The Impact of Inflation on the U.S. Stock Market After the COVID-19 Pandemic

Abstract

Inflation remained quiescent for several decades and then surged in 2021 and 2022. Inflation subsequently fell in 2023 and 2024. This paper investigates how the rise and fall in inflation after 2019 affected the U.S. stock market. To do this, it estimates a fully specified multi-factor model that measures the exposure of 54 assets to inflation, monetary policy, and other macroeconomic variables over the 1994 to 2019 period. This paper then uses the inflation betas to investigate how investors’ perceptions of inflation changed between 2020 and 2024. The results indicate that concerns about inflation roiled the stock market over this period. The Fed’s anti-inflationary policies whipsawed markets even more. These findings highlight the dangers that arise when monetary policy allows inflation to accelerate.

1. Introduction

U.S. inflation in 2022 reached its highest level in more than 40 years (see Figure 1). When did investors begin responding to news of inflation? Did they expect fiscal stimulus such as the USD 1.9 trillion American Rescue Plan Act (ARPA) to raise inflation?

If investors perceived that inflation would rise, they would seek to sell assets that are harmed by inflation and purchase assets that benefit from inflation. This would lower the prices of assets that are harmed by inflation and raise the prices of assets that benefit from inflation. Following this intuition, inflation betas are estimated for 54 assets over the January 1988 to December 2019 period. This paper then examines the relationship between inflation betas and changes in returns over the August 2020 to September 2024 period.

The results indicate that there was not a systematic positive relationship between inflation betas and returns, indicating that investors anticipated higher inflation, until April 2021. At this point, the consumer price index (CPI) inflation rate had reached its highest level in 13 years and was growing rapidly. Thus, investors ignored incipient inflation until the actual rate began to accelerate. There were then four subsequent months when investors expected higher inflation before they anticipated inflation to fall again. On average, for the three most exposed assets, expectations of inflation caused returns to fall by 3.4% per month over these five months.

The Federal Reserve was also slow to respond to inflation. To investigate the Fed’s reaction, this paper investigates the relationship between monetary policy betas and changes in returns. As with inflation, the intuition is that if investors perceived that monetary policy would tighten, they would seek to sell assets that are harmed by contractionary monetary policy and purchase assets that benefit from contractionary policy. This would lower the prices of assets that are harmed by tighter policy and raise the prices of assets that benefit from tighter policy. Monetary policy betas are estimated for 54 assets over the January 1988 to December 2019 period, and the relationship between monetary policy betas and changes in returns is examined over the August 2020 to September 2024 period.

The results indicate that there was not a systematic positive relationship between these two variables, indicating that investors anticipated anti-inflationary monetary policy, until November 2021. At this point, the inflation rate was at its highest level in more than 39 years. There were then four subsequent months when investors anticipated anti-inflationary policy before they anticipated any policy easing. On average, for the three most exposed assets, expectations of contractionary monetary policy caused returns to fall by 9.2% per month over these five months. After this, changing perceptions of monetary policy caused massive swings in asset prices.

Gomez-Cram et al. (2024), De Soyres et al. (2022), and Di Giovanni et al. (2023) presented evidence that expansionary fiscal policy contributed to inflation after the pandemic started. This paper uses Gomez-Cram et al.’s (2024) daily fiscal news series to investigate whether investors expected fiscal policy to affect inflation. The findings reveal little response of inflationary expectations to expansionary fiscal policy news.

Greenspan noted that when fiscal and monetary policy turned inflationary in the 1960s and 1970s, investors initially reacted little (see, e.g., Greenspan, 1994; and Woodward, 2001). They trusted U.S. institutions to keep inflation under control. Then, as inflation accelerated, they demanded high risk premia in stocks and bonds to compensate for the risk of inflation. The findings in this paper indicate that neither investors nor policymakers should be complacent about the dangers of inflation. This is especially true as tariffs, tax cuts, and other policies advocated by the Trump administration could cause inflation to re-accelerate (Smith, 2024).

The next section reviews the literature. Section 3 discusses the data and methodology. Section 4 presents the results. Section 5 discusses the findings and conclusions.

2. Literature Review

2.1. Inflation During the COVID-19 Era

Observers have highlighted several shocks affecting inflation after the pandemic began (see, e.g., Bernanke, 2024). In the 2020 lockdowns, fears of catching COVID-19, rising unemployment, and other factors suppressed consumption (Goolsbee & Syverson, 2021). These may have lowered inflation. The availability of COVID-19 vaccines in December 2020 and massive vaccinations over the next several months then increased consumer confidence and contributed to the easing of lockdowns. President Biden signed the ARPA in March 2021, providing USD 1.9 trillion of fiscal stimulus. Consumer spending soared beginning in the first quarter of 2021. Increased demand may have interacted with supply constraints arising from supply chain bottlenecks to raise inflation. Then, in February 2022, Russia invaded Ukraine, raising prices for energy, food, and other commodities.

How have researchers explained the surge in inflation after the pandemic arrived? Harding et al. (2023) investigated this issue using a structural general equilibrium model with monopolistic competition in the labor and goods markets and sticky wages and prices. They included a quasi-kinked goods demand function that generates a nonlinear Phillips Curve. In their nonlinear formulation, both cost-push and demand shocks impact inflation more when inflation surges. They reported that their nonlinear model forecasts one-year-ahead inflation better during the post-COVID-19 period than linear models do. Their framework explains why, after COVID-19 began, shocks produced accelerating inflation.

Jordà and Nechio (2023) also presented evidence that the slope of the Phillips Curve steepened when supply–demand imbalances raised inflation after the pandemic. For the pre-pandemic period (before 2020Q1), they found that an increase in one-year-ahead Consensus Economics inflation forecasts was unrelated to nominal wage growth across a sample of 17 OECD countries. During the post-pandemic period (beginning in 2020Q3), they reported a one-to-one pass-through from expected inflation to nominal wage growth. They concluded that, with the labor market tight and inflation high during the post-pandemic period, workers focused on expected inflation when renegotiating contracts.

Cavallo (2024) reported that the inflation that emerged during the pandemic was worse than that reported in Figure 1. This is because the pandemic changed consumers’ spending patterns. For instance, Chetty et al. (2020) found, using daily data that consumers cut back spending on travel, hotels, restaurants, transportation, and other services requiring close contact and increased spending on consumer goods. The U.S. Bureau of Labor Statistics, on the other hand, used weights from before the pandemic to calculate inflation in 2020. Cavallo, using daily data on credit and debit card spending to update the weights, found that the actual inflation rate in 2020 was much higher than the official data indicated.

There is evidence that consumer habits are returning to pre-pandemic patterns. Spending on travel, tourism, and other services has revived. Genc and Arzaghi (2024) also investigated whether consumers who started shopping online during the COVID-19 outbreak returned to in-store shopping after the pandemic. Their results indicated that there was not a structural break in e-commerce spending following the pandemic. Rather, it reverted towards its long-run trend. They concluded that consumers returned to the shopping patterns they had followed before the pandemic.

2.2. Fiscal Policy and Inflation During the COVID-19 Era

Summers (2021) warned of the inflationary risks of the ARPA. He said the package could produce inflationary pressures of a kind not seen in a generation. He also stated that obtaining legislative support for spending reductions or tax increases would be difficult. As a result, he highlighted the danger of inflation expectations rising sharply. Blanchard et al. (2022) observed that unemployment in early 2022 was exceptionally low and job vacancies were exceptionally high. They argued that this combination multiplied inflationary risks.

Gomez-Cram et al. (2024) examined the relationship between fiscal debt regimes, bond yields, and inflation. Under a regime of fiscal dominance, surprise spending increases are not offset by future tax increases. Instead, in their framework, unfunded fiscal expansions lead to inflation. They collected a sample of days between 2020 and 2023 when news indicated that fiscal policy was expansionary. They reported that Treasury yield hikes were concentrated on these days with news of higher deficits. They interpreted their findings as implying that investors began pricing Treasury bonds as if the U.S. had entered a risky fiscal regime.

De Soyres et al. (2022) investigated the relationship between fiscal stimulus and excess inflation during the COVID-19 pandemic. They documented that U.S. fiscal spending in 2020 was almost 20% more than was predicted before the pandemic. They then projected quarterly real consumption and production growth based on mobility during the pandemic and country-specific fiscal support. They reported that, for the U.S., fiscal stimuli raised inflation by 2.5 percentage points.

Di Giovanni et al. (2023) examined the impact of fiscal spending in the U.S. on inflation over the December 2019–June 2022 period. They employed a multisector macroeconomic network model with 66 sectors and both Ricardian and hand-to-mouth consumers. They reported that aggregate demand shocks caused two-thirds of the U.S. inflation in their model and that half or more of the aggregate demand shocks were due to the fiscal stimulus.

Jordà and Nechio (2023) exploited differences in fiscal plans across 17 OECD countries during the pandemic to investigate how fiscal policy affected inflation. They measured fiscal stimulus during the pandemic by real disposable income per capita relative to trend (RDI). Using a difference-in-difference approach, they reported that a 5-percentage-point increase in RDI led to a 3-percentage-point increase in inflation after four quarters. Since RDI in the U.S. increased by more than 5 percentage points, their findings indicate that the COVID-19 fiscal expansion stoked inflation in the U.S.

2.3. Monetary Policy and Inflation During the COVID-19 Era

When the economy was on the cusp of reflating, the Federal Reserve, in August 2020, announced a new monetary policy framework called Flexible Average Inflation Targeting (FAIT). As Boocker and Wessel (2024) discussed, FAIT implies that the Fed will react more aggressively when inflation falls below the 2% target than when it rises above the target. The new framework also indicates that the Fed will react more aggressively to labor market shortfalls than to excess demand in the labor market. Employing a New Keynesian model and the Fed’s FRB/U.S. model to investigate alternative strategies, Kiley (2024) showed that these asymmetric responses produce an inflationary bias. Using narrative evidence, Eggertsson and Kohn (2023) reported that the new framework contributed to inflation by causing the Fed to delay tightening as inflation emerged in 2021.

Della Chang et al. (2023) used a vector autoregression to investigate the impact of aggregate demand and supply shocks on the growth of the GDP deflator and real GDP. They reasoned that a supply shock would cause GDP growth and inflation to be negatively correlated and a demand shock would cause these to be positively correlated. They thus imposed sign restrictions to identify their model. They found that over the 2021–2023 period, demand and supply shocks contributed equally to inflation. They concluded that the Federal Reserve should not be absolved from blame for adopting an overly easy policy stance that permitted inflation to rise.

Gagliardone and Gertler (2023) investigated whether expansionary monetary policy and oil price increases contributed to the increase in inflation starting in 2021. They used a simple model with sticky wages and prices and unemployment. Oil was both an intermediate good and a consumption good in their framework. They obtained the important parameters in their model from a structural vector autoregression. They reported that both rising oil prices and the Fed’s slow response to inflation increased inflation in 2021.

Aguirre and Casares (2024) employed a New Keynesian model with wage stickiness and a Phillips Curve to investigate the sources of inflation after the pandemic. They measured monetary policy using a rule where the nominal interest rate responds to unemployment and inflation and using the Wu and Xi (2016) shadow interest rate. They reported that excessively easy monetary policy was the primary cause of inflation in 2021 and 2022.

2.4. Inflation and Asset Prices

In theory, inflation impacts assets such as Treasury bonds, whose payoffs are fixed in nominal terms. Assuming the real log of the stochastic discount factor (SDF) is conditionally normal and the SDF and inflation are jointly log-normal, Cieslak and Pflueger (2023) showed that the nominal interest rate on a two-period default-free bond depends positively on the covariance between the SDF and expected inflation. When there is a negative covariance, when the SDF is high, states are associated with lower expected inflation. Investors value payoffs more when the SDF is high. Lower expected inflation in these states raises long-term bond prices. In this case, nominal bonds are a hedge against inflation, and investors are willing to accept a negative risk premium to hold them.

On the other hand, when there is a positive covariance between the SDF and expected inflation, states when the SDF is high are associated with higher inflation. Higher expected inflation in these states lowers long-term bond prices. In this case, investors will require a positive risk premium to hold bonds. Cieslak and Pflueger (2023) label the case where there is a negative covariance between inflation and the SDF as good inflation and the case when there is a positive covariance as bad inflation.

Cieslak and Pflueger (2023) showed that good inflation in New Keynesian models is driven by demand shocks. These will cause inflation to decrease when output and consumption decrease. Thus, bond prices will be higher when consumption is lower. Bad inflation is caused by supply shocks. These will cause inflation to increase when output and consumption decrease. Thus, bond prices will be lower when consumption is lower.

Unlike for bonds, the relationship between inflation and stock prices is ambiguous. In theory, stock prices equal the expected present value of future cash flows. Since nominal cash flows should increase with inflation, one might expect stock prices to increase with inflation. Nevertheless, many researchers have found a negative correlation between stock prices and inflation.

Bekaert and Wang (2010) discussed several explanations for this finding. As Fama (1981) argued, it could be that inflation occurs when economic activity is low, as in Cieslak and Pflueger’s (2023) bad inflation state. In this case, inflation is just a proxy for lower economic activity. Bekaert and Engstrom (2010) reported that risk premia are higher in recessions; so, when there is stagflation, lower stock prices may be associated with higher inflation. Lin (2009) focused on the negative relationship between inflation uncertainty and stock prices. Doepke and Schneider (2006) examined how unexpected inflation, by redistributing wealth from lenders to borrowers, can increase saving, decrease labor supply, and thus decrease output. Bekaert and Wang observed that inflation may be correlated with stock prices because inflation is correlated with economic activity.

Controlling for economic activity, returns on different stocks will have different covariances with inflation and thus may have different required returns. To investigate this issue, this paper estimates a multi-factor model including returns on several assets. The model includes measures of both economic activity and inflation. The results confirm that inflation exposures and inflation risk premia differ across assets.

3. Data and Methodology

3.1. Estimating a Multi-Factor Model

In a multi-factor asset pricing framework, the return on asset i is given by:

$$R_i=E_i+{\sum }_{j=1}^K{\beta }_{ij}{f}_j +{\epsilon }_i,$$

(1)

where R_i is the ex-post realized return, E_i is the ex-ante required return, β_ij is the beta or factor loading of asset i to macroeconomic factor j, $f_j$ represents innovations in macroeconomic factor j, and ${\epsilon }_i$ is a mean-zero error term capturing the effect of idiosyncratic news on asset i. Ross (2001) assumed that the effect of idiosyncratic risk can be ignored in large portfolios and showed that arbitrage profits will exist unless there is a linear relationship between the required return and the betas. Required returns can thus be written as:

$$E_i={\lambda }_0+{\sum }_{j=1}^K{\beta }_{ij}{\lambda }_j,$$

(2)

where ${\lambda }_0$ is the return on the risk-free asset and ${\lambda }_j$ is the risk price associated with factor j.

McElroy and Burmeister (1988) employed an iterated nonlinear seemingly unrelated regression (INLSUR) technique to simultaneously estimate the factor loadings and the risk prices.

They combined Equations (1) and (2) to yield:

$$R_i={\lambda }_0+{\sum }_{j=1}^K{\beta }_{ij}{\lambda }_j+{\sum }_{j=1}^K{\beta }_{ij}{f}_j +{\epsilon }_i$$

(3)

They then estimated a system of equations by stacking Equation (3) across all the assets. This approach increases efficiency by permitting the imposition of the cross-equation restrictions that the risk prices (the ${\lambda }_j$’s) are common across assets.

The system they estimated can be represented as:

$$\left[\begin{array}{c}{R}_1-{\lambda }_0\\ R_2-{\lambda }_0\\ .\\ .\\ .\\ R_n-{\lambda }_0\end{array}\right]=\left[\begin{array}{cccccc}X(\lambda ,f)& 0& .& .& .& 0\\ 0& X(\lambda ,f)& .& .& .& 0\\ 0& 0& .& .& .& 0\\ 0& 0& .& .& .& 0\\ 0& 0& .& .& .& 0\\ 0& 0& .& .& .& X(\lambda ,f)\end{array}\right]\left[\begin{array}{c}{\beta }_1\\ {\beta }_2\\ .\\ .\\ .\\ {\beta }_n\end{array}\right]+\left[\begin{array}{c}{\epsilon }_1\\ {\epsilon }_2\\ .\\ .\\ .\\ {\epsilon }_n\end{array}\right] .$$

(4)

$R_1-{\lambda }_0$ is a 1 × T vector whose element at time t equals $R_{1,t}-{\lambda }_{0,t}$, where $R_1$ represents the ex-post return on asset i and ${\lambda }_0$ is the return on the risk-free asset. $X(\lambda ,f)$ is a T × k matrix whose tith element equals $f_{i,t}$ + ${\lambda }_i$ . $f_{i,t}$ represents news about macroeconomic factor i, and λ_i is the risk price associated with factor i. ${\beta }_i$ is a 1 × k vector containing asset i’s betas to the systematic factors. ε_i is an i × T vector. It is assumed that E(ε₁, ε₂, …, ε_n) = 0_nT, E(ε₁, ε₂, …, ε_n)’(ε₁, ε₂, …, ε_n) = ∑⊗I_T, and ∑_i,j = cov(ε_i,t, ε_j,t). INLSUR estimation is used to estimate the system (4).

This paper follows Chen et al. (1986) by using, not latent variables from a dynamic factor model, but actual macroeconomic data to measure the factors in Equation (3). The factors they employed were a measure of the term structure of interest rates (the difference between returns on 20-year and one-month Treasury securities), the default premium (the difference between returns on 20-year corporate bonds and 20-year Treasury bonds), the monthly growth rate in industrial production, unexpected inflation, and the change in expected inflation. Because each of these series is either the difference between asset returns or very noisy, Chen et al. treated them as innovations. Since Thorbecke (2018) reported that the default premium was not a priced factor, a monetary policy factor was used in this paper in place of the default premium.

The term structure measure was incorporated to account for factors affecting the discount rate and changes in the rate of time preference. The growth rate of industrial production was included to account for factors affecting cash flows.

Table 1. Input and output variables for the multi-factor model.

(1)	(2)	(3)
Variable	Type of Variable	How Calculated
Unexpected Inflation	Input	Residuals from regressing monthly change in natural log of CPI on lagged changes in CPI and on Treasury bill returns.
Change in Expected Inflation	Input	Change in fitted value from regressing monthly change in natural log of CPI on lagged changes in CPI and on Treasury bill returns.
Bauer and Swanson Monetary Policy Measure	Input	First principal component of the change in the first four Eurodollar contracts over a 30 min period surrounding Federal Open Market Committee news.
Growth Rate of Industrial Production	Input	Monthly change in natural log of industrial production.
Term Structure Measure	Input	Difference between monthly returns on 20-year and one-month Treasury securities.
Asset Returns on 54 Assets	Input	Change in natural log of asset price index.
Risk-Free Rate	Input	Return on one-month Treasury security.
Beta of Asset i to Macroeconomic Factor j	Output	Coefficient from iterated nonlinear seemingly unrelated regression of the return on asset i on news of macroeconomic factor j.
Risk Price Associated with Factor j	Output	Variable estimated from the iterated nonlinear seemingly unrelated regression of the the relationship between the required return on an asset and the asset’s beta to macroeconomic factor j.
Risk Premium of Asset i to Macroeconomic Factor j	Output	The product of the beta of asset i to macroeconomic factor j and the risk price associated with factor j. It measures the increment in ex-ante required returns to compensate for asset i’s exposure to macroeconomic factor j.

Note: The table presents the input and output variables used to estimate the multi-factor model employed in this paper.

presents the input and output variables used to estimate the system of Equation (4).

3.2. Data

To measure unexpected U.S. monetary policy, data from Bauer and Swanson (2022) (B&S) were used. They calculated monetary policy surprises by examining changes in the first four Eurodollar future contracts over a 30 min period surrounding Federal Open Market Committee (FOMC) news. They used the first principal component of the changes in the Eurodollar contracts to measure monetary policy. They constructed this variable so that an increase represents a contractionary monetary policy surprise. Their sample period extended from January 1988 to December 2019. The model estimated here thus used this sample period.

To calculate unexpected inflation, the monthly CPI inflation rate was regressed on current and lagged one-month Treasury bill returns and on lagged values of the CPI inflation rate. This approach was used by Boudoukh et al. (1994). This framework was also used to calculate the change in expected inflation.1 Data to calculate unexpected inflation, the change in expected inflation, the growth rate of industrial production, and the term structure measure came from the Federal Reserve Bank of St. Louis FRED database and from Kroll (2023).

For the variable R_i in Equations (1) through (3), the returns on 54 portfolios were used. These include the returns on sectoral stock returns and on assets related to gold and silver. Bekaert and Wang (2010) noted that many stocks are harmed by inflation, and Bampinas and Panagiotidis (2015), Hoang et al. (2016), and Sharma (2016) reported that assets related to gold and silver benefit from inflation. Including both sectoral stock returns and returns related to precious metals increases the cross-sectional variation in assets’ responses to inflation. Data on returns came from the Datastream database.2 For the variable λ₀ in Equations (2) and (3), the return on one-month Treasury bills was used. Data on the return on one-month Treasury bills came from Kroll (2023). Descriptive statistics for the variables are contained in Appendix A.

3.3. Investigating the Impact of Inflation and Monetary Policy on Asset Performance

To investigate when investors’ anticipations of inflation changed, this paper adopted an intuitive approach. If investors expect inflation to rise, they will seek to purchase assets such as gold and silver that hedge against inflation and to sell assets such as stocks that are harmed by inflation. This will raise the prices of inflation hedges and lower the prices of assets hurt by inflation. Assets that hedge against inflation have positive inflation betas, and assets that suffer from inflation have negative inflation betas. There should thus be a positive relationship between assets’ returns and their inflation betas when investors expect higher inflation and a negative relationship when they expect lower inflation. For each month from August 2020 to September 2024, returns on the 54 assets were thus regressed on their inflation betas estimated over the 1988–2019 period.

To facilitate interpretation, the regression coefficients were multiplied by the average of the betas of the three assets most harmed by inflation (these assets are airline stocks, durable household product stocks, and container and packaging stocks). Since the average of these three beta coefficients is negative, the product of this average and the regression coefficient is positive when investors expect lower inflation and negative when they foresee higher inflation. The product measures the monthly change in returns on these three assets because of their exposure to inflation.

This approach can also be used to investigate how news of expansionary fiscal policy affected inflation perceptions. This can be investigated using days when Bloomberg articles indicated that fiscal policy would be more expansionary. Gomez-Cram et al. (2024) collected a sample of stories from Bloomberg that contained news directly related to fiscal spending and specified the total amount. They only included dates that did not coincide with releases of any of the top 50 macroeconomic announcements, as calculated by Bloomberg. If investors expected expansionary fiscal policy to produce inflation, they would sell assets harmed by inflation on those days and purchase assets that benefit from inflation. Thus, the methodology discussed in the previous paragraphs using monthly data can be applied to daily data on the days highlighted by Gomez-Cram et al. (2024).

If investors expect contractionary monetary policy, they will seek to purchase assets that hedge against contractionary policy and to sell assets that are harmed by contractionary policy. This will raise the prices of assets that hedge against tighter policy and lower the prices of assets hurt by tighter policy. Assets that hedge against contractionary policy have positive B&S monetary policy betas, and assets that suffer from contractionary policy have negative B&S monetary policy betas. There should thus be a positive relationship between assets’ returns and their monetary policy betas when investors expect tighter policy and a negative relationship when they expect easier policy. For each month from August 2020 to September 2024, returns on the 54 assets were thus regressed on their monetary policy betas estimated over the 1988–2019 period.

Again, to facilitate interpretation, the regression coefficients were multiplied by the average of the three assets most harmed by contractionary monetary policy (these assets are aluminum stocks, electronic and electrical equipment stocks, and Australasian gold mining stocks). Since the average of these three betas is negative, the product of this average and the regression coefficient is positive when investors expect a policy to become easier and negative when they foresee a tighter policy. The product measures the monthly change in returns on these three assets because of their exposure to contractionary monetary policy.

4. Results

4.1. Results from Estimating the Multi-Factor Model

Table 2. Iterated nonlinear seemingly unrelated regression estimates of assets’ sensitivities to unexpected inflation and to Bauer and Swanson’s (2022) Fed policy surprises.

(1)	(2)	(3)	(2)	(3)
Asset	Inflation Beta	Standard Error	Monetary Policy Beta	Standard Error
Aerospace/Defense	−0.696	0.806	−0.161 ***	0.047
Aerospace	−1.154	0.896	−0.165 ***	0.052
Airlines	−3.658 ***	1.301	−0.218 ***	0.076
Aluminum	−0.095	1.513	−0.259 ***	0.089
Apparel Retail	0.012	1.187	0.012	1.187
Auto and Parts	−0.359	1.113	−0.167 **	0.066
Auto Parts	−0.647	0.968	−0.159 ***	0.058
Automobiles	0.226	1.383	−0.152 *	0.083
Basic Materials	−0.614	0.877	−0.182 ***	0.052
Basic Resources	0.366	1.114	−0.227 ***	0.065
Beverages	−1.174 *	0.704	−0.048	0.042
Broadcast and Entertainment	−0.472	1.117	−0.159 **	0.066
Brewers	−1.111	0.851	−0.074	0.050
Building Materials/Fixtures	−0.081	0.949	−0.101 *	0.056
Business Supply Services	−0.943	0.789	−0.206 ***	0.045
Chemicals	−1.162	0.816	−0.164 ***	0.048
Clothing and Accessories	−0.120	1.080	−0.227 ***	0.063
Commercial Vehicles/Trucks	−1.348	1.108	−0.216 ***	0.065
Computer Hardware	−0.552	1.141	−0.173 **	0.068
Computer Services	0.429	0.819	−0.162 ***	0.048
Construction and Materials	−0.285	0.933	−0.128 **	0.055
Consumer Discretionary	−0.806	0.725	−0.173 ***	0.042
Consumer Finance	−0.485	1.002	−0.111 *	0.059
Consumer Goods	−1.306 *	0.730	−0.159 ***	0.043
Consumer Staples	−1.054 *	0.578	−0.063 *	0.034
Consumer Services	−0.742	0.727	−0.187 ***	0.04
Container and Packaging	−1.935 **	0.873	−0.216 ***	0.051
Conventional Electricity	−1.197 *	0.614	0.012	0.036
Defense	0.609	0.858	−0.151 ***	0.051
Distillers and Vintners	−0.196	0.898	−0.013	0.054
Diversified Industrials	−0.591	0.843	−0.124 **	0.050
Drug Retailers	0.458	0.926	−0.066	0.055
Durable Household Products	−2.504 **	1.190	−0.163 **	0.071
Electronic Components and Equipment	−0.694	0.945	−0.216 ***	0.055
Electricity	−1.192 *	0.613	0.012	0.036
Electronic and Electrical Equipment	−0.652	0.956	−0.231 ***	0.055
Food and Drug Retail	−0.385	0.693	−0.098 **	0.041
Food Producers	−0.434	0.587	−0.043	0.036
Food Retailers and Wholesalers	−0.950	0.772	−0.075	0.046
Financial Services	−0.221	0.901	−0.129 **	0.053
Financials	−0.308	0.827	−0.105 **	0.049
Gold Bullion	1.590	1.162	−0.025	0.038
Gold Mining (Americas)	2.646 *	1.494	−0.140	0.087
Gold Mining (Australasia)	2.967 *	1.709	−0.325 ***	0.097
Gold Mining (World)	3.095 **	1.436	−0.182 **	0.082
Health Care	−0.586	0.613	−0.073 **	0.035
Oil and Gas	0.647	0.784	−0.065	0.047
Pharmaceuticals and Biochemical Products	−0.836	0.693	−0.059	0.041
Real Estate Investment Trusts	0.699	0.858	−0.068	0.051
Silver (S&P GSCI)	2.321 **	1.162	−0.081	0.069
S&P 500	−0.705	0.631	−0.118 ***	0.037
Technology	0.031	1.052	−0.183 ***	0.062
Telecom	−1.260	0.797	−0.098 **	0.047
Utilities	−0.990 *	0.600	0.019	0.036

Note: The table presents iterated nonlinear seemingly unrelated regression estimates of assets’ betas to unexpected inflation and to the Bauer and Swanson (2022) (B&S) measure of Fed policy surprises from a multi-factor model that includes returns on the 54 assets listed in column (1) (minus the return on one-month Treasury bills) on the left-hand side and the difference between returns on 20-year and one-month Treasury securities, the Bauer and Swanson (2022) (B&S) measure of monetary policy, the monthly growth rate in industrial production, unexpected inflation, and the change in expected inflation on the right-hand side. Unexpected inflation is calculated as the residuals from a regression of the consumer price index inflation rate on lagged inflation and current and lagged Treasury bill returns. The change in expected inflation is also calculated from this model. The B&S measure is constructed so that an increase represents a contractionary monetary policy surprise. The sample extends from January 1988 to December 2019. *** (**) [*] denotes significance at the 1% (5%) [10%] level.

presents assets’ exposures to inflation and to monetary policy. Forty assets in Table 2 exhibit negative inflation betas, implying that they are harmed by news of inflation. Nine of these are significant at at least the 10% level. Fourteen assets exhibit positive inflation betas. Four of these are significant at at least the ten percent level. Fifty assets in Table 2 exhibit negative monetary policy betas, implying that they are harmed by contractionary monetary policy. Thirty-eight of these are significant at at least the 10% level. Four assets have positive monetary policy betas, though none are statistically significant.

Table 3. Iterated nonlinear seemingly unrelated regression estimates of the risk prices associated with the macroeconomic factors.

(1)	(2)
Macroeconomic Factor	Risk Price
Unexpected Inflation	−0.0046 ** (0.0020)
Term Structure Measure	−0.0275 ** (0.0109)
Industrial Production Growth	−0.0133 *** (0.0046)
Change in Expected Inflation	−0.0151 * (0.0008)
Monetary Policy	0.0455 (0.0291)

Note: The table presents iterated nonlinear seemingly unrelated regression estimates of risk prices from a multi-factor model including returns on 54 assets (minus the return on one-month Treasury bills) on the left-hand side and the difference between returns on 20-year and one-month Treasury securities, the Bauer and Swanson (2022) (B&S) measure of Fed policy surprises, the monthly growth rate in industrial production, unexpected inflation, and the change in expected inflation on the right-hand side. The B&S measure is constructed so that an increase represents a contractionary monetary policy surprise. Unexpected inflation is calculated as the residuals from a regression of the consumer price index inflation rate on lagged inflation and current and lagged Treasury bill returns. The change in expected inflation is also calculated from this model. The sample extends from January 1988 to December 2019. *** (**) [*] denotes significance at the 1% (5%) [10%] level.

presents the risk prices associated with the macroeconomic variables. The value for unexpected inflation is −0.0046. For the three assets in Table 1 most harmed by unexpected inflation, the beta values average −2.70. The risk premium associated with inflation for these assets averages 0.0124 (equals −0.0046 times −2.70). This indicates that these assets must, on average, pay a premium of 1.24 percent per month to compensate for their exposure to inflation. These assets thus behave as bonds do in Cieslak and Pflueger’s (2023) bad inflation case.

For the three assets in Table 1 that benefit the most from unexpected inflation, the beta values average 2.69. The risk premium associated with inflation for these assets averages −0.0124 (equals −0.0046 times 2.69). This implies that the return required to hold these assets decreases by 1.24% per month because these assets do well when inflation increases.

4.2. Investigating Investors’ Changing Perceptions of Inflation and Monetary Policy After the Pandemic

To understand how investors’ perceptions of inflation changed after the pandemic began, Figure 2 plots the CPI inflation rate and all of the months between August 2020 and September 2024 when there is a statistically significant relationship at at least the 10% level between the returns on the 54 assets and their inflation betas. To facilitate interpretation, the regression coefficients are multiplied by the average of the betas of the three assets in Table 1 that are most harmed by inflation. Since the average of these three beta coefficients is negative, the product of this average and the regression coefficient is positive when investors expect lower inflation and negative when they foresee a higher inflation policy. The coefficients in Figure 2 show how investors’ responses to inflation news affected assets that are exposed to inflation.

Figure 2 also plots all of the months when there is a statistically significant relationship at at least the 10% level between the returns on the 54 assets and their monetary policy betas. These regression coefficients are multiplied by the average of the three assets in Table 2 that are most harmed by contractionary monetary policy. Since the average of these three betas is negative, the product of this average and the regression coefficient is positive when investors expect a policy to become easier and negative when they foresee a tighter policy.

Figure 2 shows that investors did not price higher inflation into monthly returns until April 2021. They even priced in lower inflation in February 2021. Thus, as increased mobility and the passage of ARPA contributed to soaring consumer spending in the first quarter of 2021, investors remained complacent about the risks of inflation.

As inflation rose from 1.7% in February 2021 to 4.1% in April 2021 and to 6.6% in November 2021, they priced higher inflation into the monthly data in April, May, July, October, and November. They also priced in higher inflation in February 2022. February was the month when Russia invaded Ukraine on the 22nd, raising prices for energy, food, and other commodities. Examining daily data indicates that much of the response in February 2022 was concentrated around the days when Russia invaded Ukraine.

Figure 3, using the same methodology as in Figure 2, plots the daily responses between 1 February 2021 and 30 April 2021. In February, investors foresaw lower inflation on eight days and higher inflation on only two days. In March, they expected lower inflation on six days and higher inflation on six days. In April, however, they perceived lower inflation on four days and higher inflation on eight days. These results confirm the findings in Figure 2, indicating that investors foresaw lower inflation in February 2021 and only bid up the prices of assets that benefit from inflation and bid down the prices of assets that are harmed by inflation starting in April 2021.

4.3. Investigating Investors’ Changing Perceptions of Inflation on Days When Fiscal Policy Was Expansionary

Table 4. Changes in returns for assets exposed to inflation on days when fiscal policy news was expansionary.

Date	News Item	Change in Returns for Assets Exposed to Inflation
19 August 2020	The Trump Team Sees Path to Pared-Down USD 500 Billion Stimulus Deal; White House Open to USD 25 Billion for Postal Service, McEnany Says; White House Open to USD 25 Billion for USPS.	0.0144 ***
28 September 2020	Colleges Battling COVID-19 Upsurge Seek USD 120 Billion as Costs Mount.	0.0049 *
9 October 2020	White House Preparing New USD 1.8 Trillion Stimulus Proposal.	−0.0142 **
21 October 2020	U.S. Sees Spending USD 10 Billion on Missile Defense Through 2025; Democrats Block a USD 500 Billion ’Skinny’ Coronavirus Aid Bill Identical to Another that Republicans Unveiled a Month Ago.	−0.0080 ***
1 December 2020	Bipartisan Group of Senators Prepares USD 908 Billion Stimulus Plan, Aiming to Break Partisan Logjam; Mitch McConnell Rejects Bipartisan Proposal by Senate Moderates for USD 900 Billion COVID-19 Relief Package; U.S. Bipartisan Lawmakers Propose USD 908 Billion COVID-19 Relief Bill; Bipartisan Senate Group Pitches USD 908 Billion Stimulus Plan; Mnuchin Says Congress Must Redirect USD 455 Billion, Not Biden; Senate GOP Relief Package Includes USD 332.7 Billion for PPP Plan.	−0.0149 ***
13 January 2021	Schumer Asks Biden to Seek More Than USD 1.3 Trillion in Relief; Stocks Gain, Schumer Wants Stimulus North of USD 1.3 Trillion.	0.0009
16 February 2021	Focus on Capitol Hill Turns to Passing Biden’s USD 1.9 Trillion Coronavirus Relief Bill; Biden’s USD 1.9 Trillion Stimulus Plan is Popular with Voters, but It Crashes into Strong Republican Resistance in Congress.	0.0075 **
22 February 2021	Most U.S. Stocks Fall While Copper Hits Decade High; USD 1.9 Trillion Stimulus to Reach House of Representatives Shortly; Biden Stimulus Dash; USD 1.9 T Plan a ’Bailout for Lockdowns’; Biden’s USD 1.9 Trillion Stimulus Plan Enters 3-Week Congress Dash.	−0.0161 *
19 May 2021	The Senate Weighs a Bill That Would Invest USD 120 Billion in Technology Research to Counter China; Senate Weighs Investing USD 120 Billion in Science to Counter China; Senate China Bill to Add USD 52 Billion for U.S. Chip Making; Senate Democrat proposes USD 52 billion for U.S. chip production; GOP’s USD 400 Billion Highway Bill Focuses on ‘Core Infrastructure.’	0.0054 *
26 May 2021	Bipartisan USD 304b Highway Bill Advanced by Senate Panel; Biden’s American Jobs Plan Includes USD 318 Billion for Housing.	−0.0004
22 June 2021	Biden’s USD 6.5 Billion Biomedical Agency Backed in Bipartisan Bill; Warren Leads Letter Seeking USD 700B for Child-Care Infrastructure.	0.0010
12 July 2021	Biden’s USD 579 Billion Plan Is a Tiny Step in the Right Direction.	−0.0003
9 August 2021	Democrats Unveil USD 3.5 Trillion Budget Plan.	0.0139 **
10 August 2021	Sweeping USD 550 Billion Infrastructure Bill Passes U.S. Senate; Senate Poised to Pass USD 550 Billion Infrastructure Bill; Senate Passes USD 550 Billion Infrastructure Plan in Win for Biden; U.S. Senate Passes USD 550 Billion Infrastructure Bill that Could Unleash Biggest Burst of Spending in Decades; Senate Approves Bipartisan, USD 1 Trillion Infrastructure Bill, Bringing Major Biden Goal One Step Closer; Treasury Yields End Higher as Senate Passes USD 1 Trillion Bipartisan Infrastructure Bill; Senate Passes USD 550 Billion Infrastructure Bill.	0.0109 ***
30 August 2021	Top Defense Republican to Propose USD 25 Billion Pentagon Boost; Democrats Pressured to Add USD 10 Billion to Transit in Budget Bill; Groups and Mayors Urge U.S. Congress to Back USD 10 billion in New Public Transit Funding; Modeling Impact of USD 4 Trillion Fiscal Stimulus on U.S. Outlook.	−0.0046
10 September 2021	Senators Push for USD 6 Billion Bailout for Private Bus Industry; House Panel Readies USD 150B Clean Energy Plan; Extra USD 28 Billion to Appease Farms Promised in Full Budget Bill.	0.0033
27 October 2021	Democrats Near Deal on USD 500 Billion to Fight Climate Change.	−0.0044 *
9 November 2021	BI’s Companies to Watch in Biden’s USD 1.2T Infrastructure Bill.	−0.0022

Note: The figure presents the average change in returns due to inflationary expectations for the three assets most exposed to unexpected inflation on days when there was news of expansionary fiscal policy. The three assets most exposed to inflation are airline stocks, durable household product stocks, and container and packaging stocks. To calculate the change in returns associated with inflation, assets’ inflation betas are estimated. The betas are obtained from an iterated nonlinear seemingly unrelated regression (INLSUR) of returns on 54 assets (minus the return on one-month Treasury bills) on the Bauer and Swanson (2022) (B&S) measure of Fed policy surprises, the difference in returns between 20-year and one-month Treasury securities, the monthly growth rate in industrial production, unexpected inflation, and the change in expected inflation. Unexpected inflation is calculated as the residuals from a regression of the consumer price index inflation rate on lagged inflation and current and lagged Treasury bill returns. The change in expected inflation is also obtained from this model. If investors believe that inflation will rise, they will purchase assets that benefit from higher inflation (those with positive inflation betas) and sell assets that are harmed by inflation (those with negative inflation betas). There should thus be a positive relationship between asset returns and assets’ inflation betas on days when investors foresee higher inflation. For each day when Gomez-Cram et al. (2024) identified news of expansionary fiscal policy, returns on the 54 assets are thus regressed on the assets’ inflation betas. To facilitate interpretation, the resulting regression coefficient is multiplied by the beta coefficient for the 3 assets from the INLSUR regression that are harmed the most by inflation. The change in returns associated with inflation in the table thus represents the average change for the three assets that are most exposed to inflation. Since the average inflation beta coefficient for these three assets is negative, positive values in the table indicate that investors expect lower inflation, and positive values indicate that they foresee higher inflation. *** (**) [*] denotes significance at the 1% (5%) [10%] level. Source: Gomez-Cram et al. (2024) and calculations by the author.

reports the change in returns associated with inflation on the days between 1 August 2020 and 30 November 2021 that are included in Gomez-Cram et al.’s (2024) sample.

There are six cases when news of expansionary fiscal policy increased (at the 10% level) returns on assets exposed to inflation and five cases when it lowered (at the 10% level) returns on assets exposed to inflation. There is thus no compelling evidence in the table that news of expansionary fiscal policy lowered returns on assets exposed to inflation. Rather, investors only began pricing in higher inflation starting in April 2021 when inflation began accelerating. They also began pricing in lower inflation in April, July, and October 2022 and in February 2023 as inflation was actually declining.

Greenspan noted that when fiscal and monetary policy turned inflationary in the 1960s and 1970s, investors initially reacted little (see, e.g., Greenspan, 1994; and Woodward, 2001). They trusted U.S. institutions to keep inflation under control. Then, as inflation accelerated, they demanded high-risk premia in stocks and bonds to compensate for the risk of inflation (i.e., they viewed stocks and bonds as responding to what Cieslak and Pflueger (2023), called bad inflation). The evidence in Figure 2 is consistent with what happened in the 1970s and 1980s when, after a long period of quiescent inflation, investors initially reacted little to news of expansionary policy.

While De Soyres et al. (2022), Di Giovanni et al. (2023), and others reported that fiscal policy during the pandemic stoked inflation, Eggertsson and Kohn (2023) found that the monetary policy framework that the Federal Reserve adopted in 2020 also contributed to inflation by causing the Fed to delay tightening as inflation emerged in 2021. Then, after inflation reached its highest level in more than 40 years, the Fed began what Eggertsson and Kohn called unprecedented federal funds rate hikes. Starting in March 2022, it raised the funds rate by 5 percentage points in 15 months.

Figure 2 and Table 1 and Table 2 indicate that the damage to stocks caused by inflation pales in comparison to the damage caused by this monetary policy tightening. Beginning in April 2021 for inflation and November 2021 for monetary policy, there were five months when negative news led to stock price declines in Figure 2. For inflation, the average effect over these five months equaled 3.4% per month, while for monetary policy, the average effect equaled 9.2% per month. In addition, Table 1 indicates that 13 of the 54 assets are exposed to inflation at at least the 10% level, while Table 2 indicates that 38 of the 54 assets are exposed to monetary policy. In Table 1, some assets gain and some lose from inflation, while in Table 2, almost all the assets lose from contractionary monetary policy. Figure 2 also makes clear that changing perceptions about monetary policy contributed to massive swings in stock prices after June 2022. Thus, contractionary monetary policy inflicted greater damage and uncertainty across a wide cross-section of assets than higher inflation did.

5. Discussion and Conclusions

The COVID-19 pandemic initially restricted the supply of labor in the U.S. At the same time, U.S. fiscal spending increases of 20% in 2020 and the USD 1.9 trillion ARPA stimulus in 2021 stoked demand. The Fed also implemented a new policy framework in 2020 that implied it would react more forcefully to low inflation than to high inflation. They did this partly because, as Figure 1 indicates, the economy in 2020 had experienced several decades of low inflation. Inflation then accelerated in 2021 and reached the highest level in more than 40 years in 2022.

Evidence presented here indicates that investors did not price inflation into asset prices until inflation began to accelerate. The Fed also did not respond to inflation until it reached historic highs and then responded with unprecedented tightening. The evidence presented here indicates that this aggressive response roiled financial markets.

Experience from the 1970s and 1980s highlights the devastation that inflation and dis-inflationary monetary policy can cause. High inflation necessitates contractionary monetary policy that damages both the stock market and the real economy. Many policymakers and investors who are active now have forgotten these lessons. The Fed, rather than adopting an asymmetric framework that downplays inflation, should respond to both incipient inflation and labor market shortfalls. This is all the more important as the incoming Trump administration adopts expansionary fiscal policies, tariffs, the repatriation of illegal immigrants, and other policies that can stoke inflation.