The COVID-19 pandemic fractured the global economy in ways that no single generation had experienced. When the initial shock faded and restrictions began to ease, there was hope that “recovery” would be a simple, uniform return to growth—a V-shaped bounce, as some optimists called it. It did not turn out that way. By now, we see clearly that economic recovery is deeply uneven, both across and within countries. For economists and policymakers, understanding this patchwork requires more than intuition or anecdote. It demands rigorous, structured data, and, to a surprising extent, the ISIC coding framework offers precisely the level of granularity needed to make sense of what happened.
Between 2021 and 2024, ISIC codes allowed analysts to go beyond broad aggregates and dig into the details of which sectors were rebounding, which were stagnating, and which had entered entirely new patterns of behavior. Looking at tourism, for example, the collapse was immediate and nearly total in the early pandemic months. Codes covering air transport, accommodation, and travel agencies showed declines so sharp that standard models almost broke under the strain. But the story did not end there. In some countries, as vaccination rates rose and restrictions loosened, recovery in these sectors was astonishingly rapid, at least in percentage terms, even if overall levels remained below the pre-pandemic trend.
Contrast that with manufacturing, where the story is less clear-cut. ISIC divisions dealing with food processing, medical supplies, and certain electronics rebounded early—sometimes even outperforming previous years, driven by new patterns in consumption and supply chain adjustments. Yet other branches of manufacturing, particularly those linked to automotive production or heavy machinery, continued to struggle well into 2023, hampered by persistent logistics disruptions and shortages of key inputs. Here, the codes capture not just a general rebound but a bifurcation within manufacturing itself. There’s a temptation to treat the sector as a monolith, but ISIC data pushes back against this.
Retail, another headline sector, also splits into divergent stories. E-commerce and delivery-focused businesses, newly tracked in detail thanks to recent revisions of ISIC, surged ahead. Traditional brick-and-mortar establishments saw less consistent recoveries. In some cases, the codes tell a story of complete reinvention—enterprises that switched codes as their business models shifted from physical sales to online fulfillment. There’s something slightly unsettling about the speed with which some segments adapted, while others simply vanished from the registers.
The real power in using ISIC codes during this period lies in the possibility for time-series analysis. With consistent coding, one can track monthly or quarterly employment and output by sector, overlay government support interventions, and even try to tease out which policy tools aligned with faster recovery. Yet, it must be said, there are ambiguities. Not all countries updated or harmonized their ISIC coding with equal diligence. There were lags in reporting. Some sectors, especially those at the interface of the informal economy, are still only partially captured. This is where caution—perhaps even skepticism—is required. Sharp recoveries in the data sometimes reflect statistical catch-up, or even simple reclassification, as much as real activity.
For economists seeking to incorporate ISIC-based data into forecasting models, a few methods have proven robust. First, the sectoral approach to GDP forecasting, where model weights are dynamically adjusted according to observed rebound patterns in high-frequency ISIC-coded data. Second, state-space or structural break models, where time-series discontinuities in specific codes (e.g., the dramatic drop and bounce in accommodation services) are treated as exogenous shocks and then tracked for reversion or persistence. This sounds technical, but in practice, it is a way to avoid letting outlier sectors skew headline forecasts unduly.
There’s also an opportunity for more granular, scenario-based modeling. Suppose the data show that the recovery in transport and hospitality is lagging by a year in some countries relative to others. Forecasters can construct scenarios in which those sectors never quite regain their former size, or, alternatively, that pent-up demand drives overshoots in the years ahead. ISIC-coded data is flexible enough to allow for such experimentation, but again, modelers must document the underlying assumptions and recognize the limitations of the data. Not every code tells a complete or uncontested story.
A persistent challenge is the lag in reclassification. Sometimes, businesses shift activity but do not update their ISIC code right away, meaning there is a silent churn under the surface. Analysts must be alert to sudden surges or drops that may have less to do with economic fundamentals and more to do with administrative adjustments or government-mandated re-coding. This issue complicates forecasting but is difficult to avoid.
Looking back on the past four years, perhaps the most valuable lesson is that sectoral granularity matters. Policymakers cannot afford to craft interventions using only the broadest aggregates. Whether targeting relief funds, designing training programs for displaced workers, or predicting tax revenue, the difference between, say, ISIC code 5510 (Hotels and similar accommodation) and 5610 (Restaurants and mobile food service activities) is not academic. During recovery, each behaved differently, and each faced unique pressures.
The path forward, then, involves not just continuing to use ISIC codes in the usual way but also investing in greater timeliness and accuracy of updates. A living statistical system is only as useful as its capacity to reflect change. The pandemic may have been extraordinary, but the lesson is enduring: when the world is upended, the best maps are the ones drawn with the finest lines.
If there is a final word here, it’s that data alone doesn’t confer insight. The ISIC framework provides a common language, but interpretation—always—remains the task of the economist, the policymaker, the statistician. If we learn to listen closely, these codes have more to tell us about resilience, vulnerability, and transformation than we might have expected, even a few short years ago.Top of Form