Aim: To test an implication of Becker's rational addiction theory, namely that price changes will lead both to simultaneous consumption changes as well as lagged changes (and potentially also immediate changes if future changes in prices are anticipated).

Design: Time-series analysis, first of aggregate sales of distilled spirits and prices, controlled for gross national product (GNP), and secondly of deaths from delirium tremens.

Setting: Denmark 1911-31. Price changes were very large in the period 1916-18 due to shortages during World War I, and the Danish case can be conceived as a natural experiment.

Findings: No evidence for lagged price effects in the expected direction was found. On the contrary, the evidence pointed in the opposite direction. The immediate reduction in sales following rising prices are, to some degree, counteracted by an adjustment in the opposite direction the following year. The delirium tremens data confirm this pattern.

Conclusion: Becker's theory is not confirmed. Several possible explanations are discussed. If the pattern observed in these data is representative of a more general mechanism, current price elasticity estimates may be too high, by ignoring lagged compensatory effects.