The impact of local knowledge on project performance: Analyzing UNDP project evaluations and staff data

In the realm of international organizations (IOs), the significance of staff expertise and local knowledge has recently garnered scholarly attention, with studies suggesting a positive correlation between country-specific knowledge and organizational performance. Yet, findings are limited to the few IOs for which project performance ratings and staff data are available. This study extends previous findings by examining the United Nations Development Programme (UNDP), where the composition of local versus international staff is hypothesized to impact project success. Our study utilizes a text-as-data approach, leveraging a pretrained BERT model and Named Entity Recognition, to analyze the content of over 3000 UNDP evaluation reports. This analysis provides a quantitative measure of project performance, allowing us to investigate the effect of staff composition on project outcomes. Drawing from literature that highlights the importance of local hiring in acquiring crucial country-specific knowledge, we hypothesize that projects with a higher proportion of locally recruited staff will demonstrate enhanced performance. To quantify local knowledge, we construct a new measure using subnational staff statistics from the UN System Chief Executives Board of Coordination, providing a nuanced understanding of staff composition in terms of local versus international personnel. This data, spanning from 2006 to 2021, enables a comprehensive evaluation of the relationship between local expertise and project success. The aim of this research is to add to the theoretical and empirical discourse on the impact of staff composition on IO performance by examining a highly influential but understudied IO. Furthermore, it demonstrates how employing new text-as-data approaches can address past issues with data availability, thereby broadening the empirical scope of the discipline.