How to improve SME performance using iterative random forest in the empirical analysis of institutional complementarity

Empirically investigating institutional complementarity within organizations has long been challenging in the social sciences. This study applies an iterative random forest (iRF), a machine-learning approach developed in biostatistics, to World Management Survey (WMS) data to examine complementarity among management practices in small and medium-sized enterprises (SMEs). Using 18 management quality indicators, the analysis evaluates their effects on profitability, growth, and viability. Results highlight the relative importance of rewarding high performers, reassignment and retraining of poor performers, and clearly established performance criteria. Moreover, the ability to set short-term goals based on a long-term perspective exhibits broad complementarity with many other indicators. These findings align with prior survey-based evidence on institutional complementarity, suggesting iRF is a credible and promising method for empirical research on organizational complementarities.

Atsushi Sannabe