A smartphone user allocates her time among multiple mobile applications. To study the competitive relationship between apps, I develop a discrete-continuous model of time allocation with a binding time constraint; its parameters are estimated using a weekly panel of app usage in China. If two apps are often used together, it is because either they are complements or the preferences of the two apps are positively correlated. To disentangle complementarity (substitutability) from correlation in preferences, I use the exclusion restriction wherein updates of an app should affect the utility of that app but not those of other apps. I estimate the model on three representative pairs of apps (a priori substitutes, complements, and independent apps). In each case, I recover a reasonable competition pattern. Morever, I simulate counter-factuals in which one of the two apps is shut down and decompose the effects into two parts: functional competition and budget competition (the latter of which captures the effects of a binding time constraint). With an assumption about the monetary value of time, I simulate mergers of the three pairs of apps. I find that a merger of seemingly independent apps can hurt consumers because the two apps are competing for user time. My results confirm that users and firms can both benefit from a merger of complements. I also find that usage-based pricing leads to higher profits and total surplus compared to subscription pricing, because it enables price discrimination based on usage on the part of companies.