Essays on Demand Externalities and Cross-Market Rewards

This dissertation is comprised of three essays that focus on the impact of increases in scope on a firm's pre-existing product offerings. In the first essay, titled "Demand Externalities from Co-Location," I examine the grocery sales benefits to a supermarket of introducing a co-located gas station. In doing so, I illustrate a framework with which to measure demand externalities from co-location based on household-level changes in behavior.;The size of externalities are estimated by comparing changes in household level spending of households that use the gas station with those that do not. Since the decision to use the gas station -- as well as the timing of first gas purchase -- is not exogenously imposed, the primary econometric challenge lies in controlling for observable and unobservable selection influencing the decision to buy gas. In doing so, I implement a method proposed by Altonji et al. (2005) that places bounds on the impact of unobservable selection.;I find evidence for significant demand externalities that lie within a tight range: on average a household that buys gas has a 7.7% to 8.4% increase in spending on groceries. This change in spending is accompanied by a 14% to 15% increase in grocery trips to the store. Accounting for differences in gross margins, the profit from the grocery spillovers is 130% to 140% the profit from gasoline sales. The spillovers are moderated by store loyalty, with the gas station serving to cement the loyalty of store-loyal households. The grocery spillover effects are significant for traditional grocery products, but 23% larger for products that might typically be purchased at convenience stores.;In the second essay, titled "Category Level Implications of Demand Externalities," I use the same context as the first essay, and exploit an additional level of variation in spending -- at the household-category-level -- to understand how positive demand externalities arise differentially across product categories.;The first set of hypotheses builds on the finding (from the first essay) that households make more frequent trips to the store following their first purchase of gas. Increased frequency of store visits should differentially affect categories that are amenable to more frequent purchasing. I find that product categories purchased with greater frequency before the gas station opens experience the greatest positive externalities in both absolute and relative terms. More frequent trips also increase a gas-buying household's ability to engage in temporal price search relative to non-gas buyers. While externalities are relatively larger for the majority of categories that never go on discount, among discounted categories, more frequently discounted categories show relatively larger demand externalities. Finally, externalities are also larger on perishable categories -- which are more likely to be purchased on more frequent, "fill-in" trips -- than on non-perishable categories.;A second hypothesis is that externalities may simply arise as a result of broader purchasing behavior by gas buyers -- i.e. across a wider range of categories at the store. Following their initial use of the gas station, gas-buyers do spend on more previously un-purchased categories relative to non-buyers. While externalities on newly purchased categories are significant, the relative increases in spending are greater on previously purchased products.;In the third essay, "Customer Response to Cross-Market Rewards," I investigate the effect of a reward program -- operated by a supermarket chain -- in which greater cumulative spending on groceries is rewarded with progressively larger discounts on the price of gasoline. The program is unique in that reward redemption does not affect the demand for the product on which rewards are earned -- as in the case of free flights purchased with airline miles, for instance -- and because redemptions are sufficiently frequent, I am able to observe multiple reward redemption occasions for each customer. Initial regressions show that households are cognizant of their endowment of reward points, and that they alter the timing of gas and grocery purchases in order to receive larger rewards.;In order to quantify the benefits of the cross-market reward program, I model the simultaneous -- but unobserved -- decision by customers about whether to go shopping, and if so, whether to purchase either groceries, gasoline, or both, as well as how much to buy. I find that customers increase (decrease) their purchases when they are near to (further away from) reward eligibility thresholds. Moreover, based on a measure of pre-existing customer loyalty, I find that the most loyal customers are also the ones who are most responsive in terms of "stretching" towards the next highest reward level. Policy simulations show that removing the cross-market discount program would reduce the revenues from groceries by 7%.