How much do worker skills, firm pay policies, and their interaction contribute to wage inequality? Standard approaches rely on latent fixed effects identified through worker mobility, but sparse networks inflate variance estimates, additivity assumptions rule out complementarities, and the resulting decompositions lack interpretability. We propose TWICE (Tree-based Wage Inference with Clustering and Estimation), a framework that models the conditional wage function directly from observables using gradient-boosted trees, replacing latent effects with interpretable, observable-anchored partitions. This trades off the ability to capture idiosyncratic unobservables for robustness to sampling noise and out-of-sample portability. Applied to Portuguese administrative data, TWICE outperforms linear benchmarks out of sample and reveals that sorting and non-additive interactions explain substantially more wage dispersion than implied by standard AKM estimates.
Recent literature proposed variants of estimators of conditional error variances that are robust to the presence of many regressors. One of the proposals produces unbiased estimates, which, however, are sensitive to regression parameter values. We propose its modification based on cross-fitting and averaging over random sample splits, which preserves the unbiasedness but eliminates that sensitivity. Then, we extend the cross-fitting machinery to unbiased estimation of pairwise variance products and to testing for many restrictions. We verify properties of associated estimators and tests in a simulation setup with many covariates.
Abstract This study examines whether racial outrage in the media can result from the incentives of elites who seek to divide society for political gain, focusing on the late 19th century Southern United States. At this time, the Democratic Party had strong political incentives to incite racial outrage, and local Democratic elites exercised considerable influence over newspapers, the only mass media at the time. Using high-frequency historical newspaper data and difference-in-differences and triple-difference estimation strategies, we find evidence consistent with Democratic elites orchestrating a racist propaganda campaign to sway white voters. Supporting analyses are consistent with Democrats benefiting from anti-Black propaganda in the subsequent elections.
Has broadband become a necessity good immune to price changes? Using a15-year panel of 33 countries—27 European Union (EU) member states and6 Eastern Partnership (EaP) countries (2010–2024)—and two-way fixed effectswith Driscoll–Kraay standard errors, we document a fundamental transformation in broadband demand. Pre-COVID, EaP countries exhibited relatively elastic demand (ε = −0.60, p<0.001)—a 10% price reduction increased subscriptionsby 6.0%—while EU countries showed lower elasticity (ε = −0.10, p<0.10). By 2020–2024, both regions converged to near-zero elasticity, with price changes having no detectable effect on adoption. Crucially, placebo tests reveal this transformation began in 2015, not 2020, indicating a decade-long digital integration process rather than a COVID-19 shock. We further demonstrate that price measurement critically affects inference: income-relative prices (as % of gross national income, GNI) yield significant EaP elasticities in all specifications, whereas purchasing-power-parity (PPP) adjusted prices yield significant results in only 12% of EaP specifications; for EU countries, all three price measures perform broadly similarly. These findings have immediate policy relevance: as broadband transitions from discretionary service to essential utility, policy emphasis must shift from affordability subsidies to universal infrastructure deployment.
Belt and Road Initiative (BRI) is a multi-billion-dollar investment and a ‘Going Out’ strategy of China. The major vision of BRI is to accelerate infrastructure investment, which, in turn, improves connectivity, enhances trade and foreign direct investment and spurs economic growth in the participating countries. This research seeks to examine the influence of the BRI on key economic indicators—specifically gross domestic product (GDP) growth, unemployment and foreign direct investment across 20 low-income African countries. Employing a robust panel data least squares dummy variable modelling approach, the study analyses the country-specific impact of BRI for data spanning the period from 2008 to 2022. The research outcomes indicate varying impacts of the BRI across different nations. In most countries, the BRI has had no substantial influence on GDP per capita, yet a detrimental effect is evident in Mozambique. The results show BRI has a negative effect on foreign direct investment in Benin, Congo, Tanzania and Rwanda, whereas it has a positive impact in Mozambique and Sierra Leone. Whereas variables like gross capital formation, labour force participation and regulatory indices consistently display substantial positive effects on economic growth, investment and employment. The research underscores the necessity for strong domestic policies that can optimise economic advantages and guarantee long-term growth. JEL Codes: O11, O55, H54