Internalizing market transactions has been seen by New Institutional Economics (NIE) as a means of over-coming transaction costs. Several specific theories have been put forward as to how the presence of such costs (variously defined), or other institutional factors, motivates firms to vertically integrate (Chapter 2).

However, a review of case study evidence from garment producers in Egypt (Chapter 3) shows that there are also important constraints on vertical integration, though these do not affect all firms equally. Small firms in particular suffer from an underdeveloped capital market which hinders expansion. Other features of the Egyptian setting encourage integration; e.g. problems with fabric quality for high-end garment producers, and the need for timely delivery. Chapter 4 uses the findings from these case studies to adapt the theories presented in Chapter 2 to the Egyptian context, and identifies other institutional factors relevant to the 'make or buy' decision.

This thesis presents analysis of a new data set of 257 private Egyptian garment firms collected for this research.  Chapter 5 presents the sample and survey design, and descriptive analysis of the degree and order of integration. Features of the Egyptian business environment, such as imperfect credit markets, are shown to be among the most significant determinants of vertical integration (Chapter 6).  However, prominent theories of vertical integration are also relevant, for example search and switch costs encourage integration while monitoring costs discourage it. 

The contributions of this thesis to the existing literature include the analysis of vertical integration in a developing country setting using a new data set. The empirical analysis encompasses all existing theories, rather than simply the one of interest, and introduces context-specific factors not considered elsewhere, which is shown to bias other research's results. Integration is modeled as a fractional response, rather than a dichotomous variable as has been done in other papers, using a model specification that partially gets around the endogeneity problem, which has plagued the literature.

The empirical findings have two methodological implications. First, theories are complementary rather than competitors. Second, empirical work focusing on only one theory suffers from omitted variable bias.