some of China's Measuring,Labor’s,Share,of,Income,in,China:,1993-2008

　　Abstract: Accurately measuring labor’s share of income is of great importance for China’s national policies, particularly its labor compensation policies during the 12th Five Year Plan period. This paper first analyses the measurement problems of labor’s share of income from the perspectives of definition and data and their influences on the accuracy of the measurement results. Then it sums up and appraises several adjustment methods for existing measurement problems. Finally, the author estimates the operating surplus of private, unincorporated enterprises (OSPUE) in China using data from Urban household survey (UHS), Rural household survey (RHS) and population survey, and further makes a correction to labor’s share of income in China from 1993 to 2008 with data from flow of funds accounts. Finally, the author conducted trend analysis and international comparisons using the results obtained from the first part of the paper.
　　Key words: labor’s share of income, measurement problem, correction
　　JEL Classifications: E25, J30
　　1. Introduction
　　Labor’s share of income, which is also referred to as the ratio of labor income, denotes the percentage of labor’s income in national income distribution. In the market economy, factors of production receive different compensation according to their contribution, which is seen as an effective means of resource allocation. Distribution of national income among all factors of production constitutes functional forms of national income distribution. Economics research generally presumes only two input factors (capital and labor), and capital income and labor income together constitute national income. Under this assumption, labor’s share of income also reflects the distribution of capital income; therefore, much research on functional forms of national income focuses on measuring labor’s share of income.
　　Over the course of China’s rapid, sustained economic growth since the 1990s, its national income distribution has undergone significant change: a striking feature of the current expansion is the unusually low share of national income being earned by labor. Using provincial income to approximate GDP, Li (2006) found that the share of labor in national income decreased from 53 percent in 1990 to 40 percent in 2006. With an adjusted flow of funds statement and data from the 2004 economic census, Bai and Qian (2009) found that labor’s share of income reached an historic high of 66.83 percent in 1996 and trended downward in the following years, decreasing by 16.2 percentage points cumulatively as of 2007. 　　Using the same approach to GDP, Luo and Zhang (2008) found out that labor’s share of income fell from 51.4 percent in 1995 to 46.2 percent in 2003 and dramatically to 41.6 percent in 2004, far lower than the level in most countries (55 percent to 65 percent). Xiao and Hao (2009) observed that the mean value of labor’s share of income decreased from 57.3 percent over the years 1978 to 2003 to 47.7 percent for the years 2004 to 2007, reflecting substantial changes from before and after the 2004 economic census. These two figures are lower than the average for developed countries and even lower than the average for developing countries. Based on the operating income of urban and rural self-employed individuals, Zhang and Zhang (2010) found that labor’s share of income increased from 40.15 percent in 1978 to 45.69 percent in 1995, before dropping to 39.16 percent in 2007. Zhou (2010) observed that net tax on production serves as the wedge between labor and capital, which after removed from GDP, the decrease of labor’s share of income postponed from 1995 to 1998 (from 56.0 percent in 1993 increased to 59.3 percent in 1998, and then dropped to 54.8 percent in 2003).
　　While most of the recent studies concluded that labor’s share of income in China is not only among the lowest internationally but that it also has been trending downward since the 1990s. Statistical calibers applied in these studies, however, are inconsistent and can thus result in varied conclusions. Using an in-depth measurement of labor’s share of income, including possible correction methods to account for definition and accuracy, this paper carries out a more thorough analysis of labor’s share of income in China between 1993 and 2008.
　　2. Measurement Problems of Labor’s Share of Income
　　Despite thorough exploration by scholars (Krueger 1999; Gollin 2002; Gomme 2004), the economics community has yet reached a consensus on how to accurately measure labor’s share of income. One common practice is to use the following formula in conjunction with national accounting indicators:
　　(1)
　　Formula (1) looks rather simple, but we argue in this paper that it does not reflect the below two sets of measurement issues.
　　2.1 Definition Related Issues
　　Labor compensation and value added are difficult to define accurately. Specifically:
　　(1) Accounting definition vs. economics definition of labor compensation. Labor compensation definition from accounting perspective does not equal labor income definition in Economics. The latter presumes no existence of human capital accumulation, i.e. per-unit labor income1, while the former refers to the sum of income paid to laborers including benefits from human capital accumulation. Estimation of labor’s share of income through the Cobb-Douglas production function and Mincer function differs from that of formula (1), and results from the two types of definition s should not be compared against each other. However, due to the inherent difficulties in estimating human capital stock and revenue, most estimates of labor’s share of income are based on labor compensation indicators. 　　(2) The accounting boundaries of labor compensation. The accounting boundaries of labor compensation are not so clear. As Krueger (1999) asked: who is considered “labor”? Does labor cover CEOs and proprietors? Which components belong to “labor compensation”? Are stock options given to executives considered labor compensation? How should we handle social security payments and pensions received by retired workers? Boggio (2009) also asked how should CEO and proprietor income be divided between labor and capital. Because of management’s needs for convenience when calculating, a broader caliber definition of labor and labor compensation are used in actual accounting. The System of National Accounts (SNA) in 1993 divided labor into employees and self-employed workers, with employees receiving wages from a corporation and the sum of cash or payments in kind are regarded as labor compensation. Self-employed workers are defined as the sole or joint owners of unincorporated enterprises who work for that enterprise. For this type of enterprise, it is difficult to designate income received as that of an employee or that of a proprietor; thus, SNA (1993) introduced a new concept: mixed income, which includes labor compensation and proprietor income (Xu 2001).
　　(3) How to choose value-added indicator. Disagreement on this figure is primarily the result of gross indicators and net indicators. Gross value added is equivalent to net value added plus depreciation. Should depreciation, however, be included or excluded from our measure of output? Normally depreciation is reflected in the market price of final products or services, and accounting figures for a country or sector’s gross output contain wear and tear of fixed assets (Morel 2009). In addition, as argued by Gomme and Rupert (2004), it is difficult to justify using net value added on the basis that depreciation merely compensates owners of capital for the physical wear and tear of their capital while labor is likewise subject to wear and tear, both physically and intellectually. Thus, our research uses gross value added as an indicator.
　　(4) How to choose measuring price of value-added. There are three kinds of measuring prices in SNA 1993: basic factor price, producer price, and purchaser price, of which the former two are often disputed. Since gross value added (calculated using the producer price) less gross value added (calculated using the basic factor price) is equal to the net production tax, the essence of dispute is how to handle the net production tax. From the distribution of factor income, net production taxes constitute a substantial wedge between the share of capital and labor. Therefore, when measuring labor’s share of income, net production taxes should be excluded from gross value added. In other words, the value added indicator should be measured using the basic factor price (Gomme & Rupert 2004). 　　2.2 Data Related Issues
　　In system of national accounts, manufacturing account and the primary distribution of income account2 provide the primary sources of data for measuring labor’s share of income. In different countries, due to management convenience rather than rationale in Economics, the accounting procedures for the two types of account are different. Therefore, the data collection system tends to overlook content definitions, which leads to biased measurement in labor compensation and gross value added. Unfortunately, the deeper one digs into national accounts, the more biases one will discover in the methodology (Gomme and Rupert 2004). Gollin (2002) has also proposed that deviations in labor’s share of income among nations could be reconciled to a large extent if the definition and data handling were more careful and consistent. Usually, measuring biases of labor’s share of income come from data of the following three aspects:
　　(1) Self-employment Income. Most of the bias in measuring labor’s share of income depends on self-employment. First, self-employed income is considered as mixed income in SNA1993, which includes both part of labor compensation for self-employed workers, as well as part of investment income and profits. It is quite a headache to distinguish them. Second, randomly apportioning proprietors’ income also introduces uncertainty in measuring labor’s share of income (Krueger 1999). For example, if two doctors run an unincorporated clinic in a partnership, their income is counted as self-employed income. However, if they run an incorporated clinic and pay wages, their income is counted as employee compensation. Finally, the proportion of self-employed workers varies across countries and sectors. In general, the percentage is lower in developed countries and in sectors such as manufacturing, education, medicine, and public administration. On the other hand, it is relatively high in developing countries3 and in sectors such as agriculture and constructions. Failing to account for these discrepancies when measuring labor’s share of income fails to support accurate comparisons among countries and industries.
　　(2) Non-wage compensation and pensions. Stock options are a typical form of non-wage compensation. However, stock options are excluded from labor compensation in actual accounting due to barriers to accessing information and controversy in defining it as compensation. Pensions can be thought of as deferred labor compensation. Whether to define pensions as labor compensation during a given work period or how to allocate eventual pension compensation for on-the-job workers remains a huge technical challenge. SNA1993 suggests using one of three methods: 1) directly count the pension “created” by the on-the-job workers; 2) do not include it but rather count it as a current transfer; and 3) handle it virtually. In reality, accountants tend to neglect the size and wage level of retired people and adopt the first method. Obviously, if pensions included in the calculation are oversized or inadequate, this method will lead to large deviations. 　　(3) Value added. Errors in value added data are another source of deviation in measuring labor’s share of income. In addition to the selection of value-added indicator and its measuring price, errors come from three sources. First, the omission of the non-observed economy. In practice, some activities may be missed from national accounts: these activities constitute the non-observed economy, whose size varies among countries and industries. This affects the accuracy of comparisons of labor’s share of income. Second, the omission of capital income when calculating value added in the public sector. For non-market public sectors such as government, education, and medical care, current national accounting procedures normally adopt the input method or expense method, i.e. substituting value added with current payment plus depreciation of fixed assets. This input-substitute-output method sets operating surpluses at zero, which omits capital income. Taking the public sector into consideration, the measured share of labor’s income will be pulled up; if not, it will be pulled down. We can exclude the public sector or attempt to impute its capital income in measuring labor’s share of income (Gomme 2004; Morel 2009). Finally, labor compensation data in value added of owner-occupied housing is often omitted. National accounts impute rental (that is, capital) income to owner-occupied housing, which means they do not impute any labor income. If including the owner-occupied housing component, the measured share of labor’s income will be pulled down. If not, it will be pulled up. We can exclude the owner-occupied housing component or attempt to impute its labor income in measuring labor’s share of income (Gomme, 2004; Morel, 2009).
　　3. Correction Method in Measuring Labor’s Share of Income
　　Most measurement problems of labor’s share of income depend on data quality; therefore, there has no much room for correction. For some of the problems, however, a few scholars have offered some methods.
　　3.1 Johnson’ Correction Method
　　Based on U.S. historical data from 1850 to 1952, Johnson (1954) found out that labor’s share of income stabilized at 65 percent within the American economy. Since then, common practice for calculating the labor-capital ratio within an individual economy in economic literature is to apportion two thirds to labor income and one third to capital income. The precondition of this correction method is that labor income data remain stable. However, because of national accounting, Western countries since the 1970s and China since the 1990s do not meet this precondition. 　　3.2 Gollin’ Correction Method
　　In system of national accounts, unless a self-employed worker receives wages from his own enterprise or incorporate the enterprise, or benefits received are considered to be an operating surplus. This is known as the operating surplus of private unincorporated enterprises (OSPUE). Most of the income of the self-employed worker should be included in the OSPUE. In order to properly apportion OSPUE between labor and capital, Gollin (2002) proposed three correction methods.
　　(1) The first method presumes that self-employed workers provide only net labor services, and that therefore all OSPUE imputes to labor, as below:
　　
　　
　　
　　(2)
　　This method has the advantage of being both easy and intuitive, and the presumption conforms to the reality that most self-employed workers in low income countries have only net labor input. Even in some low income countries, however, self-employed workers tend to increase capital input. Therefore, formula (2) tends to overestimate labor’s share of income.
　　(2) The second method presumes that private unincorporated enterprises have the same share as the rest of the economy, so labor’s share of income of the national economy equals to:
　　Labor’s share of income =
　　
　　(3)
　　The assumption of this method means that factors such as a firm’s size and structure and the ratio of unincorporated enterprises in organizations do not affect labor’s share of income. This stands in contrast to the fact that most of unincorporated enterprises are labor-intensive enterprises, which can lead to further underestimation in formula (3).By and large, bias in the second method is much smaller than that of the first method.
　　(3) The third method uses labor compensation and employment survey data to determine labor’s share. This method assumes that average compensation of self-employed workers equals the average of labor compensation, thus:
　　(4)4
　　This method considers the fact that the ratio of self-employed varies in different countries. Due to errors in employment statistics, however, results obtained using formula (4) may be close to 1 or even above 1. Meanwhile, if systematic discrepancies5 exist in the average compensation between employees and self-employed workers, bias may also appear (Morel 2009).
　　In practice, Gollin (2002) pointed out that if official statistics had adjusted income figures of self-employed workers6, adopting the above correction method may lead to an overestimation of labor’s share of income. Likewise, if errors exist in self-employed workers’ output or income, then using the above correction method may cause the researcher to underestimate labor’s share. 　　3.3 Gomme and Rupert’s Correction Method
　　When using the income approach of GDP to allocate income between labor and capital, the component of GDP can be divided into three parts: the “unambiguous” income belong to labor , “unambiguous” income belong to capital and ambiguous income between labor and capital. Specifically, let denote unambiguous labor income (such as employee compensation), let denote unambiguous capital income (such as depreciation), and let denote ambiguous income (such as mixed income). Then, total labor income, would be
　　(5)
　　where a is labor’s share of income, as yet undetermined. Since a is labor’s share of gross value added (Y), it follows that
　　(6)
　　These two equations can be solved for labor’s share:
　　(7)
　　The basic idea underpinning Gomme and Rupert’s correction method is exactly the same as that of Gollin’s second correction method; however, the former’s content is broader.
　　4. Measuring Labor’s Share of Income in China
　　4.1 Selection of Basic Data
　　There are three types of basic data use in the measurement. 1) Physical settlement in flow of funds accounts. Historical Data of Funds Flow Statements: 1992-20047 provides funds flow statements for the years 1992 to 2004 after the adjustment of the first national economic census, while the China Statistical Yearbook 2010 provides adjusted data for the years following the second national economic census. 2) Income approach of provincial GDP. Historical Data of Funds Flow Statement: 1952-2004 provides data supporting an income approach to GDP. China Statistical Yearbook 2006-2010 provides data supporting a provincial approach GDP for the years 2005 to 2007 and for the year 2009.8 3) Input-output table. This is used to test the labor’s share of income for the nation and for different industries. China has compiled input-output table for 1987, 1990, 1992, 1995, 1997, 2000, 2002, 2005, and 2007.
　　Of the three types of data, input-output table data has the poorest consistency and are updated after significant lag time9. Provincial income approach data used to calculate GDP can fluctuate wildly, and there is a large error margin between provincial added value and national GDP. Data from flow of funds accounts, however, do not have these shortcomings. Therefore, we use funds flow statement data for the years from 1993 to 2008 to measure labor’s share of national income in China.10
　　4.2 Measurement of China’s OSPUE
　　China’s flow of funds accounts was compiled on the basis of GDP accounting. For GDP accounting prior to 2004, labor compensation includes not only labor income of all employed persons but also earned profit of individual proprietors, making it broader than labor compensation as defined in SNA 1993 (Xue 2001). For GDP accounting after 2004, because it was difficult to distinguish labor income and operating profits of private unincorporated enterprises’ proprietors, the two components were regarded as operating surplus, while labor compensation was defined to include only employee compensation. 　　Except for operating profits of rural household, the rest of compensation of employees using the new definition is similar to that used in SNA (Department of National Economic Accounting in National Bureau of Statistics of China, 2007, 2008). In compiling flow of funds accounts 90 percent of value added in the rural household sector is used as the figure for rural household compensation.11 This ratio is close the labor’s share of income of primary industry.12 Thus, in this paper, we do not seek to further adjust the self-employment income of rural households; rather, adjust only the self-employment income of private unincorporated enterprises’ proprietors, i.e. China’s OSPUE.
　　Steps for measuring China’s OSPUE: 1) According to the Rural household survey from 1993 to 2008 and population survey data from 1992 to 2008, we obtained the annual net income data from non-agriculture operations13, as well as annual average rural population data. We then multiplied the two to obtain rural OSPUE. 2) Using the same interval data from Urban household survey and population survey, we obtained annual net business income data and population data for Urban households, and we multiplied the two to obtain urban OSPUE. 3) We combined the figures for rural OSPUE with those for urban OSPUE to obtain China’s total OSPUE. Results are presented in Table 1.
　　4.3 Measuring Labor’s Share of Income in China
　　For an easier comparison, we substituted gross value added data using producer price into formula (1) and obtained labor’s share of income using the producer price gross value added indicator. Due to data quality issues, calculating labor’s share of income using this indicator can lead to imprecision. After removing net taxes on production and using formula (1), we obtained labor’s share of income by basic factor price gross value added indicator. As this indicator does not include the OSPUE, labor’s share of income may have been seriously underestimated. Due to the dual structure characteristics of China’s economy, labor compensation varies among employees and individual proprietors, as well as between urban and rural areas. In addition, current employment and wage statistics do not precisely reflect the compensation discrepancy between employees and individual proprietors. Thus, we used Gollin’s method to correct. Results are shown in Figure 1.
　　Figure 1 shows that, using the producer price indicator, labor’s share of income shows a stable upward trend from 1993-1997 and reaches a peak of 53 percent in 1997. Subsequently, it trends downward and reaches a low value of 47.79 percent in 2008, an overall decrease of 5.21 percentage points. Regarding results when the correction was applied, adopting Gollin’s correction method did not apportion capital income to individual proprietors, which in turn means less capital input and slight overestimation. Labor’s share obtained with Gollin’s second correction method proportions too much capital income to individual proprietors, and it therefore underestimates labor’s contribution. The trends of both correction results are almost the same, and a discrepancy for results for same year is fewer than two percentage points. Therefore, we use the average of the two results as our estimate for labor’s share of income in China. 　　4.4 Trend Analysis and International Comparisons of the Result
　　Figure 1 shows labor’s share of income trending upwards from 1993 to 1999 and reaching its peak (66.97 percent) in 1999. The share then trends down to reach a low point of 60.56 percent in 2007, a drop of 6.41 percentage points. Compared with results calculated using producer price, these results show labor’s share dropping an additional 1.20 percentage points two years later.
　　Table 2 shows that compared with mean levels of labor income share in both developed and developing countries, China’s share is lower by ten percentage points and five percentage points, respectively.
　　5. Conclusion
　　Accurately measuring labor’s share of income is the basis for analyzing historical changes in functional forms of national income distribution and evaluating the rationality of factor income distribution in China. More importantly, it is the prerequisite for monitoring and commenting on labor compensation policies during the Twelfth Five Year Plan period (2011-2015). This paper first builds a comprehensive framework for calculating labor’s share of income in China and offers methods to correct for discrepancies and data quality issues. We then estimated China’s OSPUE using carefully calibrated labor compensation statistics from both urban and rural survey and census data. Finally, we made a correction to labor’s share of income in China from 1993 to 2008 with data from flow of funds accounts
　　Our findings show that labor’s share of income in China rose steadily before 1999, then decreased from 66.79 percent in 1999 to 60.56 percent in 2007. We also observed that labor’s share of income is five to ten percentage points lower than in other countries. Finally, we observed that after applying correction methods, results differed significantly: labor share was shown to decrease two years later than in previous calculations and by 1.20 percentage points more. We reemphasize that solving national accounting issues and improving data quality are the prerequisites for measuring labor’s share of income accurately.
　　References
　　[1] Bai, Chongen, and Zhenjie Qian.2009. “Who Is the Predator, Who the Prey?—An Analysis of Changes in the State of China’s National Income Distribution.” Social Sciences in China (4):179-205.
　　[2] Boggio L., Dall’Aglio V., and Magnani M. 2009. On Labor Shares in Recent Decades: A Survey.http://www.unicatt.it/Istituti/TeoriaEconomica/Quaderni/itemq0957.pdf 　　[3] Department of National Economic Accounting in National Bureau of Statistics of China. 2008. China’s GDP Accounting Method in Chinese Non-Economic Census Year. Beijing: China Statistics Press.
　　[4] Department of National Economic Accounting in National Bureau of Statistics of China. 2007. China’s Compilation Method for Flow of Funds Table in Chinese Economic Census Year. Beijing: China Statistics Press.
　　[5] Gollin, D. 2002. “Getting Income Shares Right.” Journal of Political Economy 110(2):458–74.
　　[6] Gomme, P., and P. Rupert. 2004. Measuring Labor’s Share of Income. https://www.省略/Research/PolicyDis/No7Nov04.pdf.
　　[7] Johnson, D.G. 1954. “The Functional Distribution of Income in the United States: 1850-1952.” The Review of Economics and Statistics 36(2):175-182.
　　[8] Krueger, Alan B.1999. “Measuring Labor’s Share.” American Economic Review 89(2):45-51.
　　[9] Li, Daokui, Linlin Liu, and Hongling Wang. 2009. “Changes in the Labor Share of GDP: A U-shaped Curve.” Economic Research Journal (1):70-82.
　　[10] Luo, Changyuan, and Jun Zhang. 2009. “Labor Income Shares (LIS) in Economic Development: An Empirical Study Based on China’s Sectoral-level Data.” Social Sciences in China (4):154 -178.
　　[11] Morel, L. 2009. A Sectoral Analysis of Labor’s Share of Income in Canada. http://economics.ca/2006/papers/0169.pdf.
　　[12] Xiao, Hongye, and Feng Hao. 2009. “Comparative Research on the Structure of Primary Distribution of Income among China and Foreign Countries.” Finance & Trade Economics (2):13-21.
　　[13] Xu, Xianchun. 2001. “Some Differences between China’s GDP Accounting and GDP Accounting of SNA in 1993.” Economic Research Journal, (11):63-68.
　　[14] Zhang,Juwei, and Shibin Zhang.2010. “Changes in Primary Income Distribution and Resulting Problems: A View of Labor Share in GDP.” Chinese Journal of Population Science (5):524-35.
　　[15] Zhang, Shangfeng, and Bing Xu.2010. “Labor Shares Estimation in the Initial Distribution.” Statistical Research (8):74-78.
　　[16] Zhou, Minghai, Wen Xiao, and Xianguo Yao. 2010. “China’s Unbalanced Economic Growth and National Income Distribution.” China Industrial Economics (6):35-45.

推荐访问:Labor Measuring Share China