Conclusion
This paper links the current literature on capital misallocation with a classic literature on investment-cash flow sensitivity. It provides a simple accounting device to compute the aggregate productivity loss due to capital misallocation in the presence of financial frictions. We make use of the differences in the stage of financial development of U.S. and China, and interesting institutional features within China, to apply various sample-splitting tests using an errorcorrection investment model. Our estimated investment-cash flow sensitivities imply an aggregate TFP loss around 5% for the balanced panel and 15% for the unbalanced panel of the Chinese manufacturing firms. Thus on the one hand, our finding echoes the literature on the importance of financial frictions on efficiency loss by deterring entry and exit. On the other hand, our results are in line with Midrigan and Xu (2014) and Gilchrist et al. (2013), who find that financial frictions are unlikely to cause substantial efficiency loss among existing and ongoing firms.
This of course raises an interesting question, when we consider those large TFP losses identified in Hsieh and Klenow (2009), Brandt et al. (2013) and Song and Wu (2015) from capital misallocation in China. Banerjee and Dufflo (2005) offer a discussion on various causes of capital misallocation in addition to financial frictions. One possible candidate is studied in Wu (2018), who finds that the vast majority of capital misallocation in China is due to policy distortions instead of financial frictions. Another explanation, which is not specific to China thus more general, lies in the role of technology adoption. Midrigan and Xu (2014) conclude that the impact of financial frictions on technology adoption is more important than its impact on the allocation of capital across plants for explaining TFP. The role of financial frictions for technology adoption is the focus of the work of Cole et al. (2016).