Although corruption is a widespread phenomenon, and one with significant implications for both growth and equity, relatively little is yet known with confidence about its macroeconomic consequences. The aspiration in this thesis is to provide both theoretical and empirical work which contributes to an improvement of that understanding. Corruption is itself difficult to define and is by nature multidimensional. In consequence, it is difficult to measure. One task undertaken therefore is a review of presently available measures of corruption so as to uncover which dimensions of corruption have been accurately measured by those indicators. Current indicators of corruption are successful in capturing types of administrative corruption, especially in the form of bribery, and limited success has also been achieved in measuring aspects of "state capture", a form of corruption which influences the content of laws and rules within an economy. Other dimensions of corruption nonetheless remain unmeasured, and there is a potential for some aspects of corruption to remain empirically unmeasurable.
Another problem in the empirical context is that extensive data sets on corruption across countries and across time are not readily available, thus restricting the econometric work yet possible. One contribution of this thesis is to work up a data set which substantially improves upon those data sets previously employed by other researchers in the corruption context. After undertaking an extensive search, a data set of 110 countries for the 20 year period 1984-2004 was assembled. Such coverage in terms of countries and time represents a substantial improvement over previous studies. For example, the number of countries is more than double that used by Mauro (1995), in what is a seminal paper in the recent surge in empirical corruption literature.
After offering a review and critique of the existing literature, empirical work based on the newly extended data set is provided. This empirical work is based on a structural model which seeks to account for the multidimensional nature of corruption through explicit modelling of the transmission channels through which corruption indirectly affects growth. These transmission channels are hypothesised to represent the main channels through which the real effects of corruption on growth can be found. Thus, the new data set and the more complete modelling of the nexus between corruption and growth provide the thesis with two significant differences from previously available work.
Results suggest that corruption's main effect on growth can be found through investments in physical capital, human capital and political instability. However, the importance of each channel is found to be dependent on the economy's level of development and degree of excessive regulations. No one-size-fits-all conclusion appears supportable. Conclusions regarding policy implications and suggestions for further research are also offered.