by
Azizur Rahman Khan
Faculty of Economics
University of California, Riverside
U.S.A.
ILO East Asia Multidisciplinary Advisory Team (ILO/EASMAT)
ILO Regional Office for Asia and the Pacific
Bangkok
August 1996
________________________
* Unpublished document issued without formal editing by ILO. The responsibilities for opinions expressed in this document rest solely with the author and its inclusion here does not constitute an endorsement by the ILO of the opinions expressed therein.
Table of Contents
Poverty Income Thresholds
Poverty Indices
Method of Estimation
A Summary of Poverty Trends
The Location of the Poor
III. Poverty and Development Strategies: Growth and Distribution
IV. Growth in Personal Income and Asymmetrical Sectoral Growth
Growth in Personal Income
Growth of Rural Income
Rural Inequality
V. Export-Led Industrialization, Globalization and Poverty
The Transition from Agriculture-Led Development to Export-Led Industrialization
Growth in Employment During the Period of Globalization and Export-Led Development
Export-Led Development, Globalization and Regional Concentration
VII. The Access of the Poor to Education and Health
VIII. Specific Macroeconomic Policies and Poverty: A Summary
Macroeconomic Policies for High Savings and Investment
The Stop-Go Policy Concerning Overall Economy Activity
The Composition of Macroeconomic Instruments
Employment Policy
Trade, Exchange Rate and Export-Led Industrialization
Migration
Why China's Growth Since the Mid 1980s Has Not Alleviated Poverty?
The Focus of Official Poverty Alleviation Strategy
The Need for a Comprehensive Poverty Alleviation Strategy
References
Annex: Estimating Trends in Poverty Incidence in China
Notes
Notes to Annex
I am very grateful to Carl Riskin who helped me understand many details of China's current economic scene especially about the on-going poverty alleviation activities. I am also very grateful to Keith Griffin, who gave valuable comments on an earlier draft and to Martin Ravallion, who cleared up a number of issues concerning the computation of poverty estimates which were made with the help of a computer programme that he created in collaboration with Gaurav Datt at the World Bank. Zhang Ping rendered much help in getting access to some unpublished data. Piyasiri Wickramasekara of ILO/EASMAT coordinated the study. Mr Liu Siwu of ILO, Beijing made arrangements for the field trip to China. I received a great deal of assistance from many officials of the Government of the People's Republic of China and others in Beijing. In particular I would like to thank Mr. Ma Yongtang who organized the programme of my visit to Beijing in July 1996.
In the final years of the twentieth century the international community appears to have reached the consensus that growth in income and output by itself is not an adequate measure of economic progress; it must at least be accompanied by a reduction in poverty.(1) In this context poverty is measured by the number, or the magnitude of some characteristic, of persons who fall below some absolute standard of living (e.g., a threshold level of per capita income). A change in the incidence of poverty, thus defined, depends on (a) the rate of growth in income and (b) the change in the distribution of income. Formally stated, the incidence of poverty is uniquely determined by the average level of income (or some other indicator of living standard that is used to define poverty) and the Lorenz distribution of income (or the relevant indicator), given the poverty threshold. Thus a change in the incidence of poverty is completely determined by the change in average income and the change in the Lorenz distribution of income, once the poverty threshold is kept unchanged. With an unchanged distribution of income, a rise in average income reduces the incidence of poverty and the higher the rate of growth in income the greater is the reduction in poverty. With a given rate of growth, the greater the improvement (deterioration) in the distribution of income, the greater (smaller) is the reduction in poverty.(2) The positive effect of rapid growth on the incidence of poverty can be largely offset if growth is accompanied by a deterioration in the distribution of income. Similarly, the positive effect of a sharp improvement in the distribution of income on the incidence poverty can be weakened if increased egalitarianism is accompanied by a fall in the rate of growth.
Economic development does not represent a unique combination of growth and distributional change. Widely varying combinations of growth and distributional change can be generated by different development patterns and policies. Macroeconomic and sectoral policies are among the principal components of development policy that determine the combination of growth and distributional change. These policies do not only determine the rate of growth in income but also affect the distribution of income. Some examples help illustrate the point. In a poor society in which much of poverty is concentrated in the rural economy, a development strategy that is driven by growth in agriculture is likely to combine overall growth with a favourable change in the distribution of income. This should contribute to a rapid reduction in poverty. A rise in the volume of public expenditure targeted to enhance the productivity and entitlement of the poor, especially when it is combined with non-inflationary budgetary policies and an equitable system of raising public revenue, should have similar results. On the other hand, fiscal support to industrialization in a poor agrarian society, through the promotion of capital-intensive technology by means of generous credit to industrialists and the underpricing of capital equipment, is likely to combine growth with an adverse change in the distribution of income. Thus the impact on poverty is powerfully influenced by the nature of macroeconomic and sectoral policies that are included in the development strategy.
A rather narrow definition of macroeconomic policy is to consider only the direct determinants of aggregate economic activity, namely, aggregate public revenue and expenditure and aggregate volumes of credit and money supply. We shall go beyond these aggregates and include their composition and other related instruments that affect the levels of output, investment, consumption, savings and employment and their broad sectoral composition. For the purpose of the present study macroeconomic policy instruments will include: fiscal policy, taxation, monetary and credit policies and public expenditure policies; price policy and the policies concerning sectoral terms of trade; trade and exchange rate policies and policies towards foreign investment; employment policies; and policies concerning the system of ownership and organization of economic activities. Among sectoral policies we shall focus on public policy in social sectors, e.g., education and health, that represent inseparable dimensions of poverty as well as access to productive resources. Macroeconomic policies and sectoral policies are closely intertwined in such areas as the system of ownership, employment policies, sectoral distribution of growth and access to productive resources.
The incidence of poverty in a society is influenced by a wide variety of economic and social policies. This study will not be concerned with all of them. It will concentrate on macroeconomic and sectoral instruments and largely leave out those policies that have no more than a remote link with macroeconomic and sectoral changes. Thus it is widely documented that household size is significantly higher for the poor than for the non-poor. This indicates that policies that help the poor reduce their family size will help reduce the incidence of poverty. The present study will not consider these policies, or will consider them only marginally, due to their remote link with the kind of macroeconomic and sectoral strategies that are our primary concern.
The next section of the study is concerned with the estimation of poverty trends in rural and urban China during the period of reform. The third section explains the relationship between poverty trends and the broad strategies guiding China's economic development during different phases of reform. The fourth section is concerned with a more specific identification of macroeconomic policies which have helped or hindered poverty alleviation in rural China. The fifth section analyzes the effects of China's export-led development strategy on poverty. The sixth section summarizes the condition of women in China and looks for evidence of the effect of macroeconomic and sectoral policies on their welfare. The seventh section provides a brief analysis of recent changes in the social sectors that have an impact on the welfare of the poor. The eighth section summarizes the role of specific macroeconomic policy instruments on poverty. The concluding section briefly considers China's official poverty alleviation strategy, which is mainly focused on programmes in areas other than macroeconomic and sectoral policies, and provides a summary of the main findings of the study.
Poverty Income Thresholds
Our starting point is to use as the minimum per capita daily requirement 2,100 kilocalories for urban China and 2,150 kilocalories for rural China. Since there is no way to distinguish differences in nutritional levels among members of a household, all members of a household that achieves an average food energy consumption level that is at least as high as the figures specified above are considered non-poor. The higher requirement for rural China is in recognition of the fact that the proportion of household members who are in the labour force is higher in rural areas than in urban areas. Furthermore, it is likely that the average intensity of work is higher in rural China than in urban China.
For urban China the poverty income threshold (PIT) is determined for the year 1994 to be the disposable per capita household income at which an average consumer actually bought a basket of food that provided 2,100 kcalories per day. This was done by fitting an Engel function relating food energy to per capita income. For rural China data are not available to estimate the Engel function. It was assumed that the cost of food energy per kcalorie for the poor was 85 per cent of the cost of food energy per kcalorie for an average consumer. PIT was arrived at by assuming that food accounted for two-thirds of expenditure at poverty threshold as compared to the rural average of 59 per cent for 1994. The 1994 PITs were converted to current price PITs for previous years by using the respective consumer price indices (CPIs). These indices are from official sources for the period up to 1990. Between 1990 and 1994 China experienced a very sharp change in relative food prices. Prices of grain increased much faster than the prices of other food and food prices generally increased much faster than other prices. Since a household at poverty threshold spends a much higher proportion of expenditure on food, and a much higher proportion of food expenditure on grains, than does an average household, the use of official CPIs, which are based on average consumption weights, would significantly understate the rise in cost of living for the poor. We have therefore made adjustments in the CPIs for the households at poverty threshold for the period after 1990 to reflect the higher proportion of their consumption on food. Urban and rural PITs thus estimated are called the "standard" PITs. An additional, lower, PIT has been used in each of urban and rural China representing 80 per cent of the standard PIT. Detailed calculations of the PITs are shown in the Annex.
Although the PITs have been formally anchored to food energy in the benchmark year, there is no stipulation that the same level of food energy will be ensured at poverty threshold in other years. Indeed, food energy consumption at PIT will be significantly lower in years after 1990 than in the years up to 1990. The reason is a sharp upward rise in relative food prices after 1990. Between 1990 and 1994 the food energy consumption in urban China declined by 14 per cent both for the average consumer and for the bottom decile. For rural China per capita consumption of food energy between those years remained unchanged for the average consumer and almost certainly declined for the low-income households who experienced a lower than average rise in real income. This happened in spite of the fact that between 1990 and 1994 per capita real disposable income increased by 38 per cent for an average urban household and by 18 per cent for an average rural household. To require constant actual kcalories of food energy consumption between 1990 and 1994 one would have to settle for a PIT for 1994 which represents a much higher level of real income than the PIT for 1990 to offset the negative effect of relative price increase on food consumption. This does not seem appropriate. Instead our PITs over time represent the same constant real income in the sense that consumers could buy the benchmark-year bundle of goods in other years if they wanted to.
One might argue that the requirement that food energy consumption in 1994 is adequate ensures that in previous years, with lower relative price for food, the constant real income PIT will represent a higher level of food energy consumption than is stipulated for poverty threshold. The use of 1994 as the benchmark year is dictated by the consideration that, especially for urban China, information on food consumption is not available in necessary detail for prior years. It is indeed true that the constant real income standard PIT will mean a higher level of food energy consumption in previous years and this is one of the reasons that additional lower PITs have been used.
Our PITs, especially for urban China, are much higher than the PITs used by the World Bank in its poverty report of 1992 (see World Bank, 1992). As shown in the Annex, the World Bank PITs are abysmally low and simply can not provide for an adequate food energy consumption. Our standard urban PIT is about the same as the official Chinese PIT. Furthermore, what matters is the measurement of change in the incidence of poverty over time, not so much the level of absolute poverty at any particular time which is essentially an arbitrary concept. The absolute level of the PIT is not of critical importance for the comparison of the incidence of poverty over time.(3) In any case, the use of an additional PIT, substantially lower than the standard PIT, provides a safeguard against an overestimation of PIT from the standpoint of the adequacy of food energy consumption.
It may be noted that the anchoring of the PITs to the minimum food energy requirement results in an urban PIT which is more than twice as high as the rural PIT at actual prices. Before discussing the logic behind this, it is useful to recognize that the difference between the official urban PIT and the official rural PIT in China is substantially higher than the difference between the urban and rural PITs used in this study.(4) Although average kcalories of food consumption in urban China is lower than in rural China, per capita food expenditure in urban China in 1994 was 138 per cent higher than in rural China. Even for the poorest 5 per cent of the urban households, per capita food expenditure was 56 per cent higher than the per capita food expenditure of an average rural household (food expenditures for different rural income groups are not reported).(5) This is due to the difference in unit cost per kcalorie of food energy between two areas which in turn derives from both a difference in the composition of food and a difference in food prices. For ten major items of food consumption that supply approximately three-quarters of food energy, the unit cost is 42 per cent higher in urban China than in rural China at comparable prices.(6) Actual unit cost of a kcalorie in urban China in 1994 was about 150 per cent higher than the same in rural China implying that the difference in unit cost in kcalorie is greater for the sources of food energy other than the ten major items(7) and that unit food prices are higher in urban China than in rural China. Urban per capita income in 1994 was 2.6 times the rural per capita income at actual prices. We do not know what the relative levels of these PITs would be at comparable prices although it appears that they would still be significantly higher for urban China. This means that food energy threshold is reached in rural China at a lower level of real income than in urban China. This is a phenomenon that has been widely observed in other countries as well. It is perhaps due both to the difference in consumer preferences between two areas and to the commitment of a higher proportion of income to other uses in urban China than in rural China. It is also possible that there are additional constraints in urban China on the composition of food consumption, e.g., a higher proportion than in rural China that must be purchased in processed form. There is no stipulation that urban and rural PITs represent comparable real incomes. It is not obvious that one should use comparable real income PITs for populations who differ so much in terms of average standards of living as do the urban and the rural Chinese. Clearly, at the threshold real income level at which a rural resident achieves adequate food energy consumption, an urban resident would be seriously malnourished.
Poverty Indices
Three different indices of poverty have been used in this study. The first is the well known head count index which shows the proportion of population in poverty, the number of persons belonging to households which have per capita incomes lower than the PIT divided by total population.
The second measure, the poverty gap index, is estimated as follows: the gap between PIT and income as a proportion of PIT is summed for all the poor. This sum is then divided by total population. It is the product of the head count index and the average proportionate shortfall in income from PIT for the poor.
The third measure, the "weighted poverty gap" index, is estimated as follows: the gap between PIT and income as a proportion of PIT is squared and summed for all the poor. This sum is then divided by total population.
Each of the three indices provide useful information about the nature of change in the incidence of poverty. The head count index, the most widely used measure, hardly needs an interpretation. In spite of its generally useful and simple meaning, it suffers from obvious limitation. A change in the incidence of poverty may leave the head count ratio exactly the same as before while increasing or reducing the average income of the poor. Such a change, of great importance to the policy maker, is not captured by the head count index. The poverty gap index is capable of measuring changes of this kind.
It is possible for the number of poor and the average income of the poor to remain unchanged while the distribution of income among the poor improves (the extreme poor gain at the cost of the moderately poor) or deteriorates (the moderately poor gain at the cost of the extreme poor). Policy makers need to monitor this kind of change which is not captured by the head count index or the poverty gap index. The weighted poverty gap index reflects such a change accurately. By itself the weighted poverty gap index is inadequate because it is not subject to easy interpretation and the common-sense illumination that the other two indices provide. The algebraic derivation of the three indices and a fuller discussion of their meaning can be found in the Annex.
Method of Estimation
The procedure adopted for the estimation of various indices of poverty may be described as follows. Parametric Lorenz functions are fitted to the distributional data separately for rural and urban China.(8) Combined with the specification of average income and the PIT, this determines all three indices of poverty.
Distributional data showing class intervals of income and the percentage of population in each class interval are published by the SSB. They have been described in the Annex and shown in Annex Tables 1 and 2.
Table 1. Poverty Indices for Rural China
Standard Poverty Income Threshold
Year Gini Average PIT Poverty Indices
Ratio Income -----------------------
HC PG WPG
1980 0.248 191 197 59.83 18.96 8.13
1981 0.245 223 202 48.34 13.66 5.44
1983 0.258 310 211 26.41 6.32 2.16
1985 0.307 398 242 24.51 6.29 2.31
1988 0.319 545 339 26.95 7.53 3.02
1990 0.300 686 400 23.94 5.98 2.05
1992 0.315 784 432 22.90 5.91 2.09
1993 0.336 922 496 23.70 6.65 2.78
1994 0.338 1221 647 23.05 6.41 2.68
Low Poverty Income Threshold
Year PIT Poverty Indices
------------------------
HC PG WPG
1980 158 40.80 11.16 4.25
1981 162 30.34 7.30 2.64
1983 169 14.32 2.89 0.82
1985 194 14.04 3.13 0.98
1988 271 16.09 4.07 1.45
1990 320 13.87 2.77 0.75
1992 346 13.63 2.85 0.81
1993 397 14.11 3.61 1.47
1994 518 13.62 3.47 1.42
HC = Head Count Index
PG = Poverty Gap Index
WPG = Weighted Poverty Gap Index
Note: Average income and PIT are in current yuan.
Table 2. Poverty Indices for Urban China
Standard Poverty Income Threshold
Year Gini Average PIT Poverty Indices
Ratio Income --------------------
HC PG WPG
1981 0.184 458 415 44.36 9.05 2.78
1985 0.226 685 497 26.22 6.09 2.08
1989 0.228 1261 810 17.72 3.76 1.45
1990 0.233 1387 821 15.04 3.20 0.98
1991 0.236 1544 879 12.18 2.45 0.91
1994 0.294 3179 1448 11.99 2.45 0.69
Low Poverty Income Threshold
Year PIT Poverty Indices
----------------------
HC PG WPG
1981 332 20.08 3.30 0.94
1985 398 12.70 2.61 0.98
1989 648 7.42 1.67 0.77
1990 657 7.39 1.25 0.30
1991 703 4.73 1.04 0.48
1994 1158 5.90 0.85 0.17
HC = Head Count Index
PG = Poverty Gap Index
WPG = Weighted Poverty Gap Index
Note: Average income and PIT are in current yuan.
A Summary of Poverty Trends
Tables 1 and 2 respectively summarize poverty indicators estimated for rural China and urban China. These tables also report Gini ratios, average incomes and PITs.
The main trends for rural China may be summarized as follows:
(i) There was a rapid decline in the incidence of poverty according to all three indicators up to the mid 1980s. Reduction in poverty gap and weighted poverty gap were proportionately larger than reduction in head count index especially in the earlier years and the reduction in an indicator was proportionately larger for lower PITs than for higher PITs. These suggest that there was a greater than average reduction in more severe kind of poverty in the early years.
(ii) Since 1985 the reduction in poverty has been particularly slow. In the early 1990s the reduction in the proportion of rural population in poverty came to a halt. Given the rise in rural population, this means that the number of rural Chinese in poverty rose. Thus, for example, using the lower of the two sets of PITs, the head count of rural poor increased from 113 million in 1985 to 117 million in 1994.(9) The particularly poor performance in reducing the more abysmal kind of rural poverty is represented by the actual rise in poverty gap and weighted poverty gap between mid 1980s and 1994 for the lower PIT.
Trends for urban China may be summarized as follows:
(i) All three indices of poverty declined rapidly through the 1980s. This is the case for both the PITs.
(ii) In the 1990s the reduction in the incidence of poverty appreciably slowed down. This change is more clearly discernible in the case of the lower PIT, representing more extreme kind of poverty. Thus, according to the lower PIT, the proportion of urban population in poverty rose and the head count of urban poor rose by about six million between 1991 and 1994.(10) It is probable that the estimated indicators do not adequately reflect the serious deterioration in China's performance in reducing urban poverty in the 1990s due to two reasons. The first of these is related to the exclusion from China's income accounts of income transfers in the form of ration coupon subsidies and other subsidies.(11) Between 1990 and 1994 there was a sharp reduction in these subsidies in urban China. A proper system of income accounting would have revealed a slower rise in average income and, hence, a slower reduction in poverty. The other reason for an understatement in the incidence of urban poverty in the 1990s is the exclusion of the "floating migrants" from the survey data in China. It is widely believed that the extent of poverty among these migrants is higher than among the officially registered urban population. As a proportion of urban population floating migrants increased in the early 1990s.
To summarize: the rate of reduction in the incidence of poverty was very high in both rural and urban China in the early years of reform. In rural China there was a sharp decline in the rate of reduction in poverty from about 1985 and in the 1990s the reduction of poverty was halted. Indeed the number of rural population in poverty rose between mid 1980s and mid 1990s. In urban China the decline in the rate of reduction in poverty began later, around the turn of the decade. The reduction in more extreme kind of poverty came to a virtual halt as a proportion of urban population and was actually reversed in absolute numbers. These are rather unexpected and striking findings for an economy that has experienced a growth in per capita GDP that is historically almost unprecedented.
A word should be said on how our estimates of above trends compare with the trends estimated by others, notably the World Bank and the official Chinese sources. World Bank, 1992 estimates trends in poverty only up to the year 1990 and their conclusions about trends are generally consistent with our findings even though their estimates of levels of poverty are obviously different from ours because of their use of far lower PITs. In China official estimates of rural poverty are made by the SSB. For 1985 the SSB estimates the rural poor to consist of 125 million, a figure that falls between our two estimates and is 11 per cent higher than our low estimate.(12) The SSB estimates over time are consistent with our finding that the rate of reduction in poverty sharply slowed down after 1985. The SSB estimates however suggest that the absolute number of the rural poor continued to decline after 1985, albeit at a slower rate of about 8 million per year during 1985-90 and 4 million per year during 1990-96 as compared to an annual average reduction of poverty by 18 million per year between 1978 and 1985. The fact that the SSB has already come up with a number for 1996 suggests that they must be making more of a projection than a careful estimate for this year and it is possible that for other years as well their figures are less than rigorous estimates. And yet it is of some concern that their estimate of trend in the reduction of rural poverty in recent years is so different from ours because we use the same distributional data that must be the basis for the SSB estimates. The most important reason for the difference between the SSB estimates and our estimates appears to be that the PITs used by the SSB have been inadequately adjusted upwards in recent years for changes in cost of living. As shown in the Annex, for recent years the SSB adjustment of the nominal PIT has been smaller than even the rise in the average rural CPI. As we have argued, the adjustment in recent years should be at a significantly higher rate than the rise in the average rural CPI.(13)
The Location of the Poor
We have emphasized the difficulty in measuring rural and urban poverty by using a common standard of living. We have also noted in the Annex the asymmetry between rural and urban areas in the format in which distributional data are available. Due to these reasons we have avoided the aggregation of rural and urban poverty indices into national indices. And yet it should be abundantly clear that poverty in China is an overwhelmingly rural problem. There are three distinct aspects of this diagnosis of imbalance in locational concentration of poverty in China: first, most of the poor live in rural areas; second, the severity and intensity of poverty of those who are poor is greater in rural areas than in urban areas; and third, obstacles to the reduction of poverty have been more stubborn in rural China than in urban China.
Let us consider the estimated indices of poverty in rural and urban China for 1994 for the lower PITs. There were 117 million poor in rural China as compared to 20 million in urban China. But, as we have demonstrated beyond all reasonable doubt, the low urban PIT represents a higher standard of living for urban China than the low rural PIT does for rural China. It therefore appears that more than 85 per cent of China's poor live in the countryside.
Again, let us consider the different indices of poverty for urban and rural China for 1994 for low PIT. The rural head count index is 130 per cent higher than the urban head count index. But the rural poverty gap index is 308 per cent higher than the urban poverty gap index and the rural weighted poverty gap index is 735 per cent higher than the urban weighted poverty gap index. Thus the rural-urban difference in the depth and intensity of poverty are far greater than similar difference in the head count of poverty. In spite of all the changes in recent years, the social safety net seems to provide a more effective protection to the poor from falling far below the poverty threshold in urban China than in rural China.
It has also been noted that for nearly a decade the indices of poverty in rural China have hardly moved down at all. In urban China the slowdown in poverty reduction has occurred much more recently.
III. POVERTY AND DEVELOPMENT STRATEGIES: GROWTH AND DISTRIBUTION
What caused the halt in China's remarkable success in achieving a high rate of poverty-reducing growth in the early years of reform? Was the transition to growth without poverty reduction a consequence of a change in development strategy? This section and the next two sections try to answer this question.
The sharp reduction in rural poverty in the early years of reform was associated with a remarkably rapid increase in rural income and a modest rise in the inequality of income distribution. The slowdown and halt in poverty reduction since the mid 1980s was associated with a sharp reduction in the rate of growth in rural income and a continued increase in the inequality of income distribution. In the earlier years, the negative impact of increased inequality was dominated by the positive impact of growth. In the period since the mid 1980s growth was not high enough to offset the effect of continuing increase in inequality.
In urban China the sharp decline in the incidence of poverty during the 1980s was associated with a fairly high rate of growth in per capita income and a rather small increase in inequality of income distribution. Since the late 1980s the growth in per capita urban income actually accelerated. In spite of that the rate of reduction in poverty weakened substantially mainly because of a significantly sharper increase in inequality of income distribution.
The extent to which changes in income distribution and growth performance affected the incidence of poverty in China can be gauged by simulating the outcome of alternative scenarios combining different income distribution and growth outcomes. Table 3 shows the results of these simulations. All these results are shown for the standard PITs.
For rural China we ask what the different indices of poverty would have been in 1994 if: (a) the distribution of income had remained the same as in 1980; and (b) if per capita rural income had continued to grow after 1985 at just half the annual trend rate at which it grew between 1978 and 1985.(14) As discussed later, the trend rate of growth of rural income is estimated to have fallen from 13.1 per cent between 1978 and 1985 to only 2.4 per cent between 1985 and 1994. Our alternative scenario postulates an annual growth rate of 6.6 per cent since 1985.
Table 3. Poverty Counterfactual for 1994
Rural China
HC PG WPG
-------------------
Actual incidence of poverty 23.05 6.41 2.68
What the incidence of poverty
would have been if:
A. Income distribution remained
the same as in 1980 13.42 2.47 0.63
B. Growth since 1985 had been at
just half the pre-1985 rate 9.28 2.29 0.96
Urban China
Actual incidence of poverty 12.89 2.71 0.79
What the incidence of poverty
would have been if:
A. Income distribution remained
the same as in 1981 1.61 0.33 0.16
B. Growth had not accelerated
since 1989 17.05 3.97 1.29
Note: All the indicators are for standard PITs.
Had the distribution of income been the same as in 1980, with growth no different from what actually took place, the head count index of poverty would have been 42 per cent lower than what it actually was in 1994, the poverty gap index would have been 61 per cent lower and the weighted poverty gap index would have been 76 per cent lower. Thus the increased inequality not only greatly exacerbated the problem of poverty reduction in rural China, it made the task of reducing the more extreme and abysmal kind of poverty proportionately more difficult.
Had per capita rural income in China grown since 1985 at just half the trend rate at which it grew in the early years of reform, with inequality increasing as it actually did, the head count index of poverty would have been 60 per cent lower, and the poverty gap index and the weighted poverty gap index would each have been 64 per cent lower than what they actually were in 1994. Higher growth in income in rural China would thus have benefited the poor at different levels of intensity of poverty roughly equally.
Had urban China experienced no change in the distribution of income since 1981 with growth taking place at the rate at which it did, there would virtually be no problem of urban poverty! The head count index of poverty would have been 88 per cent lower than what it was and the poverty gap and weighted poverty gap indices would respectively have been 88 per cent and 80 per cent lower. It is worth noting that, unlike in the case of rural China, the worsening distribution of income in urban China did not disproportionately affect the groups that are more severely poor.
As discussed later, the trend rate of growth in per capita urban income accelerated after 1989, from 5.6 per cent between 1980 and 1989 to 8.1 per cent thereafter. Had there been no acceleration in growth rate since 1989 (i.e., had the rate of growth in per capita urban income continued to be 5.6 per cent), with the distribution of income changing the way it actually did, the head count index of poverty would have been 32 per cent higher than what it was; the poverty gap index would have been 46 per cent higher and the weighted poverty gap would have been 63 per cent higher. Thus higher growth not only protected the urban poor in general; it provided proportionately greater protection to the more severely poor.
A reduction in growth in income and a deterioration in the distribution of income are both detrimental to poverty reduction. In the case of rural China the reduction in growth was the major obstacle to poverty reduction in the period after the mid 1980s. In the case of urban China the increased inequality in the distribution of income was the main obstacle. We shall next try to analyze the causes, including macroeconomic and sectoral policies, responsible for these problems.
IV. GROWTH IN PERSONAL INCOME AND ASYMMETRICAL SECTORAL GROWTH
Growth in Personal Income
The positive relationship between growth and poverty reduction has been widely recognized in development literature. It is however a nearly universal presumption that growth in this context is well represented by growth in per capita GNP. With the annual trend rate of per capita GNP growth in China approaching a nearly unprecedented 8 per cent, it would appear puzzling to worry about inadequate growth as an obstacle to poverty reduction.
Part of the answer to the riddle lies in the fact that per capita disposable incomes of households (hereafter designated by the commonly used term "personal income"), rather than per capita GNP, is the determinant of poverty in the model that underlies the identification of the poor and the estimation of the incidence of poverty.(15) In the case of China the annual growth rates of per capita rural and urban personal incomes are each insignificantly correlated with annual growth rates in per capita GDP/GNP.(16)
Not only are year-to-year growth rates in personal incomes and GDP uncorrelated, the trend rates of growth in personal incomes have been lower than the trend rate of growth in GDP/GNP. There was a clear break in the growth rate in per capita rural personal income in 1985 and the trend rates of growth have been estimated as follows:(17)
Between 1978 and 1985 13.1 per cent per year
Between 1985 and 1994 2.4 per cent per year
There was a similar break in the growth rate in per capita urban personal income in 1989 and the trend rates of growth have been estimated as follows:(18)
Between 1980 and 1989 5.6 per cent per year
Between 1989 and 1994 8.1 per cent per year
It appears that in the early years of reform, i.e., until about the mid 1980s, the growth in personal income was as high as, or higher than, growth in GDP. Thereafter growth in personal income has lagged significantly behind growth in GDP.
China's national accounts do not provide details that are necessary to fully explain the difference between growth in personal income and growth in GDP. But a good part of the difference is perhaps explained by the policy with respect to an important macroeconomic variable, namely the saving rate (i.e., domestic savings as a proportion of GDP).
Changes in the saving rate are shown in Table 4. In the early years of reform the rate of saving declined significantly, allowing the rate of growth of personal income to exceed the rate of growth of GDP, especially in rural China. It was only in 1985 that the saving rate caught up with, and exceeded, the previous peak. Thereafter it continued to grow, exceeding a staggering 40 per cent of GDP in 1993.
Table 4. Domestic Saving Rate
1978 34.1 1986 36.3
1979 31.7 1987 38.0
1980 30.0 1988 37.5
1981 27.9 1989 36.5
1982 31.6 1990 37.3
1983 31.3 1991 37.0
1984 32.0 1992 37.0
1985 34.7 1993 40.2
Source: World Bank, 1995.
There is a view that it was perhaps less urgent for China to push for a further rise in its already high rate of saving and instead to concentrate on a more efficient use of its investible resources. Had China followed such a strategy, the growth in personal income would have been higher with a greater reduction in the incidence of poverty.
The Growth of Rural Income
Perhaps the most remarkable aspect of the evolution of income distribution in China is the drastic decline in the growth of rural incomes since the mid 1980s despite an undiminished growth of GDP and an acceleration in the growth of urban income later at the turn of the decade. Estimates based on a detailed survey for 1988 show that the Gini ratio for income distribution is higher for China as a whole than either for rural China or for urban China.(19) This is because the inequality between rural and urban China dominates intra-rural and intra-urban inequality. Although we do not have precise estimates, the rise in the Gini ratio in personal income distribution must have been greater for China as a whole than for either rural or urban China. In this sense the problem of inadequate rural growth has been the most serious source of increased inequality in China.
Table 5. Indicators of Growth and Related Variables
(Incomes & GDP in yuan at 1990 Prices)
Year Per Capita Income GDP Per Agricultural Terms of
----------------- Capita Output Index Trade Index
Rural Urban
1978 271.5 - - 100.0 100
1979 - - - 107.5 122
1980 365.1 890.5 813 109.0 130
1981 418.4 907.0 836 115.3 136
1982 497.4 960.2 889 128.3 137
1983 563.2 999.9 967 138.3 141
1984 627.8 1125.1 1091 155.4 143
1985 656.1 1132.8 1216 160.6 150
1986 659.0 1279.6 1299 166.1 155
1987 677.2 1301.1 1421 175.7 165
1988 679.5 1318.4 1555 182.6 177
1989 629.2 1277.3 1596 188.2 171
1990 686.3 1387.3 1636 202.5 159
1991 692.6 1469.4 1749 210.0 159
1992 741.4 1600.4 1976 223.4 152
1993 757.3 1763.4 2217 240.9 154
1994 812.9 1918.6 2451 261.6 184
Note: Rural per capita income is "per capita net income" reported in Statistical Yearbooks of different years deflated by the rural CPI. Urban per capita income is "per capita income available for living" reported in Statistical Yearbooks deflated by the urban CPI. Per capita GDP series is based on the data reported in World Bank, World Tables, 1995 converted to 1990 prices. Agricultural output index measures changes in gross output value of agriculture and is from Statistical Yearbooks. Terms of trade index refers to agriculture's terms of trade and is measured as the ratio of "overall farm and sideline products purchasing price index" to "overall industrial products rural retail price index"both reported in Statistical Yearbooks.
Why did growth in income for rural China fall so suddenly and drastically since the mid 1980s? Table 5 shows some of the indicators that we use in explaining the phenomenon. The growth in gross output value of agriculture at constant prices (agricultural output) itself slowed down after 1985 and remained low before picking up again in 1992.(20) More dramatic was the trend in agriculture's terms of trade which improved rapidly in the early years of reform but showed no trend between the mid 1980s and 1993. It registered a fairly sharp rise in 1994.
Together per capita agricultural output and terms of trade for agriculture account for 90 per cent of the variation in per capita rural personal income between 1978 and 1994.(21) The elasticity of personal income is 0.69 with respect to agricultural output and 0.81 with respect to terms of trade.
The decline in agricultural growth is partly explained by the failure of the terms of trade to continue to improve. In addition there appears to have been an inadequacy of resources for investment. The latter is hard to quantify with precision due to the inadequacy of information. It is however known that government expenditure on agriculture as a proportion of total expenditure fell from 13.6 per cent in 1978 to 8.3 per cent in 1985 and stayed there in 1993 (SSB, 1995).
The initial sharp rise in agriculture's terms of trade, together with the systemic reform of agriculture, was instrumental in unleashing rapid rural growth in China. This could not be sustained in years after 1985 due mainly to macroeconomic considerations. Rising procurement prices were contributing to public sector deficit, given the preference for a continuously rising rate of investment in non-agricultural sectors.
The reduction of the share of agriculture in public expenditure and the halt to an improvement in agriculture's terms of trade signalled a basic shift in China's development strategy. Until the middle of the 1980s China was following an agriculture-led strategy of development. Since the middle of the 1980s its development strategy has distinctly been oriented towards export-led industrialization. The consequent decline in emphasis on the development of the rural sector led to a rate of growth in rural personal income that was too inadequate to offset the adverse effect of continued increase in inequality on the incidence of rural poverty.
Rural Inequality
Even with the reduction in the rate of growth of rural income it would have been possible to reduce poverty if the distribution of income had remained unchanged. But the distribution of income continued to deteriorate through the mid 1990s. Indeed, the rise in the Gini ratio was modest in the early years of reform. It became sharper after 1983. What caused the increased inequality in rural China to continue unabated and to accelerate over time?
Some increase in inequality was warranted by systemic reform as the artificial equality created by the inefficient system of incentives under collectivized agriculture was dismantled. But this need not have been more than an once-for-all increase as long as the egalitarianism of land distribution in the post-reform period was maintained and matched by an equality of access to complementary resources. There were two other changes that exacerbated the effects of systemic change on the distribution of rural income. Results of the detailed survey for 1988 shows that income from farm production has an equalizing effect on the distribution of rural income while income from most other sources - notably wage income, income from non-farm production activities and income from property - have a disequalizing effect.(22) After the mid 1980s the share of rural income derived from farm production appears to have declined due to a decline in the growth of agricultural production and agriculture's terms of trade and a relatively rapid growth of rural non-farm activities. The other probable source of inequality was the decline in the proportion of public expenditure directed to agriculture, resulting in the weakening of a major instrument of improving egalitarian access to productive resources. By 1988 rural taxes on households exceeded rural subsidies for households and net taxes on rural households were highly regressive, i.e., they had a disequalizing effect on the distribution of rural income.(23)
V. EXPORT-LED INDUSTRIALIZATION, GLOBALIZATION AND POVERTY
The Transition from Agriculture-Led Development
to Export-Led Industrialization
China's development strategy in the reform period, which was initially led by agricultural growth and rural reforms, gradually changed into a strategy that came to be led by the growth of manufactured exports in a globalizing world economy during the 1980s. Table 6 shows some of the indicators of transition from one strategy to the other.
Changes in direction in different indicators were not exactly synchronized. But it is clear that a basic change in the pattern of development took place in the mid 1980s. The most outstanding indicator of this is the change in the ratio of urban per capita income to rural per capita income which fell steadily after the beginning of reforms, bottomed out during 1983-85, and rose steadily thereafter. By 1994 the disparity between urban and rural income had substantially exceeded the level that obtained at the beginning of reforms.
The rise in export propensity, begun earlier, had a decisive acceleration after 1986. More significantly, the share of manufactured exports began to rise sharply after 1985. There was also a gradual rise in the volume of foreign direct investment (FDI) which later, in the 1990s, reached explosive proportions.
Table 6. Indicators of Transition from Agriculture-Led
Growth to Export-Led growth
Year Urban/Rural Export as Manufacturing FDI ($ bn)
Income Ratio % of GDP Export as % of
Exports
1978 2.37 6.5 - -
1979 - 8.7 - )
1980 2.29 10.1 49.7 )
1981 2.05 12.9 53.4 ) 0.36
1982 1.83 12.6 55.0 )
1983 1.70 11.3 56.7 )
1984 1.71 11.7 54.3 1.26
1985 1.72 10.5 49.4 1.66
1986 1.95 12.4 63.6 1.87
1987 1.98 17.3 66.5 2.31
1988 2.05 18.1 69.7 3.19
1989 2.10 17.6 71.3 3.39
1990 2.02 19.1 74.4 3.49
1991 2.18 20.9 77.5 4.37
1992 2.33 22.5 80.0 11.01
1993 2.54 24.0 81.8 27.52
1994 2.60 - 83.7 33.77
Note: Column 1 shows the ratio of urban per capita personal income to rural per capita personal income (source: SSB, 1995). Column 2 shows exports of goods and non-factor services as per cent of GDP (source: World Bank, World Tables, 1995). Column 3 is the share of manufactured exports in merchandize exports and is from SSB, 1995. The last column, dollar value of foreign direct investment, is also from SSB, 1995.
Together these indicators represent China's transition to an export-led industrialization strategy of development in a rapidly globalizing world economy. The focus of public investment and fiscal incentives were deliberately shifted in favour of the coastal growth poles in which export industries, especially industries financed by FDI, were concentrated. We have already analyzed that this shift constituted the principal explanation of the halt to poverty reduction in rural China. In this section we shall analyze the consequences of this strategy on the incidence of urban poverty.
Growth in Employment During the Period
of Globalization and Export-Led Development
The principal defence of the strategy of export-led growth for an LDC rests on the argument that a move towards free trade, which is a precondition for the success of this strategy, promotes activities and techniques that use labour, the abundant factor, intensively. The result is a rapid growth in employment. High output-elasticity of employment makes growth egalitarian. China's export-led growth during the last decade of increasing globalization does not conform to this expectation. Indeed growth in China during the last decade was anything but employment friendly. The low output-elasticity of employment was a principal cause of the rise in inequality and the decline in the rate of reduction in poverty.
During this decade the proportion of total workers employed fell by less than 10 percentage points (from 64.0 per cent to 54.3 per cent) in agriculture, increased by a mere 2.7 percentage points (from 20 per cent to 22.7 per cent) in industries and rose by 7 percentage points (from 16 per cent to 23 per cent) in services.(24) Output-elasticity of employment was 0.31 for agriculture, 0.27 for industries and 0.58 for services (Table 7). The reduction is agriculture's share of employment was modest by the standard of the other rapidly-growing East and South-East Asian (ESEA) countries. At the present level of China's development, agriculture is expected to begin shedding labour in absolute terms and this has started happening in recent years, a fact that is hidden by the estimated output-elasticity of employment for the decade as a whole. The growth in employment in services has also been reasonably rapid. It is the industrial sector that had an apparently dismal employment performance. Output-elasticity of employment for the sector was only 60 per cent of what it was for Korea in the same period, an economy with a much higher level of real wages and technology.(25) This low absorption of labour characterized the Chinese industries in a period of extremely rapid industrial growth and growth of manufactured exports which, according to the findings of a World Bank study, were labour intensive.(26)
Table 7. Growth of Output and Employment
1978-84 1984-94
Growth Rates (per cent per year)
GDP 9.25 9.00
Agricultural GDP 7.23 3.90
Industrial GDP 8.91 11.63
Services GDP 11.88 9.30
Employment 3.09 2.42
Agricultural Employment 1.45 1.20
Industrial Employment 5.52 3.12
Services Employment 7.97 5.41
Elasticities
Employment with Respect to GDP 0.33 0.27
Agricultural Employment with
Respect to Agricultural GDP 0.20 0.31
Industrial Employment with Respect
to Industrial GDP 0.62 0.27
Services Employment with Respect
to Services GDP 0.67 0.58
Note: For 1984-94 all growth rates are estimated by fitting semi-logarithmic regressions and all elasticities are estimated by fitting double-logarithmic regressions. All these estimated growth rates and elasticities are significant at 1 per cent level. For 1978-84 only three of the annual observations are available. Hence all the estimates of growth rates are annual compound rates of change between 1978 and 1984 and the elasticities are ratios of growth rates in employment to growth rates in output. All data are from SSB, 1995.
How might one explain the paradox that industrialization based on the expansion of labour-intensive exports during the era of globalization resulted in such a slow growth in employment? The explanation seems to lie in the initial condition of Chinese industries. These industries, almost exclusively state and collectively owned until the beginning of globalization, had for years employed labour far in excess of requirement. This is because of the past "iron-rice-bowl" principle of guaranteed employment to all members of the labour force. Even after the beginning of reforms, state and collective enterprises continued to avoid shedding surplus labour. In effect this meant a concealed system of unemployment insurance in a society which lacked formal institutions for the protection of the unemployed and officially adhered to a doctrine that did not admit the possibility of unemployment.
Table 8. Employment in Urban State, Collective and Other
Industrial Enterprises and Township and Village Enterprises
(Million Workers)
Urban Industries Township and Village ---------------------------------- Enterprises (TVEs)
State Collective Other Total
1990 43.64 18.76 1.37 63.78 92.65
1991 44.72 18.98 1.82 65.51 96.09
1992 45.21 18.62 2.38 66.21 106.25
1993 44.98 17.00 4.28 66.26 123.45
1994 43.69 16.04 6.07 65.80 120.17
Source: China SSB, 1995, pp. 85, 376.
Note: Urban industries include manufacturing; excavation; and the production and distribution of electric power, gas and water. Manufacturing accounted for 83 per cent of urban industrial employment. The ownership category of other urban industries includes private enterprises, individual enterprises, joint ownership enterprises, share-holding enterprises and foreign-owned enterprises.
Economic reform and globalization made it impossible for the state and collective enterprises to continue to implement this concealed system of social protection. Reforms brought into existence industrial enterprises under private ownership and a variety of other forms of ownership. These enterprises were not subject to strict restrictions on hiring and dismissing workers that applied to state and collective enterprises. Compared to state and collective enterprises, the enterprises under private and other forms of ownership thus had a great competitive advantage. Furthermore, China's integration in the globalizing world economy concentrated the attention of the enterprises on keeping down the unit labour cost which was the critical determinant of external competitiveness and the principal instrument of attracting FDI.
Thus it became imperative to find a way to liberate the state and collective enterprises from the burden of carrying the system of concealed unemployment insurance. A 1986 regulation seriously dented the "iron-rice-bowl" system of life-time guaranteed employment by requiring that all new employment in state enterprises be based on fixed-term contracts for three to five years without an obligation of renewal on the part of the enterprises. State enterprises were simultaneously granted power to dismiss workers for inefficiency. In 1988 State Bankruptcy Law was introduced to make it possible to liquidate or drastically restructure state enterprises with consequent reduction in employment. Armed with these changes in the legal and institutional framework, state and collective enterprises have started shedding some of their surplus labour (Table 8). The decline in employment was sharpest in collective enterprises, 15.5 per cent between the peak year of 1991 and 1994. In state enterprises the decline was much more modest, 3.4 per cent between the peak year of 1992 and 1994. In private and other enterprises employment grew rapidly, but not by as much as would be necessary to offset the fall in employment in state and collective enterprises. During the 1990s urban industrial employment increased very little and actually declined absolutely after 1993 in spite of an explosively rapid growth of output.
Indeed the process is likely to accelerate in the future. So far state enterprises have used, or have been allowed to use, the opportunity to reduce employment to a far lesser extent than is warranted by economic efficiency. It is common for the redundant workers to be kept on the payroll and temporarily sent home for a fraction, approximately a half, of the wage that is paid the working employees. The urban unemployment rate, which had declined to 1.9 per cent by 1984, went up to 2.8 per cent in 1994. But this estimate, excluding all but the officially unemployed among those holding residence permits (hukou), is just the tip of the iceberg. At least two other groups should be included in this category: those who have been sent home on partial pay (concealed "unemployment benefit") and those among the estimated 72 million floating migrants from rural areas who remain unemployed.(27) Rudimentary estimates, based on local data on the incidence of these categories, suggest that the rate of urban unemployment, adjusted for these two factors, might be well above 10 er cent.(28)
It is possible to argue that export-led industrialization and globalization should not be blamed for the problem of China's urban unemployment. Industrial employment, measured at constant intensity of employment per worker, has almost certainly increased quite rapidly. But the head count rate of employment does not reflect this due to the reduction in concealed unemployment in state and collective enterprises. The validity of the argument is undeniable. It is however important to recognize that the dismantling of the concealed system of unemployment insurance needed to be offset by the institution of alternative systems of social protection, either in the form of comprehensive and formal unemployment insurance schemes or in the form of transitional public works programmes, if the social cost of the employment consequence of globalization were to be avoided.(29) Globalization forced a pace of adjustment which has proved much faster than the capacity of the Chinese government and society to set up these alternative institutions and programmes. The increased unemployed members of labour force, including those who have been sent home from state enterprises on partial pay and the unemployed and low-paid among the migrants, are the principal categories of the new urban poor.
Export-Led Development, Globalization
and Regional Concentration
Export-led development strategy and globalization have been associated with the aggravation of the problem of poverty in China in other ways as well. During the entire history of pre-reform development China had little inflow of external resources. By 1993 China had become by far the largest recipient of external resources among all LDCs. It accounted for a fifth of all net resource flows into LDCs: a tenth of all official development assistance, nearly a fifth of all borrowing from private sources, five per cent of portfolio equity and, most remarkably, two-fifths of all FDI.(30) FDI continued to increase and reached $38 billion in 1995, 42 per cent of the total for all LDCs and about 4.5 per cent of China's GDP.(31)
Investment under FDI is almost exclusively attracted to the urban centres of the twelve rich provinces in the east - the eleven coastal provinces and Beijing - which have the best infrastructure and easiest access to export and lucrative domestic markets. The eleven coastal provinces, inhabited by 40 per cent of China's population, account for 80 per cent of China's exports. Eight of these twelve rich eastern provinces are ranked as the eight top provinces in terms of both per capita GDP and manufacturing wage. These provinces also achieved the highest growth in manufacturing wage rate and in per capita GDP in recent years. Most of the remaining four provinces in this category also achieved faster rates of growth in per capita GDP than did China as a whole. According to a recent official report the rate of growth of GDP in these twelve eastern provinces was 78 per cent higher than the rate of growth of the central and western provinces during 1991-95 (China Daily, March 22, 1996, p.4). The poorest 5 per cent of China's 2,148 counties averaged no income growth at all during 1985-91 according to Howes and Hussain, 1994.
Distributional data for the provinces are not published. It is therefore not possible to estimate the indices of poverty for the provinces in the way that we have done for rural and urban China. The World Bank, which obtained access to provincial rural distributional data for 1989 and rural distributional data for Guangdong, Guangxi, Guizhou and Yunnan for the period 1985-90, made estimates that are of considerable interest. (It should be noted that the World Bank used a much lower PIT than we did and therefore found that the incidence of poverty for rural China was lower than our estimates in Table 1.) The World Bank estimates show that in 1989 the head count indices of the incidence of poverty in the twelve coastal and eastern provinces were much lower than the average for rural China:(32)
Province Per cent of rural
households in poverty
in 1989
National Average 11.3
Shanghai 0.0
Guangdong 1.1
Beijing 0.1
Zhejiang 2.3
Tianjin 0.4
Fujian 2.1
Hainan 3.8
Jiangsu 3.6
Guangxi 18.1
Liaoning 7.1
Hebei 11.7
Shandong 5.9
With the exception of Guangxi and Hebei, the incidence of rural poverty was far lower than national average in the twelve coastal and eastern provinces in which export-led industrialization was concentrated. It is likely that they had even lower an incidence of urban poverty relative to national average.
An on-going World Bank research project has also found that the incidence of rural poverty in rich Guangdong declined over 1985-90 while it increased in poor Yunnan and Guizhou. It also increased in Guangxi, the only coastal province with a higher than national incidence of rural poverty.(33) It appears that poverty alleviation record in the richer provinces has been more favourable (less unfavourable) than elsewhere in China because of the far higher growth that they experienced.
The failure to reduce poverty during a period of unprecedented growth was in large part due to the concentration of benefits of growth in the coastal areas. Globalization, with its emphasis on FDI, provided a major impetus for this concentration. Public investment was not directed to offset the effect of concentration due to globalization. Had it done so, by shifting investment to improve infrastructure in backward provinces to enhance their attractiveness to investors and to strengthen their linkage with the coastal growth poles, China might have succeeded in achieving a significant reduction in poverty during the period of globalization. Clearly the policy makers felt that dispersed public investment for the decentralization of infrastructural development would make China less attractive to industries under FDI.
In estimating the indices of poverty it was assumed that all members of a given household are simultaneously poor or non-poor. This assumption is made in this study, as in most quantification of poverty of this kind, because the distributional data do not report intra-household inequality in the distribution of income, consumption or other characteristics. This of course does not mean that there is no difference in these characteristics between male and female members of households; it only means that the present state of statistical information does not allow us to estimate the common indices of poverty separately for male and female members.
Some information about the relative well-being of women as members of labour force is however available from a detailed survey for 1988 (Table 9).(34) In rural China women represented about 32 per cent of wage employment. On the average they earned about a fifth less than the male workers. This lower payment derived from a combination of lower remuneration in the same kind of employment and the concentration of women in lower-paid employment although there are exceptions to both. In urban China the difference in wage and bonus payment is less; but urban women receive only 74 per cent of retirement income and only 56 per cent of income from private enterprise that men receive. While both urban and rural women work significantly shorter hours in wage employment, they spend longer hours working on "economic work", including work on cooking, washing and other domestic activities.(35) Widely prevalent social problem of violence against women, especially in rural China, and the low representation of women in political and administrative leadership further contribute to the overall gender inequality in China.
Very little is known about how the condition of women has changed in recent years. Women clearly shared the benefits of economic reforms although there is little solid evidence to determine if they benefitted proportionately more or less than men did. Nor is it possible to document if the recent decline in the rate of reduction of poverty has more or less adversely affected the women. Several points however can be made with varying degrees of confidence.
Table 9. Gender Difference in Earnings and Work
A. Rural China: Annual Earning (Yuan) Female Employment as
Male Female % of Total Employment
Agriculture 1325 1006 22.5
Industry 1695 1263 44.4
Handicraft 1601 1751 52.8
Trade 1842 2172 31.0
State (Centre/
Province) 1998 1322 23.2
State (Local) 1790 1991 14.8
Cooperative 1790 1319 38.7
Private 1677 1190 27.4
Household Responsi-
bility Farms 1226 1132 25.4
All Wage Employment 1660 1334 31.9
B. Urban Income Per Person
(Yuan per month)
Income Category Male Female
Regular Wage 88.7 73.8
Bonus 30.1 26.0
Subsidies Received 26.1 23.0
Other Cash Income
from Work Unit 12.9 11.0
Retirement Income 116.1 85.4
Private Enterprise
Income 352.7 196.2
C. Hours of Work Per Day
National Rural
Male Female Male Female
Work 7.28 5.97 7.21 5.74
Travel to Work 0.62 0.47 0.62 0.44
Shopping 0.36 0.40 0.36 0.34
Cooking 0.50 1.84 0.46 1.93
Washing 0.18 0.83 0.16 0.85
Other Domestic 1.16 1.94 1.25 2.05
Total "Economic"
Work 10.10 11.45 10.06 11.35
Source: A: Khan, 1993; B: Khan et al., 1992; C: Tao and Jiang, 1993.
As already noted, over the last decade a sizable proportion of rural workers - an estimated 86 million - have migrated outside, 72 million to urban areas and another 14 million to rural areas in other counties or provinces. Much of this migration consisted of male workers who left their families behind. This has undoubtedly increased the proportion of rural households headed by women and has added to the burden of women in these households. Whether this increased economic responsibility has also meant an increase in economic power of women is impossible to know.
In recent years households have come to share a much enlarged burden of education and health expenditure due to reduced government allocation and increased cost recovery (see next section). This has meant an increase in the relative deprivation of female members from access to education and health services. Thus in 1991-92 girls represented about 70 per cent of the primary school age children (between 7 and 11 years) who were not enrolled. At the same time the primary school completion rate was significantly lower for girls (89 per cent) than for boys (94 per cent).(36)
The estimated life expectancy in the early 1990s was 70 for women and 67 for men. The gender difference in life expectancy in China is significantly lower than what it is in countries at comparable levels of life expectancy.(37)
It is not possible to document conclusively whether women were more or less adversely affected by the reduced output-elasticity of employment in Chinese industries during the last decade. Wage data are not separately published for men and women. Data on urban employment are available for men and women separately and they show that between 1984 and 1994 there was a slight increase in the share of women, from 36.4 per cent to 38 per cent (SSB, 1991 and 1995). The trend in women's share in urban manufacturing employment was also the same. This is apparently in contradiction with the commonly held belief that women are far more likely to be fired when the need to reduce employment arises. It is possible, though far from documented, that many of the rapidly growing export industries are intensive in the use of skills in which female workers specialize. A conclusive judgement on whether women have relatively gained or lost out in the period of "employment-hostile" industrialization would require a good deal of additional information on the earnings and composition of female employment.
VII. THE ACCESS OF THE POOR TO EDUCATION AND HEALTH
The method of estimating the incidence of poverty in this study, as in most other cases, determines the poverty status of individuals on the basis of the real per capita incomes of the households to which they belong. There are of course other characteristics of the poor that affect both their current well-being and their ability to pull themselves out of poverty in the future. Access to education and health are the principal elements of these other characteristics. Macroeconomic and sectoral policies - e.g., the level and composition of public expenditure, the extent and method of cost recovery and the system of delivery - are among the critical determinants of the entitlement of the poor to these services.
In China a sea change has occurred in the manner of producing and delivering these services. There are reasons to believe that many of these changes have had unfavourable effects on the capabilities of poor households. China's remarkable progress in the provision of basic education and health services in the decades before the launching of reforms is well documented. Among the principal features of China's strategy of providing education and health services were the emphasis on basic education and preventive health care and the financing of the cost of the services mainly through the resources of the collective economy with supplement from the state budget. With the abolition of the collective economy in rural China, the responsibility of providing these services was largely delegated to the newly-created local government institutions which have suffered from a perennial shortage of resources. In urban China the ability of the state and collective enterprises to fund these services also came under severe pressure with the progress of reforms as competition from enterprises under alternative ownership systems, not burdened with these responsibilities, put state and collective enterprises at a serious disadvantage. State budget, under severe pressure due to the erosion of the power of the government to generate surplus by the direct regulation of prices, procurement and distribution, was simply not able to make up for the reduction of resources from other sources. An increasing share of the costs of these services came to be shifted to the households.
Information on this subject is limited but sufficient to indicate the broad trend.(38) Between 1980 and 1989 the shares of government, insurance agencies and the users in the cost of funding health services changed as follows (per cent of total):
1980 1989
Government 31 21
Insurance 46 42
User 23 37
With the sharp rise in the share of the individual in the cost of health services, it is highly likely that the access of the poor to health service became relatively limited. China continued to make gains in health indicators. Life expectancy continued to rise, albeit at a much slower rate than in the past, which may simply be due to the increasing difficulty of achieving further progress when a society reaches a relatively high level of life expectancy as China has. What is not documented is how changes in life expectancy has been distributed among different income groups.
Evidence is a little clearer in the case of certain other indicators of health. Maternal mortality rate has also continued to decline, from 50 per 100,000 live births to 39 in urban areas and from 115 to 85 in rural areas between 1989 and 1993. But the average rate over the same period (1989-93) has varied widely between richer and poorer provinces as follows:
Range of maternal Provinces in the range
mortality rate
28 to 49 Beijing, Tianjin, Shanghai, Zhejiang
Guangdong, Heilongjiang, Liaoning,
Jiangsu
100 to 200 Hubei, Qinghai, Ningxia, Shanxi, Henan, Sichuan, Yunnan, Gansu
200 to 344 Xinjiang, Guizhou, Tibet
Seven of the eight provinces with the lowest maternal mortality rates are among the richest eastern/coastal provinces (Heilongjiang is the exception) while the ones with higher maternal mortality rates are among the poorer provinces.
In a remarkably candid statement made at a national conference held in July 1996 the Minister of Health, Chen Minzhang, identified the individual burden of health service cost as the cause of poverty for half the rural poor. The report of his statement reads as follows ("Rural Medical System Revived", China Daily, July 16, 1996, p.2):
The rural co-operative medical insurance system experienced a boom in rural areas in the late 1960s but was largely abandoned in the 1980s...
"We can no longer let farmers be responsible for dealing with serious diseases themselves; their incomes are too low", said Chen.
Nearly 90 per cent of farmers have to pay all of their medical costs. About half of the rural poor are reduced to poverty by disease, according to the Ministry of Health.(39)
Trends in education appear to be similar to trends in health services. A 1993 examination of poor villages in Shaanxi showed that school fees and other educational costs came to between 5 and 25 per cent of the average per capita income of the poor households.(40) Shortage of fund has been identified by the UNICEF to be the leading cause of non-enrollment in schools. We have already noted that girls represent a disproportionately high ratio of the children who stay out of schools. Also the poorer regions have lagged behind the national average in terms of primary school enrollment. In 1992 in 121 of the 2,148 counties the enrollment rate was below 90 per cent as compared to the national average of 98 per cent. Most of these low-enrollment counties are located in five poor provinces: Sichuan (37), Qinghai (28), Guizhou (20), Gansu (16) and Yunnan (12).
Available evidence thus seems to point to the conclusion that in recent years China's poor have suffered from a reduced access to education and health relative to the access that population on the average had. It also seems highly likely that many among the poor have experienced an absolute reduction in access to these services. Changes in macroeconomic and sectoral policies have been the main reason for this increased relative and absolute deprivation of the poor.
VIII. SPECIFIC MACROECONOMIC POLICIES AND POVERTY: A SUMMARY
So far our discussion of the effects of macroeconomic policies on poverty has been intertwined with the effects of the shift in the overall development strategy. This seems to be the right approach to the analysis of the issue because it is hard to argue that a particular macroeconomic policy has a unique effect on income distribution and poverty. Its effect is conditioned by the initial circumstances and the overall strategy of development of which it is a component. This is the point that will be brought out in this section by specifically focusing on the role of individual macroeconomic policies.
Macroeconomic Policies for High Savings and Investment
One of the basic objectives of China's macroeconomic policies has been the promotion of a high rate of saving and investment. Domestic saving rate in China, estimated to be 44 per cent in the World Development Report of 1996 (World Bank, 1996a), is the highest of all countries of the world (with the exception of the unusual and minuscule Gabon with rich mineral resources). Even if allowance is made for the possible overstatement of the saving rate, due to the often-noted underestimation of GDP, China's saving rate would easily be about the highest in the world and completely out of line with the saving rates in other LDCs. This quest for a high and rising rate of savings has created a large wedge between the rate of growth in GDP and the rate of growth of personal income in China. Since the latter is the principal determinant of the rate of reduction in poverty, this aspect of macroeconomic policy has been detrimental to poverty alleviation.
To respond to the argument that a high rate of accumulation would help reduce poverty more quickly in the future, one needs to consider two issues. First, China might give greater attention to improving the efficiency of investment while allowing the drive to increase the rate of investment to slow down a little. Secondly, a country that has been growing at more than 10 per cent per year should perhaps allow its inter-temporal welfare valuation to be a little more tilted in favour of the current generation.
The Stop-Go Policy Concerning Overall Economic Activity
During the reform period China has been gradually learning the need for and the significance of managing the overall level of demand and economic activity by using macroeconomic policy instruments, e.g., fiscal, monetary and credit policies. During the period of central planning these policies had no independent role. The level of economic activity and prices were managed by physical control. China's learning process has so far succeeded in avoiding runaway inflation and stagnation. It can not however be said that macroeconomic management in China has achieved a harmonious balance between growth and stability. It appears that the policy has always veered towards as high a rate of growth as possible until the threat of high inflation became imminent. This was the case in 1988 and 1989 and again in 1994. China's response to the threat of high inflation has so far been orthodox and decisive. Fiscal, monetary and credit brakes are pushed down hard until price stability is restored by bringing the rate of growth down to a more sustainable level. It is however noteworthy that in more recent years China has managed to bring down the rate of inflation with a relatively small reduction in growth rate. In this sense the use of the term "stop" in the present context does not mean quite the same thing as in ordinary usage.
The exact effect of this stop-go policy on the poor is hard to determine. It however seems that the years of high inflation have by and large coincided with a reduction in the alleviation of poverty. In years of high inflation, adjustments in money incomes of the poor have obviously lagged behind increases in cost of living. The course of poverty reduction might have been steadier with more even macroeconomic management.
The main problem seems to consist of an inadequate central control over macroeconomic instruments. Provincial and local governments and agencies are often powerful enough to persuade local credit institutions to lend in excess of national guidelines. It is only when things go so completely out of control as to make runaway inflation a real threat that decisive action is taken. These are often acts of desperation and not the most efficient methods of stabilizing the level of economic activity.(41)
Why macroeconomic control is lost and how it is restored is perhaps more important than the stop-go nature of control itself. Thus it seems that the restoration of stability in the late 1980s hurt poverty alleviation in rural China especially because one of the main methods employed was to depress agriculture's terms of trade in 1989, 1990 and 1991.
The Composition of Macroeconomic Instruments
More than the level of economic activity, the shift in the composition of fiscal and credit instruments appears to affect the poor. For example, the shift of public expenditure, credit and incentives away from the rural economy in the middle of the 1980s was the major cause of the halt to China's earlier success in reducing rural poverty. Similarly, the abolition of ration coupon subsidies in urban China in the early 1990s must have hurt the poor. While overall urban subsidies had a disequalizing effect on urban income distribution, ration coupon subsidies had an equalizing effect.(42) This of course is not a sufficient justification for the retention of these subsidies which may have had detrimental effects on the efficiency of economic performance. But their abolition should have been accompanied by compensating action, e.g., a targeted income subsidy of adequate magnitude. This does not appear to have taken place and one of the main reasons for the omission may be the concern for macroeconomic stability, given the unrelenting drive for high accumulation. A further aspect of the composition of public expenditure and credit that hurt the cause of poverty alleviation in China is their concentration in the urban areas of coastal and eastern provinces in which export-led industrialization came to be localized.
Employment Policy
As we have seen, the low growth in the head-count rate of employment in the period of export-led industrialization was almost certainly associated with a high growth in employment at given intensity of work. The discrepancy between the two was due to a reduction in concealed unemployment in state and collective enterprises. In the past this concealed unemployment was a form of unemployment insurance and social security. In a globalizing world economy, with a strong emphasis on keeping the unit wage cost of production low, this system was not viable. State and collective enterprises, in their quest for collaboration with foreign investors, wanted to get rid of this burden and succeeded in removing many of the legal obstacles to their elimination. In the long run the transition will prove healthy for the economy. The problem arises because China was unable to replace the past concealed system of unemployment insurance and social security by more transparent alternatives. Once again, a major obstacle was the concern for macroeconomic stability, given the thirst for high accumulation.
Trade, Exchange Rate and Export-Led Industrialization
We have noted that China's shift towards the strategy of export-led industrialization, with the attendant reform of the trade and exchange rate regime and the quest for FDI, coincided with a reduction of poverty alleviation. And yet it is not appropriate to blame these policies for the outcome on poverty. These policies have generally shifted resources towards activities that enhance employment and help reduce poverty under favourable conditions. As noted above, the outcome during the transition period was vitiated due to the initial distortions that characterized China's economy and the tardiness in putting in place appropriate social institutions.
We have also wondered if China went overboard in its quest for FDI as a component of the strategy for export-led growth in a globalizing world economy. The principal manifestation of this took the form of concentration of public expenditure and incentives in areas which held potential attraction for FDI. There are other ways in which the single-minded pursuit of FDI has caused problem for China's poverty alleviation. In recent years China's currency has appreciated significantly in real terms and it is possible that the surge of external capital has been a factor behind this steady appreciation.(43) The loss of flexibility of adjusting the exchange rate has increased the pressure on domestic policies - e.g., reduction of labour input per unit of output and containing wage rate - as instruments for maintaining international competitiveness. These policies have negative effects on poverty alleviation in the short run.
Migration
Before reforms Chinese policies strongly discouraged migration from rural to urban areas. Urban population as a proportion of total population increased very little in the decades before reforms began. It rose from 12.5 per cent in 1952 to only 17.9 per cent in 1978. Since the launching of reforms the rate of migration has been more rapid. The proportion of the Chinese population living in urban areas increased to 28.6 per cent by 1994.(44) The actual rate of urbanization in the reform period has of course been even higher because the official figures exclude the so-called "floating migrants", those who have migrated but not received official residence permits. As we have noted above, these migrants are estimated to be about 72 million according to the findings of a 1995 survey. Official resistance to migration has softened in recent years and an estimated half of the floating migrants have received temporary residence permits.
The effect of migration on poverty is a subject of debate in China. It is almost certain that the inclusion of these migrants in the household surveys would increase the estimate of urban poverty in China. This is because these migrants must be worse off than the official urban residents. The 1995 survey referred to above does not provide information on the earnings of the migrants. It however reports that 20 per cent of them have set up businesses of their own, 24 per cent are wage labourers and 56 per cent have had multiple occupations. The high proportion of multiple occupations indicates that the migrants have a high turnover of employment, which probably means that they are typically employed in residual categories of urban work. They also suffer from the disadvantage of being excluded from public housing and access to social services, e.g., health benefits and schooling for children at low cost that official urban residents are entitled to. These discriminatory measures make their cost of living considerably higher than the cost of living of the official residents, especially when the migrants are accompanied by the members of their families. This last consideration makes a high proportion of these migrants leave their families behind in villages and send income remittances to them.
It seems virtually certain that the incidence of rural poverty would have been greater if these migrants had stayed on in rural China. Their departure has increased the incomes of the rural residents both because of a reduction in family size per unit of productive resources and because of the income remittance that the migrants make to their families in rural areas. Since poverty is a much more serious problem in rural China than in urban China, it seems that the overall effect of migration has been a favourable factor in alleviating poverty in China.(45) An accelerated dismantling of the existing obstacles to migration should therefore help alleviate the problem of poverty.
Why China's Growth Since the Mid 1980s
Has Not Alleviated Poverty?
Reforms ushered in a period of very high growth and rapid reduction in poverty in both rural and urban China. In rural China there was a sharp decline in the rate of reduction in poverty since the mid 1980s and in the 1990s the reduction of poverty was halted. The number of rural population in absolute poverty appears to have increased between the mid 1980s and the mid 1990s.
In China there is little official concern about urban poverty and indeed the incidence of urban poverty is far lower than the incidence of rural poverty even when the urban poverty threshold represents a higher level of real income than the rural poverty threshold. In urban China the decline in the rate of reduction in poverty began later, around the turn of the decade. The reduction in the more extreme kind of urban poverty virtually halted as a proportion of urban population and was reversed in absolute magnitude.
The decline in poverty reduction in rural China was due to both an increase in the inequality of income distribution and a sharp reduction in the rate of growth of personal income. Of the two factors, the latter was more decisive. The decline in growth in personal income in rural China since the mid 1980s can be traced to three sets of causes in each of which macroeconomic and sectoral policy changes featured prominently: (a) a large wedge between growth in GDP and growth in personal income due to an overwhelming emphasis on further raising the already high rate of accumulation; (b) a reduction in the growth of agricultural output; and (a) a reversal of the improvement of agriculture's terms of trade that characterized the early years of reform. The rise in rural inequality was due partly to the forces unleashed by institutional reforms in agriculture and rural non-agricultural sectors and partly due to the absence of offsetting public action to enhance the access of the poor to services, skill and resources. Strict restriction on migration from rural to urban areas was a further hindrance to the reduction of rural poverty. In recent years these restrictions have been widely violated and the tolerance shown by the authorities to floating migration has helped the avoidance of further exacerbation of the problem of rural poverty.
In urban China the decline in poverty reduction began at a time when the rate of growth in personal income accelerated. Thus the reduction in poverty alleviation was by and large due to a rise in the inequality of income distribution. This in turn was largely due to the apparent "employment hostility" of industrial growth. Although the absorption of labour in industries at constant intensity of employment appears to have been quite reasonable, the head-count rate of employment growth was dismally low because the state and collective enterprises began to shed the surplus labour that they traditionally carried. The income distribution effect of enterprise reform that led to this shedding of labour was particularly unfavourable in the absence of compensating institutions and programmes of social protection.
The shift towards an export-led industrialization strategy and the quest for FDI in a globalizing world economy provided China the impetus for labour shedding in state and collective enterprises. Preoccupations with these objectives also led to a concentration of public expenditure and incentives in the urban centres of rich coastal/eastern provinces which contributed to an increase in both urban-rural and interregional inequality with adverse implications for poverty reduction.
Macroeconomic and sectoral policies dictated a pattern of expenditure and cost recovery for health and education services which adversely affected the access of the poor to these services in recent years.
In China women have traditionally been discriminated against both as members of the society and as members of the labour force. Evidence on how recent changes in macroeconomic and sectoral policies have affected them is rudimentary although it strongly suggests that the access of women to health and education services has been unfavourably affected.
The Focus of Official
Poverty Alleviation Strategy
Chinese policy makers are acutely aware of the sharply reduced impact that the remarkable growth of their economy has had on poverty after 1985. The elimination of absolute poverty by the year 2000 is an important target of China's economic and social policy. To oversee the implementation of this target The Leading Group for Development of Poor Areas, directly reporting to the State Council, was established in 1986.
Official poverty alleviation (PA) strategy in China starts from the assumption that poverty is a rural phenomenon. It has designated 592 counties, a high proportion of them in north-west and south-west China, as poor. A broad programme of encouraging a variety of agricultural and non-agricultural activities, backed by funding provided by the central and local governments and technical assistance provided by both government ministries and non-government agencies, has been adopted in these poor counties. Since 1993 this PA strategy has come to be known as the 8:7 programme, indicating the target of eliminating poverty, then estimated to consist of 80 million people, in the remaining seven years of the present century.(46)
An evaluation of this programme is outside the scope of this study which is concerned with macroeconomic and sectoral policies. It is however worth noting that China's PA strategy does not start from the recognition that the principal determinant of poverty in China is the development strategy itself and its constituent macroeconomic and sectoral policies. Instead poverty is viewed as a localized problem, concentrated in predominantly remote areas with poor resource endowment. If our analysis of the causes of the decline in poverty reduction is at all valid, then this approach is inadequate. Its inadequacy is also demonstrated by the commonly admitted fact that between 30 per cent and 50 per cent of China's poor live outside the 592 designated poor counties.(47) Indeed many in urban China are poor too and the composition of the urban poor appears to be changing fast as their ranks are being joined by the redundant workers in state and collective enterprises and the poor among the floating migrants.
Another interesting feature of the official PA strategy is its rejection of a targeted income subsidy approach. According to the information provided by the office of The Leading Group for Development of Poor Areas, in 1995 the central government alone provided 10.8 billion yuan for poverty alleviation programmes. Assuming that the official estimate of 65 million poor is correct; assuming further that the average income gap of the poor is 25 per cent (which is the average gap in our low PIT alternative for 1994 in rural China(48)); and using the official PIT of 530 yuan for 1995, only 8.6 billion yuan was necessary to eliminate poverty through a perfectly targeted income subsidy programme. Given that funds were also available from local government sources, a targeted income subsidy approach would have removed poverty, as officially defined, even with very substantial leakages. Whether the information base and the organizational capability of the government were adequate for the implementation of a reasonably well-targeted income subsidy programme is something on which one must reserve ones judgement. What is known is that China's PA strategy started out by rejecting the income subsidy approach (the "relief approach") and instead focused on the development of the capacity to produce income in poor areas. In doing so, targeting was not a criterion that was insisted upon. The PA strategy encourages the creation of groups consisting of non-poor and poor - and led by the non-poor - to engage in income enhancing activities in poor areas. That the gain made by the non-poor may exceed the gain made by the poor, both absolutely and in percentage terms, is not a consideration that discourages public support for projects of this kind. Not only the overall development strategy, but also the directly poverty alleviating programmes in China seem to be premised on a strong endorsement of the "trickle-down approach".
The Need for a Comprehensive
Poverty Alleviation Strategy
It is however unfair to conclude that there has been no change in macroeconomic and sectoral policies affecting the poor in China. A number of important changes have been made although they have not been formulated as an integral part of an overall poverty- alleviating development strategy. We shall conclude by enumerating some of these changes in macroeconomic and sectoral policies affecting the poor.
In 1994 agricultural purchase prices were sharply adjusted upwards resulting in a significant improvement in the terms of trade of the sector. For the first time in more than a decade there was an appreciable reduction in the ratio of urban personal income to rural personal income in 1995.(49) The improvement in agriculture's terms of trade appears to have been partly reversed in 1995. There are reports of another round of rise in agricultural purchase prices in 1996.
One however notes with some concern that official policy in China merely aims at a reduction in the rate of increase in urban-rural disparity, not a reduction in the disparity itself. Thus Prime Minister Li Peng's report on the Ninth Five-Year Plan delivered at the National People's Congress on March 5, 1996 states (Li Peng, 1996, emphasis added):
During the Ninth Five-Year Plan period, the urban residents' per capita income spent on living expenses after allowing for price rises is expected to increase by about 5 per cent annually and the per capita net income of the peasants is expected to increase by about 4 per cent annually.
In the autumn of 1995 the Plenary Session of the Central Committee of the Communist Party of China decided to make adjustment in the policy of extreme concentration of incentives in the coastal/eastern provinces. It has been decided to shift the focus of the incentive system and public expenditure to attract FDI to central and western provinces as well.
In recent years China has made major de-facto changes in its policy of restricting migration from rural to urban areas. Fear of social instability has led the policy makers to avoid more radical measures such as the abolition of the residence permit requirement (hukou). The official PA programme encourages migration of workers out of poor regions. There is however a clear policy of dispersing the migrants to smaller cities. Temporary residence permits are reported to have been given to many of the migrants while the presence of the remainder in urban areas has been tolerated. These policies have eased the problem of rural and overall poverty in China.
Recently the government has decided to revive the rural co-operative medical insurance system of the collective era. From the current 10 per cent level, its coverage is planned to rise to 80 per cent of the rural population by the year 2010 (China Daily, July 16, 1996, p.2). This is intended to improve the access of the rural population, especially the poor, to health service.
It is essential for China to incorporate these policies - along with other high-priority macroeconomic and sectoral policies, e.g., passing a higher proportion of GDP growth on to the households in the form of steady and regular increases in personal income; a reduction of urban-rural inequality through a combination of accelerated rural growth and orderly migration; social protection for redundant workers during the transitional period of enterprise reform; and adjustments in the level and composition of public expenditure and in the policy of cost recovery to improve the access of the poor and women to education and health - as an integral part of a poverty-alleviating growth strategy. The reliance on PA programmes in designated poor areas will remain an inadequate approach unless they are integrated with an overall strategy encompassing macroeconomic and sectoral policies for poverty alleviation.
REFERENCES
Allen, R.G.D., 1975, Index Numbers, theory and Practice, Macmillan, London.
Boltho, Andrea, 1994, China's Emergence - Prospects, Opportunities and Challenges, International Economics Department, The World Bank.
Cambodia, 1994. Socio-Economic Survey of Cambodia, 1993/94, Department of Statistics, Ministry of Planning, Phnom Penh.
China Statistical Information and Consultancy Centre, 1996. The Collection of Statistics on China Economy and Social Development, Beijing.
Foster, James, J. Greer anad E. Thorbecke, 1984. "A Class of Decomposable Poverty Measures", Econometrica, Vol. 52.
Howes and A. Hussain, 1994. Regional Growth and Inequality in Rural China, STRICTED Working Paper EF11, London School of Economics, London.
Khan, A.R., 1993. Employment and Wages in Rural China, Asian Regional Team for Employment Promotion, New Delhi.
Khan, A.R., 1996. Employment, Growth and Liberalization: China's Growth in a Globalizing World Economy, ILO East Asia Multidisciplinary Advisory Team, Bangkok (Forthcoming).
Khan, A.R., 1996a. "Grobalization and Urban Employment: Some Issues in an Asian Perspective", Paper presented at UNDP/Habitat II Round Table "The Next Millenniun: Cities for People in a Globalizing World", Marmaris, Turkey.
Khan, A.R., Keith Griffin, Carl Riskin and Zhao Renwei, 1992. "Household Income and its Distribution in China", The China Quarterly, No. 132
Khan, A.R., Keith Griffin, Carl Riskin and Zhao Renwei, 1993. "Household Income and its Distribution in China", in Keith Griffin and Zhao Renwei (eds), The Distribution of Income in China, Macmillan, London.
Knight, John and Mo Rong, 1996. Report on the 1996 Survey of Migrants, Chinese Academy of Labour Science, April (Processed in Chinese), Beijing.
Li Peng, 1996. Report of the Ninth Five-Year Plan for National Economic and Social Development and the Long-Range Objectives to the Year 2010, March 5, Xinhua News Agency, Beijing.
Ravallion, Martin, 1992. Poverty Comparisons, A Guide to Concepts and Methods, Living Standards Measurement Study Working Paper No. 88, The World Bank, Washington, D.C.
Ravallion, Martin, Shaohua Chen and Jyotsna Jalan, 1996. Dynamics of Poverty in Rural China, Some Results of a World Bank Research Project, World Bank Resident Mission, May, Beijing (Processed).
State Statistical Bureau (SSB), 1995. China Statistical Yearbook 1995, Beijing. Statistical Yearbooks for other years are cited similarly as: SSB followed by the relevant year.
Tao Chungfang and Jiao Yongping (eds), 1993. A General Survey on the Social Status of Women in China, China Women's Publishing House, Beijing (in Chinese).
United Nations Children's Fund (UNICEF), 1995. Children and Women of China, a UNICEF Situation Analysis, June, Beijing (Processed).
United Nations Development Programme (UNDP), 1995. The Human Development Report 1995, Oxford University Press, New York.
Wong, Christine (ed), 1996. Financing Local Government in the People's Republic of China, Chapters 1 and 6. Oxford University Press (Forthcoming).
World Bank, 1992. China: Strategies for Reducing Poverty in the 1990s, Washington, D.C.
World Bank, 1992a. World Tables, 1992, The John Hopkins University Press, Baltimore.
World Bank, 1995. World Tables, 1995, The John Hopkins University Press, Baltimore
World Bank, 1995a. World Development Report, 1995, Oxford University Press, New York.
World Bank, 1996. The Chinese Economy, Fighting Inflation, Deepending Reforms, Washington, D.C.
World Bank, 1996a. World Development Report, 1996, Oxford University Press, New York.
ANNEX
ESTIMATING TRENDS IN POVERTY INCIDENCE IN CHINA
Poverty Income Threshold (PIT)
The first step in estimating trends in the incidence of poverty is to determine a poverty income threshold - a minimum acceptable level of income - below which a person would be designated poor. In LDCs, especially the poor ones, a widely used method is to define PIT in relation to some minimum acceptable standard of basic nutrition, e.g., the recommended kilocalories of food energy per capita. Those who accept this basic principle differ in its application, two opposing methods being: (a) the estimation of a minimum cost diet, with various degrees of concession to consumer preference as constraints, that achieves the recommended level of nutrition and a further allowance for "minimum necessary" non-food consumption; and (b) the observation of actual consumption behaviour, as revealed by cross-sectional data on household expenditure, to determine the level of income at which the specified nutritional target is achieved. We have preference for the second method because it avoids arbitrary assumptions about the behaviour of consumers. We can however fully apply this method only for urban China for which cross-sectional data on detailed expenditure on food is available. For rural China we need to make some additional assumptions because of the absence of data about consumer expenditure on food at different levels of income.
Urban PIT
For the year 1994 the State Statistical Bureau provides information on per capita disposable income ("per capita income available for living") and per capita purchase of 25 categories of food by the following eight groups of urban households: the poorest five per cent; the next five per cent; the second decile; the second, third and fourth quintiles; and the ninth and tenth deciles (SSB, 1995). These 25 food items account for 1863 kcalories on the average and 1640 kcalories for the poorest decile of households. To estimate total food energy consumption we need information about the proportion of total kcalories supplied by these 25 major items. We do not have this information for China. The nearest source that we can find is a very detailed household expenditure survey for urban Cambodia (Phnom Penh) for 1993 which shows that these 25 items accounted for 85 per cent of kcalories on the average and a declining proportion of kcalories as income rises, for example, 93 per cent for the poorest decile, 91 per cent for the second poorest decile and only 79 per cent for the richest decile (Cambodia, 1994). We surmise that the pattern of declining proportion would hold for urban China although the average proportion of food energy supplied by these 25 items would be lower in urban China both because of the greater variety of Chinese diet and because of the fact that the "purchases" of food exclude consumption of self-produced food which, though small in urban China, may still be significant. The reason behind this hypothesis is that if the average Cambodian ratio is applied to China, the per capita urban daily consumption of food energy turns out to be just under 2,200 kcalories which appears low. Reliable independent estimate of average kcalories of food energy consumed by the urban Chinese is not available; but a figure of about 2,500 seems to be a reasonable guess.(50) In consideration of all these we assume that the 25 items of food referred to above account for 75 per cent of food energy consumed by an average household and 78 per cent of the food energy consumed by a household with poverty threshold income. Making the further assumption that the daily food energy requirement for urban China is 2,100 kcalories, we find that for a household at poverty threshold these 25 items must provide 1,638 kcalories. Our requirement for urban China is thus a little less than 2,150 kcalories used by World Bank, 1992 for both urban and rural China. We use a slightly lower per capita energy requirement for urban China than the 2,150 kcalories for rural China in recognition of the fact that in rural China a higher proportion (64 per cent) of the members of a household are in the labour force than in urban China (57 per cent). The average intensity of physical labour may also be higher in rural areas than in urban areas.
Next, using the information for the eight income groups we fit the following Engel function for food energy consumption for urban China:(51)
Kcal = -672.68 + 317.50 log Y Adjusted R2 = 0.98
(-8.31) (31.47)
Where: Kcal = Daily per capita kilocalories of food energy
Y = Per capita income
and figures in parentheses are t-values.
Using this function we find the PIT that would satisfy the minimum food energy requirement to be 1,448 yuan per capita for 1994.
The next step is to obtain PITs for other years. It would appear that the obvious procedure is to use the urban consumer price index (CPI) to deflate the PIT for 1994. The problem is that the CPI represents the rise in the cost of living of the average consumer which is likely to be different from that of the consumer at poverty threshold. We do not have the information necessary to estimate a separate CPI for the poor; but there are ways to make some adjustment for the different rate of increase in the cost of living of the poor. There are two important differences in the consumption pattern of the poor from that of the average consumer in China: (a) the poor spend a higher proportion of their expenditure on food (e.g., the bottom decile in urban China spent 61 per cent as compared to 50 per cent spent by an average urban household in 1994); and the poor obtain a higher proportion of their food energy from grain (57 per cent of the total from the 25 items referred to above in 1994) than the average (53 per cent). Adjustment is necessary for the period between 1990 and 1994 because during this time food prices rose much faster than CPI and grain prices rose much faster than food prices due to the reform of retail price and distribution of food. For example, the prices of wheat and rice rose respectively by 243 per cent and 302 per cent as compared to the average rise in the CPI of 66 per cent between those years. For 1994 we obtain the CPI for PIT as follows (all CPIs below are for 1994 with 1990 as base):
Pp = apPpf + (1-ap)Pnf
Where: Pnf = (P-aPf)/(1-a) = CPI for non-food expenditure
P = CPI for the average consumer
Pf = CPI for average consumer for food expenditure
Pnf= CPI for average consumer for non-food expenditure
a = the average proportion of expenditure on food
and CPIs and the parameter "a" with superscriptp are corresponding items for the consumer at poverty threshold.
From information about prices of major sources of food energy we estimate Pf (=196.65) and Ppf (=203.07). We have P = 165.68, a = 0.5 and ap = 0.61 which enables us to estimate Pp, the CPI for poverty threshold, as 176.41, higher than the CPI for the average consumer. Note that this adjustment is made only for the period between 1990 and 1994, a period during which there was a sea change in relative food prices in China. The need for such an adjustment in the previous period is far less urgent and, in any case, we do not have the price information to make a similar adjustment for the earlier years.
In addition to the "standard" PIT derived in the way shown above, we also use a "low" PIT (80 per cent of the standard PIT). The CPI appropriate for PIT and the two sets of PITs for urban China for different years are as follows (for pre-1990 years the CPI is based on that for urban China shown in SSB, 1995):
Year CPI for PIT PITs in current yuan
Standard Low
1981 50.5 415 332
1985 60.5 497 398
1989 98.7 810 648
1990 100.0 821 657
1991 107.1 879 703
1993 138.5 1137 910
1994 176.4 1448 1158
This may also be the time to note that it is almost certain that the PIT for 1990 and preceding years would represent a higher per capita food consumption than the PIT in 1994. The reason is that there was a very sharp rise in relative prices of food between 1990 and 1994. Although the PIT in 1994 is sufficient to buy the actual bundle of food and non-food goods that the PIT in 1990 could buy, consumers would not, actually did not, buy the same bundle in these two years. Between 1990 and 1994 Kcalories from the 25 purchased items referred to above fell by 14 per cent for both the average urban consumer and the poorest decile of urban consumers while average urban real income rose by 38 per cent and the real income of the poorest decile rose by 14 per cent. This is because of the sharp rise in relative food prices. It would however not be justified to insist that the PIT over time must represent constant actual kcalories of food energy. This would mean a much higher real income for the 1994 PIT than for the PIT for 1990 and previous years in order to offset the price effect of much more expensive food in 1994. Our PITs over time represent the same constant real income in the sense that the consumers could buy the benchmark-year bundle of goods in other years if they wanted to. It is only in the benchmark year that specified food energy is actually consumed. In other years a different level of food energy might be achieved by the households at poverty threshold depending on relative food prices and price elasticities of different food items. Minimum required food energy is a concept that helps us anchor the PIT to something concrete in the benchmark year. PITs in other years represent the same level of real income as in the benchmark year only in the sense that income in other years is sufficient to enable the consumers to buy the bundle of goods that are actually purchased in the benchmark year without any stipulation that the consumers actually do so.(52)
It is worth noting that the income elasticity of kcalories of food energy (from the 25 items) is low, only 0.19 at the standard poverty threshold and gently declining thereafter. The income elasticity of food expenditure is much higher, about 0.68 at the standard poverty threshold level of income.(53) This means that even the households at the poverty threshold shift towards a higher quality of food while increasing the intake of food energy (from the 25 items) by only a fifth of the proportionate rise in income.(54) This perhaps raises questions about the validity of the calorie norms that are widely used, but does not necessarily constitute a serious problem for the formal anchoring of the benchmark poverty threshold to a specified level of food energy consumption.
How do our PITs compare with the ones used by the others? World Bank, 1992 uses an urban PIT for 1990 of 321 yuan which is only about half of our lower PIT. On the other hand, our standard PIT for 1994 is almost exactly the same as the official Chinese PIT of 1440 yuan. There is little doubt that the World Bank PIT is dismally low. It is 30 per cent below the actual food expenditure of the poorest 5 per cent of the urban households in 1990 and 60 per cent below the total expenditure of the same group of households! It is based on the transparently false assumption that the poor in China obtain 90 per cent of their food energy from grain, a heavily subsidized item during the period for which they make use of their PIT.(55) To use such a low PIT is to start from the presumption that there is no problem of poverty. Our estimates appear to make much better sense since they are both closer to the actual expenditure of the poorest five per cent of households and are much more meaningfully anchored to the minimum food energy requirement. Furthermore, one needs to note that what matters is the change in the incidence of poverty over time. For this purpose the absolute level of PIT is not of critical importance.
Rural PIT
For rural China we do not have cross-sectional information on variation in food consumption for different income groups. We therefore use the following method. We start by using 1994 as the benchmark year and estimate for an average household total kcalories of food energy and the expenditure on food. From these estimates we calculate the average cost of a kcalorie of food energy for an average household. We set the daily per capita food energy requirement for rural China at 2,150 kcalories and assume that the unit cost of food energy for a household at poverty threshold is 85 per cent of the unit cost of food energy for an average consumer. In urban areas the unit cost of kcalories for the poorest decile of households is 80 per cent of the unit cost for an average household. We surmise that in rural China the difference is less. We further assume that the share of food in total expenditure of a household at poverty threshold is two-thirds. This compares with the actual share of 59 percent for an average household. These steps lead us to an estimate of PIT of 647 yuan for rural China for 1994. In addition to this "standard" rural PIT, we use a second, lower, PIT which is 80 per cent of the standard PIT.
To convert these into current price PITs for other years we need a rural CPI for poverty threshold. For years up to 1990 we use the rural CPI from SSB, 1995 and 1990. For years after 1990 the official rural CPI appears to be understated even for an average consumer. The official CPI for rural China increased by 50.2 per cent between 1990 and 1994. We make an estimate of CPI for poverty threshold as follows. Household survey data show that grain prices (expenditure divided by quantity) increased by 80 per cent and non-grain food prices increased by 69 per cent over this period.(56) The weighted average rise in food prices was 73 per cent. For non-food prices we use the "overall industrial product rural retail price index" (which rose by 39 per cent). By weighting food price relative by 0.67 and the non-food price relative by 0.33 we arrive at an estimate of a 62 per cent increase for CPI for households at poverty threshold. We surmise that our adjusted CPI also understates the increase in cost of living for the households at poverty threshold because the latter almost certainly have a greater weight for grain in their consumption than the average. As we report above, grain prices rose faster than average prices. Had we assigned higher than average weight to grain in estimating the food price relative, the CPI for PIT would have been higher. For 1992 and 1993 the estimation of the CPI for PIT is based on exactly the same method as for 1994. The rural PITs and the corresponding rural CPIs for different years are shown below:
Year CPI for PIT PITs in Current Yuan
Standard Low
1980 49.3 197 158
1981 50.5 202 162
1983 52.6 211 169
1985 60.5 242 194
1988 84.9 339 271
1990 100.0 400 320
1992 108.0 432 346
1993 124.0 496 397
1994 162.0 647 518
Even our low PIT is significantly higher than the official rural PIT which, for 1994, was 440 yuan. As we have often said, this by itself is not a reason for worry since our interest lies entirely in establishing the trend in the incidence of poverty over time. We did not accept even lower a PIT than our low estimate because it would render the formal anchoring of PIT to food energy consumption meaningless. Also it would further magnify the difference between the urban and the rural PITs which is already rather high.
What one should however worry about is the relative levels of the official rural PIT for different years, which have been as follows:
1988 260
1989 285
1994 440
Between 1988 and 1989 the average rural CPI increased by 19.3 per cent while the official PIT was adjusted upwards by 9.6 per cent. Again, between 1989 and 1994 the average rural CPI increased by 57.1 per cent while the official PIT was adjusted upwards by 54.4 per cent. According to our adjustment, the CPI for poverty threshold should have increased by a minimum of 65 per cent between 1989 and 1994. Thus the official rural PIT over time has been progressively representing declining levels of real income. As we have argued, this is a major explanation of continued reduction of rural poverty after mid 1980s according to official estimates.
It would be noted that the standard urban PIT is more than twice the standard rural PIT. While we do not have information necessary to estimate urban and rural PITs that would represent comparable levels of real income, it is clear that our urban PIT represents a higher level of real income than the rural PIT does. Why do we opt for different levels of real incomes for PITs in two locations? The first reason is obvious, namely that we could not have estimated PITs that represented same levels of real income if we wanted to due to the lack of information. Secondly, it does not seem justified to use the same real income PITs for societies that differ so greatly in terms of average standard of living as urban and rural China do. Urban per capita net income in the most recent year reported above, 1994, was 2.6 times the rural per capita net income and this may be an underestimate of the actual urban-rural difference in living standard (as argued by Khan et al., 1993).(57) Given consumer preferences and relative prices of goods and services, food energy threshold appears to be attained at a lower level of real income in rural areas than in urban areas. As discussed in the main text, the use of constant real income PITs would therefore mean that if the rural PIT just satisfies food energy need, the corresponding urban PIT would leave the household at poverty threshold seriously undernourished. Finally, it should be noted that the ratio of official urban PIT to official rural PIT for 1994 was 3.27, far greater than the corresponding ratio for our PITs.
Data
Once the PIT is established, one needs information on the distribution of net income to estimate the incidence of poverty. The only available source of such information over a relatively long time period is the SSB which publishes grouped data on the distribution of urban and rural income. These data, showing proportions of households falling within specified income class intervals, are presented in Annex Tables 1 and 2 below.(58)
We adjusted the data from proportions of households to proportions of population by using information about household size for different groups. Where information on this was not directly available, we made estimates from whatever alternative source was available.
This type of distributional data does not completely specify the Lorenz function. One needs to know the average income of the population within each group. This latter information being available only in a few cases, one needs to make assumptions about the average income of each group. The computational programme that we used puts it at the mid-point of each class interval unless it is known. This introduces some error in the estimation of both the Lorenz function and the incidence of poverty. Our experiment for the few years for which we have information on group average incomes shows that this error was of limited magnitude.
There are other inadequacies of these distributional data. The class intervals do not coincide for rural and urban China so that we were unable to aggregate rural and urban poverty indices into national indices. Class intervals also vary from one year to another. For some years class intervals are rather small in number and occasionally the class intervals are not balanced in the sense that a few of them have concentrations of population in them.
The general procedure that we followed is to fit two alternative parametric Lorenz functions and, combining the function that gives better fit with the estimate of average income and PIT, compute each of the indices of poverty.
Indices of Poverty
Three different indices of poverty have been estimated. The first is the Head-Count Index, HC = h/n, where h = the number of persons belonging to households with per capita incomes below PIT and n is total population.
In the event that HC remains unchanged while the average income of the poor falls or rises, it is quite inappropriate to consider that the incidence of poverty has not changed. In this situation the HC index of poverty is misleading. The second of our indices, the so-called Poverty Gap Index (more accurately, the mean proportionate poverty gap across the entire population, the gap for the non-poor being defined as zero), captures this kind of change. It is defined as follows:
h h
PG = (1/n) [(PIT - Yi)/PIT] = [ (PIT - Yi)]/[n(PIT)]
i=1 i=1
where Yi is the net income of the i-th person below PIT. The numerator of this measure is the sum of the shortfall in income from PIT of all the poor and its denominator is the total income necessary to guarantee that all members of the society have just adequate income to avoid poverty. The former represents the cost of removing poverty in a perfectly targeted program of income subsidy and the latter represents the cost of removing poverty in a situation of complete ignorance of who the poor are. Thus the intuitive interpretation of this index is that it represents the ratio of minimum income transfer needed to remove poverty (the case of perfect targeting) to maximum income transfer needed to remove poverty (the case of complete absence of targeting).
It is possible for both HC and PG to remain unchanged while the distribution of income among those below the PIT becomes worse, i.e., the not-so-poor become better off at the cost of the extremely poor or vice versa. In these circumstances it would again be inappropriate to consider that the incidence of poverty has remained unchanged. The "Weighted Poverty Gap" index (our alternative to the more usual, if unmanageable, name "Foster-Greer-Thorbecke P2 measure of poverty"), defined as the mean of squared proportionate poverty gaps, captures this change:
h
WPG = (1/n) [(PIT - Yi)/PIT]2
i=1
The squaring of the proportionate poverty gap is a special formulation of the more general case of raising it to a non-negative power . With replacing 2, the measure represents a generic class of poverty indices. For =0 the measure collapses to the HC index and for =1 it becomes the PG index.(59)
Annex Table 1. Distributional Data for Rural China
Annual Income Range Per cent of Population in the Range
(Current Yuan) 1980 1981 1983 1988
Below 100 11.07 4.9 1.5 0.6
100 - 150 26.04 15.5 6.7 1.6
150 - 200 27.26 23.9 14.2 3.6
200 - 300 24.12 36.2 35.6 14.6
300 - 400 7.65 13.5 23.1 18.3
400 - 500 2.49 4.0 10.5 16.9
500 and above 1.37 1.9 8.4 44.3
1985 1990 1992 1993 1994
Below 100 1.09 0.25 0.37 0.51 0.34
100 - 150 3.85 0.58 0.56 0.43 0.21
150 - 200 8.90 1.47 1.16 0.89 0.45
200 - 300 27.04 7.47 5.68 3.89 1.75
300 - 400 24.17 13.15 9.89 7.23 3.31
400 - 500 15.19 14.89 12.52 9.95 5.20
500 - 600 8.35 14.08 12.91 10.96 6.75
600 - 800 6.90 20.57 21.16 20.80 16.43
800 - 1000 2.52 11.75 13.59 15.30 16.17
1000 - 1500 1.63 10.95 14.40 17.96 25.60
1500 - 2000 0.22 3.01 4.27 6.21 11.94
Above 2000 0.14 1.83 3.49 5.87 11.85
Note: Data have been compiled from different issues of SSB, China Statistical Yearbook, which show proportions of households for given income ranges. See note to Annex Table 2 about the conversion of proportions of households to proportions of population.
Annex Table 2. Distributional Data for Urban China
(Income Ranges are Annual in Current Yuan)
Income Range Per cent of Income Range Per cent of
Population Population
1981 1985
-------------------------- --------------------------
Below 240 2.72 Below 420 14.21
240 - 300 6.75 420 - 600 26.67
300 - 420 35.11 600 - 720 18.72
420 - 600 40.07 720 - 840 14.93
600 - 720 10.43 840 - 960 10.06
Above 720 4.92 960 - 1080 6.27
1080 - 1200 3.64
1200 - 1320 2.30
Above 1320 3.20
Income Per cent of Income Per cent of
Range Population Range Population
1989 1990 1991
-------------------- -------------------------
Below 600 5.74 Below 600 4.71 2.66
600 - 720 5.54 600 - 720 4.40 2.67
720 - 840 7.53 720 - 840 6.52 4.16
840 - 960 9.92 840 - 960 7.68 6.04
960 - 1080 11.29 960 - 1080 9.30 7.38
1080 - 1200 11.39 1080 - 1200 9.01 8.02
1200 - 1320 10.17 1200 - 1320 9.74 8.59
1320 - 1440 8.34 1320 - 1440 8.94 9.02
1440 - 1560 7.04 1440 - 1560 7.67 8.46
1560 - 1680 5.53 1560 - 1680 7.01 7.38
1680 - 1800 3.95 1680 - 1800 5.46 6.51
Above 1800 13.56 1800 - 1920 4.37 5.81
1920 - 2040 3.19 4.40
2040 - 2160 2.44 3.53
2160 - 2280 1.88 2.91
2280 - 2400 1.30 2.33
Above 2400 6.39 10.14
1994
---------------------
Below 1200 4.53
1200 - 2400 31.82
2400 - 3600 30.95
3600 - 4800 16.88
4800 - 6000 8.00
6000 - 7200 3.53
7200 - 8400 1.80
8400 - 9600 1.03
9600 - 10800 0.49
10800 - 12000 0.34
Above 12000 0.62
Note: All data, except those for 1994 which are from unpublished SSB sources, have been compiled from different issues of SSB, China Statistical Yearbook. The latter shows percentages of households for each income range. These are converted into percentages of population by using average household size for households in each income range. Direct data on the latter is shown for only a few years for urban China. For all other cases household sizes for different income ranges have been estimated from a variety of data sources.
NOTES
1. Many would consider additional features of growth, e.g. its sustainability, as a necessary condition of acceptability of progress.
2. This statement needs to be qualified if changes in distribution are measured by summary indices, e.g., the Gini ratio of income distribution. An improvement or deterioration of these summary indices can leave the incidence of poverty unchanged as long as the Lorenz distribution in the relevant range, i.e., the range affecting the poor, remains unchanged. This can happen if the change in Lorenz distribution involves only a redistribution from one non-poor group to another non-poor group. for China which has been moving away from public provision of consumption.
3. The case is clearly different for official Chinese policy which seems to be rather single-mindedly focused on the elimination of absolute poverty defined in a completely arbitrary manner. Current government policy in China puts an inordinate emphasis on eliminating poverty, in 592 designated poor rural counties, by the year 2000. For the success of this politically sensitive target, the criterion of defining the poor is of course of paramount importance. The irony is that the criterion is completely arbitrary and impossible to justify in so far as it is well known that many households that are poor according to the same criterion live outside the designated counties. This fetishism about an absolute PIT (320 yuan at 1993 prices), applied in counties which themselves have average incomes below some "county PIT" (400 yuan at 1993 prices), can lead to public policies and programmes that are inconsistent with an improvement of the welfare of the poor across the nation. Also, as shown later, nominal rural PITs that China has adopted in different years do not appear to satisfy the principle of either constant real income or constant level of food energy consumption.
4. For 1994 the official urban PIT was 1440 yuan and the official rural PIT was 440 yuan, making the urban PIT 3.27 times the rural PIT. The corresponding ratio of PITs used in this study is 2.24.
5. These figures are from SSB, 1995, chapter 9.
6. This was estimated by valuing urban and rural consumption at urban prices.
7. This seems plausible in view of the likelihood that in urban China much of the remaining food energy is obtained from eating out. Also the ten major items represent a lower proportion of food expenditure in urban China than in rural China.
8. A computational programme developed by Gaurav Datt and Martin Ravallion at the World Bank was used for estimation. It fits two different types of parametric functions, tests the validity of each type of fitted function and identifies the one that fits the data better.
9. These are based on official estimates of rural population of 807.6 million in 1985 and 855.5 million in 1994 (SSB, 1995).
10. This is based on official urban population of 305.4 million in 1991 and 343.0 million in 1994 (SSB, 1995).
11. See Khan et al., 1992.
12. It is puzzling that the SSB should end up with a higher estimate of the poor with a lower PIT than our low estimate and the same distributional data that we use. See below for possible explanations of this puzzle. Official estimates of poverty were obtained from the office of the Leading Group for Development of Poor Areas.
13. When his attention to this point was drawn by the author of this report, a senior official of The Leading Group for Development of Poor Areas admitted that in recent years PITs over time often represented consecutively declining levels of real income. He justified this by arguing that a lowering of the PIT in real terms allows the targeting of limited government resources to those who are more severely poor and are more urgently in need of assistance. While we are inclined to conclude that the difference in the method of adjusting PITs over time explains much of the difference between our trend and the SSB trend, one can think of other possible reasons for the discrepancy. Thus a second reason may be the use by the SSB of a simpler method of interpolation of the distributional data to determine the number of the poor than ours. A third possibility is that the SSB bases its estimates on an aggregation of regional estimates, something that we can not do because the SSB does not publish regional distributional data. Our efforts to find out the methodology underlying the SSB estimates did not succeed.
14. Various indices of poverty are determined by: (i) the Lorenz distribution of income and (ii) the ratio of PIT to per capita income. Thus simulation (a) is generated by combining the 1980 Lorenz distribution with the 1994 income level and PIT; and simulation (b) is generated by combining 1994 Lorenz distribution and PIT with what per capita income would be in 1994 had the rate of growth been higher.
15. In principle, it should be possible to provide individuals with higher levels of food energy through public expenditure. But this would be an unrealistic assumption for most societies andespecially so
16. The adjusted R2 between growth rate in per capita rural personal income and per capita GDP has a negative sign and has the very low value of -0.08. The adjusted R2 between growth rate in per capita urban personal income and per capita GDP also has the very low value of 0.07. Growth rates in per capita rural personal income and per capita urban personal income are nearly perfectly uncorrelated with an adjusted R2 = -0.007. These results remain unaffected if growth rates in GNP, instead of GDP, are used.
17. To test if rates of growth in per capita rural personal income were different before and after 1985, we fitted the following trend equation:
Log Yr = a + bZ + gT + h(ZT)
where: Yr = per capita rural personal income
T = time (= 1 for 1978, 3 for 1980 etc.)
Z = 0 for years up to 1985 and 1 for years thereafter
The estimated equation is as follows:
Log Yr = 4.509 + 0.733Z + 0.131T - 0.107 (ZT)
(17.107) (-11.104)
The adjusted R2 is 0.98 and the growth coefficients are highly significant (t-values shown in parentheses). The trend growth rate for the period 1978-85 is g = 13.1 per cent and the trend growth rate for the period 1985-94 is (g+h) = 2.4 per cent.
18. For urban China the fitted equation for income growth is as follows:
Log Yu = 6.715 - 0.380Z + 0.056T + 0.025(ZT)
(16.803) (3.613)
where: Yu = per capita urban personal income
Z = 0 for years up to 1989 and 1 thereafter
Adjusted R2 = 0.99 and the growth coefficients are highly significant. The coefficient of T (i.e., 5.6 per cent per year) is the trend growth rate for 1980-89 and the sum of the coefficients of T and ZT (i.e., 8.1 per cent per year) is the trend growth rate since 1989.
19. See Khan et al., 1992.
20. Agriculture accounts for only a part of rural production activities. It would be better if we could obtain indicators of change in output of non-agricultural activities. This is not available.
21. The fitted equation showing the relationship between the index of per capita rural personal income (YIND) on the one hand and the index of per capita agricultural output (QIND) and the index of agricultures's terms of trade (TOT) is as follows:
Log YIND = -2.174 + 0.685 log QIND + 0.814 log TOT
(4.087) (2.697)
Adjusted R2 is 0.90 and the elasticity coefficients are highly significant.
22. The results of this survey of 19,000 individual households - the only source of ungrouped, household-level, detailed data to date - are reported in Khan et al., 1992. The statement that a particular component of income has an equalizing effect on income distribution means that its percentage contribution to the Gini ratio is lower than its percentage share of income.
23. See Khan et al., 1992.
24. These data are from SSB, 1995. For further details about the arguments in this section see Khan, 1996 and Khan 1996a.
25. For the Republic of Korea the output-elasticity of employment in manufacturing industries was 0.67 during 1970-80 and 0.45 during 1980-90. By the 1980s Korea had become a relatively high-wage economy, particularly in comparison with China during the late 1980s and the early 1990s. It would therefore appear that the elasticity for the 1970s is the relevant one for comparison with the elasticity for China during the decade of globalization. On this basis the Chinese elasticity was only 40 per cent of the Korean elasticity. The Korean elasticities, estimated from the value added and employment data reported in World Bank, 1992a and 1995, are discussed in Khan 1996.
26. 26. See Boltho, 1994.
27. 27. This estimate is based on the ratios from a survey of 4,000 households (9,481 workers) in eight provinces in 1995. Estimates based on the results of this survey show that by that date 86 million rural workers had migrated. Of this, 14 million migrated to other rural areas and the remaining 72 million migrated to urban areas: 16 million to county towns, 10 million to larger cities, 11 million to provincial capitals, 2 million to centrally controlled cities and 33 million to multiple urban destinations. Of course large cities also had migrants from other urban areas. Most of these migrants have found employment in informal activities and in less attractive formal jobs for which nearly half of the migrants are estimated to have been granted temporary residence permits. The survey results are reported in an unpublished internal document by John Knight and Mo Rong at the Institute for Labour Studies, Beijing dated April 1996.
28. 28. The sources of data on which all these estimates and statements have been based are to be found in Khan, 1996.
29. China has indeed instituted a system of unemployment insurance; but it clearly falls short of protecting those who have become unemployed in urban China in recent years. The total number of workers on unemployment insurance in 1995 is reported to be about 2 million, well below half the reduction in employment in state and collective enterprises during 1992-94. Also the amount of unemployment insurance is far below the wage rate. It is impossible to guess the extent to which the workers on unemployment insurance succeeded in supplementing their earnings from other sources.
30. For sources of these data see Khan 1996.
31. This figure is not reported in Table 6 in which the FDI figures have been shown from official Chinese sources for the period up to 1994. The 1995 figure is from the World Bank sources. For past years the estimates from the two sources were not always exactly the same.
32. See World Bank, 1992, Annex 1, Table 2. These estimates use provincial PITs that reflect provincial differences in cost of living.
33. These results are reported in Ravallion et al., 1996. It is worth noting that even in rich Guangdong the proportion of rural population below PIT stopped showing decline after 1987.
34. The date for the data on working hours from Tao and Jiang refers to the early 1990s.
35. It is noteworthy that, according to the source quoted, women work only about 13 per cent longer hours than men. One would have expected a greater difference in view of the widely discussed "triple burden" - the burden of work outside; care of children and the aged; and housework - that Chinese women bear. Indeed some unpublished research suggest that the gender difference in total hours spent in all "economic work" is greater in South Asia than what is reported for China in the data quoted.
36. These data, and the following ones on gender difference in life expectancy, are from UNICEF, 1995.
37. For example, Korea and Mexico, with male life expectancy of 68 years in 1993 had female life expectancy of respectively 75 and 74 years. Thailand had male life expectancy of 66 years and female life expectancy of 72 years. These data are from World Bank, 1995a.
38. Data in this section, unless otherwise stated, are from UNICEF, 1995.
39. The statement that 90 per cent of the farmers pay all of their medical costs need not contradict the data reported above from UNICEF source that 37 per cent of health service costs was borne by the users in 1989 since the latter figure presumably refers to the national average.
40. Christine Wong (ed), 1996 (forthcoming).
41. The latest round of macroeconomic management brought inflation down from 25.2 per cent in the twelve-month period to October 1994 to 8.3 per cent in the twelve-month period ending in December 1995. It is remarkable that during this most recent episode China brought inflation down so substantially with a reduction in GDP growth rate from the peak of 13.5 per cent in 1993 to a still very high rate of 10.2 per cent in 1995. This time China's orthodox macroeconomic management through restricting credit, reducing budgetary deficit and cutting down investment by state enterprises was greatly helped by a surge in agricultural output. See World Bank, 1996 for an account of this episode. Earlier episodes of macroeconomic stabilization were less smooth.
42. See Khan et al., 1992.
43. See Khan 1996 for details. There is however an alternative point of view that China's exchange rate is still "administered" and not sufficiently market determined for its value to be affected so significantly by the flow of external capital. It is nevertheless difficult to accept that the recent appreciation of the Renminbi yuan in real terms would have taken place in the absence of the surge of external capital.
44. These data on urbanization are from SSB, 1995.
45. We are saying that estimated national indices of poverty would have been higher in the absence of migration if urban and rural PITs represented same real income, other things remaining the same. Whether our estimates, using a higher real income PIT for urban China than for rural China, would have shown the same result is not certain.
46. In China the most commonly used numerical units are 10,000 (wan) and 10,000,000 (yichen wan or a thousand wan). The figure 8 in 8:7 programme is in thousand wan units and means 80 million.
47. This range was given to the author by a senior official of The Leading Group for Development of Poor Areas.
48. This gap is PG/HC as can be seen from the algebraic formulations of these indices in the Annex. Also see Ravallion, 1992.
49. China Statistical Information and Consultancy Center, 1996 reports, on the basis of the results of the SSB survey, that this ratio declined from 2.60 in 1994 to 2.47 in 1995. It should be noted that in 1990 there was a small reduction in urban/rural disparity in personal income which was more than reversed in the following year.
NOTES TO ANNEX
51. Recognizing that the eight groups are unequal in size, we use a system of weighting to make the contribution of the observation for a group roughly proportionate to its size: the observation for each of the two bottom groups, representing five per cent of households, is counted once; the observation for each of the decile groups is counted twice; and the observation for each quintile group is counted four times. We thus generate 20 observations. It may be noted that the PIT predicted by the weighted Engel function is approximately the same as the PIT predicted by the unweighted Engel function based on eight observations, one representing the average for each group.
53. - -
55. The point has been noted in Jamal, 1995. Jamal's alternative estimate also appears to be on the low side.
56. The latter is found by dividing expenditure on non-grain food by kcalories from non-grain food.
57. We recognize that these incomes are at actual prices which differ between urban and rural areas. These estimates are therefore not true indicators of differences in living standards between urban and rural China.
58. A second kind of distributional data, showing average incomes of ascending fractions of households classified in eight groups (poorest five per cent; the next five per cent; the second and third deciles; the second, third and fourth quintiles; and the ninth and tenth deciles) are available only for urban China for the period 1988 to 1994. These data are not quite consistent with the class interval data shown in Annex Tables 1 and 2. For selected years we have both types of data for urban China. For these years the second kind of data yield lower estimates of inequality and poverty than the first kind of data although their trends over time are the same. We surmise that the basis of the two sets of data may be different although our efforts to determine the source of the difference has not succeeded. In any case, the first kind of data is the only kind of distributional data available for both rural and urban China for the entire period under review, a consideration that was decisive in making this set the basis of all estimates reported in this study. It should be noted that the publication of the first type of data was discontinued for urban China after 1991. Fortunately, we succeeded in obtaining it for 1994 from unpublished SSB sources.
59. See Foster, Greer and Thorbecke, 1984 and Ravallion, 1992 for further discussions of properties of this class of poverty measures.
Back to the list of papers