In my thesis I develop a theoretical model of interdependent choices and an estimation strategy which I apply to model patent renewal. The model and the estimation are not confined to my application, but rather can have other applications in which firms or people are making strategic and simultaneous decisions. Chapter 1 is the introduction which contains a brief description of the structure of the thesis.

Chapter 2 provides a literature review of studies that have focused on spatial dependence with discrete choice dependent variables; recent contributions include Pinkse and Slade (1998), LeSage (2000), Kelejian and Prucha (2001), Beron and Vijverberg (2004), and Wang et al. (2009). A major difficulty in the estimation of spatially dependent discrete choice models is computational intensity.

Chapter 3 is a Monte Carlo study that investigates the small sample properties of an estimator for spatially dependent discrete choice models which is computationally simple. The analogue of a linear probability can be formulated as a spatial autoregressive Cliff and Ord (1973, 1981)-type model. The sets of Monte Carlo experiments show that the parameters of the model can be estimated without bias using a spatial 2SLS estimator.

Chapter 4 is a study is on the determinants of patent renewal, using US patents for Computer Hardware and Software granted between 1994 and 1997. Patent protection is important in that it encourages innovation by allowing firms to rely on patents to appropriate the returns to their R&D efforts. Returns to patents are modeled to depend on the firm's willingness to pay the patent renewal fees by, e.g., Harhoff et al. (2003), Serrano (2006), and Bessen (2008, 2009), and typically ignored potential interdependences in the decision making. Liu et al. (2008) showed that patent renewal was more likely if the patent was part of a firm's sequence of citing patents. I elaborate on their result and formulate a model in which the decision to renew a patent is dependent on the decisions of other firms to renew technologically similar patents. The theoretical model implies for the probability to renew a patent to depend on the probabilities to renew other patents, where the extent of interdependence is modeled based on a measure of similarity for patents. By making use of the estimation strategy from Chapter 3, I find that indeed the decision to renew a patent is dependent on the decision to renew related patents. Results in the literature which ignored this interdependence may hence suffer from specification biases. One plausible explanation for the interdependence I find is defensive patenting in the form of patent fencing, patent blocking and patent thickets. In the latter case, litigation and negotiation can impose high costs to society and their anticipation can lead to a hold up problem, which could deter investment in R&D.