Changes in Fertility Intentions Due to Multiple Crises in Austria: A Machine Learning Approach

Maria Winkler-Dworak , Vienna Institute of Demography

The recent multiple crises, including the COVID-19 pandemic, a period of historically high inflation, and the repercussions of the Russian invasion of Ukraine, have contributed to economic uncertainty, heightened financial strain, and growing concerns about the future. Fertility decisions, which are inherently future-oriented, are particularly sensitive to such conditions. Previous studies show that uncertainty can lower fertility intentions or increase uncertainty about having children. However, the factors shaping these responses vary across social groups and may involve complex interactions that are difficult to capture using traditional regression models. This study examines changes in fertility intentions during the recent crises using data from the Austrian Generations and Gender Survey (GGS-II) collected in 2022–2023. We apply a Random Forest classification model to identify the most important predictors of whether respondents revised or became unsure about their family plans due to the crises. The model accounts for class imbalance and uses permutation-based variable importance to assess predictor relevance. Preliminary results indicate that the perceived burden of the war in Ukraine and inflation are the strongest predictors of changes in fertility intentions, followed by subjective well-being and financial strain. The perceived burden of the COVID-19 pandemic, by contrast, plays a limited role in this late-pandemic period. Future work will refine the model through hyperparameter tuning, extend the outcome to distinguish between different types of change—including uncertainty in fertility intentions—and compare performance with alternative machine-learning algorithms.

See extended abstract

 Presented in Session 95. Fertility Responses to War and Crisis