Misallocation Losses and Financial Frictions in India

• In 2014, misallocation losses from dispersion in borrowing costs ranged from 7% to 12.5% of TFP in Indian manufactory sector.

• Misallocation losses from financial frictions decreased substantially between 1998 and 2004, but there has been no improvement from 2004 to 2014.

Hsieh and Klenow (2009) estimated output losses due to misallocation of resources in the Indian manufactory sector to be 60% more than what is observed in the U.S. manufactory sector. In this note, we provide an assessment of how much of these misallocation losses can be attributed do financial frictions (dispersion in borrowing costs) faced by Indian manufactory firms using the same conceptual framework that Gilchrist et al. (2013) employed to estimate the misallocation loses       

We use plant level data from the Annual survey of industries (ASI) conducted by the Indian government's Central Statistical Organization.  The time period we study ranges from 1998 to 2014. The dataset covers an universe of around 50.000 establishments, providing detailed information  on plants balance sheet and income statements. 

Gilchrist et al. (2013) propose a partial equilibrium framework where establishments produce according to a Cobb-Douglas production function with decreasing returning to scale in the combination of capital and labor: 

In their framework,  financial frictions generate dispersion in the interest rate charged from each establishment, creating wedges that prevents the marginal productivity of factors of production o be equalized across establishments. The marginal productivity of capital is not equalized due to dispersion in interest rates and the same is true for the marginal productivity of labor. This later effect happens due to he assumption that establishments have to finance their payroll borrowing working capital. In such framework, it can be shown that TFP losses from misallocation can be estimated by:

Where σwl and σwk   represents he dispersion in the marginal productivity wedge that arises due to the dispersion in interest rates, ri . 

Below we document the evolution of dispersion in interest rates paid by establishments in the ASI.

We use the series which drops 2% observations as the tail of the interest cost distribution likely capture reporting errors or firms in severe financial distress - Gilchrist et al. (2013) - and we are interested in the dispersion related to market inefficiencies. Below we report  a summary of the interest cost distribution and estimated misallocation losses from financial frictions for the years 1998, 2014 and 2014:  

The last column show estimated losses due to interest rate cost dispersion in percentage point. The second number in parenthesis show estimated losses if we drop the assumption that firms have to finance they payroll with short term working capital borrowing, a scenario in which losses would come only from misallocation of capital. 

Interest rates until 2014 were above 11% for manufactures in India, much higher than the 2.4% documented by Gilchrist et al. (2013) for the US economy. Naturally, dispersion of borrowing costs is also much higher for India. Whereas in the US the standard deviation of borrowing cost is 1.66% (63.7 % of the mean cost), in India for 2014 the dispersion is 10.9 % (82% of the mean). Misallocation losses form dispersion in borrowing costs for India, in 2014 were 12.61% (or 6.94% if one consider only losses steaming form capital misallocation), much higher than what 1.75% (1.4%) estimated by Gilchrist et al. (2013) for the US economy. 

One last point noteworthy, from 1998 to 2004 dispersion in borrowing costs decreased substantially in India, indicating a reduction in the degree of financial frictions in that market and, consequentially, a decrease in losses due to misallocation. But apparently progresses was halted as dispersion in borrowing costs have not changed much from 2004 to 2014. 

**Caveats: we take dispersion in interest expenses reported in the establishments' financial statements as a measure of dispersion in borrowing costs. The interest expenses reported might not truly reflect borrowing costs that establishments face in the market for new borrowings.  Still, we believe the dispersion in interest expenses in the aggregate should be a good proxy for dispersion in borrowing costs.