We propose a simple and direct test of in-couple income pooling and provide a typology of couples that pool resources and those who do not. For this purpose, we performed a five-round experiment with 95 established real-life couples in Germany. In each round, the couples received the same total amount of money, but the relative allocation to the spouses differed and they had to agree on an irreversible private goods consumption pattern. We find that for the majority of the sample, the consumption choices depend on the spouses' relative resources. While this suggests a rejection of income pooling at the mean, a substantial share of the couples do indeed pool their resources. Pooling behavior is more common among couples in which the spouses' socioeconomic characteristics are more similar. In particular, traditional couples with distinct work division between the spouses are significantly less likely to pool their individual resources and treat any spouse's income symmetrically. Conventional variables used to approximate income pooling are only loosely related to the behavior observed in our experiment.
We propose a simple and direct test of in-couple income pooling and provide a typology of couples that pool resources and those who do not. For this purpose, we performed a five-round experiment with 95 established real-life couples in Germany. In each round, the couples received the same total amount of money, but the relative allocation to the spouses differed and they had to agree on an irreversible private goods consumption pattern. We find that for the majority of the sample, the consumption choices depend on the spouses' relative resources. While this suggests a rejection of income pooling at the mean, a substantial share of the couples do indeed pool their resources. Pooling behavior is more common among couples in which the spouses' socioeconomic characteristics are more similar. In particular, traditional couples with distinct work division between the spouses are significantly less likely to pool their individual resources and treat any spouse's income symmetrically. Conventional variables used to approximate income pooling are only loosely related to the behavior observed in our experiment.