[MUSIC] To what extent has the misallocation of resources in the economy affected the level and the growth rate of GDP in China? In this presentation, we are going to talk about the problem of resource misallocation. And we're going to first present a framework for defining what is the optimal resource allocation, so that we can then think about how to measure the degree of misallocation in the economy. Then we'll present some evidence about the extent of misallocation of resources across manufacturing firms in China. And also the extent of misallocation of resources across different provinces and sectors of the economy. Now when we think about the optimal resource allocation to a firm, it's helpful to think about the marginal revenue product being a function of the amount of input that is being put into a production process for a firm. Specifically, we usually think of this curve as being downward sloping. Because there are decreasing returns to scale, which simply means that as I increase the amount of input X by an extra unit, the amount of additional output I get will eventually start to decline. And in this framework, if a given firm, a, has a amount of inputs Xa, then that's going to correspond to a marginal revenue product a here, based on this curve. And similarly, if there's another firm, b, which has much fewer inputs, then it's going to have a much higher marginal revenue product. It might be obvious from just looking at this picture that we could improve or increase overall output in this economy by simply taking away some of the resources from firm a, which has a lower marginal product and giving those resources to firm b, which has a higher marginal product. To maximize efficiency, if these two firms are identical in their technologies, we should allocate the exact same amount of inputs to each of the firms, to the point where they have the same marginal revenue product. Only when the marginal revenue products are the same is it the case that there's no gain from reallocating resources from one firm to the other. This also suggests that an efficient economy is one where most firms have the same marginal revenue product, or most activities have the same marginal revenue product. And in a firm or an economy that's inefficient will be characterized by a large variance or kind of wide distribution of marginal revenue products in the economy. So there are some very efficient firms and some very inefficient firms coexisting, and there are large potential gains from reallocating resources. Now, why would we actually observe misallocation? Why do some firms have more inputs than other firms, even if they basically are facing the same technology? Well, we usually think that firms have an incentive, if they're profit maximizing, to allocate resources to the point that marginal revenue product equals the cost of that production input. So I'm going to keep hiring workers until the marginal product of that labor is exactly equal to the current market wage. And so everyone should be hiring workers to the point where the marginal product equals weight. And it therefore stands to reason that if there is misallocation, it could be that firms simply are facing different prices. This could be the case if the government is subsidizing some firms, but not others. For example, they're giving state enterprises a cheaper interest rate than non-state enterprises. It could also be because there are costs of moving resources from one place to another. So for instance, for migration to get a worker to move from province A to province B, it might be the case that there's a cost of moving that person. He has to actually buy a train ticket, set up a new place to live, etc. And he also may need to be compensated for leaving behind his friends and relatives. So there's going to be a wage gap that is going to then occur in the different regions. Now, there also could be firm characteristics that would affect the prices for inputs. For example, a larger firm might have more bargaining power and so get, for instance, a cheaper interest rate from the bank. If we're going to try to look at this issue empirically, and look at the distribution of marginal revenue products in the data, we also are going to be concerned with some measurement issues. One is that there could be some omitted factor or inputs that we're not really taking into account. So we might not really understanding them the margin of productivity completely accurately of each firm. And also, there could simply be measurement error in the revenues or profits of firms, which will make it look like there's a wider distribution of margin of revenue of products than is actually the case. Now, there's been an interesting research paper which has actually documented the extent of the variance of the marginal revenue products across firms, manufacturing firms. And they did this for three countries, India, China, and the United States. And you can see visually that there's a very stark difference between the distribution of marginal revenue products in India and China compared to the US. The US distribution is much narrower. And so there's less variance in these marginal revenue products, reflecting a more efficient allocation of resources. China has many more firms, and India too. And the left-hand tail of the distribution, these are really low return, low productivity firms. If we do that same analysis for different periods of time, we find an interesting result. If we think of what would happen if China and India could match the distribution of marginal products in the US, and take the US as kind of an efficient benchmark. Then what we'd see is that the gains from matching US efficiency are large for both countries, but they are declining in China and increasing in India. So that China's distribution is becoming narrower, and its resource allocation is becoming more efficient over time, whereas the opposite is the case in India. So some implications here are that more efficient allocation of resources in Chinese manufacturing can actually explain about one third of the total factor productivity growth in China over the period being studied. And the total factor productivity was about 6% per year. So just increasing efficiency increased TFP by about 2% per year. 40% of this improvement can actually be attributed to shrinking the gap between state-owned enterprises and other firms. So this inefficiency in allocation between the state and non-state sector has actually improved over time in China. Misallocation can explain, on average, nearly half of the productivity gap between the US and China, which is a really huge potential gain from reducing misallocation. And finally, as noted, China contrasts to India, in that China is showing improving efficiency over time, wherein it's the opposite in India. Now, there's another interesting empirical analysis. We just studied the distortions in the allocation of labor and capital in the non-state and state sectors in different provinces in China. And this helps us try and understand the extent of misallocation across broad sectors of the economy and broad regions of the economy. And what we can learn from this type of analysis is we can see clearly the large gaps between the state and non-state sectors, what it means for growth. Also, it reveals the extent to which there may be existing barriers to mobility of capital labor across provinces and between sectors. And we can see how distortions may lead to lower, not only to lower TFP in each grid, but that changes in the distortions could be affecting the growth rates of TFP and overall growth in different periods. So this figure shows this large difference in productivity growth in the two sectors, the state nonagricultural sector and the non-state nonagricultural sector. And here these are applying the levels of TFP. So what you can see is that the state sector shows no growth. It's very flat. Whereas the non-state sector shows quite substantial growth and productivity over time. And the bars actually reflect the variability across the provinces. And you can see also that the variability in this total factor productivity across provinces is also much smaller in the non-state sector. The analysis also can separate the degree of distortions or misallocation within provinces, across sectors, and between provinces. And the difference in the results are quite interesting, because if we look at the left hand panel here, the within-province distortions, we see that there's a big distortion of capital allocation between the non-state and state sectors within provinces. But very little distortion of labor between sectors within provinces. In contrast, if we look at the allocation of capital labor across provinces, we see kind of the opposite result. That now we see a large inefficiency in the allocation of labor across provinces. But very little inefficiency in the allocation of capital across provinces. So capital seems to flow freely across space, but doesn't seem to be allocated in a reasonable way between the non-state and state sectors. Whereas labor can go where it wants, to state or non-state sectors, but moving geographically is costly. If we look at the extent of the distortions as a percent of the total factor productivity in the economy, we find that over the total period studied here, 1985 to 2007, the average distortion of TFP is about 20%. If we also look at how the distortions are changing over time, we can see that there's this difference between the average efficient TFP growth and the average actual TFP growth. And for the earlier period, '85 to 1997, actual TFP growth is faster than efficient TFP growth, which means that there's increasing efficiency, or a decline in distortions, that add kind of an extra 0.5% to TFP growth over that period. But in the more recent period, 1997 to 2007, there is actually a slower average actual TFP growth to the growth in efficient TFP. And that means that resource allocation is becoming less efficient and this is reducing growth in total factor productivity by about a half a percent a year. Now in conclusion, what have we learned about the level of misallocation and its relationship to growth in China? Well, first of all, we know now that the misallocation of resources is a substantial problem, that across manufacturing firms it explains about 30% of total factor productivity. And the distortions account for about 20% of total factor productivity in the allocation between the state and non-state sectors within the nonagricultural sector. Reducing barriers to mobility in both labor and capital thus can contribute positively to growth going forward. And in particular, the results suggest that much can be gained by reducing the barriers to inter-regional labor mobility. Which is probably related to institutional differences in the treatment of rural residents to urban residents, and various types of hardships that migrants may be facing in urban areas. And also, a lot can be gained from reducing the biases in the financial system, and the allocation of capital that seem to favor the state sector.