Lockdowns Weren’t the Real Job Killer: Türkiye’s Pandemic Labor Market Reexamined

When COVID-19 swept across the world in early 2020, governments faced an agonizing trade-off: shut down sectors where people crowd together, or let the virus run through them. The economic damage that followed was often attributed directly to those shutdown orders. But a new study of Türkiye’s labor market, published in the open-access journal Heliyon, suggests that the story is more complicated. Economists Aslı Dolu and Hüseyin İkizler analyzed monthly administrative data spanning the pandemic and found that the sharpest disruptions to hiring and productivity were tied less to the formal restrictions themselves than to the broader, economy-wide shock of the pandemic. The finding carries weight well beyond Türkiye, because it speaks to a question that has divided economists since the first lockdowns: how much of the economic pain actually came from policy, and how much from fear, uncertainty, and voluntary behavioral change?

The research exploits a natural experiment embedded in Türkiye’s pandemic response. After the country confirmed its first COVID-19 case on 11 March 2020, the government suspended university and secondary education, closed shopping malls, cafes, and entertainment venues, restricted restaurants to takeaway service, imposed travel limits on air and public road transport, and eventually introduced curfews across all provinces. Crucially, these measures targeted specific sectors—transport and storage, accommodation and food services, real estate, and professional, scientific, and technical activities—while leaving manufacturing, mining, construction, trade, information technology, finance, and other industries formally untouched. That sector-specific pattern allowed the authors to designate four restricted sectors as a treatment group and compare their trajectories against a set of unaffected comparison sectors, before and after the restrictions took effect.

The method at the heart of the study is difference-in-differences, a workhorse of empirical economics. The logic is straightforward: if the treated and control sectors were moving along similar paths before the pandemic, then any divergence afterward can plausibly be attributed to the treatment—in this case, the restrictive measures. The authors estimated models with sector fixed effects, which absorb all time-invariant characteristics of each industry, and month fixed effects, which soak up macroeconomic shocks hitting every sector simultaneously. The coefficient of interest, an interaction between sectoral treatment status and the post-March 2020 period, captures the differential change in outcomes experienced by restricted sectors relative to everyone else. Standard errors were clustered at the sector level, and the parallel-trends assumption was tested formally using pre-treatment data.

The data came from two of Türkiye’s official statistical institutions. Monthly job vacancies and job placements were drawn from administrative records of the Turkish Employment Agency, known as ISKUR, covering seventeen sectors from March 2019 to March 2021. Productivity was harder to pin down at the monthly frequency, so the authors constructed a novel proxy: the consumer-price-index-adjusted turnover index published by TURKSTAT, divided by the number of paid employees in each sector. This revenue-based measure, covering ten sectors over thirty-four months from September 2018 to June 2021, captures short-run changes in real revenue per worker rather than fully adjusted labor productivity—a distinction the authors are careful to emphasize, since turnover and value added can diverge, and the measure ignores capital intensity.

The headline result is striking for its restraint. Across the full sample, the difference-in-differences estimates show no statistically significant association between the restrictive measures and job vacancies, job placements, or the revenue-based productivity proxy. In other words, once common pandemic dynamics were accounted for, sectors that were formally restricted did not fare measurably worse than sectors that were not. The descriptive data do reveal two distinct phases: a temporary dip in logged job vacancies during the initial wave, and a more prolonged but ultimately modest association between restrictions and job placements, with the larger magnitude appearing in the second wave beginning in November 2020. But the aggregate regressions point to a consistent conclusion—the immediate labor market contraction looks more like the footprint of the pandemic itself than of the policy curbs.

The picture becomes more textured when the authors split the economy by formality. Using sectoral informality estimates from earlier research, they classified mining, water supply, construction, wholesale and retail trade, accommodation and food services, and administrative services as high-informality sectors. In these subsamples, the point estimates told a story of vulnerability: job vacancies, placements, and productivity all appeared to decline more sharply in restricted informal sectors, while formal sectors showed relative resilience, with even a marginally positive association between restrictions and job vacancies in the preferred specification. The interpretation is intuitive. Informal firms rely on cash flows, face-to-face interaction, and short-term labor arrangements, leaving little room for remote work or organizational restructuring. Formal firms, by contrast, could pivot to digital tools and telework.

Yet the authors subject their own findings to a demanding robustness check, and the results temper the narrative. With so few sector clusters—as few as four in some specifications, and only a single treated sector in the informal subsamples—conventional cluster-robust standard errors can understate uncertainty. Applying a wild-cluster bootstrap procedure, the study finds that every informal-sector estimate that appeared significant under conventional inference loses statistical significance, with bootstrap p-values ranging from 0.17 to 0.20. The formal-sector job vacancy estimate weakens from conventional significance to a marginal bootstrap p-value of 0.073. Parallel-trends diagnostics also fail for informal-sector job vacancies under both the seasonally adjusted and unadjusted specifications. The authors are transparent about this: the informal-sector heterogeneity is a descriptive pattern warranting further investigation, not definitive statistical evidence.

The methodological care extends to the vacancy data themselves. The authors discovered that seasonal adjustment of short administrative series during an unprecedented structural break can distort inference, potentially misinterpreting shock-driven variation as seasonal movement. They therefore report two specifications for job vacancies—one seasonally adjusted, one raw—and place greater weight on the unadjusted model, which lets the sector and time fixed effects absorb temporal variation without the risk of adjustment artifacts. This kind of methodological self-scrutiny is rare and valuable, particularly in a literature where pandemic-era data are short, noisy, and prone to structural breaks that standard statistical machinery was never designed to handle.

The findings align with a growing international consensus. Research on South Korea, where lockdowns were minimal, documented job destruction comparable to that in the United States and the United Kingdom, suggesting that fear of contagion and voluntary distancing drive much of the economic damage. Real-time tracking studies in the United States similarly found that spending and employment collapsed before or regardless of formal orders. Türkiye’s sector-level evidence adds an emerging-market dimension to this picture, indicating that even in an economy with substantial informality, the aggregate labor market response was shaped more by the pandemic shock than by the regulatory response to it.

The policy implications are sobering. If restrictions themselves were not the primary driver of labor market damage, then lifting them quickly is not a reliable route to economic recovery—and imposing them need not be economically catastrophic if paired with the right support. But the suggestive evidence of informal-sector fragility points to where policy should focus: targeted income support for informal workers and micro-enterprises during closures, and investments in digital and operational resilience that would help the most exposed firms weather future shocks. As the authors conclude, uniform containment measures can generate uneven consequences across sectors with different levels of formality, and economies like Türkiye—where informality and structural dualism remain prevalent—need crisis-response tools calibrated to that heterogeneity. In a world preparing for the next pandemic, that may be the study’s most enduring lesson.

Subject of Research: The impact of COVID-19 restrictive measures on the labor market and productivity in Türkiye

Article Title: Impact of COVID-19 restrictive measures on labor market and productivity in Türkiye

Article References: Dolu, A., & İkizler, H. (2026). Impact of COVID-19 restrictive measures on labor market and productivity in Türkiye. Heliyon, 12(15), Article e45535. https://doi.org/10.1016/j.heliyon.2026.e45535

Image Credits: AI Generated

DOI: 10.1016/j.heliyon.2026.e45535

Keywords: COVID-19, labor market, Türkiye, difference-in-differences, job vacancies, job placements, labor productivity, informal economy, lockdowns, emerging markets, econometrics, public policy