Professor George John

Many firms sell tied products in the sense that “the purchase of a primary 'tying' good requires consumers to purchase complementary 'tied' or 'aftermarket' goods from the same manufacturer." (Hartmann&Nair, 2009). Examples include razor and blades, printer and toners, electric toothbrush and brush heads, etc. Typically, the primary good is a durable good and the aftermarket goods are consumable. In many cases, firms enjoy a much higher margin in the aftermarket. Therefore, for firms selling tied products, what matters is not just whether a consumer is willing to buy the primary good, but also how many aftermarket goods she will consume once the primary one has been purchased. This requires an understanding of the demand for both parts of the tied products and the relation between the two. Here, several questions are of interest: first, how are the preferences for the two tied products correlated? In other words, do consumers who purchase the primary good earlier also demand more in the aftermarket? Second, how long will a customer hold the primary product before she replaces it? These are important questions because different answers imply distinct pricing and bundling strategies, which will in turn affect the profitability of a firm selling such tied products.  

Ideally, to answer the previous questions, one would like to have individual level data so that all customers’ consumption habits and replacement decisions can be directly observed. However, in most cases, individual level data is either too hard or too expensive to get. This research seeks to show that the questions above can still be answered using only aggregate-level data, which is relatively easier to obtain. The researchers build an empirical model of demand for tied products that account for consumer heterogeneity and forward-looking behavior, as well as replacement. The model is applied to the industry of single-serve coffee systems, where firms produce both coffee machines (the primary tying good) and coffee pods (the aftermarket tied goods). The researchers are showing how consumer preference and replacement can be identified from sales data at the machine model level. The results indicate that the correlation between consumers’ willingness to pay for the coffee machine and their consumption rate for coffee pods differs across brands. While for some brands people who purchase the coffee machine earlier also consume more coffee pods, it is not necessarily the same case for others. The group also found that there is significant coffee machine replacement going on during the latter period of the data. Ignoring this process might lead to biased estimation of consumers’ drinking habits. Further, using the demand estimates, the researchers evaluate the impact of entry of the store brand which occurred during the period studied as well as the impact of a quality improvement of coffee machines.