Daniel E. Rigobon's research while affiliated with Princeton University and other places

... This strategy is useful when the carmaker collaborates with dependable suppliers that provide stable excellence and support long-term partnerships that decrease price volatility, prevent overproduction errors, and lower freight costs, especially during times of modest demand growth and change. Despite the fact that the JIC inventory methods are based on forecasted sales and entail the carmaker to undertake more proactive purchases of components that can accommodate shifts in demand while remaining within clearly defined limitations (Jiang, Rigobon, & Rigobon, 2021). Companies which employ the JIC strategy may be capable of mitigating the negative effects of frequent inventory controlling hurdles, like bottlenecks or disconnections in supply or unanticipated declines or surges in demand for commodities/supplies and the cost of those items. ...

... The first hypothesis validates the basic premise of our analyses: Is it possible to attribute headline success to the linguistic features of headlines? Success can be the consequence of many complex factors at play, many of which are not observable or subject to unpredictable external shocks [35,36]. It has been shown that even a fully-described complex system can be so prone to the accumulated effects of random behavior that reasonable predictability is impossible [26,27,37]. ...

... For example, conventional regress models, such as discrete-time event history models, perform poorly with a low out-of-sample predictive accuracy, while random survival forest models provide considerably superior predictive accuracy in predicting marital union dissolution (Arpino et al., 2022). Another example is that machine learning models of random forest, elastic net, and gradient-boosted trees have improved out-of-sample predictions for three life outcomes in GPA, grit, and layoff compared to baseline models (Rigobon et al., 2019). However, when operating without theoretical guidance, researchers should not mistake data-driven machine learning models as causal models because neither their parameters nor their predictions necessarily have a causal interpretation. ...