Does Retail Improve Office Building Performance?
- Office buildings with retail absorb more space without commanding higher rents or recording higher move-outs.
Retail ranks among the amenities most valued by office occupiers. It would therefore be reasonable to expect buildings with retail space to be more occupied and command higher rents. But SiiLA’s data in Mexico tell a different story.
An analysis of 248 office buildings found that those with retail space have approximately eight percentage points higher office vacancy rates than buildings without retail, even after controlling for differences in size, age, class, and submarket. However, no statistically significant differences were identified in market rents¹.
The higher vacancy, however, does not appear to be explained solely by the presence of retail, but also by where many of these buildings stand in their real estate life cycle. On average, office buildings with retail are 5.6 years newer² than those without commercial space and, like most recently completed developments, remain at...
On the absorption side, office buildings with retail averaged 63.9 square meters per 1,000 square meters of gross leasable area annually, compared with 37.4 square meters for buildings without retail. The difference was statistically significant and is consistent with the fact that these assets also tend to be newer and concentrated in higher-quality buildings and specific submarkets⁴.
Move-outs, by contrast, were virtually identical in both groups: 36.3 annual square meters per 1,000 square meters of gross leasable area in buildings with retail versus 37.0 square meters in those without retail, a difference that was not statistically significant.
Taken together, these results suggest that greater tenant losses do not drive the higher vacancy rate observed in office buildings with retail. Instead, although they begin with higher levels of vacancy, they add new tenants at a faster pace while maintaining move-out levels similar to the rest of the market, resulting in higher net absorption. If that dynamic persists, the vacancy gap should narrow as these assets progress through the stabilization process.
The findings also show that comparing office buildings without considering their stage of maturity can lead to misleading conclusions. In this case, an indicator that initially appeared to signal weaker performance ultimately described a more active leasing dynamic among a different generation of assets.
To learn more about the trends shaping 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 office buildings in Mexico City, Guadalajara, Monterrey, and Querétaro with sufficient information to identify the presence of retail space. Differences were estimated using regression models controlling for gross leasable area, building age, class, and submarket. Robustness was assessed using heteroskedasticity-robust standard errors (HC3), 5,000-bootstrap resampling, Welch, Wilcoxon, and permutation tests, age-sensitivity analyses, winsorization of extreme values, influence diagnostics (Cook’s distance), propensity score overlap weighting, and an Augmented Inverse Probability Weighting (AIPW) estimator to evaluate the sensitivity of the estimates across alternative econometric specifications.
² Buildings with retail differ systematically from the rest of the sample. On average, they are 5.6 years newer (17.1 versus 22.7 years; medians of 15 and 22.5 years), have greater gross leasable area (17,660 versus 12,426 m²; medians of 12,624 and 10,073 m²), and are distributed differently across building classes and submarkets, with a greater concentration in Class A+ assets and corridors such as Polanco and Reforma. These differences were explicitly incorporated into the estimation. Gross leasable area showed no statistically significant association with vacancy, either on its own or after controlling for age, class, and submarket.
³ The estimated relationship between retail and vacancy was evaluated under alternative specifications for the relationship between building age and occupancy. Under the linear model, the estimated retail effect was approximately 8.0 percentage points (p≈0.031). Allowing for a quadratic relationship between age and vacancy reduced the estimate to 6.2 percentage points (p≈0.084). Finally, modeling both age and size flexibly using splines reduced the estimate to 4.5 percentage points (p≈0.193). The systematic decline in the estimated effect is consistent with the hypothesis that part of the relationship initially attributed to retail is actually associated with the younger age of buildings that incorporate retail.
⁴ The simple comparison showed higher absorption among office buildings with retail. However, after controlling for age, gross leasable area, class, and submarket, the difference was no longer statistically distinguishable (p = 0.067; adjusted p = 0.199), indicating that part of the observed gap coincides with structural differences between the two groups of buildings.
Subscriber? Sign In
