Early effects of COVID-19 pandemic-related state policies on housing market activity in the United States

Abstract

I use daily and weekly data from 100 metropolitan areas in 2020 to investigate the effects of state-level policies to combat the COVID-19 pandemic on various indicators of U.S. housing market activity. Measures of housing market activity include change in new listings, total inventory, newly pending sales, median list price, web traffic to for-sale homes, and average number of days to pending sale status. Using event study and difference-in-differences models, I find that the closure of non-essential businesses in certain states was associated with up to an 11-percentage point decrease in new home listings and a 3.5 percentage point decrease in total inventory relative to the same period in 2019. I also find that school closures may affect some outcomes.

1. Introduction

COVID-19 is an infectious disease that had spread to 219 countries by April 2021. The World Health Organization (WHO) on March 11, 2020, declared the COVID-19 outbreak a global pandemic. The early cases of COVID-19 were confirmed in the United States by January 2020. As of April 2021, there were more than 135 million confirmed cases of COVID-19 globally, with approximately 3 million related deaths. Approximately 20 percent of all confirmed cases and deaths related to COVID-19 were reported in the United States.

Several recent studies document the negative short-run impact of the coronavirus pandemic on the U.S. economy (Baker et al., 2020; Atkeson, 2020; Lewis et al., 2020). This paper provides the first estimates of the short run impact of the state level policies to mitigate the spread of the coronavirus on various indicators of housing market activity in selected U.S. metropolitan areas.1

During the early pandemic (2020), the majority of U.S. states implemented several policies to slow the spread of the COVID-19. Examples include shutting down certain businesses deemed non-essential, closing schools, ordering people to stay-at-home, and imposing restrictions on large gatherings. Early papers show that these policies were effective in slowing down the spread of the virus among the population (Martin-Calvo, et al., 2020) and reducing the movement and social interactions of individuals (Dave et al., 2021; Friedson et al., 2020; Gupta et al., 2020).

A first step towards understanding the short-run effects of these policies on U.S. economy requires a more detailed analysis of their effects on numerous areas of economic activity, including the housing market. To the best of my knowledge, no previous study has investigated the early short-run effects the state level policies to control the spread of the COVID-19 on housing market activity. This paper thus analyzes detailed data covering change in new listings, total inventory, newly pending sales, median list price, web traffic to for-sale homes, and average number of days to pending sale status. The results presented here can help us understand how the state level policies played out, along with providing evidence that could be relevant to other times and other places.

I use daily and weekly housing market data from 100 major U.S. metropolitan area housing markets from January 1, 2020, before any U.S. COVID-19 cases were confirmed, to December 12, 2020, when there were more than 16 million COVID-19 related cases and 297,000 deaths in the U.S.

The publicly-available aggregate data indicate that the early collapse and partial recovery of housing market activity was broad- based. Using event study and difference-in-differences models, I find that many of the policies that were introduced to combat the pandemic had no significant short-run impact on many indicators of housing market activity. The strongest effects stem from the closure of non-essential businesses. I find that the closure of non-essential businesses in certain states was associated with up to an 11-percentage point decrease in new home listings and a 3.5 percentage point decrease in total inventory relative to the same time period in 2019. I also find that school closures may affect some outcomes.

2. Background and existing literature

The coronavirus epidemic hit the United States relatively late. The first case of COVID-19 in the United States was confirmed on January 20, 2020, when parts of China were already under a quarantine order. The first COVID-19 related death in the United States was reported on February 29. Washington was the first state to issue a state of emergency on the same date, with New York and California following the next week. There were 100 cases by March 4, 2020, 10,000 cases by March 18, and more than one million cases by April 28.

From mid-March to early April 2020, many states adopted policies to mitigate the spread of the coronavirus. These included stay-at-home orders, non-essential business closures, bans on large gatherings, school closures, and closings of certain businesses such as restaurants and bars. By April 20, 2020, all states except Arkansas, Iowa, Nebraska, North Dakota, South Dakota, and Wyoming had issued some form of a stay-at-home order. By the same time, all but seven states had implemented some form of non-essential business closures. Initial implementation of these policies across different states occurred at different times. Kong and Prinz (2020) demonstrate that although announcement dates of the policies across different states are positively correlated, the correlation is weak in most cases. For each state in the analysis sample, Appendix Table A1 indicates when these policies were first implemented and when they were subsequently lifted.

Several papers investigated the early effects of restrictive state-level policies on economic and social outcomes in the United States. Engle et al. (2020) found that local infection rate and stay-at-home-orders are significantly associated with reduction in mobility. Painter and Qiu (2020) documented that political beliefs present a significant limitation to the effectiveness of state-level social distancing orders. In particular, residents in Republican counties were less likely to completely stay at home after a state order has been implemented relative to those in Democratic counties. Allcott, et al. (2020) also demonstrated that Republicans engage in less social distancing, even after controlling for other factors including state policies, population density, and local COVID-19 cases and deaths. Kong and Prinz (2020) found that restaurant and bar limitations and non-essential business closures could explain 4.4% and 8.5% of unemployment insurance (UI) claims in the United States respectively, while the other state level policies did not have an impact on UI claims.

There is also a broader literature which investigates the effects of health shocks and pandemics on housing markets. Ambrus et al. (2020) examine the impact on housing prices of a cholera epidemic in one neighborhood of nineteenth century London. They show that ten years after the epidemic, housing prices are significantly lower just inside the catchment area of the water pump that transmitted the disease. Moreover, differences in housing prices persist over the following 160 years. Francke and Korevaar (2020) investigate the effects of historical outbreaks of the plague in the 17th century Amsterdam and cholera in the 19th century Paris on housing prices. They find that outbreaks resulted in large declines in house prices, and smaller declines in rent prices, with large reductions in house prices during the first six months of an epidemic in heavily affected areas. They also show that price shocks were transitory, and both cities quickly reverted to their initial price paths. Wong (2008) finds that SARS epidemic led to a 1.6% fall in home values and 72% fall in transactions volume in Hong Kong. Davis (2004) investigates the effect of a health risk, a cluster of cases of pediatric leukemia in an isolated county of Nevada, on housing prices. He finds that compared with unaffected counties, housing prices in the affected county declined significantly during the period of maximum risk.

As broad background to examining the 2020 COVID period, Figs. A1 and A2 provide summary trends for the U.S., comparing January 2020 through mid-December 2020 with the same period in the previous year. Figs. A1 indicates that there was considerable aggregate movement in numerous measures of housing market activity in 2020. On February 22, 2020, changes in new home listings and pending sales were considerably higher than in the same time period in 2019. By March 28, 2020, both indicators were below their 2019 levels. On April 18, changes in newly listed homes and pending sales in the United States were more than 30% below their 2019 levels. Both indicators started to exhibit and upward trend after May 9, 2020 and recovered by early July 2020. Not surprisingly, total inventory exhibited a decreasing trend throughout the year.

Fig. A1.

Housing market activity in the United States during the COVID-19 pandemic (relative to the same time period in 2019). (A) Percentage change in total inventory, pending sales, and new listings (B) Average number of days to pending sale status, and percentage change in web page views and median list price.

Notes: Weekly data. Percentage changes in outcome variables are relative to the same time period in the previous year. Percentage change in web traffic to for-sale listings is relative to the same period in the previous year, calculated as a 7-day moving average. Author's calculations from Zillow data (Zillow Research, 2020).

Changes in web traffic to for-sale listings initially declined, but then quickly rose to a considerably higher level for the rest of 2020 (Fig. 1 B). Average number of days to pending sale status was declining until March 28. From that date until late November, it exhibited a relatively steady trend. Change in median list price of homes was mostly stable until early July 2020, but then exhibited a moderately increasing trend after that date, remaining positive relative to its level during the same time period in the previous year.

Fig. 1.

Change in new listings for selected metropolitan areas.

Notes: Daily data from January 1 to December 12, 2020. Percentage change in new listings is relative to the same time period in the previous year. Author's calculations from Zillow data (Zillow Research, 2020).

Intensity of the spread of the virus and total confirmed cases and deaths in 2020 differed considerably across metropolitan areas. For example, as of mid-December 2020, New York City had more than 427 cases per 10,000 people. On the other hand, several cities in my sample had less than 100 cases per 10,000 people on the same date. Fig. 1, Fig. 2, Fig. 3, Fig. 4, Fig. 5, Fig. 6 present trends in various indicators of housing market activity relative to the same time period in 2019 for selected metropolitan areas.

Fig. 2.

Change in median list price for selected metropolitan areas.

Notes: Weekly data from January 4 to December 2, 2020. Percentage change in median list price is relative to the same time period in the previous year. Author's calculations from Zillow data (Zillow Research, 2020).

Fig. 3.

Change in newly pending sales for selected metropolitan areas.

Notes: Weekly data from January 4 to December 12, 2020. Percentage change in newly pending sales is relative to the same time period in the previous year. Author's calculations from Zillow data (Zillow Research, 2020).

Fig. 4.

Change in total inventory for selected metropolitan areas.

Notes: Weekly data from January 4 to December 12, 2020. Percentage change in total inventory is relative to the same time period in the previous year. Author's calculations from Zillow data (Zillow Research, 2020).

Fig. 5.

Average number of days to pending sale status for selected metropolitan areas.

Notes: Weekly data from January 4 to December 12, 2020. Author's calculations from Zillow data (Zillow Research, 2020).

Fig. 6.

Percentage change in web traffic to for-sale homes for selected metropolitan areas.

Notes: Daily data from March 1 to December 12, 2020. Percentage change in web traffic to for-sale homes is relative to the same time period in the previous year. Author's calculations from Zillow data (Zillow Research, 2020).

3. Data

I use data on daily confirmed coronavirus cases and deaths for each state from the New York Times COVID-19 database (New York Times, 2020). I use state population data from 2019 Census estimates to calculate per capita number of cases and deaths for each state. Since recent papers show that residents in Republican states are less likely to comply with restrictive policies, I estimate several models that exclude states with Republican governors.2

There are few sources that track the social distancing policies at the state level. I use the data assembled by The Institute for Health Metrics and Evaluation (IHME) at the University of Washington because it is comprehensive and provides precise information on timing of the social distancing orders. In particular, I use information on the timing of the introduction and lifting of bans on mass gatherings; that is, any business closures that only affected certain businesses such as restaurants and bars, school closings, closures of all non-essential businesses, and stay-at-home orders

Daily housing data come from Zillow, an online real estate database company. Zillow reported for 2020 the daily percentage change in the number of newly listed homes and web-traffic to for-sale homes relative to the same time period in the previous year for 100 major U.S. housing markets.3 Zillow also reported weekly data on percentage change in newly pending sales, total inventory, median list price and average number of days to pending sale status for selected markets. The markets used in the empirical analysis, summary statistics of variables, and data availability period for each market are described in the appendix Tables A2 and A3.

4. Empirical models

In order to formally estimate the short run effects of the spread of the coronavirus-related state-level social distancing policies on housing market activity, I use two empirical models. First, I estimate an event study model that exploits differential policy announcement dates across different states. The main specification for this model for the metropolitan areas that adopted a particular policy is of the form

$$Y_{ist}=\sum _{j=-7}^7{\delta }_j\times 1\left\{day=j\right\}+{\gamma }_i+{\varsigma }_s+{\mu }_t+{\epsilon }_{ist}$$ (1)

where $Y_{ist}$ is one of the indicators of housing market activity for metro area i, located in state s, for the calendar date t, day denotes days relative to the date that a particular policy was announced (day = 0); and ${\gamma }_i$, ${\varsigma }_s$, ${\mu }_t$ are city, state, and calendar fixed effects. Standard errors are clustered at the state level. This model thus contains series of dummy variables that control for each day of the week before and after the policy announcement day. I exclude the day prior to the announcement date (day = -1). The coefficients of interest ${\delta }_j$ estimate the differential change in outcome variables for each day j relative to the day prior to the announcement date. Following Kong and Prinz (2020), for periods j > 7 and j < -7, I assign j = 7 and j = -7 respectively with the underlying assumption that the dynamic effects of the policy are constant 7 days before or after the policy announcement.4 Eq. (1) not only enables one to estimate the daily effects of a particular policy on these outcomes during the first week of the policy announcement date but also acts as a falsification test since it includes controls for the periods before the policy change.

Next, I use a difference-in-differences model to estimate the joint impact of state policies on different indicators of housing activity, while controlling for the spread of the coronavirus in certain cases. The main specification for this model is

$$Y_{ist}=\beta \prime Polic{y}_{st}+\alpha X_{st}+{\gamma }_i+{\varsigma }_s+{\mu }_t+{\nu }_{ist}$$ (2)

where $Polic{y}_{st}$ is a set of binary variables for policies adopted by state s at calendar date t. These binary variables take the value of one for the calendar dates for which day≥ 0. In several models, I control for the spread of the coronavirus at the state level ($X_{st}$), either as total number of cases or deaths per 10,000 people. I present the summary statistics for these variables in Appendix Table A3. Unlike the event study model, I estimate Eq. (2) also for the outcome variables for which weekly data are available (change in total inventory, median list price, newly pending sales, and average number of days to pending sale status). The key identification assumption behind Eq. (2) is that alternative indicators of housing market activity follow a common trend for treatment and control cities (cities that adopted a particular policy at a given date vs. those did not) in the absence of the policies to control for the spread of coronavirus. The common trend assumption may fail when a policy endogeneity problem exists. However, it is not plausible that states that are experiencing relatively strong or weak housing markets are correspondingly more or less likely to adopt these policies. The event study model in Eq. (1) also provides a pre-trend analysis of each policy since it includes terms that control for 2 to 7 days before the true adoption dates. In these models, dummy variables that capture the time periods before the policy change should not be significant.5

5. Results

5.1. Event study models

I estimate Eq. (1) for the outcome variables for which daily data are available and report the results under different model specifications in Table 1, Table 2, Table 3, Table 4, Table 5 . My purpose here is to estimate the short-run effects of certain state-wide policy changes. In each table, five columns of results are presented for a given policy restriction: (1) results without considering COVID activity; (2) results taking into account COVID activity; (3) dropping New York given that it was an early COVID state; (4) also dropping California and Washington as early COVID states; (5) dropping states with Republican governors as potential slow responders. All results compare a daily result with the same date one year earlier.

Table 1.

Event study estimates: restrictions on mass gatherings.

	Percentage change in new listings					Percentage change in page views
No. of days to policy	(1)	(2)	(3)	(4)	(5)	(1)	(2)	(3)	(4)	(5)
-7	-0.031	-0.047**	-0.049**	-0.057**	0.005	-0.049	-0.078*	-0.065	-0.104***	0.010
	(0.020)	(0.022)	(0.022)	(0.022)	(0.043)	(0.033)	(0.042)	(0.040)	(0.025)	(0.055)
-6	-0.036**	-0.041**	-0.045**	-0.053***	-0.036	-0.021**	-0.031***	-0.027***	-0.036***	-0.007
	(0.018)	(0.016)	(0.017)	(0.016)	(0.028)	(0.008)	(0.011)	(0.010)	(0.008)	(0.022)
-5	-0.024	-0.028*	-0.031*	-0.038**	-0.028	-0.016**	-0.024***	-0.021**	-0.028***	-0.008
	(0.017)	(0.016)	(0.016)	(0.016)	(0.026)	(0.007)	(0.008)	(0.008)	(0.007)	(0.016)
-4	-0.007	-0.010	-0.013	-0.018	-0.014	-0.011**	-0.017**	-0.014**	-0.019***	-0.006
	(0.015)	(0.014)	(0.014)	(0.014)	(0.023)	(0.005)	(0.007)	(0.006)	(0.006)	(0.012)
-3	0.002	0.000	-0.001	-0.004	-0.006	-0.007*	-0.010**	-0.009*	-0.013***	-0.006
	(0.011)	(0.010)	(0.011)	(0.012)	(0.019)	(0.004)	(0.005)	(0.004)	(0.004)	(0.008)
-2	0.001	0.000	-0.000	-0.000	-0.005	-0.004*	-0.005**	-0.004**	-0.006**	-0.004
	(0.006)	(0.006)	(0.006)	(0.006)	(0.011)	(0.002)	(0.002)	(0.002)	(0.002)	(0.004)
0	-0.003	-0.002	-0.002	0.001	0.003	0.002	0.004**	0.004**	0.005**	0.005
	(0.005)	(0.005)	(0.005)	(0.005)	(0.008)	(0.002)	(0.002)	(0.002)	(0.002)	(0.003)
1	0.003	0.006	0.006	0.009	0.009	0.006*	0.010**	0.009**	0.011***	0.012*
	(0.009)	(0.008)	(0.009)	(0.009)	(0.017)	(0.003)	(0.004)	(0.003)	(0.003)	(0.006)
2	0.003	0.006	0.008	0.012	0.011	0.009	0.014**	0.012**	0.017***	0.017
	(0.014)	(0.013)	(0.014)	(0.015)	(0.028)	(0.005)	(0.006)	(0.006)	(0.005)	(0.010)
3	0.003	0.007	0.009	0.016	0.016	0.010	0.017**	0.015*	0.022***	0.021
	(0.018)	(0.017)	(0.017)	(0.019)	(0.037)	(0.007)	(0.008)	(0.008)	(0.006)	(0.014)
4	0.002	0.007	0.009	0.020	0.028	0.012	0.021**	0.019*	0.028***	0.029
	(0.021)	(0.020)	(0.021)	(0.022)	(0.045)	(0.009)	(0.010)	(0.010)	(0.008)	(0.018)
5	0.003	0.010	0.011	0.026	0.039	0.016	0.028**	0.024*	0.037***	0.037
	(0.025)	(0.024)	(0.025)	(0.027)	(0.055)	(0.012)	(0.013)	(0.013)	(0.009)	(0.022)
6	0.003	0.011	0.012	0.031	0.047	0.021	0.035**	0.031*	0.046***	0.045
	(0.029)	(0.028)	(0.028)	(0.031)	(0.067)	(0.015)	(0.016)	(0.016)	(0.011)	(0.027)
7	-0.026	-0.007	0.005	0.031	0.083	0.038	0.071	0.069	0.108***	0.083
	(0.067)	(0.061)	(0.061)	(0.068)	(0.128)	(0.046)	(0.043)	(0.045)	(0.032)	(0.076)
N	34,700	34,700	32,965	28,801	16,656	27,700	27,700	26,315	22,991	13,296
Daily new cases & deaths per capita	N	Y	Y	Y	Y	N	Y	Y	Y	Y
Drop NY	N	N	Y	Y	N	N	N	Y	Y	N
Drop NY, CA, WA	N	N	N	Y	N	N	N	N	Y	N
Drop states with Republican governor	N	N	N	N	Y	N	N	N	N	Y

Notes: The day prior the announcement of the policy is the excluded category. All models include city, state, and calendar date fixed effects. States that did not implement the policy are excluded from the sample. Standard errors are clustered at the state level and reported in parentheses. The signs *, **, and *** indicate statistical significance at 10, 5, and 1 percent levels.

Table 2.

Event study estimates: school closures.

	Percentage change in new listings					Percentage change in page views
No. of days to policy	(1)	(2)	(3)	(4)	(5)	(1)	(2)	(3)	(4)	(5)
-7	0.065	0.055	0.051	0.083	-0.008	0.083	0.064	0.060	0.075	-0.077
	(0.065)	(0.069)	(0.068)	(0.065)	(0.086)	(0.059)	(0.059)	(0.060)	(0.058)	(0.073)
-6	0.019	0.013	0.009	0.030	-0.004	0.057	0.046	0.043	0.050	-0.040
	(0.036)	(0.039)	(0.037)	(0.035)	(0.074)	(0.035)	(0.035)	(0.035)	(0.033)	(0.044)
-5	0.011	0.006	0.004	0.021	-0.011	0.047*	0.038	0.035	0.041	-0.032
	(0.029)	(0.031)	(0.030)	(0.028)	(0.066)	(0.028)	(0.028)	(0.028)	(0.025)	(0.037)
-4	0.001	-0.002	-0.003	0.013	-0.015	0.036*	0.029	0.028	0.032	-0.023
	(0.023)	(0.024)	(0.024)	(0.021)	(0.055)	(0.021)	(0.021)	(0.021)	(0.019)	(0.028)
-3	-0.001	-0.004	-0.003	0.008	-0.012	0.023*	0.019	0.018	0.020*	-0.015
	(0.015)	(0.016)	(0.016)	(0.014)	(0.040)	(0.013)	(0.013)	(0.013)	(0.012)	(0.018)
-2	-0.004	-0.006	-0.006	-0.000	-0.010	0.011*	0.009	0.008	0.009*	-0.008
	(0.008)	(0.009)	(0.009)	(0.008)	(0.021)	(0.006)	(0.006)	(0.006)	(0.005)	(0.009)
0	-0.000	0.001	0.001	-0.004	0.011	-0.010	-0.007	-0.007	-0.008	0.010
	(0.007)	(0.007)	(0.007)	(0.007)	(0.018)	(0.006)	(0.006)	(0.006)	(0.006)	(0.009)
1	0.004	0.007	0.007	-0.000	0.029	-0.017	-0.012	-0.011	-0.014	0.022
	(0.013)	(0.013)	(0.014)	(0.013)	(0.037)	(0.012)	(0.013)	(0.013)	(0.011)	(0.020)
2	0.004	0.008	0.009	0.003	0.038	-0.025	-0.018	-0.016	-0.021	0.034
	(0.019)	(0.020)	(0.020)	(0.020)	(0.057)	(0.018)	(0.018)	(0.018)	(0.015)	(0.031)
3	0.004	0.008	0.010	0.005	0.046	-0.034	-0.024	-0.022	-0.031	0.046
	(0.026)	(0.026)	(0.026)	(0.027)	(0.077)	(0.024)	(0.024)	(0.024)	(0.020)	(0.042)
4	-0.000	0.006	0.008	0.004	0.049	-0.039	-0.027	-0.025	-0.039	0.064
	(0.035)	(0.035)	(0.036)	(0.037)	(0.101)	(0.031)	(0.031)	(0.031)	(0.025)	(0.054)
5	-0.004	0.003	0.008	0.006	0.048	-0.043	-0.029	-0.026	-0.044	0.082
	(0.044)	(0.044)	(0.044)	(0.045)	(0.124)	(0.037)	(0.037)	(0.036)	(0.029)	(0.067)
6	-0.009	-0.001	0.007	0.010	0.040	-0.047	-0.031	-0.027	-0.049	0.096
	(0.052)	(0.052)	(0.051)	(0.053)	(0.147)	(0.044)	(0.043)	(0.042)	(0.035)	(0.078)
7	-0.018	-0.006	0.005	0.002	0.055	-0.062	-0.038	-0.033	-0.067	0.138
	(0.075)	(0.075)	(0.074)	(0.078)	(0.208)	(0.066)	(0.065)	(0.063)	(0.054)	(0.107)
N	34,700	34,700	32,965	28,801	16,656	27,700	27,700	26,315	22,991	13,296
Daily new cases & deaths per capita	N	Y	Y	Y	Y	N	Y	Y	Y	Y
Drop NY	N	N	Y	Y	N	N	N	Y	Y	N
Drop NY, CA, WA	N	N	N	Y	N	N	N	N	Y	N
Drop states with Republican governor	N	N	N	N	Y	N	N	N	N	Y

Notes: The day prior the announcement of the policy is the excluded category. All models include city, state, and calendar date fixed effects. States that did not implement the policy are excluded from the sample. Standard errors are clustered at the state level and reported in parentheses. The sign * indicates statistical significance at 10 percent level.

Table 3.

Event study estimates: closure of non-essential businesses.

	Percentage change in new listings					Percentage change in page views
No. of days to policy	(1)	(2)	(3)	(4)	(5)	(1)	(2)	(3)	(4)	(5)
-7	0.075	0.060	0.056	0.032	0.175***	0.094***	0.057	0.065	-0.008	0.127**
	(0.048)	(0.049)	(0.051)	(0.062)	(0.046)	(0.032)	(0.054)	(0.051)	(0.039)	(0.049)
-6	0.044	0.037	0.036	0.033	0.095**	0.059***	0.041	0.043*	0.010	0.085***
	(0.030)	(0.030)	(0.031)	(0.038)	(0.033)	(0.018)	(0.026)	(0.024)	(0.023)	(0.023)
-5	0.031	0.025	0.023	0.024	0.068**	0.049***	0.034	0.036	0.009	0.071***
	(0.025)	(0.025)	(0.026)	(0.031)	(0.028)	(0.015)	(0.022)	(0.021)	(0.019)	(0.021)
-4	0.019	0.014	0.011	0.014	0.040*	0.039***	0.027	0.028	0.009	0.057***
	(0.019)	(0.019)	(0.019)	(0.023)	(0.022)	(0.012)	(0.017)	(0.016)	(0.015)	(0.018)
-3	0.011	0.008	0.007	0.008	0.026*	0.028***	0.020	0.021*	0.007	0.040***
	(0.012)	(0.011)	(0.011)	(0.015)	(0.014)	(0.008)	(0.012)	(0.012)	(0.010)	(0.012)
-2	0.005	0.004	0.004	0.003	0.015**	0.014***	0.011	0.011	0.003	0.021***
	(0.006)	(0.005)	(0.006)	(0.007)	(0.006)	(0.005)	(0.006)	(0.006)	(0.005)	(0.006)
0	-0.015	-0.013	-0.015	-0.015	-0.031***	-0.016***	-0.012*	-0.014**	-0.005	-0.023***
	(0.009)	(0.009)	(0.009)	(0.011)	(0.009)	(0.005)	(0.007)	(0.007)	(0.005)	(0.008)
1	-0.036**	-0.032**	-0.032**	-0.035*	-0.070***	-0.034***	-0.026*	-0.029**	-0.010	-0.047**
	(0.015)	(0.014)	(0.015)	(0.019)	(0.018)	(0.009)	(0.014)	(0.013)	(0.009)	(0.016)
2	-0.059***	-0.055**	-0.051**	-0.057*	-0.108***	-0.051***	-0.039*	-0.042**	-0.012	-0.070**
	(0.021)	(0.020)	(0.021)	(0.027)	(0.031)	(0.014)	(0.021)	(0.020)	(0.014)	(0.024)
3	-0.088***	-0.082***	-0.074***	-0.078**	-0.153***	-0.066***	-0.050*	-0.053*	-0.011	-0.094**
	(0.028)	(0.027)	(0.026)	(0.036)	(0.044)	(0.020)	(0.029)	(0.028)	(0.019)	(0.033)
4	-0.113***	-0.105***	-0.092**	-0.103**	-0.195***	-0.081***	-0.062*	-0.062*	-0.015	-0.113**
	(0.039)	(0.037)	(0.034)	(0.045)	(0.061)	(0.023)	(0.033)	(0.032)	(0.022)	(0.038)
5	-0.131**	-0.122**	-0.105**	-0.119**	-0.229***	-0.096***	-0.072*	-0.073*	-0.020	-0.128**
	(0.048)	(0.045)	(0.041)	(0.055)	(0.074)	(0.025)	(0.037)	(0.036)	(0.026)	(0.044)
6	-0.156**	-0.145**	-0.125**	-0.143**	-0.260***	-0.109***	-0.082**	-0.082**	-0.025	-0.145***
	(0.056)	(0.053)	(0.047)	(0.062)	(0.086)	(0.028)	(0.040)	(0.039)	(0.031)	(0.047)
7	-0.248***	-0.230***	-0.205***	-0.219***	-0.344***	-0.180***	-0.135***	-0.133**	-0.064	-0.205***
	(0.067)	(0.059)	(0.049)	(0.071)	(0.108)	(0.037)	(0.047)	(0.048)	(0.049)	(0.044)
N	23,249	23,249	21,514	17,350	15,268	18,559	18,559	17,174	13,850	12,188
Daily new cases & deaths per capita	N	Y	Y	Y	Y	N	Y	Y	Y	Y
Drop NY	N	N	Y	Y	N	N	N	Y	Y	N
Drop NY, CA, WA	N	N	N	Y	N	N	N	N	Y	N
Drop states with Republican governor	N	N	N	N	Y	N	N	N	N	Y

Notes: The day prior the announcement of the policy is the excluded category. All models include city, state, and calendar date fixed effects. States that did not implement the policy are excluded from the sample. Standard errors are clustered at the state level and reported in parentheses. The signs *, **, and *** indicate statistical significance at 10, 5, and 1 percent levels.

Table 4.

Event study estimates: Any business closures.

	Percentage change in new listings					Percentage change in page views
No. of days to policy	(1)	(2)	(3)	(4)	(5)	(1)	(2)	(3)	(4)	(5)
-7	-0.016	-0.041	-0.049*	-0.046*	-0.055	0.003	-0.046	-0.029	-0.032	-0.127
	(0.027)	(0.026)	(0.026)	(0.027)	(0.064)	(0.035)	(0.044)	(0.041)	(0.042)	(0.114)
-6	-0.003	-0.016	-0.022	-0.019	-0.043	0.008	-0.018	-0.009	-0.012	-0.076
	(0.019)	(0.019)	(0.018)	(0.020)	(0.047)	(0.020)	(0.025)	(0.023)	(0.022)	(0.087)
-5	-0.000	-0.011	-0.016	-0.010	-0.042	0.007	-0.013	-0.006	-0.008	-0.058
	(0.018)	(0.017)	(0.017)	(0.018)	(0.042)	(0.016)	(0.020)	(0.018)	(0.018)	(0.073)
-4	0.010	0.002	-0.002	0.004	-0.031	0.008	-0.006	-0.002	-0.003	-0.040
	(0.014)	(0.014)	(0.013)	(0.014)	(0.034)	(0.012)	(0.015)	(0.014)	(0.013)	(0.056)
-3	0.012	0.007	0.003	0.008	-0.019	0.007	-0.003	-0.001	-0.003	-0.022
	(0.011)	(0.011)	(0.010)	(0.011)	(0.024)	(0.008)	(0.009)	(0.009)	(0.009)	(0.037)
-2	0.012*	0.009	0.007	0.007	-0.006	0.004	-0.001	0.000	-0.001	-0.011
	(0.006)	(0.006)	(0.006)	(0.006)	(0.012)	(0.004)	(0.005)	(0.004)	(0.004)	(0.018)
0	-0.004	-0.001	-0.001	-0.000	0.021**	-0.006*	-0.001	-0.002	-0.001	0.011
	(0.004)	(0.005)	(0.005)	(0.006)	(0.007)	(0.003)	(0.004)	(0.004)	(0.004)	(0.019)
1	-0.004	0.001	0.002	0.006	0.040**	-0.009	0.001	-0.001	0.000	0.028
	(0.009)	(0.009)	(0.010)	(0.012)	(0.017)	(0.006)	(0.008)	(0.007)	(0.007)	(0.038)
2	-0.002	0.006	0.007	0.011	0.060**	-0.013	0.003	-0.001	0.000	0.047
	(0.012)	(0.013)	(0.014)	(0.015)	(0.025)	(0.009)	(0.011)	(0.010)	(0.010)	(0.057)
3	-0.010	0.000	0.001	0.006	0.067*	-0.017	0.003	-0.001	-0.001	0.070
	(0.016)	(0.016)	(0.017)	(0.018)	(0.034)	(0.011)	(0.014)	(0.013)	(0.013)	(0.077)
4	-0.019	-0.006	-0.005	-0.001	0.070	-0.020	0.006	0.001	-0.000	0.098
	(0.019)	(0.019)	(0.020)	(0.021)	(0.045)	(0.014)	(0.017)	(0.016)	(0.016)	(0.098)
5	-0.035	-0.018	-0.017	-0.012	0.068	-0.022	0.009	0.004	0.002	0.125
	(0.024)	(0.024)	(0.026)	(0.028)	(0.055)	(0.018)	(0.021)	(0.020)	(0.019)	(0.118)
6	-0.039	-0.020	-0.019	-0.014	0.079	-0.024	0.013	0.006	0.004	0.151
	(0.028)	(0.028)	(0.030)	(0.033)	(0.064)	(0.022)	(0.025)	(0.023)	(0.023)	(0.136)
7	-0.092*	-0.060	-0.046	-0.048	0.086	-0.050	0.009	0.002	-0.005	0.196
	(0.046)	(0.044)	(0.045)	(0.048)	(0.092)	(0.038)	(0.039)	(0.038)	(0.038)	(0.171)
N	34,700	34,700	32,965	28,801	16,656	27,700	27,700	26,315	22,991	13,296
Daily new cases & deaths per capita	N	Y	Y	Y	Y	N	Y	Y	Y	Y
Drop NY	N	N	Y	Y	N	N	N	Y	Y	N
Drop NY, CA, WA	N	N	N	Y	N	N	N	N	Y	N
Drop states with Republican governor	N	N	N	N	Y	N	N	N	N	Y

Notes: The day prior the announcement of the policy is the excluded category. All models include city, state, and calendar date fixed effects. States that did not implement the policy are excluded from the sample. Standard errors are clustered at the state level and reported in parentheses. The signs *, **, and *** indicate statistical significance at 10, 5, and 1 percent levels.

Table 5.

Event study estimates: Stay at home orders.

	Percentage change in new listings					Percentage change in page views
No. of days to policy	(1)	(2)	(3)	(4)	(5)	(1)	(2)	(3)	(4)	(5)
-7	0.008	0.002	-0.011	-0.047	0.019	0.006	-0.019	-0.010	-0.066*	0.040
	(0.031)	(0.030)	(0.035)	(0.052)	(0.099)	(0.040)	(0.049)	(0.048)	(0.036)	(0.073)
-6	-0.014	-0.016	-0.024	-0.043	-0.030	0.010	0.000	0.001	-0.028	0.019
	(0.024)	(0.023)	(0.027)	(0.038)	(0.072)	(0.017)	(0.021)	(0.022)	(0.019)	(0.040)
-5	-0.016	-0.018	-0.025	-0.041	-0.036	0.010	0.002	0.002	-0.022	0.017
	(0.021)	(0.020)	(0.023)	(0.032)	(0.060)	(0.014)	(0.017)	(0.018)	(0.015)	(0.034)
-4	-0.016	-0.018	-0.023	-0.033	-0.038	0.008	0.002	0.002	-0.017	0.015
	(0.016)	(0.016)	(0.018)	(0.025)	(0.047)	(0.011)	(0.013)	(0.014)	(0.011)	(0.026)
-3	-0.016	-0.017	-0.020	-0.026	-0.033	0.008	0.004	0.004	-0.009	0.013
	(0.011)	(0.011)	(0.012)	(0.016)	(0.033)	(0.008)	(0.009)	(0.010)	(0.007)	(0.019)
-2	-0.006	-0.007	-0.008	-0.010	-0.013	0.006	0.004	0.004	-0.003	0.007
	(0.007)	(0.007)	(0.007)	(0.008)	(0.017)	(0.004)	(0.005)	(0.005)	(0.003)	(0.010)
0	-0.001	-0.001	-0.000	0.001	0.002	-0.005	-0.003	-0.004	0.002	-0.006
	(0.007)	(0.007)	(0.008)	(0.009)	(0.017)	(0.004)	(0.005)	(0.005)	(0.004)	(0.010)
1	-0.008	-0.007	-0.003	-0.002	-0.009	-0.008	-0.004	-0.005	0.008	-0.010
	(0.012)	(0.012)	(0.013)	(0.015)	(0.032)	(0.008)	(0.009)	(0.010)	(0.007)	(0.021)
2	-0.016	-0.014	-0.005	-0.007	-0.014	-0.012	-0.007	-0.007	0.012	-0.016
	(0.016)	(0.016)	(0.016)	(0.020)	(0.045)	(0.012)	(0.014)	(0.015)	(0.011)	(0.030)
3	-0.032	-0.030	-0.016	-0.019	-0.032	-0.017	-0.009	-0.008	0.018	-0.024
	(0.022)	(0.022)	(0.020)	(0.026)	(0.060)	(0.016)	(0.019)	(0.020)	(0.014)	(0.040)
4	-0.044	-0.042	-0.023	-0.028	-0.046	-0.021	-0.011	-0.009	0.022	-0.027
	(0.027)	(0.027)	(0.024)	(0.032)	(0.072)	(0.019)	(0.022)	(0.023)	(0.018)	(0.048)
5	-0.058*	-0.055*	-0.033	-0.034	-0.065	-0.022	-0.010	-0.008	0.028	-0.029
	(0.031)	(0.032)	(0.027)	(0.037)	(0.085)	(0.022)	(0.026)	(0.027)	(0.021)	(0.058)
6	-0.073**	-0.069*	-0.044	-0.044	-0.076	-0.021	-0.006	-0.004	0.036	-0.028
	(0.035)	(0.035)	(0.029)	(0.041)	(0.095)	(0.025)	(0.029)	(0.030)	(0.025)	(0.065)
7	-0.142**	-0.133**	-0.093**	-0.087	-0.084	-0.046	-0.018	-0.007	0.046	-0.028
	(0.057)	(0.056)	(0.045)	(0.059)	(0.152)	(0.049)	(0.048)	(0.050)	(0.047)	(0.111)
N	29,148	29,148	27,413	23,249	14,921	23,268	23,268	21,883	18,559	11,911
Daily new cases & deaths per capita	Y	N	N	Y	Y	Y	Y	Y	Y	Y
Drop NY	N	N	Y	Y	N	N	N	Y	Y	N
Drop NY, CA, WA	N	N	N	Y	N	N	N	N	Y	N
Drop states with Republican governor	N	N	N	N	Y	N	N	N	N	Y

Notes: The day prior the announcement of the policy is the excluded category. All models include city, state, and calendar date fixed effects. States that did not implement the policy are excluded from the sample. Standard errors are clustered at the state level and reported in parentheses. The signs * and ** indicate statistical significance at 10 and 5 percent levels.

Table 1 indicates that restrictions of mass gatherings did not have a significant impact on the percentage change in new listings relative to the same period in 2019. In particular, while the estimates suggest that introduction of this restriction was associated with a small (0.3 to 0.9 percentage points) increase in this impact, this result is statistically insignificant. This result continues to hold during the following week. Table 1 also does not provide evidence of effects of restrictions on mass gatherings on web traffic to for sale homes. In this case, there are positive and increasing results after imposition of the policy (specifications 2 through 4). However, the coefficient estimates for the previous week are also increasing. I thus cannot rule out the possibility that this is a case of a continuation of pre-existing trends rather than effects of the policy itself.

Table 2 demonstrates that school closures did not have a significant impact on the percentage change in new listings and page views to for sale homes.

The first specification in Table 3 indicates that the effects of closure of non-essential businesses on percentage change in new listings are negative and highly significant. In particular, column (1) indicates that one day after this policy adoption date, there was about a 3.6 percentage point decrease in the change in new listings relative to the same time period in the previous year. This effect grew during the week: after one week, this number was down by about 24.8 percentage points relative to its level in 2019. This effect is large, representing about a 1.2 standard deviation change from the mean of this variable. Importantly, there were no effects in the time periods before the adoption date. Results reported in specifications (2) – (4) are quite similar. Inclusion of total cases and deaths per capita (2), exclusion of New York (3), or inclusion of the first states where cases were seen (California and Washington) do not change the basic result. Specification 5 of Table 3, where states with a Republican governor are excluded from the sample, shows what appear to be substantive post-policy effects. However, for this particular specification, coefficient estimates for the pre-policy period are also larger but positive. They decline throughout the pre-policy period, implying that the post-policy decreases could well be a part of a continuing trend and not a response to the policy change.

The results presented in the second part Table 3, columns (2) and (3), show evidence of a negative impact of the closure of non-essential businesses on the percent change in web traffic to for-sale homes. For specifications 2 and 3, the closure of non-essential businesses led to a 13.3 to 13.5 percentage point decrease in page views to for-sale homes relative to the same period in 2019 one week after the policy adoption date (a 0.65 standard deviation change from the mean of this variable). Column (4) (NY, CA, and WA dropped from sample) shows no statistically significant evidence. Interpretation of column (5) is more complicated. While there are statistically significant positive coefficients estimates in column (5) for pre-closure, they are positive. This could reflect underlying change from the prior year, or perhaps some anticipation of closures.

Tables 4 and 5 present the effects of any business closures and stay at home orders on the percentage change in new listings and page views to for-sale homes. Here, I find no significant effects.

5.2. Difference-in-differences models

In this section, I report the results from difference-in-differences models which I used to estimate the impact of state policies on several additional indicators of housing market activity, while controlling for the spread of the coronavirus over time. Table 6 reports the estimated effects on (1) the percentage change in total inventory and (2) the number of newly pending sales relative to the same time period in the previous year. The results show that none of these policies had a statistically significant impact on the change in newly pending sales.

Table 6.

The effect of state level policies on total inventory and newly pending sales.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
A. Percentage change in total	inventory
Ban on mass gatherings	0.000	-0.008	0.003	-0.006	-0.009	-0.016	0.006
	(0.024)	(0.024)	(0 0.025)	(0.024)	(0.024)	(0.022)	(0.024)
School closures	-0 046	-0.057*	-0.050*	-0.058*	-0.056*	-0.055*	-0.091*
	(0.028)	(0.030)	(0 0.027)	(0.029)	(0.031)	(0.033)	(0.044)
Closure of any business	0.024	0.022	0.024	0.023	0.020	0.009	0.002
	(0.027)	(0.025)	(0.027)	(0.026)	(0.026)	(0.025)	(0.033)
Closure of non-essential business	-0.035***	-0.027***	-0.034***	-0.028***	-0.024**	-0.021	-0.025*
	(0.011)	(0.010)	(0 0.011)	(0.009)	(0.010)	(0.016)	(0.014)
Stay at home order	0.007	0.009	0.006	0.008	0.014	-0.030	0.039
	(0.034)	(0.032)	(0 0.032)	(0.031)	(0.032)	(0.018)	(0.030)
Total cases per 10,000		-0.000**		-0.000*	-0.000*	-0.000	-0.000
		(0.000)		(0.000)	(0.000)	(0.000)	(0.000)
Total deaths per 10,000			-0.004	-0.001
			(0 0.003)	(0 0.003)
N	4719	4719	4719	4719	4469	3919	2300
B. Percentage change in newly pending sales
Ban on mass gatherings	-0.010	-0.022	-0.009	-0.028	-0.023	-0.039	0.009
	(0.047)	(0.050)	(0.047)	(0.050)	(0.053)	(0.060)	(0.053)
School closures	0.032	0.002	0.031	0.001	0.005	0.006	0.155**
	(0.072)	(0.064)	(0 0.075)	(0.058)	(0.061)	(0.058)	(0.069)
Closure of any business	0.105	0.092	0.107	0.085	0.091	0.080	0.303***
	(0.065)	(0.068)	(0.065)	(0.067)	(0.070)	(0.077)	(0.099)
Closure of non-essential business	-0.079	-0.053	-0.079	-0.051	-0.043	-0.127	-0.024
	(0.074)	(0.065)	(0.074)	(0.064)	(0.065)	(0.098)	(0.058)
Stay at home order	-0.092	-0.087	-0.094	-0.083	-0.084	-0.141	-0.101
	(0.074)	(0.068)	(0 0.075)	(0.070)	(0.075)	(0.105)	(0.082)
Total cases per 10,000		-0.001**		-0.001**	-0.001**	-0.001**	-0.001***
		(0.000)		(0.000)	(0.000)	(0.000)	(0.000)
Total deaths per 10,000			-0.003	0.006
			(0.007)	(0.008)
N	3333	3333	3333	3333	3083	2633	1838
Drop NY	N	N	N	N	Y	N	N
Drop NY, WA, and CA	N	N	N	N	N	Y	N
Drop states with Republican Ngovemors	N	N	N	N	N	N	Y

Notes: All models include city, state, and calendar date fixed effects. Standard errors are clustered at the state level and reported in parentheses. The signs *, **, and *** indicate statistical significance at 10, 5, and 1 percent levels.

On the other hand, Table 6 indicates that school closures and closure of non-essential businesses had a statistically significant and negative impact on the percentage change in total inventory. School closures led to a 5 to 9.1 percentage point decrease in this inventory measure relative to the same period in the previous year. This effect is robust under alternative models and large (0.3 to 0.7 standard deviation change from the mean of this variable) – although significance comes at a marginal 10% level. Closure of non-essential businesses is associated with up to a 3.5 percentage point decrease in the total inventory, compared with the same time period in the previous year.

In all models presented in Table 6, inclusion of controls for the spread of the coronavirus has only a modest impact on the change in newly pending sales: a one-point increase in the total number of cases per 10,000 people at the state level decreases the change in newly pending sales by only 0.1 percentage points. Table 6 also indicates that the control variables have no significant impact on the total inventory.

Table 7 reports the results from a similar DD analysis for the percentage change in new listings and web traffic to for-sale homes. Since daily data are available for these outcome variables, the number of observations is large, yielding more precise estimates that can be compared to event study results. The benchmark model results reported in the first specification of panel A show that closure of non-essential businesses is associated with a 11 percentage point decrease in percentage change in newly listed homes relative to the same time period in 2019. This effect is consistent with the event study results presented in Table 3. The estimated impact of this closure policy on the change in new listings is also robust under alternative model specifications. Except for specification 4, the coefficients on variables that control for the spread of the coronavirus are relatively small and not statistically significant.

Table 7.

The effect of state level policies on new listings and web traffic to for-sale homes.

A. Percentage change in new litings	(1)	(2)	(3)	(4)	(5)	(6)	(7)
Ban on mass gatherings	0.043	0.037	0.039	0028	0.035	0.031	0.042
	(0.027)	(0.027)	(0.026)	(0 0.026)	(0.027)	(0 0.026)	(0.034)
School closures	0.022	0.013	0.029	0.019	0015	0.011	0.023
	(0.025)	(0.026)	(0.025)	(0.027)	(0.027)	(0.028)	(0.029)
Closure of any business	0.029	0.029	0.029	0028	0028	0.022	0.033
	(0.029)	(0.027)	(0.028)	(0.026)	(0.028)	(0 0.028)	(0.049)
Closure of non-essential business	-0.110***	-0.105***	-0.110***	-0.102***	-0.095***	-0.115***	-0.090***
	(0.025)	(0.025)	(0.026)	(0.025)	(0.024)	(0.038)	(0.025)
Stay at home order	0.033	0.034	0.035	0.039	0.042	0.005	0.060**
	(0.029)	(0.027)	(0.03 1)	(0.029)	(0.027)	(0 0.027)	(0.026)
Total cases per 10,000		-0.000		-0.000*	-0.000	-0.000	-0.000
		(0.000)		(0 0.000)	(0.000)	(0 0.000)	(0.000)
Total deaths per 10,000			0.005	0.008*
			(0.004)	(0.005)
N	34,700	34,700	34,700	34,700	32,965	28,801	16,656
B. Percentage change in page views
Ban on mass gatherings	0.035*	0.031*	0.033*	0.024	0.024	0.016	0.027
	(0.020)	(0.018)	(0.020)	(0.017)	(0017)	(0.016)	(0.036)
School closures	0019	0.011	0.035	0025	0.016	0.016	0.059**
	(0.026)	(0.023)	(0.026)	(0.022)	(0.024)	(0 0.024)	(0.023)
Closure of any business	0.005	0.005	0.008	0.008	-0.002	-0.006	-0.040
	(0.036)	(0.034)	(0.035)	(0 0.030)	(0.035)	(0 0.034)	(0.063)
Closure of non-essential business	-0.054	-0.049	-0.048	-0 036	-0 033	-0.076*	-0.045
	(0.045)	(0.043)	(0.042)	(0.038)	(0.039)	(0.042)	(0.048)
Stay at home order	-0.008	-0 006	-0 009	-0 004	0.013	0.012	-0 004
	(0.029)	(0.028)	(0.025)	(0.022)	(0.025)	(0 0.031)	(0.041)
Total cases per 10,000		-0.000		-0.000**	-0.000	-0.000	-0.000
		(0.000)		(0.000)	(0.000)	(0.000)	(0.000)
Total deaths per 10,000			0.012***	0.016***
			(0.004)	(0.004)
N	27,700	27,700	27,700	27,700	26,315	22,991	13,296
Drop NY	N	N	N	N	Y	N	N
Drop NY, WA, and CA	N	N	N	N	N	Y	N
Drop states with Republican governors	N	N	N	N	N	N	Y

Notes: All models include city, state, and calendar date fixed effects. Standard errors are clustered at the state level and reported in parentheses. The signs *, **, and *** indicate statistical significance at 10, 5, and 1 percent levels.

Table 7 also shows that consistent with the results from the event study analysis, none of the policies that were introduced to mitigate the spread of the coronavirus had a statistically significant impact on the web traffic to for-sale homes. Furthermore, the effect of the other control variables on this outcome remains relatively small and statistically insignificant in most of the models.

In Table 8 , the results reported in panel A indicate that while most of the state level policies did not have a significant impact on the median list price, school closures had a positive impact on this outcome. In particular, this policy is associated with approximately 2 percentage point increase in change median list price relative to the same time period in the previous year. This effect corresponds to a 0.33 standard deviation change from the mean of this variable. The positive impact of school closures on housing prices is surprising but can possibly be explained by the timing of the introduction of this policy. School closures were adopted by the majority of states between March 13 and 20. During that period, housing prices were increasing in the United States (appendix Figs. A1 , panel B). Therefore, the coefficient on this policy could possibly be capturing the positive trend in housing prices during that time rather that the true effect of the policy. However, since daily data are not available for median list prices, it is not possible to estimate an event study model to formally test this hypothesis.

Table 8.

The effect of state level policies on median list price and days to pending sale status.

A. Percentage change in median list price	(1)	(2)	(3)	(4)	(5)	(6)	(7)
Ban on mass gatherings	-0.011	-0.008	-0 013	-0.012	-0.010	-0 007	-0.019 *
	(0.012)	(0.011)	(0.012)	(0.011)	(0.011)	(0.011)	(0.009)
School closures	0.020**	0.023**	0.024***	0.025***	0.023**	0.022**	0.021*
	(0.008)	(0.009)	(0.007)	(0.008)	(0.009)	(0 0.009)	(0.010)
Closure of any business	0.009	0.010	0.009	0.009	0.009	0.011	0.018
	(0.012)	(0.011)	(0.011)	(0 0.011)	(0.012)	(0 0.012)	(0.011)
Closure of non-essential business	-0.004	-0 006	-0 004	-0 006	-0.008	0.004	-0.017*
	(0.007)	(0.007)	(0.007)	(0.006)	(0.007)	(0.006)	(0.008)
Stay at home order	-0.002	-0.003	-0.001	-0.001	-0.003	0.008	-0.010
	(0.008)	(0.007)	(0.006)	(0.006)	(0.008)	(0 0.006)	(0.010)
Total cases per 10,000		0.000*		0.000	0.000	0.000	0.000
		(0.000)		(0.000)	(0.000)	(0.000)	(0.000)
Total deaths per 10,000			0.003***	0.003*
			(0.001)	(0.002)
N	4750	4750	4750	4750	4500	3950	2300
B. Average number of days to pending sale status
Ban on mass gatherings	0.380	0.253	1.013	1.292	0.898	0.459	3.206**
	(1600)	(1.632)	(1.330)	(1285)	(1.487)	(1.509)	(1.380)
School closures	3.969*	3.798*	2.921	3.151*	3.852*	4.001*	-4.123
	(2.057)	(2.057)	(1.892)	(1.848)	(1.965)	(2.015)	(2.404)
Closure of any business	0.442	0.422	0.612	0.661	0.781	0.498	0.612
	(1.595)	(1.631)	(1.395)	(1.324)	(1.589)	(1.656)	(2.805)
Closure of non-essential business	0.123	0.248	0.288	0.074	0.164	-1.308	2 0.530***
	(0.839)	(0.838)	(0.825)	(0.836)	(0.862)	(0 0.775)	(0.739)
Stay at home order	1.569	1.579	1062	1.003	1.173	0.029	3.134*
	(1133)	(1.115)	(0.817)	(0.804)	(1.067)	(0 0.918)	(1.735)
Total cases per 10,000		-0.004		0.008	-0.002	0.001	-0.004
		(0.008)		(0 0.007)	(0.008)	(0 0.008)	(0.013)
Total deaths per 10,000			-0.863***	-0.930***
			(0.190)	(0.206)
N	4624	4624	4624	4624	4379	3829	2240
Drop NY	N	N	N	N	y	N	N
Drop NY, WA, and CA	N	N	N	N	N	Y	N
Drop states with Republican governors	N	N	N	N	N	N	Y

Notes: All models include city, state, and calendar date fixed effects. Standard errors are clustered at the state level and reported in parentheses. The signs *, **, and *** indicate statistical significance at 10, 5, and 1 percent levels.

Nonetheless, I further investigate this possibility by estimating a model similar to the one presented in the fourth column of Table 8. In this model, I replace the actual school closures and other state level policy dummies with placebo dummies which take the value of one at one period (week) before the actual adoption date of the policy. If the coefficients on the school closures dummy in panel A of Table 8 represent the true effect of this policy on the percentage change in the median list prices, the coefficient on the placebo dummy should be insignificant. This is not the case, however. The coefficient on the placebo dummy is 0.025 (p<0.008), which is virtually the same as the coefficient presented in the fourth specification of Table 8. This provides further indication that the coefficients on the school closures dummy in panel A of Table 8 may be capturing the trend rather than representing the true effect of the policy itself.

The results in panel B of Table 8 indicate that under certain models, school closures are associated with up to a 4-day increase in average number of days to pending sale status relative to the same period in 2019. This effect corresponds to 0.3 standard deviation change from the mean of this variable, but marginally significant at 10 percent level. On the other hand, none of the remaining state level policies had a significant impact on the duration of the number of days to pending sale status.

6. Conclusion

This article explores the effects of policies adopted early in the COVID pandemic to mitigate the spread of the virus on numerous measures of housing market activity. From mid-March to early April 2020, a majority of U.S. states adopted such policies. Using daily and weekly data from 100 major U.S. metropolitan areas from January 1 to December 12, 2020, this paper provides the first estimates of the short run impact of these policies on various indicators of housing market activity in a large number of U.S. metropolitan areas.

Examples of these policies include shutting down certain businesses deemed non-essential, closing schools, ordering people to stay-at-home, and imposing restrictions on large gatherings. The changes in housing market activity covered are the change in new listings, total inventory, newly pending sales, median list price, web traffic to for-sale homes, and average number of days to pending sale status. The results presented here can help us understand how the state level policies played out, along with providing evidence that could be relevant to other times and other places.

Estimates from the event study and difference-in-differences models suggest that the response of housing market activity was complex. I find that the strongest policy effect stemmed from the closure of non-essential businesses. The closure of non-essential businesses was associated with up to an 11 percentage point decrease in new home listings and a 3.5 percentage point decrease in total inventory relative to the same time period in 2019. This result is not surprising since majority of homes in the United States are sold by real estate agents who were considered as non-essential workers in many states that adopted the closure of non-essential businesses policy. Some states such as New York later relaxed their policy and allowed real estate agents to conduct their business. In reality, however, during the early months of the pandemic, real estate agents were not allowed to meet with their clients in person and only allowed to show properties virtually, especially if the state had a stay-at-home order in place. An increasing trend in the web traffic to for-sale listings after the majority of state level policies were adopted by late March 2020 may also be explained by this policy and stay-at-home orders.

Under certain specifications, I show that school closures led to a 5 to 9.1 percentage point decrease in the change in total inventory relative to the same period in the previous year. This effect is economically large but marginally statistically significant. School closures also led to an increase in the average number of days to pending sale.

Future research should investigate the long-run effects of COVID-19 on U.S. housing market activity when housing market data for the extended post-pandemic period become available.

Appendix

Table A1, Table A2, Table A3

Table A1.

Policy adoption and end dates for sample states.

State	Mass gatherings ban	School closures	Closure of non-essential business	Any business closures	Stay at home orders
AL	3/19	3/19	3/28 - 4/30	3/19 - 6/15	4/4 - 4/30
AR	3/27 - 6/18	3/17 - 8/24		3/19
AZ	3/30 - 5/16	3/16		3/30 - 5/16	3/30 - 5/16
CA	3/11	3/19	3/19 - 9/2	3/19	3/19
CO	3/20	3/23	3/26 - 5/9	3/17	3/26 - 5/9
CT	3/12	3/17	3/23 - 5/20	3/16
DC	3/13	3/16	3/25 - 5/29	3/16	3/30 - 5/29
FL	4/3 - 6/5	3/17 - 10/5		3/17 - 9/25	4/3 - 5/18
GA	3/24	3/18		3/24	4/3
HI	3/17	3/19	3/25 - 5/1	3/17	3/25 - 6/10
IA	3/17 - 6/12	4/4	3/17 - 5/8	3/17 - 10/16
ID	3/25 - 5/1	3/23	3/25 - 5/1	3/25 - 6/13	3/25 - 5/1
IL	3/13	3/17	3/21 - 5/1	3/16	3/21 - 5/29
IN	3/12	3/19	3/24 - 5/18	3/16 - 9/26	3/25 - 5/18
KS	3/17 - 5/22	3/17		3/30 - 6/8	3/30 - 5/4
KY	3/19	3/20	3/26 - 5/11	3/16 - 6/29
LA	3/13 - 5/15	3/16	3/22 - 5/1	3/17	3/23 - 5/15
MA	3/13	3/17	3/24 - 5/18	3/17
MD	3/16 - 6/10	3/16	3/23 - 5/15	3/16	3/30 - 5/15
MI	3/13	3/16	3/23 - 5/7	3/16	3/24 - 6/1
MN	3/28	3/18		3/17 - 6/10	3/28 - 5/18
MO	3/23 - 5/4	3/23		3/23 - 6/16	4/6 - 5/15
MS	3/24	3/19	4/3 - 4/27	3/24 - 6/1	4/3 - 4/27
NC	3/14	3/14	3/30 - 5/8	3/17	3/30 - 5/8
NE	3/16	4/2		3/19 - 7/6
NM	3/12	3/13	3/24 - 5/15	3/16
NV	3/24	3/16	3/21 - 5/9	3/18	3/31 - 5/9
NY	3/12	3/18	3/22 - 6/8	3/16	3/22 - 6/8
OH	3/12	3/16	3/23 - 5/4	3/15	3/23 - 5/20
OK	3/24 - 5/24	3/17	4/1 - 4/24	4/1 - 6/1
OR	3/12	3/16		3/17	3/23 - 6/19
PA	4/1 - 9/14	3/17	3/23 - 5/8	3/18 - 7/3	4/1 - 6/5
RI	3/17	3/16		3/17	3/28 - 5/9
SC	3/18	3/16		3/18 - 8/3	4/7 - 5/4
TN	3/23	3/20	4/1 - 5/26	3/23	4/2 - 5/26
TX	3/21 - 6/4	3/19		3/21	4/2 - 5/1
UT	3/19 - 5/1	3/16		3/19
VA	3/15	3/16	3/24 - 5/15	3/17	3/30 - 6/5
WA	3/11	3/13	3/25 - 7/3	3/16	3/23 - 7/3
WI	3/17	3/18	3/25 - 5/11	3/17	3/25 - 5/13

Table A2.

List of sample housing markets.

Akron, OH	Daytona Beach, FL+	Melbourne, FL*	Salt Lake City, UT+
Albany, NY	Denver, CO	Memphis, TN	San Antonio, TX+
Albuquerque, NM	Des Moines, IA+	Miami-Fort Lauderdale, FL	San Diego, CA
Allentown, PA+	Detroit, MI	Milwaukee, WI+	San Francisco, CA
Atlanta, GA	El Paso, TX+	Minneapolis-St. Paul, MN	San Jose, CA
Augusta, GA	Fort Myers, FL+	Nashville, TN+	Scranton, PA*
Austin, TX	Fresno, CA+	New Haven, CT	Seattle, WA
Bakersfield, CA	Grand Rapids, MI	New Orleans, LA	Spokane, WA*
Baltimore, MD	Greensboro, NC	New York, NY	Springfield, MA
Baton Rogue, LA+	Greenville, SC	North Port-Sarasota-Bradenton, FL+	St. Louis, MO
Birmingham, AL	Harrisburg, PA	Ogden, UT+	Stamford, CT
Boise City, ID	Hartford, CT	Oklahoma City, OK	Stockton, CA
Boston, MA	Houston, TX	Omaha, NE+	Syracuse, NY
Buffalo, NY	Indianapolis, IN	Orlando, FL+	Tampa, FL+
Charleston, SC	Jackson, MS+	Philadelphia, PA	Toledo, OH
Charlotte, NC	Jacksonville, FL	Phoenix, AZ	Tucson, AZ
Chattanooga, TN*	Kansas City, MO	Pittsburgh, PA	Tulsa, OK
Chicago, IL	Knoxville, TN	Portland, OR	Urban Honolulu, HI*
Cincinnati, OH	Lakeland, FL+	Providence, RI	Ventura, CA
Cleveland, OH	Las Vegas, NV	Provo, UT+	Virginia Beach, VA+
Colorado Springs, CO	Little Rock, AR+	Raleigh, NC	Washington, DC
Columbia, SC+	Los Angeles-Long Beach-Anaheim, CA	Richmond, VA+	Wichita, KS
Columbus, OH	Louisville-Jefferson County, KY	Riverside, CA	Winston-Salem, NC+
Dallas-Fort Worth, TX	Madison, WI	Rochester, NY	Worcester, MA
Dayton, OH	McAllen, TX+	Sacramento, CA	Youngstown, OH*

Notes: The sign * indicates the markets with no data on days to sale pending status, newly pending listings, total for sale inventory, and median listing price. The sign + indicates the markets with no data on newly pending listings.

Table A3.

Summary statistics.

	No. of Obs.	Mean	Standard Deviation
Outcome variables
Percentage change in new listingsa	34,700	-0.010	0.205
Percentage change in newly pending salesb	3333	0.109	0.366
Percentage change in total inventoryc	4719	-0.228	0.144
Percentage change in web traffic to for-sale homesd	27,700	0.243	0.205
Percentage change in median list pricec	4750	0.069	0.061
Average number of days to pending sale statusc	4624	18.19	13.87
Controls for spread of the COVID-19
Total number of cases (per 10,000 people)	34,700	119.12	137.99
Total number of deaths (per 10,000 people)	34,700	3.32	4.12
State level policies to combat COVID-19
Bans on mass gatherings	34,700	0.583	0.493
Any business closures	34,700	0.647	0.478
School closures	34,700	0.752	0.432
Closure of non-essential business	34,700	0.131	0.337
Stay-at-home orders	34,700	0.200	0.400

Notes: Percentage changes in outcome variables are relative to the same time period in the previous year. Summary statistics for outcome variables: Data available for (a) Daily from January 1 to December 12, 2020 for 100 markets. (b) Weekly from January 4 to December 12, 2020 for 70 markets. (c) Weekly from January 4 to December 12, 2020 for 94 markets. (d) Daily from March 1 to December 2, 2020 for 100 markets.

Data Availability

• Data will be made available on request.