Electronic Thesis and Dissertation Repository


Doctor of Philosophy




Dr. James MacGee


My thesis consists of three chapters that contribute to the study of input misallocation and TFP growth in China.

In Chapter 2, I compare the misallocation of intermediate goods to those of capital and labor, which have been extensively studied in the literature. To measure misallocation, I compute the dispersion of marginal products of intermediate goods across firms, and the potential output gains by eliminating this dispersion in China Industrial Enterprise Survey (CIES) data. Although the within-industry dispersion of marginal products of intermediates is smaller than that of capital and labor, gross output and value added gains from reallocating intermediate goods are 6 and 14 times those from capital and labor reallocations. If intermediate goods, capital and labor are reallocated to equalize their marginal products, the total value added gain in the CIES is 550%, much greater than the 98% obtained under the value added approach in the literature (i.e. Hsieh and Klenow, 2009). This suggests that with its 74% revenue share and input complementarity, distortions and frictions through intermediate goods could be a promising channel to account for sizable misallocation in China’s data. I further find suggestive evidence of preorder friction: intermediate goods need half a year to pre-order, which gives rise to borrowing constraints in paying for intermediates. Similar to capital, marginal products of intermediates are found to be more dispersed among potentially constrained firms with low net worth, as one would expect if borrowing constraints exist.

Motivated by the findings in Chapter 2, Chapter 3 quantifies the novel role of pre-order friction and borrowing constraints on intermediate goods in accounting for misallocation in the CIES data (Hsieh and Klenow, 2009; Brandt, Van Biesebroeck, and Zhang, 2012). With a gross output production function, I incorporate intermediate goods frictions into the firm investment model of Cooper and Haltiwanger (2006). Firms order and prepay for a fraction of intermediate goods one period in advance (pre-order), and face one borrowing constraint on capital and intermediate goods. Firms also face capital adjustment costs. I measure misallocation by the potential gross output gain as a percentage of actual gross output, if intermediate goods, capital and labor are hypothetically reallocated to equalize marginal products across firms. Over 1998-2007, gross output misallocation in the CIES data averages 140 percent. The model accounts for around 70 percent of this misallocation, when calibrated to key moments in firm-level debt, productivity and market share distribution in the CIES data. Half of the misallocation in the model is attributed to intermediate goods frictions: 34 percent from borrowing constraints, and 16 percent from pre-order. While borrowing constraints on capital induce small misallocation, capital adjustment costs account for the other half. Larger misallocation with intermediate goods frictions than without arises from its large gross output revenue share and recurrent need of financing. This tightens the borrowing constraint and interrupts the self-financing mechanism for capital accumulation. Further, as in Chapter 2 for the data, I find that value added approach in literature also underestimates misallocation by ignoring misallocation of intermediate goods for the mode. The importance of intermediate goods frictions in misallocation could be applicable to other countries with an underdeveloped financial system.

Chapter 4 decomposes China’s fast aggregate manufacturing productivity growth into firm-level technological growth, intensive reallocation of inputs across existing firms and extensive reallocations through net entry. Following Baily, Hulten, and Campbell (1992)’s approach of aggregate productivity decomposition, I find that extensive reallocation accounts for 93% and 144% of 5-year aggregate productivity growth in 1998-2003 and 2002-2007 in the CIES. In contrast, intensive reallocation contributes -10% and -93% to the growth. These estimates are however biased by a left-censoring problem, since CIES does not survey non state owned firms with sales less than 5 million yuan. I propose a methodology accordingly to recover the three sources of growth in China’s manufacturing sector. I find that when China’s data is taken as the manufacturing universe, the role of extensive reallocation in aggregate productivity growth is overstated by a quarter to two thirds during 1998- 2003 and 2002-2007. Most of the overstated magnitude in extensive reallocation is picked up by the intensive reallocation among existing firms. Compared to U.S., intensive reallocation in China’s manufacturing sector is still smaller, but larger than that directly implied in the CIES. This also indicates that the left-censoring problem in other countries should be taken into account when analyzing micro-level sources of aggregate productivity growth across countries.