Information about Asia and the Pacific Asia y el Pacífico

Chapter 2. Achieving Pro-Poor Growth in Cambodia

Sumio Ishikawa, Sibel Beadle, Damien Eastman, Srobona Mitra, Alejandro Lopez Mejia, Wafa Abdelati, Koji Nakamura, Il Lee, Sònia Muñoz, Robert Hagemann, David Coe, and Nadia Rendak
Published Date:
February 2006
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details
Sònia Muñoz

Growth has not been pro-poor in rural areas where the overwhelming majority of the poor live, and the poorest 50 percent of the population would be the main beneficiaries of an increase in rural incomes. Using household data for 1993/94 and 1999, this chapter aims at (1) analyzing whether the strong growth in Cambodia has been pro-poor, and (2) identifying the factors that can improve the “anti-poverty effectiveness” of growth in Cambodia. Section A presents the evolution of poverty in Cambodia. Section B analyzes the impact of growth on poverty. Section C discusses the two main factors that have reduced the benefit of growth to the poor, and, in Section D, poverty reduction measures are suggested.

A. Stylized Facts

Cambodia’s troubled history exacerbated poverty and perpetuated economic inequities. Decades of destructive conflict, civil war, and economic, political, and social instability have contributed to the widespread poverty that currently exists in the country, especially among rural dwellers. The conflict resulted in the destruction of infrastructure, human capital, and institutions, as well as a large proportion of Cambodians being displaced, maimed, orphaned, or widowed. Not surprisingly, these conditions created deep poverty, and the aftermath has been accompanied by widespread economic and social inequities.

The proportion of the population classified as poor declined by only 3 percent between 1994 and 1999 despite strong growth, and is suspected to have increased since then. Cambodia’s economy grew at an average of 6 percent during 1994–2002, while the population with income below half a dollar a day, measured by the head count ratio, fell from 39 to 36 percent. The modest decline in poverty is corroborated by the slow increase in real per capita private consumption. Since 2000, real private consumption per capita has actually declined, implying a likely rise in poverty (Figure 2.1).

Figure 2.1.Real Private Consumption and Poverty

Neighboring countries’ starting positions were worse, but their poverty has declined much faster than Cambodia’s. For example, the poverty ratio in Lao P.D.R. declined from 45 percent to 39 percent between 1993 and 1998. Vietnam has been much more successful, reducing the very high initial ratio of 58 percent in 1993 to 38 percent in 1998, and to 29 percent in 2002 (Figure 2.2).

Figure 2.2.Percentage of Population Living Below the Poverty Line

(In percent)

B. Analysis of the Poverty Impact of Growth

To assess the impact of growth on poverty, we use household level data from the Cambodia Socio-Economic Survey (CSES). The CSES collects expenditure data from roughly 6,000 households. This analysis uses data for 1993/94 and 1999, the first and last surveys available, and the poverty line for rural areas, Phnom Penh, and other urban areas.1 Poverty measures based in 1993/94 were derived by deflating expenditures in each year. We use a head count ratio of 43 percent for rural households and 25 percent for urban households. The calculations for rural and urban households are done separately, since there are striking differences between the rural and urban sectors in Cambodia.

We use a measure of the growth rate consistent with the Watts index for the level of poverty developed by Ravallion and Chen (2003). The Watts index (Wt) as a measure of poverty is defined as the mean growth rate of the poor:

where yt(p) is the quantile function (obtained by inverting the cumulative distribution function of expenditure p=Ft(y) at the pth quantile) and z is the poverty line.

Equation (1) can be written as follows:


is the mean of log censored expenditures, where the censored expenditure is min[yt (p), z], that is, actual expenditure when located below the poverty line, and the poverty line itself otherwise; and yt* is a stable monotonic decreasing function of the actual value of the Watts index, that is, yt* is the exact money metric of the Watts index. The growth rate in yt* is the aggregate growth rate in the expenditures of the poor, and Ht = Ft(z) is the head count index.

Differentiating equation (3) with respect to time, we get

This is the measure of the growth rate consistent with the Watts index for the level of poverty.

Growth incidence curves (GIC) are used to assess whether recent growth in Cambodia has been pro-poor. Following Ravallion and Chen’s (2003) methodology, the GIC is defined as follows:

The equation shows how the growth rate varies by percentile of the distribution ranked by y. By normalizing equation (4) by the head count index, one obtains the mean growth rate of the poor as follows:

Assuming that all expenditure levels grow at the same rate (leaving distribution unchanged), equation (6) collapses to the growth rate of the mean expenditure or the ordinary rate of growth, γt, and the change in the Watt index (–dW*t/dt) equals γtHt (from equation (4)). The GIC plots the cumulative share of the population (depicted on the x-axis) against the growth rate of expenditure in the pth percentile (depicted on the y-axis) between two periods.

The rate of pro-poor growth, given by equation (5), can be rewritten as

and is defined as the growth rate of the mean (of daily expenditure per person) times the ratio of the actual change in poverty to the change that would have been observed under distribution neutrality (i.e., growth that would have impacted each percentile equally). If the distributional shifts in expenditure favor the poor, then the rate of pro-poor growth exceeds the rate of growth in the mean and the growth benefits the poor more than the average population, and vice versa.

Between 1994 and 1999, economic growth in urban areas appears to have been pro-poor. The rates of pro-poor growth exceeded the growth rate of the mean expenditure, suggesting that economic growth was accompanied by falling inequality. The highest growth rates were observed at around the 30th percentile (Figure 2.3).

Figure 2.3.Growth Incidence Curve: Urban

Sources: Cambodia, National Institute of Statistics; and IMF staff estimates.

By contrast, growth in rural areas was strongly anti-poor. Between 1994 and 1999, there was a distributional shift unfavorable to the poor, since the rate of pro-poor growth was appreciably lower than the rate of growth in the mean. The 20 percent poorest households experienced a growth rate that was not relatively favorable to them, resulting in increased poverty in this group. The growth rate tends to rise along the distribution, slowing around the seventh decile and peaking at the high end (Figure 2.4).

Figure 2.4.Growth Incidence Curve: Rural

Sources: Cambodia, National Institute of Statistics; and IMF staff estimates.

The distributional impact in the rural areas is magnified when viewed in the context of the overall economy. The growth rate in the mean in the urban areas was higher than in that of the rural areas by 1 percent annually. At the same time, as noted earlier, growth in the urban areas was pro-poor while it was anti-poor in the rural areas. Thus, income disparity between the poor in the rural areas and the rich in the urban areas has widened substantially.

C. Factors Affecting Pro-Poor Growth

While economic growth is the basic vehicle for reducing poverty, the extent to which the poor benefit from overall growth varies among countries depending on each country’s income and asset distribution. Ravallion (2004) emphasizes that the initial degree of inequality as well as its evolution are the two factors that make growth more or less pro-poor. Unequal access by the poor to physical assets, infrastructure, and social services makes it harder for them to partake in the opportunities afforded by the overall economic growth. Moreover, recent studies show that the sectoral structure of growth influences the effect that growth has on poverty, and emphasize that rural and agricultural growth have direct effects on poverty alleviation.2

Initial Conditions

The reduced economic opportunities in the rural areas with respect to limited access to land, infrastructure, and financial resource assets mitigated poverty alleviation. Cambodia has a highly unequal distribution of income, caused to a substantial extent by highly unequal land ownership (Figure 2.5). Most land in the country is not yet registered, and only 10 percent of farmers have formal title to their farming land. The majority of the land is suspected to be in the hands of a few powerful groups. At the more aggregate level, demarcations between land for different uses—forests, agriculture, urban areas, and so forth—have yet to be made, complicating and delaying any land redistribution initiatives. Furthermore, inadequate infrastructure has limited farmers’ access to markets.3

Figure 2.5.Agricultural and Nonagricultural GDP Per Capita1

(In U.S. dollars)

Sources: Cambodia, Ministry of Planning and National Institute of Statistics (NIS); and IMF staff estimates.

1 In the absence of population data in these two sectors, for illustrative purposes, agricultural GDP and nonagricultural GDP were divided by total population.

Sectoral Growth Pattern

Recent economic growth has benefited from favorable external factors, notably the bilateral trade agreement with the United States. In particular, the agreement contributed to a strong growth of garment exports and the creation of over 200,000 jobs. In addition, construction activities related partly to large aid inflows, and to a lesser extent tourism, also contributed to buoyant overall GDP growth. However, with the exception of the strong rebound in 2003, agriculture barely kept up with population growth.

Low growth of the agricultural sector has had an adverse impact on the poor. On the one hand, agricultural GDP divided by total population (used as a proxy for per capita) has been falling since 1995, while 80 percent of the poor depend on agriculture for their livelihood. In turn, income per capita in the rural areas started to fall from 1995 onward.

Lack of available land and investment led to substantial underutilization of human resources in the rural areas. Timmer (2003) points out that short work days at wage-paying jobs, disguised unemployment, and long hours spent on low-productivity tasks suggest that marginal productivity of rural labor is very low. In such circumstances, he notes that new resources such as capital to build local irrigation systems or rural roads to allow farmers access to markets, new agricultural technology that raises yields, or higher rural household income enable rural inhabitants to spend and invest in education, further raising their marginal productivity.

Worsening income distribution in rural areas offset the positive effect of overall growth. Table 2.1 shows the change in poverty between the two household surveys undertaken in 1993/94 and 1999 decomposed into three components: the growth component (the difference between the two poverty indices keeping the distribution constant), the redistribution component (the change in poverty if the mean of the two distributions is kept constant), and the residual component (the change in poverty due to interaction of growth and inequality). Supporting the previous results from the GICs, worsening redistribution almost offset the positive effect of economic growth in rural areas. However, redistribution alleviated poverty in urban areas and it was quantitatively more important than growth.

Table 2.1.Growth and Inequality Poverty Decomposition(In percent)
Poverty rate in 1993/940.4270.246
Poverty rate in 19990.3890.184
Change in poverty:–0.038–0.062
Due to:
Growth component–0.339–0.025
Redistributional component0.301–0.037
Residual component00
Source: Author’s calculations based on Ravallion and Datt (1992).
Source: Author’s calculations based on Ravallion and Datt (1992).

D. Suggested Measures for Poverty Reduction

Countries that foster higher farm incomes and encourage rural investment benefit from higher total factor productivity in addition to the higher rural output itself. Timmer (2002) and Mellor (1999) argue that increased farm production leads to higher employment and lower basic food prices, both of which reduce poverty.4 Furthermore, increased farm incomes stimulate demand for goods and services in the rural areas, provide food, and can generate savings that contribute to industrialization.

Land reform is a key measure that will allow the poor to benefit from higher returns in agriculture. Agricultural growth will not reduce poverty significantly if increased farm income accrues to wealthy people who tend to spend on imports or capital-intensive goods and services. Besley and Burgess (2000) analyze the impact of land reform on rural poverty and growth by coding land reform legislation amendments of India’s states between the 1950s and 1992. They find that poverty, as measured by the poverty gap and the head count ratio, was reduced as a result of land reform achieved during the previous four years. However, they also find that this poverty reduction may have come at the cost of lower agricultural growth.

Adequate rural infrastructure is critical to profitable farming, and, hence, to poverty alleviation. Public provision of rural infrastructure such as roads to markets, market centers themselves, communication networks, and air and sea port facilities help farmers with marketing surpluses. Timmer (2003) notes that the effects of higher agricultural productivity also spread to subsistence-oriented farmers, especially if rural infrastructure is constructed by the poor themselves through labor-intensive public works programs.

Based on different assumptions about growth of agriculture and the garment sector, it can be shown how different sectoral growth could benefit the poorest 50 percent of the population. For the purpose of this analysis, households sampled in the 1999 survey are divided into deciles of equal size, from the group with the lowest consumption expenditure (labeled D1) to the group with the highest (D10). As the second column in Table 2.2 shows, the lowest decile D1 accounted for less than 3 percent of total expenditure recorded in the survey; the highest group D10 accounted for 35 percent. Growth rates of 3 and 6 percent are assumed depending on whether agricultural productivity improves along with different scenarios for urban growth. Specifically, the subsequent columns show the result of the simulations in terms of percentage change in expenditure for each decile—given the particular spending pattern of each of the households in each group.

Table 2.2.Simulation Results
DecilePercentage of ExpenditureRural Incomes Increase by 6 Percent and Urban Incomes by 3 PercentRural Incomes Increase by 3 percent and Urban Incomes by 6 percentRural Incomes Increase by 6 Percent and Urban Incomes Do Not IncreaseRural Incomes Increase by 3 Percent and Urban Incomes Do Not IncreaseRural Incomes Do Not Increase and Urban Incomes Increase by 3 Percent
Expenditure (percent change)
Source: Author’s calculations based on Cambodia Socio-Economic Survey.
Source: Author’s calculations based on Cambodia Socio-Economic Survey.

The simulations show the importance of rural income growth for reducing poverty. More rapid growth of rural incomes allows a faster growth in expenditure of the lower deciles. Moreover, higher rural incomes allow farmers to invest in farm and human capital leading to further poverty alleviation.

E. Conclusion

The poor in rural areas have not benefited from economic growth, while growth in urban areas has been clearly pro-poor. Highly unequal distribution of income, and asset inequality, including limited access to land, infrastructure, and financial resource assets, have prevented growth in rural areas. By contrast, the redistributional effect of growth in urban areas has helped to improve the welfare of the poor in the cities.

Investment that mobilizes underutilized resources, or that provides funds to increase human and physical capital among the rural population, will have high returns for the poor. A new growth strategy that alters investment priorities in favor of rural growth, like those pursued in Indonesia after 1966, China after 1978, and Vietnam after 1989, will improve factor productivity because of improved resource allocation. China’s strategy was to use world markets to access basic food staples and keep food costs low to provide a competitive advantage to its labor-intensive industries and producers of high-value agricultural commodities. Low grain prices can encourage livestock production and small and medium-sized enterprise activities in rural areas, and allow farmers to specialize in higher-value products (Timmer, 2003). Consequently, a strategy that raises the productivity of staples and uses these low-cost products to diversify into high value-added agricultural products will generate pro-poor growth.

A simple simulation has illustrated the importance of rural growth to the poor. A successful structural transformation of the agricultural sector could raise rural wages. The emphasis should turn to land reform, diversification into crops and livestock, and access to supply chains.


A poverty line is the line below which a given population is believed to live in poverty. It is a line taken to imply an income that is adequate for a person to consume a food basket that provides at least 2,100 calories of energy per day with a small allowance for nonfood items such as shelter and clothing.


Empirical evidence linking measures of poverty to agricultural output can be found in Ravallion and Datt (1996 and 1998) and Timmer (2002).


Rural roads, with only 16 percent of the total road network paved, and the rail system are in poor condition.


Lower staple food prices can stimulate livestock production and provide local markets with high-quality animal protein. In addition, the low cost of these staple foods could promote agricultural diversification into other crops such as fruits and vegetables with better demand prospects. Finally, small and medium-sized enterprises in rural areas could benefit from low nominal wages, made possible by low basic food prices, and speed the absorption of surplus labor.

    Other Resources Citing This Publication