Appendix
The basic data sources are Zhongguo Guding Zichan Touzi Tongji Ziliao 1950–1985 (China Fixed Investment Statistical Materials), 59, 64, 218–19; 1988 Tongji Nianjian (Statistical Yearbook), 564, 566, 605; 1988 Tongji Zhaiyao (Statistical Abstract), 67. In the table below, technical transformation includes "other" investment. Column (5) is the sum of columns (1) and (2), divided by net material product , and column (6) is the sum of (3) and (4), again divided by NMP.
Data on budgetary and extrabudgetary investment . The capitalconstruction figures are consistent and readily available. However, I wish to include foreign borrowing disbursed through the central budget in budgetary investment. This is available through 1985 in the first source cited, and 1985–86 figures are available in

1987 Tongji Nianjian , 467n. The 1985 figure thus given is 4.103 billion, and the 1986 figure is 3.003. The latter figure can be confirmed with reference to 1988 Tongji Nianjian , 564. However, the 1987 figure is not available. Since total state investment funded by foreign capital increased 31.2 percent during 1987, and budgetary foreign borrowing of all kinds increased 36 percent, while foreign direct investment increased 33 percent in renminbi terms (1988 Tongji Nianjian , pp. 565, 732–33, 762), I have assumed that foreignfunded investment channeled through the budget increased by onethird to 4.0 billion.
Figures for technical renovation are more complicated because data on "other" investment are sometimes included and sometimes separated, but separate data are never presented with the breakdown we require. Data are good for technical transformation, not including "other," for 1981–85 in the first source cited. However, the 1980 data are incomplete and seem unreliable. Figures for 1978–84 on financing for the category of technical transformationincluding"other" are available in the earlier Statistical Yearbooks, e.g., 1983, English, 360, and 1985, 452. These are used for those years. These show that budgetary finance of "other" investment was very small but not zero. The composition of "other" investment financing 1985–87 thus is the only remaining financing problem. Overall, the composition of "other" investment changes considerably through 1983, but is quite stable in the 1984–87 period. In 1984 it can be calculated that 1.25 billion, or about 10 percent, of "other" investment was funded from the budget. This proportion has been applied for the years 1985–87. In the calculation of budgetary and nonbudgetary shares of investment some approximation is required, but the possible margin of error is very small.
Data on Investment Included in Central and Local Plans . Data for capital construction are clearly presented in the first source cited. It is less clear what proportion of technical transformation and other investment should be counted as central government. The figures from 1981 through 1987 for technical transformationexcluding"other" are good, but there are no figures at all for "other." Unfortunately, there is no alternate source where the proportion of technical transformationplus"other" that is included in the central plan is given. I have therefore assumed that no "other" investment is central. An alternate assumption would be that petroleumdepletion allowances, which account for a little over 50 percent of "other" in 1984–87, are also "central." This alternate assumption would raise the proportion of total investment included in the central plan in every year, but would have virtually no impact on the trends that are the subject of the analysis.
The other problem is determining the proportions of technical transformation that are incorporated in the centralgovernment plan during 1978–80 (there was no "other" investment during those years: thus, this is implicitly technical transformationplus"other"). I simply assumed central projects were equal to 27 percent of technical transformation in those years, as they were in 1981. This is plausible because the proportion of capital construction accounted for by central projects is quite stable during the 1978–80 period, and we know that technical transformation investment at this time was predominantly under local control. The maximum error involved in this procedure would be if we undercounted centralgovernment technical transformation projects by 2–3 bil
lion in 1978 through 1980, which would amount to at most 1 percent of NMP in 1978, or 1/2 percent of NMP in 1980 (a more likely error, since bank lending for technical transformation starts to become important at that time, and that might fund centralgovernment projects). Thus, there is a margin for error in the 1978–80 figures larger than that involved in the calculation of budgetary shares. Nevertheless, the margin is not sufficient to materially alter the qualitative conclusions presented here.