GMO approvals under pressure: How climate shocks shape policy across nations

PLOS One | https://doi.org/10.1371/journal.pone.0332298 October 22, 2025 7 / 20 Empirical hypothesis. Climate shocks lead to less GMO approvals in countries with a comparative disadvantage in GMOs. Our theoretical results provide a transparent framework for linking theory to data and guiding the empirical analysis, which proceeds in three parts. First, climate shocks (ωi )—treated as exogenous in the theoretical model—are empiri- cally measured using standardized event counts from the EM-DAT disaster database. Second, GMO approvals (θi) in the theory are proxied in the empirical analysis by the number of GMO events approved in each country-year. Third, comparative advantage in GMOs ( 1 γi ) in the theory is operationalized in two empirical ways. The first proxy is GDP per capita, which reflects a country’s capacity to invest in and adopt GMO innovations. The second is a Revealed Compar- ative Advantage (RCA) index, which measures whether a country’s export specialization in GMO products exceeds its overall export performance [3]. This index closely aligns with the theoretical concept of specialization and capability in GMO development ( 1 γi i). 3. Data Climatic shock data is collected from the International Disaster Database (EM-DAT) for the period 2000–2018. For a shock to be entered into the database at least one of the following criteria must be fulfilled: ten or more people reported killed, one hundred or more people reported affected, declaration of a state of emergency, or call for international assis-

tance [19]. The dataset has been widely used in previous studies on climatic shocks [20–23]. The shock types include geophysical, meteorological, hydrological, climatological, biological, and extra-terrestrial. In this study, climate shocks include droughts, heat waves, cold waves, severe winter conditions, floods, and storms. Cold wave includes a period of abnormally cold weather. For example, according to EM-DAT, heat wave refers a period of abnormally hot and/or unusu- ally humid weather that lasts for two or more days. Severe winter conditions are damage caused by snow and ice. Winter damage refers to damage to buildings, infrastructure, traffic (esp. navigation) inflicted by snow and ice in the form of snow pressure, freezing rain, frozen waterways etc. Drought refers an extended period of unusually low precipitation that pro- duces a shortage of water for people, animals, and plants. GMO approval data is collected from the ISAAA, containing all GMO events at the commodity and country levels each year. Since GMO and non-GMO crops cannot be distinguished and not available in the dataset, this paper considers industries that include the most commonly GMO crops—namely, cotton, maize, soybeans, and rapeseed—which con- tinually develop advanced GMO traits. Thus, the GMO use in the selected crops reflects countries’ adoption behavior of GMOs. Based on prior work [24,25], the revealed comparative advantage (RCA) index is calculated by collecting data from the United Nations (UN) Comtrade, which consists of the value of bilateral exports at the commodity and country levels in

current (rather than constant) dollars. The products are defined according to the SITC Revision 3 aggregates: soybeans (2222), maize (044), cotton seeds (2223), and rape or colza seeds (22261). The control variables include agricultural share, arable land per capita, GMO area, infrastructure conditions, an interna- tional agreement indicator, infrastructure conditions, and a developing country dummy. Many countries adopt alternative adaptation strategies such as fertilizer use and infrastructure investment. In the theoretical model, these are treated as fixed, non-technology-specific costs that do not affect the marginal returns to capital and labor across GMO and non-GMO sectors, and thus do not affect the optimal GMO allocation decision derived from the first-order conditions of Equation (4). Empirically, we control for adaptive capacity using the ND-GAIN infrastructure index. Fertilizer use is excluded due to lim- ited disaggregated data and potential endogeneity, as it may correlate with unobserved factors such as crop composition or technological capacity. Data on agricultural share and arable land per capita are collected from World Bank Open Data. Annual biotech acre- age in a million hectares is obtained from existing work [26]. The infrastructure score is generated by the Notre Dame Global Adaptation Index (ND-GAIN), which measures the vulnerability of the infrastructure sector. The infrastructure score,