ENTRY INTO PHARMACEUTICAL SUBMARKETS: A BAYESIAN PANEL PROBIT ANALYSIS GIANNI AMISANO a * AND MARIA LETIZIA GIORGETTI b,c a DG-Research, European Central Bank, Frankfurt, Germany b Dipartimento di Economia, Management e Metodi Quantitativi, Università degli Studi di Milano c Università Cattolica del Sacro Cuore, Milano, Italy SUMMARY We study entry into pharmaceutical submarkets by using a dynamic panel probit model. We develop a Bayesian version of Wooldridge’s approach to dealing with unobserved heterogeneity. We distinguish between greenfield entry (entry with first product) and undifferentiated entry, which may take the form of introducing further products. We extend the standard reduced form model to consider company size in the submarket and across all submarkets. We find that greenfield and undifferentiated entries have different determinants, that global and submarket size measures have different effects and that sunk costs often have positive effects on entry due to a commitment mechanism. Copyright © 2012 John Wiley & Sons, Ltd. Received 15 January 2007; Revised 22 August 2011; Supporting information may be found in the online version of this article. 1. INTRODUCTION The literature has emphasized the role of submarkets in explaining industry market structure (Sutton, 1998) and evolution (Klepper and Thompson, 2006). In particular, entry, diversification, exit, firm’s growth and firm size distributions are better understood when analysed at a submarket level. There are several interesting papers on aspects of entry and innovation in the pharmaceutical sector (see, for example, Scott Morton, 1999; Bottazzi et al., 2001; Magazzini et al., 2004; Ching, 2010a,2010b; Danzon et al., 2005; Lanjouw, 2005; Reiffen and Ward, 2005; Kyle, 2006). None of these studies is conducted at a firm level, distinguishing across submarkets. In our paper we focus on entry in pharmaceutical submarkets and we take a novel approach: having data on sales of different drugs for different companies in different years and in different submarkets, we study entry as a dynamic process in a reduced-form discrete-choice panel data framework where the unit of in- terest is the single firm. We chose to use a reduced-form model as a flexible and robust explorative device to capture regu- larities in the observed entry processes. The results obtained using reduced-form models can naturally be used to provide useful guidelines for the construction of structural entry models, whose estimation and validation are naturally more complex than those of reduced-form models. For the issues connected to structural discrete-choice model estimation, see Aguirregabiria and Mira (2007), Ackerberg et al. (2007) and Imai et al. (2009). Following Bresnahan and Reiss (1993), Hendricks et al. (1997) and Netz and Taylor (2002), we use a standard reduced-form model to investigate entry decisions in each submarket. As is done in these studies, we take into account demand-side conditions, costs measures, intensity of competition * Correspondence to: Gianni Amisano, DG-Research, European Central Bank, Frankfurt, Germany. E-mail: gianni.amisano@ecb.europa.eu Copyright © 2012 John Wiley & Sons, Ltd. JOURNAL OF APPLIED ECONOMETRICS J. Appl. Econ. (2012) Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/jae.2267