The Economy and Square Footage Don’t Run on the Same Clock

The Mexican economy is moving forward, but barely. Since 2018, tertiary activity has grown nearly 10%, while secondary activity is just over 2% above its level at the time; and for July, INEGI estimates that both grew just 0.1% from the previous month¹.

For the real estate market, the underlying question is when—and how—those economic movements ultimately translate into square footage. Economic changes do not necessarily translate immediately into real estate decisions, which means there may be a lag between an increase in activity and the leasing of additional space.

In the industrial market, that lag appears fairly clearly. If we leave aside the pandemic and its immediate recovery, when the same shock first drove down and then lifted both economic activity and demand for space, secondary activity and industrial absorption have practically stopped moving together within the same quarter since 2022.

This means something simple: the economy can change today and the industrial market may reflect that change months later. In fact, SiiLA data show that the relationship, virtually nonexistent in the same quarter, strengthens over time and reaches its greatest intensity close to a year later². And it makes sense. Before leasing more space, a company can increase production at the facilities it already occupies, use available capacity or wait until the change in its operations is sufficiently durable to justify a real estate decision.

However, that lag does not appear to result simply from a sequence in which space is built first and occupied months later. Economic activity appears to precede both inventory expansion and absorption separately, without the data indicating that one is simply a lagged consequence of the other³.

Offices work differently. Since 2022, movements in a single quarter also say little about demand for space. In fact, unlike in the industrial sector, waiting several months does not strengthen the relationship with the economy. Expanding the period we observe makes that relationship visible: when we compare each quarter’s tertiary activity with all office absorption recorded over the previous 12 months, the two move closely together⁴. This suggests that office demand does not appear to respond with the same lag as industrial demand, but instead tracks tertiary activity more closely; a relationship that is easier to see over longer periods than in the small movements of individual quarters.

The contrast also occurs in markets that are physically very different. Between 2020 and 2026, industrial inventory in the markets analyzed grew 53.5%, compared with 13.3% for offices; and while industrial vacancy averaged just 3.2%, office vacancy was around 18.6%⁵. In other words, over the period, the industrial market expanded much faster and operated with considerably lower availability than the office market.

This suggests that the same expansion in economic activity faces different conditions as it translates into square footage: in industrial, with little space available, greater demand for capacity may ultimately require new space; in offices, greater availability may allow some of that demand to be accommodated within existing space.

The differences between sectors matter, then, because an industrial square foot and an office square foot do not appear to run on the same economic clock. In that sense, the mere 0.1% increase INEGI estimates for July does not, by itself, mean real estate demand will slow or grow by the same proportion. It is, however, a useful signal: not to determine how many square feet will be leased or when a specific transaction will occur, but to identify earlier the economic pulse that, through different timing and mechanisms, ultimately reaches the real estate market. Because square footage responds not only to how much activity grows, but also to when and for how long it does.

Want more information and analysis on the commercial real estate market? Visit SiiLA Market Analytics or email us at contacto@siila.com.mx.

***

¹ SiiLA calculations using INEGI data. The recent reading uses the Timely Indicator of Economic Activity (IOAE), which anticipates the evolution of the Global Indicator of Economic Activity (IGAE) through nowcasting techniques and does not replace the observed figures subsequently published by the Institute. The comparison with 2018 uses the annual average of that year’s seasonally adjusted indexes as a reference. For July 2026, the IOAE estimates a monthly change of 0.1% in both secondary and tertiary activities; the 95% confidence intervals are −2.3% to 2.4% and −0.9% to 1.2%, respectively. On an annual basis, the estimates are 1.8% and 2.8%, with intervals of −0.5% to 4.2% and 1.7% to 3.9%. The estimates are subject to revision as new information becomes available.

² SiiLA estimates using quarterly INEGI data and industrial net absorption in Mexico City, Guadalajara and Monterrey. For the period following the regime change identified around Q3-22, each quarter’s secondary activity was compared with industrial absorption observed between zero and six quarters later. The association is virtually nonexistent contemporaneously and peaks around +4Q (Pearson = 0.546; Spearman = 0.440; n = 13). The result remains positive after removing time trends (r = 0.652; p = 0.016) and retains its sign in first differences, although with lower statistical precision (r = 0.475; p = 0.118). Additional tests using block bootstrap, Newey-West errors, winsorization, Cook’s distance and temporal-direction placebos do not alter the central interpretation. Given the sample size and the observational nature of the analysis, the result should be understood as a temporal association concentrated around one year, not as causality or evidence of a fixed four-quarter lag.

³ To distinguish the lag associated with demand from the one related to the market’s physical expansion, secondary activity was compared with the subsequent addition of new industrial inventory and, separately, new inventory was compared with future absorption. Since Q2-22, the relationship between secondary activity and new supply is weak in the same quarter and strengthens around three to four quarters later. In the observed-data window from Q2-22 to Q1-26, excluding 2Q26, Pearson rises from −0.061 at Q0 to 0.621 at +3Q and 0.670 at +4Q; Spearman reaches 0.544 and 0.629, respectively. At +4Q, both coefficients are statistically significant (p = 0.017 and p = 0.028). By contrast, the relationship between new inventory and industrial absorption is primarily contemporaneous and fades quickly at subsequent horizons. Accordingly, the data are consistent with economic activity preceding both inventory expansion and absorption, but do not support a mechanical sequence in which new supply by itself explains absorption several quarters later. These associations do not establish causality or identify the underlying economic mechanism.

⁴ Since Q2-22, the correlation between tertiary activity and quarterly absorption has been virtually nonexistent (r = −0.047); when four-quarter cumulative absorption is used, it rises to 0.759. The association remains after removing a linear time trend (r = 0.667). In a regression for Q2-22–Q2-26, the tertiary activity coefficient remains significant with heteroskedasticity- and autocorrelation-consistent errors (HAC; p = 0.022). When Q2-26, a quarter that incorporates estimated information, is excluded and only observed data from Q2-22 to Q1-26 are used, the model reaches R² = 0.767 and p = 0.00062. When absorption is shifted into future periods, the association progressively declines, so no lag pattern equivalent to the one observed in Industrial is identified.

⁵ Between Q3-20 and Q2-26, industrial inventory increased from 29.69 million to 45.57 million square meters (+53.5%) in Mexico City, Guadalajara and Monterrey, while office inventory increased from 10.08 million to 11.42 million (+13.3%). Over the same period, average vacancy was 3.16% in Industrial and 18.6% in Offices. These differences reflect the structure of the two markets and do not prove that they caused the different temporal patterns identified in the analysis.