Behavioral studies and recent empirical research suggest higher levels of inventory on hand can lead consumers to increase consumption. Inventory on hand is therefore posited to exert two countervailing forces on the probability of purchase incidence. First, higher levels of inventory reduce the likelihood of purchase as the consumer feels less pressure to buy. At the same time however, theory suggests higher levels of inventory may drive up the rate of consumption, thereby increasing the probability of incidence. We develop an empirical model that explicitly captures these two effects. The elasticity of purchase incidence with respect to inventory derived from the model is shown to capture these opposing forces in a simple and intuitive way. The analytical expression allows calculation of a threshold below (above) which the net effect is positive (negative). The model is estimated on ten product categories from the Stanford Market Basket database and is shown to fit better than both the standard nested logit approach and an alternative formulation developed by Ailawadi and Neslin (1998). The threshold values have plausible magnitudes and are intuitive across categories: butter, margarine and crackers have relatively low thresholds implying that inventory build up does not drive consumption; ice cream and soft drinks have relatively large thresholds (below which the inventory pressure to consume more outweighs the effect to delay purchase). Implications for retail management are discussed.