. Policy implications and discussion
In this study, we conduct a conjoint survey to analyze switching costs in the smartphone market by using the hierarchical Bayesian multinomial logit model, which considers consumers' heterogeneous preference. The switching costs are categorized into two types: those derived from a device change and those derived from an OS change. We empirically estimate the two types of switching costs, and find that the levels of switching costs depend on the users' characteristics. In terms of switching costs from smartphone device changes, the results demonstrate that the coefficient for changing smartphones is not negative, but positive, unlike our expectation. This means that consumers prefer to replace, rather than keep, their old smartphones, ceteris paribus. The results are in opposition to those in previous studies (Kim et al., 2004; Lee et al., 2006) that analyze switching costs from changing telecommunications service providers. However, it is not irrational for consumers to prefer a new device because the value of devices deteriorates over time due to obsolescence. Koo (2012) also concludes that consumers benefit from switching their smartphone devices to new ones. The switching cost from device change is dependent on consumers' education level, period of possession of current smartphone, early adopter tendency, and the uncertainty of previously purchased applications' compatibility.
