What's good for the goose ain't good for the gander: Cock-eyed counterfactuals and the performance effects of R&D

We investigate the effects of R&D investment on performance outcomes (sales growth and relative profitability) for Indian manufacturing firms. Previous research shows contradictory results - while some studies find a positive effect of R&D on firm performance, some find that firms investing in R&D do not perform significantly better, in some cases, even perform worse than their non-investing counterparts. We claim that the effects of R&D on performance are often mis-specified: The contradictory results are likely due to 1) inverse causality, i.e., firms invest in R&D as a function of sales growth and/or 2) a bias caused by censored data (i.e. R&D investment has a lower bound at zero). We apply endogenous switching regression to tackle the issue of selection and censored data, and the results we observe are sharp: firms investing in R&D would have had less growth and less relative profitability if they had not done so. Interestingly, firms that did not invest in R&D would not have benefited had they done so.