The Empirical Analysis of Economic Development: FsQCA as an Intermediate Approach Between Case Studies and Regression Analyses

Empirical analyses of economic development usually apply cross-country growth regressions. These regressions have quite firmly established robust correlations between several variables, usually on investments, initial GDP or institutions, and economic growth. After several decades of an almost abusive use regressions, however, several dead ends have been reached. Firstly, there are several contentious variables, changing sign or significance depending on the model or on their operationalization. Secondly, there are outliers which remain outliers despite the inclusion of new variables. In my paper, I have picked out the states of the Gulf Cooperation Council, six notorious outliers, and analyzed them with a special focus on two contentious variables of cross-country growth regressions: democracy and natural resource wealth. In addition to case studies of the six countries and several large-n regression, I conducted two fsQCA analyses: The first one analyzes the GCC states in a separate sample, while, secondly, they merge into a sample of oil producing countries. This methodological triangulation brought several methodological advantages over sole case studies or regressions. For the smaller sample, the conditions could be chosen with respect to the particularities of the sample and operationalizations could be adapted to their special background. Due to rent inflated GDPs, several very common variables turned out to have little explanatory power in the context of resource dependent countries. Here, new and more appropriate proxies could be used. Other conditions were given a new interpretation by calibration: In general, regressions implicitly assume a proportional relationship between resource abundance (or dependence) and economic development. However, there may be a turning point between modestly resource abundant countries which cannot expect to live on their resource wealth and immensely resource rich countries, having no hurry to diversify. Calibration made it possible to test several turning points and, accordingly, several hypotheses of the heavily debated resource curse. Also democracy could be calibrated with a special view on the region and the smaller sample allowed for a institutional approach towards the influences of Islam on economic growth instead of using the percentage of Muslims in the entire population as a proxy. At a closer look, it also turned out that a country is not necessarily a case: The UAE, a federation without common economic policy, is in fact very heterogeneous. This became clear during the analysis for sufficient solutions and splitting up the UAE into the most important emirates, Abu Dhabi and Dubai, considerably improved the quality of the analysis (coverage and consistency). Data for the single emirates, then, had to be estimated. Here, fuzzy sets seem to be preferable to inventing seemingly precise raw data. In a second step, I analyzed the GCC states in a sample of 25 oil producing countries. The resulting rather heterogeneous sample required several tradeoffs concerning the above-mentioned operationalizations: In several cases, the “usual” operationalization had to be adopted despite better knowledge. Nevertheless, path dependence effects became clearer in this sample: Both resource abundance and democracy were shown to depend on further conditions and to appear in both sufficient solutions for a negative outcome and a positive outcome. I could demonstrate that assumed “universal” factors of economic development triggered different mechanisms in this sample of outliers. The two contentious variables were shown to be dependent on further factors. Furthermore, other apparently “robust” conditions like political stability, displayed ambiguous relationships to economic development depending on their background. The use of fsQCA as an complementary methodology to regressions and case studies has opened a new path towards a more thorough analysis of economic development, which could also contribute to a reformulation and specification of growth and development theories. It could respond to those voices in development studies that argue that every case is special. Using regressions as a first step, however, remained an advantage: It helped using a two-step approach in order to reduce the number of conditions in the fsQCA-analysis: the “robust” variables were always tested in a “basic model”, while the contentious variables were added separately.