No Shopping Center Competes Against an Entire City

The most intuitive explanation would be to attribute that concentration to the characteristics of the properties themselves. The data, however, reject that hypothesis. Size shows virtually no relationship with vacancy rate, and age does not consistently explain which shopping centers concentrate available space. Format appears to matter, but those differences almost disappear once two outlets with exceptionally high vacancy rates are excluded. In other words, the concentration of availability persists even after the most obvious explanations have been ruled out.¹

The analysis, however, begins to make sense once shopping centers are no longer viewed as part of a homogeneous metropolitan market and are instead analyzed at the submarket level. At that scale, it becomes clear that shopping centers do not compete against every property in a city, but against the handful consumers actually consider as alternatives. As a result, incorporating that immediate competitive environment significantly improves the explanatory power of the models,² suggesting that availability responds less to an asset’s physical characteristics than to the corridor in which it competes.

San Pedro in Monterrey, along with Santa Fe and Norte in the Mexico City Metropolitan Area (MCMA), for example, are among the submarkets that contribute the most to the vacancy rate within their respective markets. Yet the evolution of local absorption over the past three years shows that the same concentration of availability can arise from very different competitive dynamics.

According to SiiLA, Santa Fe lost nearly seven percentage points of its share of MCMA absorption between the second quarter of 2023 and the second quarter of 2026. Norte followed the opposite path, gaining just over three percentage points, largely offsetting the new inventory delivered over the same period. San Pedro also increased its share within Monterrey—by more than nine percentage points—although it still posted slightly negative cumulative net absorption over the past three years.

The implications are significant. Two submarkets may exhibit similar levels of availability while requiring entirely different strategies to reduce it. In some cases, the challenge is to recover demand from competing corridors; in others, to absorb recently delivered supply; and in others still, to fill space that has remained vacant since previous market cycles.

Viewed this way, the vacancy rate describes the outcome, while absorption trends reveal the process that produced it. The concentration of available space therefore ceases to be understood as a uniform market phenomenon and instead emerges as the sum of distinct competitive dynamics unfolding simultaneously within each city.

To learn more about the performance of Mexico’s commercial real estate market, visit SiiLA Market Analytics or contact us at contacto@siila.com.mx.

***

¹ The analysis covers 154 shopping centers located in the Mexico City Metropolitan Area, Guadalajara, and Monterrey as of Q2 2026. The vacancy rate was calculated as vacant area divided by each property’s gross leasable area (GLA). To evaluate whether the observed concentration was driven by structural property characteristics, linear regression models with heteroskedasticity-robust (HC3) standard errors were estimated, using the logarithm of gross leasable area, age, shopping center format, and market as explanatory variables. Size showed virtually no association with the vacancy rate: Pearson and Spearman correlations were -0.107 and -0.003, respectively, while the coefficient on the logarithm of GLA was not statistically significant in any of the estimated specifications (p≥0.68 in the main models). Age also showed no consistent relationship, with Pearson and Spearman correlations of -0.110 and -0.096, respectively, remaining statistically insignificant across linear, quadratic, and natural spline specifications. As for format, results proved sensitive to model specification because of two outlet centers with a combined vacancy rate of 26.1%; once excluded, none of the remaining formats differed significantly from the reference category in the robust model (p-values between 0.24 and 0.62). Concentration was also not driven by a handful of extreme observations: removing the five properties with the largest amount of available space reduced the Gini coefficient from 0.668 to 0.628, while the share of properties required to account for 80% of available space increased from 29.2% to 32.2%. A bootstrap with 5,000 resamples placed the latter estimate within a 95% confidence interval ranging from 24.7% to 34.4%. Taken together, these tests indicate that neither size, age, nor—once the effect of the two outlet centers is isolated—format consistently explains the observed concentration.

² After ruling out the structural characteristics of individual properties, the next step was to evaluate whether location explained the observed concentration. To do so, the metropolitan market variable was replaced by each property’s specific submarket while controlling for gross leasable area, age, and shopping center format. Adjusted R² increased from 0.0266 to 0.1178, while R² rose from 0.0838 to 0.291. Under this specification, San Pedro (Monterrey) showed an estimated effect of +13.5 percentage points on the vacancy rate (p=0.0065), Norte in the Mexico City Metropolitan Area +11.2 percentage points (p=0.0096), and Santa Fe +11.1 percentage points (p=0.0469), relative to the reference category. Overall, the results indicate that submarket-level location substantially increases the model’s ability to explain observed availability.