We use cookies on our website to give you the best online experience. Please know that if you continue to browse on our site, you agree to this use. You can always block or disable cookies using your browser settings. To find out more, please review our privacy policy.


Policy Hub: Macroblog provides concise commentary and analysis on economic topics including monetary policy, macroeconomic developments, inflation, labor economics, and financial issues for a broad audience.

Authors for Policy Hub: Macroblog are Dave Altig, John Robertson, and other Atlanta Fed economists and researchers.

Comment Standards:
Comments are moderated and will not appear until the moderator has approved them.

Please submit appropriate comments. Inappropriate comments include content that is abusive, harassing, or threatening; obscene, vulgar, or profane; an attack of a personal nature; or overtly political.

In addition, no off-topic remarks or spam is permitted.

June 1, 2021

COVID Lasted Longer Than Expected. What Happened to Retail Space?

In our last post, we observed some deterioration in office market fundamentals in the face of the COVID-19 pandemic. Office market vacancy rates increased notably while asking rents remained relatively unchanged from 2019. Did the retail market respond similarly? Due to mandatory business closures and the increase in e-commerce revenue, one might expect retail market vacancy rates to increase and asking rents to decrease. However, the retail property market overall has been much less affected during the COVID-19 pandemic than the office market. As in our previous post, we explore the retail real estate market vacancy rate and asking rent in the 10 cities with the highest retail space stock square footage (according to CBRE-EA's market ranker).

Unlike the office market, we observe that retail market vacancy rates did not behave nearly as uniformly (see chart 1). The overall retail market vacancy rate was 6.2 percent in the fourth quarter of 2019 (pre-COVID) and increased to 6.6 percent in the fourth quarter of 2020. The Dallas retail market experienced the largest change in the vacancy rate, increasing by 21.31 percent from the fourth quarter of 2019 to the fourth quarter of 2020 (from 6.1 percent to 7.4 percent). Washington, DC exhibited the next largest jump, where the vacancy rate increased by 16.98 percent during the same period (from 5.3 percent to 6.2 percent). While retail markets in Los Angeles, New York City, and Philadelphia saw an 8–10 percent increase in vacancy rates during the pandemic, other markets such as Houston and Phoenix experienced moderate vacancy rate increases—3.03 percent and 2.44 percent, respectively. In more positive scenarios, retail markets in cities such as Atlanta and Chicago saw only a slight increase in vacancy rates during the pandemic, with vacancy rates returning to their pre-COVID level at 5.9 percent and 9.1 percent in the fourth quarter of 2020, respectively.

Chart 1: Vacancy Rates

Chart 2 illustrates that the asking rent in most markets increased during the pandemic. The biggest jump in gross asking rents occurred in Dallas where, perplexingly, vacancy rates also increased the most. Dallas's gross asking rents increased from $13.26 per square foot (psf) to $14.79 psf from the fourth quarter of 2019q4 to the fourth quarter of 2020. New York City also experienced a noticeable increase in gross asking rents, with a 6.3 percent increase from the fourth quarter of 2019 ($33.51 psf) to the fourth quarter of 2020 ($35.62 psf). Other markets experienced moderate increases or mirrored rent rate changes experienced between 2018 and 2019. Atlanta was the only market where gross asking rents decreased. However, Atlanta's downward trend in retail rent rates started in the fourth quarter of 2019, so it is difficult to determine to what extent this trend was preexisting or influenced by the pandemic.

Chart 2: Gross Asking Rent

These observations all beg the question: why did the retail market experience only moderate changes in vacancy rates relative to the office market during the pandemic?

We cannot say for certain why the retail market and office market fundamentals for the top 10 markets followed such different paths. However, just as office-using employment drives demand for office space, we also know that the retail market relies heavily on household consumption. The relatively modest changes in vacancy rates and increased asking rental rates in the retail market could potentially be explained by the rebound in retail sales as shoppers returned to reopened stores (as you can see hereOff-site link). Chart 3 illustrates that retail trade sales temporarily decreased in the early stages of the COVID-19 pandemic (February–April 2020), but they rebounded strongly in May and June 2020, returning the total dollar value of retail trade sales (excluding nonstore retailers) to its pre-COVID level.

Chart 3: Monthly Retail Trade Sales

However, surveys conducted by our Atlanta Fed colleagues (discussed here) show that retail and wholesale firms expressed increased optimism in August 2020 when compared to April 2020. This improved sentiment may have further supported overall retail real estate market performance.

After putting forth these observations, though, one caveat is in order. Although the overall retail market appears to have been much less affected than the office market, the effects could vary by the size and type of business. For example, larger firms often have more financing options. They also tend to have online platforms that can help to offset temporarily sluggish sales at brick-and-mortar locations. The data we reviewed for this post does not discriminate between firm size or type, and we did not explore these dimensions.

We'll continue to monitor market fundamentals and report what we find. Meanwhile, you can track trends in the commercial markets with the Atlanta Fed's Commercial Real Estate Momentum Index.

May 20, 2021

Has the COVID-19 Pandemic Affected Demand for Office Space?

Last July, our Atlanta Fed colleagues concluded that the COVID-19 pandemic would not decrease demand for commercial real estate (here). As the COVID-19 pandemic passes the one-year mark, have expectations on what a "return to the workplace" looks like changed for businesses and their employees? Many firms continue to allow employees to work from home (here and here). Anecdotes suggest that many are reimagining their office culture as they scrutinize empty or underutilized office space in an effort to reduce structural costs. Has this also resulted in a significant shift in demand for office space?

To understand how office market performance fundamentals have changed since the COVID-19 pandemic began, we examine office-using employment, vacancy rate, and rental rate trends across the 10 cities with the highest office space square footage stock. We hypothesize that markets with higher office space inventories likely have more office workers and may experience larger, longer-lasting effects resulting from massive shifts to remote work arrangements.

The United States experienced significant declines in employment due to COVID-19, but what was the effect on the office-based workforce? Using the Bureau of Labor Statistics' office-using employment series at the metropolitan statistical area (MSA) level, we observe in chart 1 that the number of employees using commercial office space has declined significantly in the pandemic. The graph on the right illustrates the monthly change in the number of employees using offices, which decreased by 6 percent on average from March 2020 to April 2020. The decline was most severe (11 percent) in the Los Angeles-Long Beach-Anaheim, CA MSA and mildest (2 percent) in the Washington-Arlington-Alexandra, DC-VA-MD-WV MSA. Although most markets saw a positive change in May 2020, the graph on the left shows that the number of employees using physical office space is still lower than it was before the start of the COVID-19 pandemic.

Chart 1 of 3: Office Using Equipment

The survey conducted by our colleagues in June 2020 (here) indicated that roughly 80 percent of respondents had no plans to change their current floor space needs. It seems reasonable, however, that the significant decline in office-using employment across metro areas with the largest office stock might translate to higher vacancy rates. Recent office market data from the CBRE Economic Advisors (CBRE-EA)1 indicate that office market vacancy rates increased significantly in 2020. Chart 2 illustrates stable vacancy rates before the COVID outbreak. Following the outbreak, in the second quarter of 2020, vacancy rates moved up. The overall office market vacancy rate was 12.2 percent in the fourth quarter of 2019 and 12.3 percent in the first quarter of 2020, then steadily increased through the year to 15 percent in the fourth quarter of 2020. All markets experienced vacancy rate increases for office space, with markedly different changes. Changes for the Denver and New York office market vacancy rates, for example, were larger than for most other cities. Denver's increased by 30.89 percent from the first quarter of 2020 to the fourth quarter (from 12.3 percent to 16.1 percent) while New York saw the largest jump (36.78 percent, from 8.7 percent to 11.9 percent). Interestingly, although Denver experienced the second highest jump in vacancy rates in 2020, the Denver and Houston MSAs did not experience a dramatic change in the number of employees using office space in 2020.

Chart 2 of 3: Office Market Vacancy Rate

Observing that office space vacancy rates increased significantly during the pandemic, we wondered if landlords renegotiated rent rates, so we explored the trend in gross asking rent. This is the estimated rent rate for a gross lease type where a landlord is responsible for expenses such as utilities and property taxes. Chart 3 illustrates trends of gross asking rent per square foot from the first quarter of 2016 to the fourth quarter of 2020. The data indicate that the asking rent rate did not change significantly during the pandemic compared to the previous period. While gross asking rent in the Dallas market increased slightly from the third quarter of 2020 to the fourth quarter, it was a relatively minor change. With such a high level of uncertainty, it's possible that real estate firms and investors were not convinced they needed to adjust asking prices. Another possibility is that tenants received longer periods of free rent in exchange for not adjusting rent prices downward. This second scenario effectively makes rents cheaper, although contract rates appear to be the same. We are unable to investigate further due to the unavailability of prerequisite data.

Chart 3 of 3: Office Market Asking Rent


We explored office market fundamentals since the onset of the COVID-19 pandemic and found that office-using employment declined while vacancy rates for office space increased significantly. How office property asking rent rates have responded to the pandemic remains unclear. Due to the long terms required for commercial leases, landlords appeared unwilling to lower rents because they expected that market conditions would soon improve.

Since firms remain uncertain as to when employees will return to the office, policymakers and investors should monitor current office market vacancy rates to best prepare to operate in an environment with sustained high vacancy rates. Will some firms have to reimagine their space and convert assets into different property types? We'll continue to monitor these market fundamentals, including property conversions, and report back. Meanwhile, you can track trends in market fundamentals with the Atlanta Fed's Commercial Real Estate Momentum Index.

1 [go back] CBRE- Econometric Advisors (EA) release quarterly data on the commercial real estate markets.

February 1, 2021

Ecommerce's Spillover Effects on the Local Retail Real Estate Market

Over the past decade, online shopping has become the primary shopping method for many. Consequently, concerns that ecommerce will replace traditional retail outlets and make physical stores obsolete have increased. However, the expansion of ecommerce drives online retailers to establish large fulfillment centers nationwide. To ensure that customers receive packages within two days, online retailers aggressively accelerated fulfillment center openings. A major online retailer, one that accounts for about 50 percent of total ecommerce sales, has opened more than 150 fulfillment centers, located across 88 U.S. counties as of 2018. Each fulfillment center hires its own workers, and thus a new fulfillment center's local economic impact might be the opposite of the typical negative belief about ecommerce's effects on brick-and-mortar stores. To estimate the overall effect of ecommerce on the local retail market, I use a fulfillment center of this online retailer as a proxy to an ecommerce local presence. Specifically, I focus in this post on retail real estate transaction prices, since store sales heavily influence property values. (Retail property sales price is commonly estimated using the direct capitalization method.)

Among the counties that received a fulfillment center, I select counties with a fulfillment center that opened between 2013 and 2015, giving clear three-year periods before and after the opening. Control counties are the matched counties that are strong potential fits for a fulfillment center establishment yet did not receive one. Fulfillment centers are more likely to be in counties where population density and education level are higher than average but median household income and median age are lower than average. These counties experience slightly faster retail establishment growth, but other demographic or industry factors did not influence whether a firm decided to establish a fulfillment center or not. Using those economic and demographic factors that may affect fulfillment center establishment, I select control counties that have very similar growth rates to the treatment counties with a fulfillment center establishment. Figure 1 indicates the average retail property price changes relative to fulfillment center establishment between treatment and control counties. Both treatment and control counties exhibit similar price movement prior to a fulfillment center opening, thus meeting the parallel trend assumption. However, the average retail property values move differently following the local fulfillment center establishment.

Impact of ecommerce on retail property transaction values
Using a difference-in-difference specification, I test how the expansion of ecommerce affects local retail property values. The model includes property characteristics and market controls. If ecommerce affects brick-and-mortar stores negatively, we should expect to see lower transaction values for local retail properties, but results exhibit contrary behavior. The entry of an ecommerce fulfillment center has a positive effect on local retail store values. Retail properties located in counties where a fulfillment center operates sold at a 5.2 percent premium relative to properties in similar counties without a fulfillment center. The premium persists through the fifth year, with the largest gains seen in year three. The natural environment of commercial real estate transactions explains the delayed effect on the retail property market. The average time on market for a retail property in the full sample is 357 days, which helps explain why retail properties' transaction prices experience a delay.

This positive pricing effect is significant in the retail and multifamily markets, but not in the office and warehousing markets. This state of affairs indicates that the local market's stronger overall economic growth drives results. A concern stems from clustering effects—that is, the pricing effect will move upward if other retailers open their warehouses in the same counties at a similar time. Using the largest big box retailer as a means to isolate this effect, I exclude the counties where this retailer also opened a warehouse within three years of the opening of the proxy online retailer's fulfillment center. The results are robust as they persist even after I account for the clustering effects involving another retailer.

Local labor market structure changes
Establishing fulfillment centers changes the composition of the local labor market of treatment counties as businesses focus on warehousing services and generate greater demand for labor. The share of employment in the warehousing-related industries (NAICS codes 48 and 49), their salaries relative to a county's average, and the number of warehouse establishments relative to total business establishments increase by 0.9 percent, 1.1 percent, and 0.5 percent, respectively, relative to control counties following a local fulfillment center opening. The positive effects may spill over into the local retail labor market as overall labor demand increases. The retail trade sector also exhibits positive spillover effects on establishment and payroll in the third year (see figure 2). However, the increase in retail employment is not statistically significant. If general economic growth of the treatment counties drives positive labor market effects, I might expect to find that the finance and insurance industry (NAICS code 52) also grows following the establishment of a fulfillment center, which is not the case.

Figure 2: Trends in Payroll, Establishment, and Employment

Transportation and Warehousing

Retail Trade

Finance and Insurance
Source: Census County Business Patterns and author's calculations

Which areas are most sensitive to these effects?
Areas where warehousing employees tend to live—that is, areas with higher population densities, lower median incomes, higher renter-occupied rates, lower levels of education, younger population, and lower rates of internet access—tend to pay higher premiums for retail properties and exhibit a stronger, positive effect on local retail property values after the establishment of a fulfillment center.

Overall, the findings in this post seem to provide another silver lining to the consolidation among retailers and the increased focus on digital rather than physical footprints. Contrary to the common belief that online retailers destroy local communities, the data appear to indicate that ecommerce could positively contribute to local economies and the value of retail property. In a county with a fulfillment center, retail properties transact 5.2 percent higher compared to properties in counties that appeared to be a strong fit for a fulfillment center but did not receive one. Growth in labor demand for the warehousing sector and a potential positive spillover effect to the retail sector also support these findings. The expansion of ecommerce anchored by the establishment of a fulfillment center might help local economies and could cause local labor markets to specialize in warehousing. Finally, a heterogeneity test suggests that the positive pricing effect on local retail values is stronger in areas where fulfillment center employees are more likely to live.

While these results perhaps cast online retailers in a positive light, the persuasiveness of this post's findings are limited. Higher property values could be the result of local government subsidies to online retailers, and I don't assess the long-term impact on warehousing and retail labor markets. As this analysis concerned only the county level, its ability to capture the overall effects of ecommerce on the servicing areas is limited. To promote a mutually beneficial relationship between ecommerce firms and brick-and-mortar retailers, I suggest that local governments and policymakers need to be aware of the spillover effects resulting from fulfillment centers and implement proactive, clearly beneficial policies for both the warehousing and retail industries.

May 27, 2020

COVID-19 Mortgage Relief—The Role of Income Support

The COVID-19 pandemic has led to a large number of furloughs, layoffs, reductions in hours worked, and wage cuts. Anticipating that many homeowners would consequently have problems paying their monthly mortgage bill, the U.S. Department of Housing and Urban Development ordered all mortgage servicers of federally backed debt to provide forbearance to any homeowners affected by the crisis. In addition, bank regulators encouraged lenders to forbear and restructure mortgages for borrowers affected by the shutdown, actions that staved off an immediate wave of foreclosures. At the end of the forbearance window, borrowers will likely be offered a series of repayment schemes: starting with a period of catch-up payments, then moving to extended terms on their mortgage or possibly even rate reductions. However, if the borrower has not returned to work, paying for what is effectively a new mortgage obviously poses a challenge. Options such as creating a modified repayment plan, lowering the mortgage interest rate, or extending the term of the loan might not be enough for a borrower who has experienced a substantial income loss.

In 2009, researchers at the Boston Fed proposed an alternative policy of supplemental mortgage payment assistance targeted to underwater borrowers experiencing a significant reduction in disposable income due to factors such as employment loss or medical costs associated with illness. That 2009 research built on earlier Boston Fed research demonstrating that—during a previous housing market downturn—most underwater households continued to pay their mortgages unless they were hit with a further reduction in earnings or increase in expenses. The idea that mortgage default is caused by both a negative house price shock and a negative income/employment shock is known as the "double trigger" theory of default. However, the empirical evidence on the double trigger theory was limited. Underwater homeowners in areas with increased unemployment appeared to default more, but this was mostly an interesting correlation, not necessarily a causal relationship.

Since the Great Recession, considerable research (here, for example) has tried to identify the central role income shocks play in default. The econometric challenge is that shocks to income from changes in employment or wages tend to be capitalized into house prices. So a community experiencing the second trigger from widespread job loss, say, will likely also experience a drop in house prices, making it difficult to isolate the real cause of default. In a forthcoming paper we consider the unique sources of changes in employment and income arising from the hydraulic fracking boom in Pennsylvania in the late 2000s to isolate the second trigger from the first.

Fracking involves injecting large amounts of water, sand, and potentially toxic chemicals underground at great pressure to break shale formations and release the trapped natural gas. The fracking process also involves piercing aquifers, storing and treating large quantities of contaminated water, and employing heavy equipment. Some evidence shows that these real or perceived negative features lower the value of homes near fracking wells. At the same time, the shale boom increased demand for middle- and low-skilled workers and generated significant royalty payments to many property owners.

Observing the performance of mortgages that originated before fracking began allows us to treat the resulting shale boom as an experiment where household incomes were sustained (or increased) even as housing prices were flat or declining. Using geological information to predict the location of fracking activity, we find that fracking wells significantly raised total household income, from both wages and royalties, and the wells appear to have increased employment in fracking-related industries. At the same time, fracking does not appear to have raised house prices or made it less likely that a household has negative equity. However, fracking does significantly reduce the probability that a mortgage becomes seriously delinquent (that is, when a borrower misses more than a few payments).

In addition, when we use only geology to predict the location of fracking wells, we get a much larger decline in mortgage delinquency, suggesting that more vulnerable communities were quicker to embrace fracking. Finally, the ameliorative effects of fracking were concentrated among borrowers who are likely to be underwater on their mortgages (the first trigger), consistent with the double trigger hypothesis, since the theory predicts that borrowers with positive equity are unlikely to default in the first place.

Our results suggest that an effective strategy for preventing a foreclosure crisis in the current situation is direct support of household income. Indeed, the Coronavirus Aid, Relief, and Economic Security Act (commonly known as the CARES Act) contains several income transfers to help sustain household budgets, including expanded unemployment insurance, direct cash payments to most households, and loans to small firms that are forgivable on the condition that they sustain employment through the shutdown. It is our view that these programs are not simply helping to sustain families during the crisis, but they're also limiting disruption to the housing market. Depending on how the crisis evolves in the coming months, further income support for affected households may forestall the need for less efficacious interventions to aid distressed borrowers.