Post-COVID-19 Real Estate Net Absorption and Social Capital in U.S. Metropolitan Areas

Abstract

The authors of this article studied the differential impact of the COVID-19 pandemic on the performance of office real estate markets using a sample of major U.S. metropolitan areas. During the pandemic, social capital became a significant determinant of net absorption and mitigated the decline in net absorption due to the impact of COVID-19. They found that social capital in U.S. metropolitan areas as measured by the density of social networks has a significant positive effect on net absorption in a post-COVID-19 period. The results suggest that social capital contributes to an increase in net absorption values by promoting trust and communication among members of a community, which increases the likelihood of a completed transaction resulting in more leasing contracts.

Keywords:

• Real estate
• social capital
• COVID-19

Introduction

The impact of the COVID-19 pandemic on the office real estate market manifested a sharp decline in net absorption, indicating the change in occupied space over time. The observed decline in net absorption that began in the second quarter of 2020 shows about 19 million square feet (msf) of negative net absorption, which is on par with the decline in net absorption reported during the financial crisis of 2007.¹

The purpose of our study was to investigate social factors that impact the commercial and, in particular, office real estate market during the pandemic. Office real estate presents a unique setting to study the role of social factors because, first, office markets tend to be concentrated in or around urban areas, the first areas to be affected by COVID-19. Second, the majority of office real estate contracts are leasing long-term contracts, which tend to lock tenants to a particular space due to substantial moving costs (Wheaton, 1987). The rapid transition to remote work during COVID-19 prompted firms to look for shorter and cheaper leases and pushed the landlords’ profits down.² Our unique contribution was to study the role of social factors on office real estate performance using COVID-19 as an exogenous shock to the office real estate market.

Social capital contributes to the economic growth of communities and, hence, impacts the local real estate market (Rupasingha et al., 2000, 2006). Recent literature has explored the role of social capital (defined by norms, values, and information common to a social network) in real estate markets. For example, Li et al. (2020) showed that social capital significantly reduces the likelihood of mortgage delinquencies through limiting opportunistic behavior among homeowners. To our knowledge, this study is the first to look at the effect of social capital (measured as the total number of social organizations in each metropolitan area³) on commercial real estate net absorption.

Social capital is broadly defined as the functioning of social groups; that is, relationships between people and their interactions within institutions and their communities. These are generally understood rules that enable people to cooperate and create reciprocity in a society (Fukuyama, 1995). Social studies use a social capital measure that is generated on a county level rather than a national level because social interactions tend to occur at the local instead of the national level. This is particularly important for activities that promote the social and economic development of a community (Putnam, 1993; Rupasingha et al., 2000, 2006). Because social capital is often found in local (metropolitan) communities, this article is focused on the impact of social capital on commercial real estate within metropolitan areas.

Recent studies have demonstrated the importance of social capital in different corporate decisions, implying the positive role of social capital in decreasing agency problems, reducing information asymmetries, and facilitating contracting (Dudley, 2021; Hasan et al., 2017, 2020, 2022; Hoi et al., 2019). In the real estate setting, social capital should facilitate contracting in commercial real estate transactions, thereby mitigating the negative impact of COVID-19. There are two ways in which social capital can affect commercial real estate. First, denser social networks might decrease search costs leading to a quicker matching of the landlord (property) to a prospective tenant. Since social capital is associated with a greater number of social interactions due to networks’ pressure to maintain social activities despite COVID-19 risks (Bai et al., 2020; Ding et al., 2020; Wu, 2021), this might lead to more touring of the properties and, hence, facilitate more deals.⁴ Second, social networks might decrease information asymmetries inherent to commercial real estate transactions, thereby facilitating the completion of more deals (Chau & Wong, 2016). In this study, we empirically evaluated the relationship between social capital and the performance of the post-COVID-19 office real estate market. We found support for the hypothesis that social capital is associated with higher net absorption in the post-COVID period and, hence, mitigates the negative impact of COVID-19.

We matched commercial real estate data from Cushman and Wakefield U.S. Office Markets datasets with county-level social capital data from the Northeast Regional Center for Rural Development (NRCRD) at Pennsylvania State University. We supplement our dataset with data on economic indicators obtained from the U.S. Bureau of Economic Analysis. Our final dataset covers 81 major U.S. metropolitan areas⁵ from the second quarter of 2019 to the third quarter of 2021. Following previous studies, our main measure of social capital was the total number of social organizations in each metropolitan area (across all types) scaled by the total population of each metropolitan area (Rupasingha et al., 2006). We used net absorption as a measure of office real estate performance, which is the net change in physically occupied space between the current measurement period and the last measurement period, taking into consideration space vacated and newly constructed space in the same area during the same period.

We started by confirming the observation that after the second quarter of 2020 there was a sharp drop in net absorption in the United States. Our main results showed that U.S. metropolitan areas with higher social capital experienced less decline in net absorption compared to U.S. metropolitan areas with lower social capital after the start of the COVID-19 pandemic. In other words, the decline in office net absorption from the start of the pandemic has been partially mitigated by denser social networks in metropolitan areas. The effect held after controlling for unobservable time-invariant differences across metropolitan areas by adding the location (MSA) fixed effects. Examining social capital on a more granular level, we showed that the observed increase in net absorption during the pandemic is driven by “P-type” or a non-rent-seeking dimension of social capital (Knack & Keefer, 1997; Rupasingha et al., 2006), which highlights the role of social connections as conduits of trust and cooperation.

Our findings enhance the understanding of the influences of the social environment and bring together the two disparate streams of literature on social capital and commercial real estate. We complement the recent literature on the role of social capital in real estate (Li et al., 2022), corporate decisions (Dudley, 2021; Hasan et al., 2017, 2020, 2022; Hoi et al., 2019), and the recent studies on the role of social capital during the pandemic (Bai et al., 2020; Ding et al., 2020; Wu, 2021). Our study is the first to introduce the social capital construct into the real estate literature and, moreover, to study the differential impact of the COVID-19 pandemic on the performance of office markets depending on local differences in social capital.

Conceptual Framework and Hypothesis

Social Capital

Social studies have shown that social capital has a positive impact on communities and individuals by promoting trust and cooperation among agents, which in turn increases socially efficient collective action (Buonanno et al., 2009; Guiso et al., 2004; Knack & Keefer, 1997; La Porta et al., 1997; Putnam, 2001).

The role of social capital has been extensively studied in finance and in social sciences generally. Social capital impacts finance literature in terms of the flow of information between analysts and corporate officers (Cohen et al., 2010), managerial entrenchment (El-Khatib et al., 2015), and the sensitivity of investments and cash (Javakhadze et al., 2016). Social capital has been shown to impact the enforcement of contracts, extending the theory of self-enforcing agreements (Kandori, 1992) and the risk-taking propensity of managers (Bloch et al., 2008). Further, Rauch and Casella (2001) showed that social capital impacts information transfer.

While social studies use different measures and definitions of social capital, the literature has concluded that social capital refers to social connections or networks among individuals that promote the norms of reciprocity, trustworthiness, and cooperation among them (Coleman, 1988; Putnam, 1995; Woolcock, 2001). Social capital has traditionally been defined as “norms, values, and trust in a social network, which enable cooperative and shared actions” (Woolcock, 1998, 2001). Putnam (1993) argued that participation in associational activities promotes cooperation and measures social capital by counting the number of such groups and associations (e.g., political, civil, and sports clubs). Rupasingha et al. (2006) took this definition further and argued that “it is possible to characterize a large portion of social capital as a collective manifestation of individual behaviors, attitudes, and values of individual members of a community.” In other words, social capital can be considered a community-level characteristic that captures the strength of social norms and associational activities at a certain geographical location. The authors developed a county-level measure of social capital based on associational density in a particular county (i.e., the total number of 10 types of social associations and organizations in a county scaled by population). Rupasingha et al. (2006) documented significant cross-sectional differences in social capital among counties; hence, metropolitan areas should also differ in terms of the degree of social capital.

Recent studies have demonstrated the importance of social capital in different dimensions of economic activities and corporate decisions. Consistent with the notion that social capital imposes strong cooperative norms, facilitates trustworthiness, and limits self-serving behaviors in transactions, studies have found that firms from high social capital regions receive financing on better terms. For example, Hasan et al. (2017) found that firms headquartered in U.S. counties with higher levels of social capital incur lower bank loan spreads. Debt holders perceive social capital as environmental pressure that constrains opportunistic firm behaviors in debt contracting. Some studies have shown that social capital is associated with better access to financing. Hasan et al. (2022) showed that borrowers from higher social capital regions have better chances of getting funded and receiving larger loans. Dudley (2021), using a sample of U.S. startups, documented that entrepreneurs from counties with higher social capital rely more on outside debt to finance a new venture. Connectivity in the markets (in certain cases, common ownership) can potentially explain co-movement in international office space markets (Zhu & Lizieri, 2021). Studies have shown a link between financial integration and dynamics of urbanization (Coën et al., 2020).

The Effect of COVID-19 on Social Capital

COVID-19 presents an exogenous demand shock to office real estate, allowing us to use the impact of the COVID-19 pandemic to study the relationship between social capital and net absorption. Industry statistics have shown a sharp decline in net absorption rates due to COVID-19. The firm Cushman & Wakefield reported a net absorption of negative 19 million square feet in the second quarter of 2020, which corresponds to the start of the pandemic. The metropolitan markets that contributed the most to the decline were San Francisco, Dallas, Los Angeles, New York City (midtown Manhattan), and Houston.⁶ According to the Wall Street Journal, during the pandemic some landlords experienced financial difficulties due to many tenants being reluctant to renew their leases and had to offer steep discounts to keep major tenants.⁷

Social studies have argued that communities with high social capital exhibit more resiliency in challenging times. Coleman (1990) and Putnam (1993) studied the role of social capital in dealing with economic and social problems; both found that communities with higher social capital can better manage economic and social problems compared to communities with a lower stock of social capital. More recent social studies have argued that social capital may affect COVID-19 responses through more social activities (Bai et al., 2020; Ding et al., 2020; Wu, 2021). Social capital might be associated with a greater number of social interactions due to networks’ pressure to maintain social activities. Indeed, recent studies have confirmed that U.S. counties with higher social capital in the form of denser social networks and higher community engagement were associated with less social distancing and compliance with stay-at-home orders (Bai et al., 2020; Ding et al., 2020). This suggests that during the COVID-19 pandemic, regions with denser social networks responded to the economic downturn by more active participation in social and business life despite apparent COVID-19 risks.

The Effect of COVID-19 on Social Capital and Commercial Real Estate

Consistent with the notion that social capital mitigates agency problems and facilitates contracting, some studies have shown a link between county-level social capital and CEO compensation (Hoi et al., 2019), and innovation (Hasan et al., 2020). Social and behavioral characteristics of the society impact various real estate decisions.⁸

Accordingly, we conjectured that social capital should facilitate contracting in commercial real estate leasing transactions, which should be reflected in higher net absorption in the post-COVID period. There are several reasons for this effect. First, denser social networks might decrease search costs, allowing for more efficient matching: the landlord is able to find a prospective tenant quicker, and the tenant is able to find property quicker. Higher community engagement (especially compared to lower capital areas) during the pandemic might lead to better communication and more touring of properties (due to less social distancing) and, hence, to more leasing deals and higher office net absorption. Prospective tenants/landlords with more social connections through various clubs, organizations, and religious groups should have an easier experience in finding a property or a future lessee, respectively.

Second, denser social networks may reduce information asymmetries and make it easier for a landlord and a tenant to make a deal. Chau and Wong (2016) showed that information asymmetries between a landlord and a prospective tenant create substantial friction in the office market. Information asymmetry might exist for both a prospective tenant (e.g., unknown quality of a building), and for a landlord (quality of a prospective tenant). Local social networks might alleviate information asymmetries, thereby facilitating the completion of more deals. Hence, social capital should alleviate the negative effects of the COVID-19 crisis on the office market through decreasing searching and matching costs.⁹ This leads to the development of our testable hypothesis, which links the level of social capital with the changes in net absorption values. If social capital facilitates connections between people, decreases information asymmetry, and increases transparency between business partners, then it should lead to higher net absorption values. Our testable hypothesis is:

Hypothesis: U.S. metropolitan areas with higher social capital should experience an increase in net absorption values (compared to U.S. metropolitan areas with lower social capital) in the post-COVID-19 period.

Data and Methodology

We obtained commercial real estate data from Cushman & Wakefield U.S. Office Markets datasets.¹⁰ These datasets include office real estate indicators such as net absorption, vacancy rates, and asking rents for major U.S. metropolitan areas. Net absorption is measured in square feet and presents the net change of the occupied commercial space over a certain period of time.¹¹ It is measured by deducting the space vacated by tenants and new space made available on the real estate market from the total space that is leased. A high net absorption shows that there is a lot of demand for the space, while a lower net absorption shows there is a lack of demand. Data provided by Cushman & Wakefield is organized by metropolitan area and measured on a quarterly basis; our sample period was from the second quarter of 2019 to the third quarter of 2021.

To measure social capital, we used the data from the Northeast Regional Center for Rural Development at Pennsylvania State University.¹² This data is measured at the county level and is based on the methodology developed by Rupasingha et al. (2006), who measured the social capital of individuals through their participation in social organizations. Membership and participation in these organizations promote information sharing and cooperation through repeated interactions. Rupasingha et al. (2006) used census data to measure the total number of establishments in each county across 10 different social organization types.¹³ Using these data, our main measure of social capital was the total number of social organizations in each metropolitan area (across all types) scaled by the total population of this area.

One challenge of our research was to match the data on net absorption by metropolitan areas to the data of social capital by county. We used the Federal Information Processing Standard (FIPS) code of a county and a combination of automated and manual matching procedures to match each U.S. office market (metropolitan area) to a corresponding set of counties. The final dataset spanned 10 quarters from 2019Q2 to 2021Q3 and covered 81 major U.S. metropolitan areas.

To measure the impact of COVID-19 on the net absorption, we used the difference-in-differences approach, where we applied the following model: (1) $$\mathit{Net}\mathrm{ }\mathit{Absorptio}{n}_i={\beta }_1*\mathit{Social}\mathrm{ }\mathit{Capita}{l}_i+{\beta }_2*\mathit{COVID}-19+{\beta }_3*\mathit{Social}\mathrm{ }\mathit{Capita}{l}_i*\mathit{COVID}-19+\mathrm{ }+{\beta }_4{X}_i+\zeta t\mathrm{ }+{\epsilon }_i$$(1) where Net Absorption_i is the net absorption of the U.S. metropolitan area; COVID-19 is the dummy variable that is equal to 1 starting from the second quarter of 2020¹⁴; Social Capital_i is measured as described above; X_i is a vector of control variables. We also included quarter fixed effects ζt to control for the common time trends across characteristics of metropolitan areas.

In the regression, we controlled for several economic determinants of net absorption. It has been established that economic growth and, more broadly, macroeconomic factors are linked to the performance of commercial real estate markets (Sagalyn, 1990; Sivitanidou & Sivitanides, 1999). Economic growth leads to an increase in demand for office space through job creation and subsequent increases in personal income, which drives retail purchases and demand for more shop space (Glickman, 2014). Therefore, in regressions, we used state-level real gross domestic product (GDP) growth (as a percentage from the previous quarter) and included the percentage of the employed population at a given MSA to account for local economic activity and overall local level of employment, respectively. We also controlled for population growth since existing literature has shown a direct link between population growth and real estate development (Capozza et al., 2002).

Considerable evidence has shown that local office market factors also affect the demand and supply for office spaces. Specifically, it has been shown that net absorption is affected by the real rents, vacancy rates, and the growth of employment in the office sectors of the economy (Wheaton, 1987; Wheaton et al., 1997). In a recession, absorption should decrease because many office spaces become vacant; so, we might observe a negative relation between net absorption and vacancy rate. Hence, we controlled for the growth of the employment in the office-consuming spaces by calculating the percentage growth in office employment in a given MSA area. In addition, we included lagged vacancy rates (defined as the ratio of unoccupied space to the total inventory) and lagged real rents (measured in 2015 dollars) to capture local current market conditions.

Several studies have shown a link between political party membership, political beliefs, and response to COVID-19 (Abel et al., 2021; Baccini et al., 2021; Murray & Murray, 2020). Republicans and Democrats are known to have different attitudes toward the response to pandemic threats, which were reflected in the severity of state lockdown measures during 2020. Barrios and Hochberg (2021) found a link between partisan differences and risk perception of contracting COVID. Their results suggested that counties with a higher percentage of votes for Donald Trump were less likely to engage in social distancing. We further used the percentage of votes that went to the Democratic Party in the 2020 presidential election (variable Voting) to capture political preferences of the metropolitan areas that could potentially affect the local mobility and office real estate market.

Empirical Results

Baseline Results

Table 1 reports the summary statistics for the major U.S. metropolitan areas as defined in Cushman & Wakefield for the period from Q2-2019 to Q3-2021. Table 1 Panel A shows that the average net absorption was negative and equaled −0.132 msf, which is consistent with the decline of office activity during the COVID-19 pandemic. The average metropolitan area had about 0.87 social organizations per 1,000 people, a mean annual population growth of 0.6%, and about 66% of population employed. The average real asking rent was $25.6 per sf (in 2015 dollars) and the average vacancy rate was 13% over the analyzed period.

Table 1. Summary statistics. Panel A: Descriptive statistics

Variable	Mean	25th perc.	Median	75th perc.	St. dev.	N
Net Absorption (msf)	−0.132	−0.252	−0.058	0.076	0.500	800
Social Capital	0.869	0.676	0.871	0.988	0.326	800
Social Capital P-Type	0.727	0.565	0.713	0.845	0.220	800
Social Capital O-Type	0.142	0.077	0.122	0.154	0.178	800
Population Growth	0.581	−0.061	0.381	1.189	0.852	800
State GDP Growth	2.578	0.250	2.800	6.500	15.570	800
Employment (% Population)	65.710	59.804	63.944	68.030	17.624	800
Office Employment Growth	0.003	−2.117	0.422	1.907	2.959	800
Asking Rent ($/SF, in 2015 dollars)	25.613	18.478	22.100	29.182	10.711	720
Vacancy Rate (%)	0.132	0.098	0.126	0.166	0.049	711
Voting (%)	57.056	49.757	54.827	62.819	10.955	800

Panel B: Univariate tests of net absorption

	High Social Capital	Low Social Capital
After COVID-19	−0.251	−0.350
Before COVID-19	0.085	0.168
Diff. (t-stat)	−0.337*** (−8.55)	−0.518*** (−9.98)

Notes: This table reports the summary statistics for the major U.S. metropolitan areas as defined in Cushman & Wakefield for the period from Q2-2019 to Q3-2021. Panel A reports the descriptive statistics, and Panel B reports the results of univariate comparisons for the subsamples of metropolitan areas with high (low) social capital before and after COVID-19. High (Low) Social Capital is a dummy variable that indicates a metropolitan area with above (below) sample median of social capital. After COVID-19 equals 1 for quarters starting from the second quarter of 2020. All other variables are defined in Appendix A. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

We started our analysis with univariate tests. To separate the U.S. metropolitan areas with high/low social capital, we defined a dummy variable High (Low) Social Capital based on whether the Social Capital variable was above (below) the sample median. Table 1 Panel B shows the mean net absorption by the subsamples defined above. The data show that in both subsamples (with high and low social capital metropolitan areas) net absorption significantly declined. However, U.S. metropolitan areas with high social capital experienced less decline in post-COVID absorption values. The average net absorption for high social capital areas was about −0.25 msf versus −0.35 msf for areas with low social capital, with the difference between these numbers significant at 5%. Additionally, the average difference between net absorptions pre- and post- COVID-19 periods was lower in metropolitan areas with high social capital (−0.337) than in metropolitan areas with low social capital (−0.518).

Figure 1 presents a graphical illustration of these results with plots averaged by quarter net absorption separate for the metropolitan areas with high and low social capital. As shown in the figure, both lines follow each other until the first quarter of 2020. Starting from the second quarter of 2020, there was a sharp decline in net absorption in both groups, which continued into the third quarter of 2020. However, consistent with our univariate results, the observed decline in the low social capital group had a greater magnitude than that in the high social capital group.

Figure 1. Net absorption before and after COVID-19.

Notes: This figure shows the averaged across quarters net absorption measured in million square feet. The data cover major U.S. metropolitan areas as defined in Cushman & Wakefield for the period from Q2-2019 to Q3-2021. High (Low) Social Capital is a dummy variable that indicates a metropolitan area with above (below) sample median of social capital. The vertical line labels the second quarter of 2020, the first quarter after pandemic was declared by the World Health Organization.

Next, we estimated the difference-in-differences regression Equation (1)(1) $$\mathit{Net}\mathrm{ }\mathit{Absorptio}{n}_i={\beta }_1*\mathit{Social}\mathrm{ }\mathit{Capita}{l}_i+{\beta }_2*\mathit{COVID}-19+{\beta }_3*\mathit{Social}\mathrm{ }\mathit{Capita}{l}_i*\mathit{COVID}-19+\mathrm{ }+{\beta }_4{X}_i+\zeta t\mathrm{ }+{\epsilon }_i$$(1) , where the dependent variable is Net Absorption. The results are presented in Table 2. First, in all models, the coefficient on COVID-19 was negative and statistically significant at 1%, suggesting a uniform decline in net absorption after the second quarter of 2020. Second and more importantly, in all models, the coefficient on the interaction Social Capital*COVID-19 was positive and statistically significant at 5% or better. Model (1) presents the results without quarter fixed effects and with only macroeconomic controls. Models (2)–(5) include quarter fixed effects and office market controls added one at a time.¹⁵

To mitigate the concern that unobservable factors across metropolitan areas may have affected our results, we added MSA-level fixed effects to all our models. Table 3 presents the results. The Social Capital variable was omitted from the models due to multicollinearity. We also excluded Voting since it did not have a time variation. As in Table 2, the coefficient on COVID-19 was negative and highly significant. The coefficient on Social Capital*COVID-19 was statistically significant in all models at 10% or better. Note, compared to Table 2 without the MSA-fixed effects, the magnitude of coefficients on both of these variables in Table 3 did not change much, suggesting that the results are robust enough after controlling for unobserved differences between MSAs.

The estimated results are economically significant. For example, according to Model (2) of Table 3, 1 standard deviation increase in social capital increased net absorption by about 0.093 msf, which offset about 22.1% of the decline in net absorption due to COVID-19.¹⁶ Overall, these results suggest that the metropolitan areas with higher social capital intensity had higher net absorption after the start of the pandemic than the metropolitan areas with a lower intensity.

The Role of Types of Social Capital

So far, we have presented evidence consistent with our hypothesis that social capital partly mitigates the negative impact of COVID-19 on office net absorption. Next, we investigated the possible channel of this effect. Social capital literature differentiates between two forms of social capital: “Olson-type” (O-Type) and “Putnam-type” (P-Type) organizations (Knack & Keefer, 1997; Rupasingha et al., 2006).¹⁷ The intuition is that membership in P-type organizations (e.g., civil, bowling, or religious groups) involves more social interaction that promotes cooperation and trust. In such groups, individuals value social connections and feel more committed to cooperating and engaging in community activities. According to Rupasingha et al. (2006), this group is more likely to go beyond the formal boundaries of these organizations and invest additional time and effort to be a valuable member of the network. This group is more likely to exhibit less social distancing due to their more active contribution to social networks (Ding et al., 2020). To the contrary, membership in O-type organizations (e.g., political, labor, or business groups) is backed by financial incentives because members are united to achieve some specific political, labor, or economic goal.¹⁸ Membership in such organizations is considered rent-seeking because it is more likely to capture the willingness of individuals to contribute to common societal objectives. Presumably, the social interactions through O-type organizations are likely to be sensitive to the core activities of these organizations and do not go beyond the formal memberships. When the pandemic slowed down all economic activities, these organizations temporarily became less active as well, as did the social interaction between members that was limited only to the activities in these organizations (due to their nature). Therefore, we hypothesized that the effect of social capital on net absorption should be stronger for P-type organizations than for O-type organizations.

To capture these two facets of social capital, we constructed two separate variables of social capital based on the total number of these two types of organizations and repeated the regressions from Table 3.¹⁹ We ran all regressions with MSA-fixed effects to account for unobserved time-invariant MSA-level factors.²⁰ The results are reported in Table 4. Column (1) of Table 4 includes the social capital variable measured using only P-type organizations. Column (2) uses an alternative definition of social capital based on O-type organizations. The estimates from these columns show that the coefficient on the interaction term Social Capital*COVID-19 was positive and statistically significant at 1% for the P-type social capital, but not statistically significant for O-type organizations. Assuming that both types of social capital matter, we included both variables simultaneously in columns (3)–(6) of Table 4. We found the same pattern: the coefficient on the interaction term Social Capital*COVID-19 remained positive and highly significant when only P-type organizations were included in the definition of social capital, while the coefficient on the interaction term for O-type social capital remained statistically insignificant. Overall, these results suggest that the observed increase in net absorption during the pandemic was driven by P-type or non-rent-seeking dimensions of social capital, which shows that the value of social connections in the form of trust and cooperation was likely the channel of the observed positive effect.

Conclusion

In this article, we studied the role of social connections in the impact of COVID-19 on the U.S. office real estate markets. Data from Cushman & Wakefield showed that there has been a decline in net absorption in the period following COVID-19. We investigated how the negative impact of the pandemic differed across metropolitan areas. We hypothesized that the density of social capital in each metropolitan statistical area influenced the way a community responded to COVID-19 and that impacted the overall business activities, commercial real estate, and net absorption. We are the first to provide new evidence that U.S. metropolitan areas with higher social capital experienced less decline in office net absorption during the COVID-19 pandemic than metropolitan areas with lower social capital. Our social capital measure was based on the total number of social organizations in each metropolitan area (across all types) scaled by the total population of this area. Our results are statistically and economically significant after controlling for standard macroeconomic factors and local fixed effects.

We also created a distinction between O-Type and P-Type organizations following the social capital literature (Rupasingha et al., 2006), and found that the observed positive effect of social capital on net absorption is driven by memberships in P-type organizations, which involve more social interaction that promotes cooperation and trust. Our findings are consistent with the view that connections between people via memberships in social organizations facilitate contracting between parties through potentially decreased searching and matching costs or information asymmetries. Overall, our study highlights the importance of the social capital of a business community in reaction to an external shock (in our case, COVID-19) and complements the recent research on the impact of social connections on real estate markets and the impact of COVID-19 on business activities.

This article provides new insights into urban development and community building. Fostering socialization in communities by developing organizations that promote social activities creates trust, improves the transfer of information, and, therefore, facilitates business activities. Our findings have policy implications. For example, government agencies and business organizations can consider creating clubs, sports facilities, and cultural centers. We found that these organizations contribute to urban development and business output in a given metropolitan area, and our results also suggest that communities with a higher intensity of social capital build more resilience for future unforeseeable shocks. Although building social capital is costly, it brings tangible benefits to communities.

Notes

1 Cushman & Wakefield. 2019–2021. U.S. Office Markets report as of the second quarter 2021.

2 Office Markets Under Pressure as Coronavirus Squeezes Cities. Wall Street Journal, August 4, 2020. https://www.wsj.com/articles/office-markets-under-pressure-as-coronavirus-squeezes-cities-11596542403?mod=article_inline.

3 For social capital, we used the data from the Northeast Regional Center for Rural Development (NRCRD) at Pennsylvania State University described later in this article.

4 According to Cushman & Wakefield, touring office spaces by firms is one of the indicators predicting leasing activity (U.S. Office Markets report Q2-2021).

5 We followed the literature (Wheaton et al., 1997) of lagging the relevant control variables, and this determined our number of observations.

6 Cushman & Wakefield, U.S. Office Market report from Q2 2021.

7 During Pandemic, Landlords Find Relying on One Office Tenant Can Backfire. Wall Street Journal, April 20, 2021, https://www.wsj.com/articles/during-pandemic-landlords-find-relying-on-one-office-tenant-can-backfire-11618920001.

8 For example, the strategic default decision—how default is viewed by one’s social network—may impact the likelihood of the event (Seiler et al., 2012; Seiler & Walden, 2014).

9 The third channel through which social capital may affect COVID-19 responses is by facilitating a common response to the COVID-19 crisis and increasing social distancing even if it incurs personal costs. We did not consider this channel because social studies that found this effect used a different measure of social capital based on individual commitment to broader social institutions (e.g., voting commitment or commitment to fill out census forms).

10 Cushman & Wakefield. 2019–2021. https://www.cushmanwakefield.com/en/united-states/insights/us-marketbeats/us-office-marketbeat-reports (last accessed November 12, 2021).

11 The official definition of net absorption is the net change in physically occupied space between the current measurement period and the last measurement period, taking into consideration both vacated and newly constructed real estate space in the same area during the same period.

12 We based our measure on 2014 social capital data because it is the latest available data. Note the variable is slow changing. For example, 90% of changes of the social capital measure from 2009 to 2014 were only within 6.4 percentage points of the initial value of social capital in year 2014; from 2005 to 2014, they were within 13.4 percentage points. Furthermore, we focused on the role of the prepandemic social connections that existed among people to study the differential response of communities to the pandemic.

13 These types include: (a) civic organizations; (b) bowling centers; (c) golf clubs; (d) recreational centers; (e) sports organizations; (f) religious organizations; (g) political organizations; (h) labor organizations; (i) business organizations; and (j) professional organizations.

14 We assigned the second quarter of 2020 as the first quarter of COVID-19 period because the WHO officially declared a pandemic from March 2020.

15 While these variables are important determinants of net absorption, we added these variables one at a time because there are some plausible endogeneity concerns about each of them.

16 The calculated decline in net absorption is equal to coefficient on Social Capital x COVID-19 times 1 standard deviation of social capital from Table 1 (0.285*0.326 = 0.093). We then divided this number by the decline in net absorption due to COVID-19 (coefficient on COVID-19 = −0.419) to get 22.1%.

17 See the detailed classification of these groups in Rupasingha et al. (2006).

18 Rupasingha et al. (2006) provide the following example of this membership: “Farmers join the Farm Bureau because it is instrumental in persuading the government to provide farm program payments.”

19 As with the initial Social Capital variable, we scaled these two variables by the population of the metropolitan areas.

20 Because of the added MSA-fixed effects, the main effect of social capital was omitted from the model.

Appendix A.

 Definition of variables

Net Absorption—The net change in occupied office space between two quarters in a given U.S. metropolitan area. Measured in million square feet (sft). Source: Cushman & Wakefield.

Social Capital—Total number of establishments in the following organizations and associations: religious, civic and social, business, political, professional, labor, recreational, bowling, sports and golf in a given U.S. metropolitan area divided by total population per 1,000 in this area. Data is obtained from 2014 dataset. Source: Northeast Regional Center for Rural Development (NRCRD) at Pennsylvania State University.

Vacancy Rate—The amount of unoccupied space (new, relet, and sublet) expressed as a percentage of total inventory in a given U.S. metropolitan area. The variable is lagged one quarter. Source: Cushman & Wakefield.

Asking Rent—Gross average asking rents weighted by the amount of available direct and sublease space in Class A, B, and C properties ($/SF). The variable is measured in 2015 dollars, and lagged one quarter. Source: Cushman & Wakefield.

Population Growth—Percentage growth of population of a U.S. metropolitan area from the previous year. The variable is lagged one year. Source: U.S. Bureau of Economic Analysis.

State GDP Growth—Annualized percentage change (from the previous period) in real GDP measured quarterly by state. Source: U.S. Bureau of Economic Analysis.

Office Employment Growth—Percentage growth in office employment in a given U.S. metropolitan area from the previous year. Office employment includes jobs in the following sectors: information; finance and insurance; real estate and rental and leasing; professional, scientific, and technical services; management of companies and enterprise; administrative and support and waste management and remediation services; educational services; health care and social assistance; other services (except government and government enterprises); government and government enterprises. The variable is lagged one year. Source: U.S. Bureau of Economic Analysis.

Employment—Total employment expressed as a percentage of total population in a given U.S. metropolitan area in a given year. The variable is lagged one year. Source: U.S. Bureau of Economic Analysis.

Voting—Percentage of votes out of total votes given to the Democratic Party in the 2020 presidential election in a given U.S. metropolitan area. Source: MIT Election Data and Science Lab, "County Presidential Election Returns 2000–2020."

Table 2. The impact of COVID-19 on net absorption in U.S. metropolitan areas.

Dep. Var: Net absorption	(1)	(2)	(3)	(4)	(5)
Social Capital	0.045	0.057	−0.010	0.033	−0.007
	(0.68)	(0.89)	(−0.13)	(0.45)	(−0.10)
COVID-19	−0.596***	−0.474***	−0.552***	−0.534***	−0.474***
	(−7.47)	(−4.40)	(−5.74)	(−5.81)	(−4.43)
Social Capital × COVID-19	0.197***	0.199***	0.225***	0.231***	0.219**
	(2.62)	(2.62)	(2.70)	(2.88)	(2.54)
Population Growth	0.052***	0.046**	0.037*	0.040*	0.023
	(2.58)	(2.16)	(1.82)	(1.75)	(0.94)
State GDP Growth	−0.003***	0.005	0.004	0.003	0.004
	(−2.80)	(1.11)	(0.99)	(0.65)	(0.91)
Employment	−0.006***	−0.006***	−0.004**	−0.007***	−0.004**
	(−3.77)	(−3.78)	(−2.29)	(−3.65)	(−2.23)
Office Employment Growth		0.004			0.009
		(0.63)			(1.20)
Asking Rent			−0.006*		−0.005
			(−1.75)		(−1.29)
Vacancy Rate				−0.915**	−0.838**
				(−2.25)	(−2.00)
Voting					−0.000
					(−0.14)
Quarter FE	No	Yes	Yes	Yes	Yes
Observations	800	800	720	711	711
Adj. R-squared	0.23	0.23	0.23	0.23	0.24

Notes: This table presents the results of difference-in-differences regressions where the dependent variable is Net Absorption in a given U.S. metropolitan area. Social Capital is the total number of establishments across the 10 types of social, professional and business organizations in a given U.S. metropolitan area divided by total population per 1,000 in this area. COVID-19 equals 1 for quarters starting from the second quarter of 2020. All other variables are defined in Appendix A. t-statistics, reported in parentheses, are based on heteroskedasticity-robust standard errors. ***. **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

Table 3. The impact of COVID-19 on net absorption in U.S. metropolitan areas: controlling for unobservable factors.

Dep. Var: Net absorption	(1)	(2)	(3)	(4)	(5)
COVID-19	−0.376**	−0.419**	−0.438***	−0.464***	−0.504***
	(−2.41)	(−2.32)	(−2.84)	(−2.91)	(−2.84)
Social Capital × COVID-19	0.270*	0.285*	0.284*	0.294**	0.309**
	(1.88)	(1.88)	(1.98)	(2.00)	(2.01)
Population Growth	0.118	0.119	0.061	0.050	0.051
	(0.80)	(0.80)	(0.37)	(0.29)	(0.28)
State GDP Growth	0.007	0.006	0.006	0.006	0.006
	(1.60)	(1.59)	(1.33)	(1.38)	(1.36)
Employment	0.037**	0.040***	0.038***	0.037**	0.041***
	(2.62)	(2.68)	(3.15)	(2.28)	(2.99)
Office Employment Growth		−0.007			−0.008
		(−0.73)			(−0.91)
Asking Rent			−0.004		−0.004
			(−0.13)		(−0.13)
Vacancy Rate				0.225	0.153
				(0.15)	(0.10)
Quarter FE	Yes	Yes	Yes	Yes	Yes
MSA FE	Yes	Yes	Yes	Yes	Yes
Observations	800	800	720	711	711
Adj. R-squared	0.35	0.35	0.36	0.36	0.36

Notes: This table presents the results of difference-in-differences regressions where the dependent variable is Net Absorption in a given U.S. metropolitan area. Social Capital is the total number of establishments across the 10 types of social, professional, and business organizations in a given U.S. metropolitan area divided by total population per 1,000 in this area. COVID-19 equals 1 for quarters starting from the second quarter of 2020. All other variables are defined in Appendix A. t-statistics, reported in parentheses, are based on standard errors that allow for clustering at the MSA-level.***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

Table 4. The impact of COVID-19 on net absorption in U.S. metropolitan areas and social capital types.

Dep. Var: Net absorption	(1)	(2)	(3)	(4)	(5)	(6)
COVID-19	−0.559***	−0.177**	−0.670***	−0.630***	−0.667***	−0.734***
	(−2.96)	(−2.11)	(−3.00)	(−3.28)	(−3.28)	(−3.33)
Social Capital P-type × COVID-19	0.559***		0.658***	0.602***	0.614***	0.647***
	(2.75)		(2.88)	(2.73)	(2.75)	(2.78)
Social Capital O-type × COVID-19		0.026	−0.222	−0.165	−0.158	−0.152
		(0.20)	(−1.27)	(−1.00)	(−0.99)	(−0.97)
Population Growth	0.097	0.135	0.091	0.039	0.032	0.030
	(0.66)	(0.90)	(0.63)	(0.24)	(0.19)	(0.17)
State GDP Growth	0.006	0.006	0.006	0.005	0.005	0.005
	(1.58)	(1.42)	(1.46)	(1.24)	(1.29)	(1.26)
Employment	0.036**	0.028*	0.035**	0.034**	0.033*	0.037**
	(2.48)	(1.75)	(2.19)	(2.50)	(1.85)	(2.50)
Office Employment Growth			−0.011			−0.011
			(−1.21)			(−1.23)
Asking Rent				−0.003		−0.003
				(−0.08)		(−0.08)
Vacancy Rate					0.554	0.481
					(0.37)	(0.31)
Quarter FE	Yes	Yes	Yes	Yes	Yes	Yes
MSA FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	800	800	800	720	711	711
Adj. R-squared	0.35	0.34	0.35	0.37	0.37	0.37

Notes: This table presents the results of difference-in-differences regressions where the dependent variable is Net Absorption in a given U.S. metropolitan area. Social Capital P-type (O-type) is the total number of establishments across non-rent-seeking (rent-seeking) organizations as defined in Rupasingha et al. (2006) in a given U.S. metropolitan area divided by total population per 1,000 in this area. COVID-19 equals 1 for quarters starting from the second quarter of 2020. All other variables are defined in Appendix A. t-statistics, reported in parentheses, are based on standard errors that allow for clustering at the MSA-level. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.