Urban Planning and the Effect of Neighborhood Characteristics and Spatial Location on the Economic Performance of Commercial Spaces

On your daily route there is a small shop you pass without noticing, until one morning you find the door shuttered and the shelves empty. You feel a strange void, as if the street had lost a phrase from its usual speech. We tend to explain it easily: the owner managed badly, or his goods did not please people. But decades of retail geography research say something more unsettling. A shop’s fate is written, to a large extent, outside its walls: in the income of its neighbors and their ages, in the number of people who work in the district, in the height of the surrounding buildings, in the number of crimes within a quarter mile, and even in the performance of the adjacent neighborhood. The store is not an isolated island but a creature that breathes the air of the place where it stands. The question that occupies researchers is not whether the neighborhood matters, but how much it matters, in which direction, and for whose benefit. And what the numbers have revealed overturns much of what we assume to be obvious.

The Market the Shopkeeper Cannot See

Most of this body of research starts from a logic attributed to Hotelling. The density of stores and their ability to survive is a function of the density of consumers, the fixed costs of the store, and the cost of the customer’s travel to reach it. If demand, cost, or the attractions of a neighborhood change, the whole retail equilibrium shifts with them. The study by Meltzer and Capperis on store turnover in New York, and the study by Schuetz, Kolko, and Meltzer on disparities between neighborhoods, both rest on this framework.

The review by Turhan, Akalin, and Zehir of store-location selection criteria shows that the field organizes the location decision around seven families of criteria. The first is performance measures: sales volume, profit, market share, patronage, and price elasticity. The rest are population structure, economic factors, competition, the level of saturation, the characteristics of the store itself, and “magnets,” meaning the neighboring elements of attraction. Performance here is not one factor among others. It is the criterion that all the other factors are used to predict.

Yet researchers repeat two reservations. The first is that causality runs in both directions: stores follow consumers, but consumers also move toward neighborhoods where services are available, so most studies settle for documenting association without establishing cause. The second is that the “neighborhood” itself is an elastic concept. It is measured at times by census tract, at times by ZIP code, at times by the blockface, at times by a quarter-mile circle, and at times by a service area drawn with a full mile’s diameter, and the results change as this choice changes.

An architect who reads this understands that the boundaries he draws around his project on the map are not innocent. They determine what will appear influential and what will vanish. Let us see what the picture looks like when we leave theory and descend to the numbers.

What Is Decided Before the Door Opens

In a study of an American discount grocery chain, Smith and Sanchez collected data on 655 stores in 135 markets, drawn from in-store surveys, observations of competitors, and demographic data. They then modeled six performance measures: sales, sales per square foot, sales per customer, market penetration, sales growth, and profit. The result was that the demographic characteristics of the trade area alone explained about 60 percent of the variation in sales, sales per customer, and market penetration. When the characteristics of the store itself and of the competition were added, explanatory power exceeded 70 percent for sales and profitability.

But the picture does not run in a straight line. The direction of the effect differs with the measure. Sales were higher where average income was lower, while sales growth and profit were lower in poorer areas. Racial composition also mattered: sales rose with a higher share of Black residents and fell with a higher share of residents of Hispanic origin, holding other factors constant. The implication is that the question “is this neighborhood good for business?” is incomplete unless we specify what we mean by good: volume, profit, or growth.

The study also added a striking tool, “gradient” variables, meaning the ratio of the characteristics of the three-mile circle to those of the one-mile circle. These variables raised explanatory power because they reveal how the near neighborhood differs from its wider surroundings. A store does not live only within a homogeneous neighborhood. It lives on the edge of a transition between two levels of income and density.

For the urban planner, this means that proximity to the richer or the poorer matters no less than the neighborhood average. But this study looks at each store individually, and the city organizes itself differently when we look at it as a network.

When One Neighborhood Wins at Its Neighbor’s Expense

In Seoul, Sung studied aggregated commercial sales across 11,182 statistical units, using a spatial autoregressive model with autoregressive disturbances and a generalized spatial two-stage least squares estimator. The aim was to separate the direct effect of the physical environment within a unit from the indirect effect transmitted from neighboring units.

The results deserve reflection. The density of workers within a unit had a positive effect: each 1 percent increase raised sales by about 0.24 percent. The density of resident population had a negative direct effect. The mix of land uses and the average height of buildings had positive effects, traditional markets a negative one as competitors, and ease of access to the subway a positive one. But the most consequential finding is that the indirect effects came, for the most part, with the opposite sign from the direct effects. That is, adjacent neighborhoods compete for sales under a zero-sum logic: what one neighborhood gains, its neighbor loses.

The study also exposes a common methodological flaw. Variables such as proximity to the city’s central business district, road accessibility, and large parks appeared statistically significant in ordinary models and in spatial lag models, then lost their significance once spatial dependence and heteroskedasticity were addressed together.

At a smaller scale, the study by Banai and Antipova in Memphis compared two neighborhood shopping centers, each anchored by a downsized big-box store from the same chain, one urban and one suburban. Using block-level analysis, they found that proximity to residential use was positively associated with sales at the half-mile and three-quarter-mile ranges, that larger establishments recorded higher sales, and that proximity to other businesses lowered sales, the competition effect expected in central place theory.

If neighborhoods compete in this way, what happens to those that lose the competition? This is what led other researchers to a different question: who gets the stores in the first place?

The Retail Desert: When the Numbers Deceive Us

A wide sector of research asks whether poor and minority neighborhoods are “retail deserts.” Schuetz, Kolko, and Meltzer studied 58 large American metropolitan areas, using establishment-level employment data and ZIP-code-level income and poverty data. They found that high-poverty neighborhoods have lower retail employment density in general, and in supermarkets, pharmacies, restaurants, and laundry services in particular, and that the main cause was a shortage of jobs in chain stores rather than in independent ones. The average size of an establishment rose with income in every category and shrank in poor neighborhoods.

The surprise is that these high-poverty neighborhoods had a larger number of supermarket establishments, but with lower employment density, smaller establishments, and fewer chains. The answer to the question “is it a desert?” therefore depends on the yardstick: counting establishments says no, while measuring employment and size says yes.

The study by Meltzer and Schuetz in New York produced consistent results with fuller detail. Lower-income neighborhoods and those with minority majorities have lower densities of establishments and employment, their establishments are smaller, the share of “unhealthy” chain restaurants is higher, gyms are fewer, and upscale grocery stores are absent. The study made an exception of neighborhoods with Hispanic majorities, which were distinguished by greater diversity in retail and food services and by better physical access to commercial streets. Between 1998 and 2007, retail grew faster in neighborhoods that began with low property values and then saw rising housing prices.

The roots of this debate go back to Hall’s 1983 study, which covered 191 food stores in 38 neighborhood markets in New York State. He found that stores in neighborhoods with high concentrations of Black and elderly residents offered higher prices, lower quality, lower cleanliness, and narrower variety. Income was not the strongest predictor. Racial and age composition and the crime rate were, and higher operating costs did not fully explain these differences.

But if the neighborhood shapes what is displayed on a store’s shelves, where does this effect appear in its clearest economic form? In the rent.

The Price That Betrays the Worth of the Neighborhood

Store sales data are mostly confidential, so commercial real estate researchers turn to substitutes: rents, vacancies, and property values. These substitutes tell the story of the neighborhood with remarkable precision.

In Quebec City, Des Rosiers, Theriault, and Menetrier studied 1,007 commercial units, of which 941 were used, in ten shopping centers. They combined lease data with an origin-destination trip survey and census income data, and built an economic potential index, which is the number of actual customers multiplied by the income of the zone, along with a gravity-type index of the center’s attractiveness. When these spatial indices were added to the baseline rent model, explanatory power rose from 0.429 to 0.476. The economic potential index proved highly significant, behind only the size of the unit and its type (the kiosk), raising base rent by about 3.5 percent per point. The researchers’ conclusion is that external location factors interweave with the internal characteristics of the center, such as its image and its mix of stores.

In Atlanta, Hardin and Wolverton modeled the rents of 248 neighborhood shopping centers together with vacancies, around a service area one mile in diameter. The purchasing power of the area raised rents strongly, the area’s vacancy rate predicted the center’s vacancy, a “dark anchor store” increased vacancies, and obsolescence lowered rent at a decelerating pace. A striking halo effect appeared for higher-order centers: proximity to a mall or a community center raised rent, but this advantage fades quickly, within about half a mile for malls and about 0.2 miles for community centers.

At the level of whole cities, Benjamin, Jud, and Winkler studied quarterly vacancy rates in 46 American metropolitan areas between 1982 and 1996, and found strong spatial correlation among neighboring areas, with coefficients between 0.68 and 0.74. The local vacancy level was governed by conditions in surrounding markets, while its changes were tied to changes in the national rate to varying degrees across markets.

Rent, then, carries the whole history of the neighborhood. But whether a shop stays open depends on another factor, one that is far more perplexing.

Does Crime Extinguish the Store or Protect It?

Before crime, let us begin with turnover. Meltzer and Capperis followed more than 2,100 census tracts in New York over twenty years, using establishment data to distinguish staying, entry, exit, birth, and death. When commercial factors and consumer factors were entered together, consumer characteristics explained turnover more than the characteristics of the commercial environment did. Larger household size and a higher share of white residents were associated with the least turnover, while population growth was the strongest predictor of higher turnover. Stores selling frequently consumed necessities were more stable, and food establishments fluctuated less than general retail.

Crime raises a more complex debate. In the study by Lens and Meltzer on New York, using precisely located crime data and instrumental variables, crime lowered commercial property values, with elasticities between −0.1 and −0.5. Within a quarter mile, one additional violent crime lowered transaction prices by about 0.5 percent and one additional property crime by about 0.2 percent, and in assessed-value models one additional crime on the same blockface lowered value by between 2 and 6 percent. The effect was concentrated in lower-income neighborhoods and those with the highest shares of Black residents, which are “harmed twice”: more crime and a stronger economic response to it. Strikingly, the ordinary estimate proved biased, even showing a positive association between crime and value at the blockface level, because commercial activity itself attracts crime.

In Southern California, Hipp and colleagues analyzed more than seven million establishment-year observations. They found that violent crime and property crime both raise the likelihood of failure and relocation, and that higher crime in a destination neighborhood lowers the likelihood that an establishment moves there. The effect of property crime was stronger on retail: a one-standard-deviation increase raised the likelihood of retail store failure by about 3.4 percent and of relocation by about 5.1 percent, and lowered the likelihood of choosing a high-crime neighborhood by about 20 percent. The researchers conclude that the largest effect of crime falls on the choice of destination rather than on exit, meaning that it prevents entry more than it expels those already there.

Not everyone agrees. The study by Bates and Robb of young small urban establishments found that those reporting a strong negative effect of crime on profitability were no less likely to survive than others. Neighborhood-oriented establishments that considered crime a major problem had higher chances of survival, which the two researchers explained with a protected-market logic: crime keeps competitors away and lowers rent.

Conclusion

Perhaps, when we pass the closed shop on our route, we should ask a question other than “who is to blame?” If the fate of a small shop is decided by the income of its neighbors and their ages, by the density of workers around it, by the proximity of a larger center, by the crime on its street, and by the performance of the adjacent neighborhood, then who holds the right to draw these conditions? The architect designs a ten-meter facade, but it hangs on a network of causes that does not appear in the drawing. And the neighborhoods where sellers stumble are the very ones that pay more: higher prices, lower quality, more crime, and a harsher economic response to it. The studies themselves also acknowledge that they know more about a few cities, namely Seoul, New York, Quebec City, Atlanta, Memphis, and Southern California, than about the rest of the world’s cities. So does the barred door you pass each morning mean that its owner failed, or that the neighborhood failed him before he could fail himself?

✦ ArchUp Editorial Insight

The research reviewed in this article starts from a measurement constraint rather than a retail question. Store-level sales are proprietary, so researchers study rents, vacancies, employment counts, and establishment counts as proxies, which means the neighborhood becomes legible mainly through the instruments that capital already uses to price it. What those instruments record is consistent across cities. Smith and Sanchez found that trade-area demographics alone explained about 60 percent of the variation in sales across 655 stores. Sung’s Seoul model showed indirect spillovers carrying the opposite sign from direct effects, so adjacent districts compete in a zero-sum arrangement. Hardin and Wolverton found that the rent premium from proximity to a mall dissolves within roughly half a mile, while Des Rosiers and colleagues found trade-area purchasing power capitalized directly into lease rates. Retail supply follows the same gradient. Schuetz, Kolko, and Meltzer located the shortfall in poor neighborhoods in chain employment rather than independent stores, and whether such a neighborhood counts as a “desert” depends on whether one counts establishments or measures employment and size. Crime operates mostly at the point of selection: Hipp and colleagues found its strongest effect on destination choice rather than exit, and Lens and Meltzer found that disadvantaged neighborhoods absorb both more crime and a sharper price response. Bates and Robb’s counterpoint, in which crime thins competition and lowers rent, shows that the same condition can reverse survival odds depending on the market’s structure. The pattern is one of sequence. The lender underwriting a trade area, the chain selecting a site, and the analyst drawing a tract boundary each decide before, or at a distance from, the consequence. The independent operator, who had no seat where catchment, rent basis, and boundary were fixed, absorbs the result. The literature cannot fully separate cause from sorting, because stores and consumers move toward each other, and its evidence comes from six places, a limit its authors acknowledge. The storefront appears last, as the point where decisions made at the scale of the tract, the lease, and the credit file become visible at the scale of a door.

References

Meltzer, R., and Capperis, S. “Neighbourhood Differences in Retail Turnover: Evidence from New York City.” Urban Studies, 2016.

Schuetz, J., Kolko, J., and Meltzer, R. “Are Poor Neighborhoods ‘Retail Deserts’?” Regional Science and Urban Economics, 2012.

Turhan, G., Akalin, M., and Zehir, C. “Literature Review on Selection Criteria of Store Location Based on Performance Measures.” Procedia – Social and Behavioral Sciences, 2013.

Banai, R., and Antipova, A. “Retail-Center Viability and Urban Form: A Micro Analysis.” The International Review of Retail, Distribution and Consumer Research, 2016.

Hardin III, W., and Wolverton, M. “Micro-Market Determinants of Neighborhood Center Rental Rates.” Journal of Real Estate Research, 2000.

Smith, L. D., and Sanchez, S. “Assessment of Business Potential at Retail Sites: Empirical Findings from a US Supermarket Chain.” The International Review of Retail, Distribution and Consumer Research, 2003.

Sung, H. “Estimating the Spatial Impact of Neighboring Physical Environments on Retail Sales.” Cities, 2022.

Hall, B. F. “Neighborhood Differences in Retail Food Stores: Income Versus Race and Age of Population.” Economic Geography, 1983.

Meltzer, R., and Schuetz, J. “Bodegas or Bagel Shops? Neighborhood Differences in Retail and Household Services.” Economic Development Quarterly, 2011.

Des Rosiers, F., Theriault, M., and Menetrier, L. “Spatial Versus Non-Spatial Determinants of Shopping Center Rents: Modeling Location and Neighborhood-Related Factors.” Journal of Real Estate Research, 2005.

Benjamin, J., Jud, G. D., and Winkler, D. “Retail Vacancy Rates: The Influence of National and Local Economic Conditions.” Journal of Real Estate Portfolio Management, 2000.

Lens, M. C., and Meltzer, R. “Is Crime Bad for Business? Crime and Commercial Property Values in New York City.” Journal of Regional Science, 2016.

Hipp, J. R., Williams, S. A., Kim, Y.-A., and Kim, J. H. “Fight or Flight? Crime as a Driving Force in Business Failure and Business Mobility.” Social Science Research, 2019.

Bates, T., and Robb, A. “Crime’s Impact on the Survival Prospects of Young Urban Small Businesses.” Economic Development Quarterly, 2008.