Comments and Discussion Nicholas Lardy and Lawrence H. Summers Nicholas Lardy: My comments on this paper by Wendy Dobson and Anil Kashyap fall into three broad areas: the political economy of bank lending, the measurement of policy lending, and the cost of bank recapitalization. The political economy of bank lending decisions is not very well spelled out in the paper. Dobson and Kashyap believe that banks in China are still under pressure to make policy loans to support state-owned enterprises. And they imply that the requirement placed upon banks to support these enterprises has been fairly persistent over the past two decades. But this approach is not satisfactory in two dimensions. First, the share of bank lending going to state-owned companies has fallen by half over the past decade, in part because the state-owned sector has shrunk relative to the economy as a whole, and in part because banks have become more selective lenders. At year-end 1995, borrowing by state-owned enterprises accounted for 83 percent of all loans from the banking system.1 By the end of November 2004 this share had fallen to 43.5 percent.2 A large part of this decline is explained by the rising importance of lending to the household sector; such lending was nonexistent before 1997. By the end of 2005, 11.2 percent of all bank loans outstanding were to retail customers. At the margin the share of households in annual credit expansion was much larger: 18 percent in 2003 and 19 percent in 2004, but falling to 8.5 percent in 2005, according to data from the People's Bank of China. Second, the authors' undifferentiated approach does not take us far in explaining the volatile pace of bank lending over recent years. Bank lending began to accelerate at the end of 2002 and hit an all-time-high expansionary [End Page 149] pace in 2003, when the stock of yuan-denominated loans outstanding increased by 2.77 trillion yuan, an amount equal to 20.3 percent of China's 2003 GDP. Lending growth moderated significantly in 2004 and 2005: loans outstanding grew by 14.1 percent and 12.8 percent of GDP in those years, respectively. Then, however, lending surged to a near-record pace in the first half of 2006, when the stock outstanding increased by 2.18 trillion, equivalent to 23.8 percent of first-half GDP as reported by the NBS. New loans extended by Chinese banks in the first half of 2006 were a stunning 50 percent more than in the same period in 2005.3 What is it about the "pressure to make policy loans" that leads to this particular temporal pattern? My second point concerns how the authors define and measure policy lending. Can one still assume, as the authors have implicitly done, that all lending to state-owned enterprises should be regarded as policy lending? Should we regard the two-fifths of loans outstanding to state-owned enterprises at year-end 2004 as a proxy for policy lending? Or is a growing share of this lending, at least in the industrial sector, driven by commercially oriented banking practices? Reported before-tax profits of state-owned industrial companies have soared in the wake of the industrial restructuring that accelerated in the mid-1990s. By the first half of 2006, before-tax profits of state-owned industrial firms were 4 percent of GDP, compared with 1 percent in 1999.4 Much of this quadrupling of profits is due to shrinkage in the financial losses suffered by the subset of state-owned industrial firms that are unprofitable. These losses fell from 115 billion yuan, the equivalent of 1.4 percent of GDP, in 1998 to 66.9 billion yuan, or 0.4 percent of GDP, in 2004.5 In part the growing profitability of state-owned manufacturing firms stems from a substantial reduction in employment in firms that remain state owned. The sharp fall over time in additions to inventory as a share of GDP also contributed. In the mid-1990s average annual additions to inventories exceeded 5 percent of GDP. In 2003-05 these additions averaged 1.4 percent of GDP...