interest rate on three-month Treasury securities, 7) the change in the spread between interest rates on ten-year and three-month Treasury security, and 8) the change in the spread between interest rates on Moody's seasoned Baa corporate bonds and ten-year Treasury securities. Column (4) presents the portion of returns in column (2) not explained by these eight factors. Column (4) thus includes the effects of other factors such as the coronavirus pandemic on returns. Column (5) - (7) and (8)

Figure 2. Adjusted R-squared Coefficients from Regressions with Eight Macroeconomic Factors versus Adjusted R-squared from Regressions with Five Fama-French Factors
Note: The horizontal axis plots the adjusted R-squared coefficients from regressing daily returns on 125 sectors on the return on the aggregate U.S. stock market index, the return on the world stock market index, the change in the Federal Reserve broad trade-weighted dollar exchange rate index, the change in the log of the spot price for West Texas Intermediate crude oil, the change in the breakeven inflation rate calculated from U.S. Treasury inflationprotected securities, the change in the interest rate on three-month Treasury securities, the change in the spread between interest rates on ten-year and three-month Treasury security, and the change in the spread between interest rates on Moody's seasoned Baa corporate bonds and ten-year Treasury securities. The vertical axis plots the corresponding adjusted R-squared coefficients from regressing daily returns on 125 sectors on the five Fama-French (2015) factors. These factors are 1) the return on the aggregate U.S. stock market index minus the return on onemonth Treasury securities, 2) the average return on nine small capitalization stock portfolios minus the average return on the nine large capitalization stock portfolios, 3) the average return on two high book value to market value portfolios minus the average return on the two low book value to market value portfolios, 4) the average return on two robust operating profitability portfolios minus the average return on two weak operating profitability portfolios, and 5) the average return on two conservative investment portfolios minus the average return on two aggressive investment portfolios. February 2020 explained by 8 macroeconomic factors. These factors are 1) the return on the aggregate U.S. stock market index, 2) the return on the world stock market index, 3) the change in the Federal Reserve broad tradeweighted exchange rate index, 4) the change in the log of the spot price from West Texas Intermediate crude oil, 5) the change in the breakeven inflation rate calculated from U.S. Treasury inflation-protected securities, 6) the change in the interest rate on three-month Treasury securities, 7) the change in the spread between interest rates on ten-year and three-month Treasury security, and 8) the change in the spread between interest rates on Moody's seasoned Baa corporate bonds and ten-year Treasury securities. The vertical axis plots the corresponding values explained by the five Fama-French (2015) factors. These factors are 1) the return on the aggregate U.S. stock market index minus the return on one-month Treasury securities, 2) the average return on nine small capitalization stock portfolios minus the average return on the nine large capitalization stock portfolios, 3) the average return on two high book value to market value portfolios minus the average return on the two low book value to market value portfolios, 4) the average return on two robust operating profitability portfolios minus the average return on two weak operating profitability portfolios, and 5) the average return on two conservative investment portfolios minus the average return on two aggressive investment portfolios.