Recent trends in poverty in PRC

Rural poverty

Though there are disagreements on the magnitude of the absolute poor in rural PRC or even on the trends of poverty changes at different time period, everyone does agree that PRC's rural poor population has been reduced substantially with the fast growth of the economy and household income in the past twenty years.

Official poverty estimates
PRC's estimate of its rural poor population is based on the poverty line defined by the National Bureau of Statistics (NBS). "Rural poverty" is defined by NBS as "difficulty in material well-being so great that a person or a family cannot reach the socially acceptable minimum standard of living." Therefore, a fundamental principle in defining the poverty line is "the minimum expense required to meet people's basic living needs for necessary goods and services under the specific conditions of time, place and social development" (Tang, 1994). The NBS divides basic personal consumption expenditures into two categories, the food consumption expenditures and non-food consumption (clothing, housing, communications, fuel, health and medical care, education, entertainment, etc.) expenditures. Food consumption to meet minimum calorie requirements is the most important factor for setting the poverty line. To determine the poverty line, one must first select the minimum caloric intake based on nutritionists' recommendations; second, define a proper food consumption bundle and set the quantity of various foods to be consumed; third, calculate the minimum food consumption expenditure based on the prices of different foods and the amounts consumed; and last, determine the Engel coefficient (food consumption as a proportion of total consumption by the poor), which can be used to calculate the non-food consumption expenditures and the poverty line.

Based on the recommendations of the China Nutrition Association, the NBS adopted a daily intake of 2,400 calories per person as the minimum nutritional standard (Wang, Xia, &Liu, 1996). The following principles are applied in defining the food bundle that meets this nutritional standard. First, all the food in the bundle should be necessities, excluding all harmful and extravagant consumption (cigarettes, alcoholic beverages, sweets, etc.). Second, the food bundle should reflect the real consumption pattern in rural areas. The NBS reckoned the farmers' real consumption ratio of grain, vegetable, fruit, meat, poultry, aquatic products, milk, edible oils, sugar, salt, etc. by using the 1984 sample survey statistics on rural households. Table 1 shows the food bundle and the corresponding caloric intake adopted by the NBS. The prices adopted by the NBS in calculating food consumption expenditures are the package (weighted) average prices of the various foods. The Engel coefficient of 60% is used in computing the non-food consumption expenditure before 1998. The reason for using this coefficient is that the Engel coefficient of 60% is usually employed internationally as the criterion for defining absolute poverty, and furthermore, the cost of food constituted about 60% of the average total living expenses of Chinese farmers in 1984, which is regarded as the year when Chinese farmers as a whole had basically emerged from absolute poverty (Tang, 1994).

Based on the above approaches, the NBS first of all calculated the 1984 poverty line in PRC's rural areas; the figure was 200 yuan per person per year.

Table 1 Food Consumption Bundle Adopted by the NBS

Over the years, the NBS has rectified the poverty line established for 1984 in keeping mainly with the changes in the rural retail price index and later with rural consumption price index. Since 1990, the statistical pricing of the farmers' self-consumed agricultural and sideline products has been changed from the state-planned purchase prices to the compound average of contract purchase prices (the weighted average of the state-planned purchase prices and the above-quota purchase prices in contract purchases). Also from 1990 onwards, the poverty line has been adjusted accordingly. Table 2 shows the poverty line in different years as determined by the NBS.

The NBS conducted a new set of poverty line calculations in 1999 using the 1998 national rural sample data (NBS, 2000). A standard food bundle of 27 items in 15 categories was established from the mean consumption pattern of households with income per capita less than 800 yuan, adjusted to meet a minimum caloric standard of 2100 kcals. The income necessary to purchase the standard bundle, or food poverty line, was 527 yuan. The non-food expenditure share was calculated using a regression method proposed by Ravallion.¹ Required non-food expenditures were estimated to be 108 yuan, leading to a poverty line of 635 yuan. However, the non-food expenditure share of 17 percent is substantially lower than the 40 percent share assumed in earlier calculations.

Table 2 Per Capita Income and Poverty Line of PRC's Rural Residents

After setting the poverty line in different years, the NBS, using the household survey system it set up in one third of PRC's counties in the early 1980s and the income data from more than 60,000 sample rural households, worked out the proportion of the rural households and population whose per capita net income level is below the poverty line, and then reckoned the total poverty-stricken population in the country according to this percentage. Table 3 provides the NBS estimates of the poverty-stricken population in PRC's rural areas. According to these figures, the poverty-stricken population in PRC’s rural areas has decreased dramatically over the past 20 years. The absolute poor population decreased from 250 million in 1978 to 28 million in 2002, and the poverty-stricken population as a proportion of the total rural population decreased from 31% to 3%.

Table 3 PRC's official rural poverty headcounts (1978-2002)

Potential bias of the official estimates

Criticism on PRC's official poverty estimates focus on the methodology used by the NBS to calculate the official poverty line (Park and Wang, 2001). The calculation may be subject to a number of different sources of potential bias, i.e. a) An unrealistic food bundle was adopted, in which consumption items viewed to be non-necessities were excluded e.g., alcohol, candy. As a result, grain accounted for 88 percent of expenditures even though grain comprises only 70 percent of actual food expenditures for poor households. Over-weighting of grain in the standard bundle leads to under-pricing of calories since grain is a relatively cheap source of calories. This puts downward bias on the poverty line and leads to an underestimation of poverty. b) Planned prices rather than market prices were used to value own-produced consumption goods before 1990. c) Decreasing share of non-food expenditure in the composition of poverty line also exaggerate the magnitude of poverty reduction in recent years². d) The implicit inflation rates evident in the official poverty lines appear much lower than the rural consumer price index. Khan and Riskin (2001) point out that even the rural CPI is likely to underestimate the growth in living costs of the poor, because their budget shares for food are higher and food-prices have grown relatively faster than other prices.³ In the initial years, the official poverty lines are consistent with the rural retail price index, and in the final several years, they are consistent with the rural consumer price index. However, in the intervening years, there are large discrepancies. Most notably, the poverty line increases only modestly during the high inflation years of 1988 and 1989, and there is a sharp increase in the poverty line 1997 that is far in excess of inflation. This helps explain why official statistics show a steady reduction in poverty in the late 1980s while other estimates show little change. If the 1985 line is inflated by the rural CPI, the 2000 poverty line reaches 721 compared to the official line of 625. This suggests that poverty reductions over time may be greatly exaggerated. e) Like poverty lines in many countries, the NBS's national poverty lines does not allow for regional price differences in calculating required food expenditures.⁴ Food prices vary greatly among different provinces due to transport costs, imperfect market integration, and other factors. Chen and Ravallion (1996) estimate that the cost of purchasing the NBS food bundle was 23 percent higher in Guangdong than in Guangxi in the late 1980s. For example, in 1992 and 1995, the NBS of Jiangxi Province calculated provincial poverty lines of 400 and 750 yuan using local prices and the national food bundle, compared to national poverty lines of 320 and 540 yuan. Failure to account for regional price differences in PRC exaggerates the concentration of poverty in poor regions. f) Some have criticized the NBS sample for under-representing households in remote areas, illiterate households, and minorities, which would lead to an underestimation of the poor population. Choosing such households can be more costly to administer, as they require greater time to reach and supervise. g) Expenditures are considered to be a better measure of both current and long-term welfare. Because individuals prefer to smooth consumption over time, expenditures tend to vary less from year to year than incomes. However, despite the availability of expenditure data, PRC has always calculated poverty rates using income data. This can lead to two types of bias. First, incomes exhibit greater dispersion than expenditures because they are more likely to be influenced by transitory factors. This increases the poverty count. Second, and more importantly, average incomes are 10 to 20 percent higher than average expenditures, so that using income data results in lower poverty rates. The poverty headcounts of the World Bank (2001) are about ten percent higher using expenditure data rather than income data (Table 4).

Alternative Estimates of Rural Poverty

Because of these different sources of bias of the official poverty estimates, it is necessary to look at other estimates using different methods and data sources. Alternative estimates of rural poverty presented in Table 4 offer a wide range of point estimates in the same year and different trends in poverty reduction during different sub-periods. Considering all of the possible sources of bias, a majority of factors, and those with the largest likely influence, lead to an underestimation of the extent of rural poverty and an over-estimation of poverty reductions over time. Improper inflation adjustments lead to a poverty line in 2000 that is 13 percent below what it should be. Use of income rather than expenditure data exaggerates average welfare by 10-20 percent. The food poverty line is overly austere before 1998 because of a standard food bundle that is not consistent with actual consumption patterns, and the non-food expenditure share is unrealistically conservative since 1998, both leading to downward bias in the poverty line. NBS sampling may exclude some of the poor. The only factors leading to upward bias in the poverty count are a high caloric standard (2400 kcals), the failure of the NBS survey to accurately record specific types of income, and a high prevalence of transitory poverty which may reduce our concerns about chronic suffering.

The recent World Bank poverty report (2001) uses a dollar-per-day poverty line, which leads to much higher poverty headcounts than the official statistics. The dollar-per-day standard was established to facilitate inter-country comparisons, but is not based on nutritional standards, consumption patterns, or social norms specific to PRC. Thus, if the dollar-per-day line in fact more accurately reflects rural poverty in PRC today, it is likely due to problems in the methodology used to calculate official statistics rather than any inherent preference for an international standard. The Khan and Riskin range of estimates exemplifies how sensitive the poverty count can be to assumptions about the poverty line.

Because of the arbitrariness of choosing any one poverty line, many feel it is more important to examine trends in poverty over time. All estimates agree that there was a spectacular reduction in poverty in the early 1980s. All estimates other than the official poverty count show little or no progress in poverty reduction in the late 1980s. Reductions in the official count are almost certainly due to insufficient inflation of the poverty line in 1988 and 1989. In the early 1990s, Khan (1996) and the World Bank (2001) show little change until after 1993. The official poverty count falls steadily throughout the 1990s. These differences are not due to differences in inflators or income definition, so must reflect different trends in different parts of the income distribution. Khan and Riskin (2001) and Riskin and Li (2001) emphasize the small magnitude of poverty reduction from 1988 to 1995. Riskin and Li (2001) report that using NBS's own income definition and poverty line, they estimate a poverty headcount of 9.4, much higher than the official 7.1. Although the mean incomes are the same in their 19-province sample and the NBS's national sample, the only plausible explanation is differences in the distribution of incomes, which should be testable even using grouped data.⁵ Khan and Riskin (2001) attribute their surprising result to their refined income measures. However, if rental income from owner-occupied housing is rapidly increasing as a share of total income, including this part of income without adjusting the non-food expenditure shares in the two survey years may bias the change in poverty downward. In any case, what is striking is the rapid fall in poverty in the mid-1990s reported by World Bank (2001). Using the income data, the Bank estimates that the poverty headcount fell from 27.1 to 10.8 in five years, a reduction as impressive as that of the early 1980s. At least for the period after 1995, this is not contradicted by alternative estimates, and deserves further confirmation.

Table 4[PDF: 375kb] | 60 pages]

Nutritional Outcomes Measures and other indicators

The poverty headcount is conceptually a nutrition-based standard of welfare, since the poverty line is constructed to reflect the income necessary to purchase a food bundle that provides a minimum acceptable number of calories per day. One way to validate official poverty statistics is to look directly at nutritional outcomes in the population. Park and Wang (2001) summarized available evidence on caloric intake and stunting in children: Zhu (2001) analyzes 1995 rural household data from 19 provinces and finds that 17 percent of the rural population had caloric intake below 2100 kcals and 28 percent had caloric intake below 2400 kcals, the caloric standard used in constructing the official poverty line. She also finds that the prevalence of inadequate calorie consumption is only weakly correlated with income, casting doubt on exclusive use of income as a poverty indicator. Using aggregate production, trade, stock, and demographic data, and a minimum energy requirement of 1920 kcals, the FAO estimated that the share of PRC’s population with insufficient calorie intake fell from 30 percent in 1979-1981 to 17 percent in 1990-1992 to 11 percent in 1996-1998. A common indicator of long-term nutrition is the prevalence of stunting in children. A national survey by WHO/UNICEF in 1992 found a stunting rate in children of 31.4 percent (FAO, 2000). A series of national surveys conducted by the Ministry of Health found stunting rates of 41.4 percent in 1990, 39.1 percent in 1995, and 22.6 percent in 1997.⁶ In officially designated poor counties, the stunting rate is much higher. The 1995 MOH survey found a stunting rate of 43 percent in poor counties and the China Rural Poverty Survey directed by the author found a stunting rate of 46.1 percent among children in 6 poor counties. These stunting rates compare with the following estimates in different parts of the developing world in 1995: 36 percent in all developing countries, 36.5 percent in Africa, 38.8 percent in Asia, and 12.6 percent in Latin America (ACC/SCN, 2000). These statistics suggest high rates of undernutrition in rural PRC. They also suggest little progress in poverty reduction in the early 1990s but substantial progress beginning in the mid-1990s. This pattern is consistent with the poverty headcount estimates.

Health indicators published by the Ministry of Health show steady progress in the quality of life PRC had made in the 1990s. Infant mortality rate dropped from 58 per 1000 live births in 1991 to 37 in 2000 in rural areas, and from 17 to 12 in urban areas. Under 5-year mortality rate reduced from 71 to 46 in rural areas and 21 to 14 in urban areas in the same time period. Maternal mortality rate decreased from 100 per 10,000 live births to 70 in rural areas and 46 to 29 in urban areas. Interestingly, the trends of these three indicators suggest remarkable progress in early 1990s but little progress in late 1990s, contradicting with the nutritional findings above.

Table 5[PDF: 375kb] | 60 pages]

Chronic and transitory poverty

PRC's official poverty estimates are based on one year of data, Thus on distinguishing has been made between chronic and transitory poverty. Using panel data for households in four provinces from 1985-1990, Jalan and Ravallion (1998a) find that the share of the poor who are not chronically poor vary from 30 to 46 percent. McCulloch and Calandrino (2001) find that in 1991 and 1995, 57 and 46 percent of the poor in Sichuan were experiencing transitory poverty. Using the data from China Rural poverty Survey in six poor counties for year 1997 and 2000, Wang and Li (2003) find that 31 percent of the poor are transitory. It is worth bearing in mind that much of measured transient poverty could be the result of measurement error. Nonetheless, whether or not that be the case, if one's goal is to measure chronic poverty, annual poverty headcounts are likely to overstate the extent of such poverty.

Regional disparity in rural poverty

PRC is a large county with wide range of differences in term of resource endorsement, climate, population, and economic and social development. Rural poverty is to a large extent a regional phenomenon. Extensive research has shown that PRC's poverty-stricken population is mainly concentrated in the southwestern, northwestern and central mountainous areas. The problem of poverty is especially remarkable in the areas inhabited by minority nationalities. Based on the official poverty line and different income data in different provinces collected from the household survey, the NBS estimated the poor population in each provinces and autonomous regions. I then calculated the provincial incidence of poverty and the proportion of the provincial poverty-stricken population in the national total for 1985, 1993 and 2001 (Table 6). These three years represent the establishment of PRC's rural poverty reduction programs, the launch of "the Eight-Seven Poverty reduction Plan" and the formulation of the new poverty reduction policy for next ten years respectively. In 1985, the incidence of poverty averaged 14.81% on the national level, and the provinces and autonomous regions having a poverty incidence higher than 14.81% were Henan, Shaanxi, Ningxia, Gansu, Guangxi, Sichuan, Guizhou and Yunnan, four located in the Southwest, three in the Northwest and one in the central. The provinces and regions whose poverty-stricken populations each accounted for more than 5% of the national total were Henan, Shaanxi, Gansu, Guangxi, Sichuan, Guizhou and Yunnan. In 1993, when PRC began to carry out its "Eight-Seven Poverty Alleviation Plan," the total poverty stricken population was 80 million and the national incidence of poverty was 8.83%, and the provinces or regions having a poverty incidence higher than the national average were Hebei, Henan, Shanxi, Inner Mongolia, Shaanxi, Ningxia, Gansu, Qinghai, Xinjiang, Sichuan, Guizhou and Yunnan, seven in the Northwest, three in the Southwest and two in the central. The provinces or regions whose poverty-stricken population accounted for more than 5% of the national total were Hebei, Henan, Shandong, Shaanxi, Gansu, Anhui, Sichuan, Guizhou and Yunnan. In 2001, the national incidence of rural poverty were 3.2%, and the provinces or regions having a poverty incidence above the national average were Liaoning, Heilongjiang, Inner Mongolia, Shanxi, Shaanxi, Gansu, Qinghai, Xinjiang, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan and Tibet. Poverty incidence in all provinces and autonomous regions in the Northwest and Southwest were higher than the national average, indicating that poverty was further concentrated in remote western regions. The provinces or regions whose poverty-stricken population accounted for more than 5% of the national total in 2001 were Henan, Inner Mongolia, Shanxi, Shaanxi, Gansu, Sichuan, Guizhou and Yunnan.

Based on the data from these three years, we can roughly conclude that the provinces or autonomous regions suffering from comparatively serious poverty are Henan, Inner Mongolia, Shanxi, Shaanxi, Gansu, Ningxia, Qinghai, Xinjiang, Guangxi, Sichuan, Guizhou and Yunnan while the provinces or regions having a comparatively large poverty-stricken population are mainly Henan, Shaanxi, Gansu, Guangxi, Sichuan, Guizhou and Yunnan. All are in the western region except Henan.

Table 6[PDF: 375kb] | 60 pages]

Vulnerability of the poor and its main causes

Since PRC's rural poor population are mainly concentrated in remote mountainous areas with fragile ecological system, harsh climate and under-development of infrastructure and social services, the poor are vulnerable to a member of internal and external shocks. High percentage of transitory poverty indicate that households around the poverty line frequently get into and escape from poverty due to changes within and outside their family. Poor households in poor areas commonly suffer from: (1) damage to the natural environment, water losses and soil erosion; (2) per capita deficiency in and poor quality of agricultural resources (infertile land and shortage of per capita arable land in southwestern rocky mountain areas and karst areas, grave shortage of water resources in northwestern regions); (3) deficiencies in such infrastructure as roads and water conservancy facilities; (4) the lack of such basic social services as education and health and medical care, which exerts an adverse influence on the ability of the local poor to cope with their situation; (5) poor agricultural productivity and underdevelopment of markets and non-agricultural industries, all caused by the above-mentioned factors.

Recent research has shown that there are notable differences between impoverished families and non-impoverished families in terms of demographic characteristics, resources, assets and ability: 1 The low-income rural household has a relatively large family size, but with relatively fewer labors, due to a youthful demographic structure; 2 basic indicators reflecting labor quality and ability (e.g., the adults' average educationallevel,the householder's educational level, the family members' non-agricultural job experience and the proportion of the households whose members have a special skill) is in direct ratio to the per capita income level, and thus there is a large difference in operation and management capacity between the low-income households and the high-income households; 3 from the perspective of owned land resources, the total cultivated area and per capita cultivated area of the low-income households are larger than those of the high-income households. On the other hand, the more other kinds of land (woodland, orchard, fish pond, etc.) a household has, the higher its income; 4 household family assets are in direct ratio to the per capita income level. Econometric analysis has also shown that a rural household’s per capita net income mainly depends on family size (negative influence), labor, the householder's educational level, its members' non-agricultural job experience, the quality of the cultivated land and the amount of productive assets (Li and Wang, 1999, Wang and Li, 2003).

Reasons for rural poverty reduction

PRC's success in rural poverty reduction has gained worldwide reorganization and greatly helped the whole international community achieve its MDG to halving the total poor population by 2015. The achievement of rural poverty reduction has been brought about by both fast economic growth and targeted poverty investments.

Economic and income growth: Overall economic growth explains much of PRC's record of success in poverty reduction since 1978. PRC‘s rapidest economic growth began in 1978 when rural economic reform began. GDP per capita increased from 379 yuan in 1978 to 2359 yuan⁷ in 2002 at constant price, with an annual increase rate of 8.7%. Though not as fast as the secondary and tertiary sectors, agricultural production also increase considerably in the past two decades. GDP from agricultural sector per rural population reached 594 yuan in 2002, which was 4.6 times as high as in 1978, increasing at an annual rate of 6.6%. In addition to agriculture, the rapid development of township and village enterprises (TVE) in rural areas in 1980s and early 1990s and the increasing labor migration from rural to urban areas in recent years has also contributed greatly to the income growth of the rural residents. Household net income per capita increased from 134 in 1978 to 586 in 2002 at constant prices, increasing 6.3% annually (Figure 1).

Simple comparison between the rates of economic growth, especially agricultural growth, and poverty reduction has shown the same downward and upward movements over the years. In their joint report on China Overcoming Rural Poverty, the Leading Group for Poverty Reduction, UNDP and the World Bank also revealed that both over time and across provinces, growth in per capita GDP had been closed associated with the pace of poverty reduction in 1990s and that the impact of aggregate growth on poverty has been substantially influenced by the regional and sectoral composition of that growth (World Bank, 2001).

Targeted poverty reduction programs: In 1986, Chinese government put forward a major rural poverty reduction initiative with the objectives of further stimulating the economic development in poor areas and lift the remaining poor in rural areas out of poverty. This initiative was reinforced in 1994 when the Eight-Seven Poverty Reduction Plan was launched⁸. More than 212 billion yuan have been invested in poverty reduction programs by the central government alone since 1986. All those investments together with other measures have helped the income growth and infrastructure improvements in term of road, drinking water and irrigation facilities, and school and health facilities (will be discussed in more detail below).

Urban poverty

Urban poverty was not an issue for the Chinese government till middle 1990, because urban residents were covered by a wide range of welfare programs from the government or state owned enterprises. Thus, the Chinese government has treated poverty exclusively as a rural problem. To date the government has released no official poverty lines or poverty counts for urban populations. State guarantees of jobs, pensions, housing, and health care for all urban workers under socialism, along with a strict residence permit system, created a large urban-rural income gap that has not been reversed by market reforms. Early estimates of urban poverty by the World Bank (1992) found insignificant poverty incidence through 1990. However, since the mid-1990s, restructuring of state-owned enterprises and substantial layoffs of workers has created significant dislocation for many workers. Growing urban poverty thus has become a very real prospect that will demand monitoring and policy response from government. Using grouped income data, Khan (1996) estimates that the urban poverty headcount fell from 20 percent in 1981 to 13 percent in 1985 to only five percent in 1991. Khan and Riskin (2001) estimate an urban poverty rate of 6.8 percent in 1988 and 8.0 percent in 1995. Using urban household survey data collected by NBS and one dollar one day poverty line, the World Bank estimated that urban poverty headcount rate for all estimated years in 1990s were equal to or below 1% either measured with income data or consumption data (table 7). Compared with rural poverty incidence which is over 10% for all years, urban poverty is much a less problem in PRC. However, Khan and Riskin (2001) argue that the World Bank’s urban poverty line is too small a percentage (23 percent) of average income to be realistic. One point worth making is that many of the potential biases in constructing rural poverty lines and poverty counts also characterize urban poverty statistics. Valuation of non-wage benefits in measuring income is a particularly difficult challenge.

Table 7 Recent Trends in urban Poverty

Download this Discussion Paper [ PDF 375KB| 60 pages ].

Comment(s)

There are [0] comment(s) for this entry. Post a comment.