In this project, we focus on the estimation of causal effects with panel data under the assumption of sequential exogeneity. Sequential exogeneity is rarely used in applied ``reduced form'' papers despite being a well-known and widely recognized econometric concept. Instead, most of the empirical work is done under a more restrictive (and often implicit) assumption of strict exogeneity. This project's primary goal is to show that sequential exogeneity is a natural and useful concept for empirical work. To achieve this goal, we follow three steps. First, using recent ideas from causal literature, we construct a flexible estimator that can be used instead of the standard ordinary least squares and provide its interpretation in a rich causal model. Second, we analyze the statistical properties of this estimator in different regimes (long and short panels) and illustrate the main trade-offs one has to face. Finally, we apply the new estimator to recently published empirical papers that use a standard OLS approach to see if sequential exogeneity changes the results in an economically important way.