The extent to which individuals commit to their partner for life has important implications. This paper develops a lifecycle collective model of the household, through which it characterizes behavior in three prominent alternative types of commitment: full, limited, and no commitment. We propose a test that distinguishes between all three types based on how contemporaneous and historical news affect household behavior. Our test permits heterogeneity in the degree of commitment across households. Using recent data from the Panel Study of Income Dynamics, we reject full and no commitment, while we find strong evidence for limited commitment.