Analyzing firm size and stage-wise efficiency of financial performance for non-fragile firms: Evidence from the Indian food processing industry

Purpose This paper investigates the non-optimizing behaviour with respect to profits for non-fragile firms using empirical evidence from dairy and grain milling sub-sectors of food processing in India. This is studied in two stages: (I) from inputs to saleable output and (II) from sales to net profits. The purpose is to identify which stage explains efficiency in profit generation. Design/methodology/approach The technique of two-stage network Data Envelopment Analysis (NDEA) is employed to find the source of inefficiency. Quantile and ordinary least squares regressions are used to identify the drivers of inefficiency. Further, the machine learning technique of regression tree is employed to check the validity of exogenously defined firm size cut-offs. Findings Stage II inefficiency drives overall inefficiency in profits for all firms in both sub-sectors from the NDEA results. While some small firms achieve full efficiency in the second stage, no firms of a larger size do so. Grain milling, marked by fewer medium sized firms than dairy, also reports lower stage II and overall profit efficiency than dairy. Originality Existing studies in food processing industries in India have used only the conventional DEA methodology or the Malmquist index. Our paper improves on this with the NDEA for firms with strictly positive profits. We also demonstrate the usage of a machine learning technique for validating exogenous firm size cut-offs.