Current time allocation models face three limitations: First, they only describe the average time allocation. Thus, information about the order of activities is lost, so influences of activities on later ones (i.e. interactions) are difficult to describe. Second, they assume a fixed effort allocation and thus are very limited in describing multitasking and labor productivity endogenously. Lastly, they require strong assumptions like perfect foresight and periodic environments, which limits their applicability in unpredictable environments like external shocks. This paper proposes a dynamical model of procedurally rational decision-making, addressing these drawbacks by incorporating a feedback loop between experienced utility, decision utility, and activities. By simulating this model, I show how neglecting work-leisure interactions biases price and time elasticities, and how the model replicates different types of multitasking. Furthermore, I show on the example of the COVID-19 lockdown how nonmarginal external shocks cause short-term demand surges, which cannot be captured by current time allocation models.
