ACS4545 A WORLD BANK COUNTRY STUDY Where Have All The Poor Gone? Cambodia Poverty Assessment 2013 Second Edition THE WORLD BANK April 2014 CURRENCY and EXCHANGE RATE EQUIVALENTS Currency Unit = Cambodian Riel (CR) Exchange rate average 2004: CR 3,906 = US$1 Exchange rate average 2011: CR 4,066 = US$1 Exchange rate April 19, 2013: CR 3,995 = US$1 FISCAL YEAR January 1 – December 31 Regional Vice President, EAPVP : Axel van Trotsenburg Country Director, EACTF : Annette Dixon (until July 2013) Ulrich Zachau (from October 2013) Sector Director, EASPR : Sudhir Shetty Country Manager, EACSF : Alassane Sow Sector Manager, EASP3 : Mathew Verghis Task Team Leader, EASPT : Carlos Sobrado Co-Task Team Leader, EASPT : Samsen Neak Where Have All The Poor Gone? Cambodia Poverty Assessment 2013 THE WORLD BANK Washington, D.C. Contents Acknowledgements.......................................................................................................................................... IX Preface.................................................................................................................................................................... XI Executive Summary.......................................................................................................................................... XIII Macro Economy................................................................................................................................................. XIII Consumption, Well-being, Inequality, and Poverty.............................................................................. XIV Drivers of Poverty Reduction........................................................................................................................ XVII Improving Human Development................................................................................................................ XVIII Lessons.................................................................................................................................................................. XX Policy Recommendations............................................................................................................................... XXI Chapter 1 : What Happened With The Poor?....................................................................................... 1 Macroeconomic Conditions....................................................................................................................... 1 Poverty and Well-being................................................................................................................................ 5 Socio-Economic Surveys............................................................................................................................. 6 Measuring Poverty........................................................................................................................................ 6 Poverty Levels and Trends.......................................................................................................................... 9 Other Well-being Indicators...................................................................................................................... 10 Housing and Durables................................................................................................................................. 12 Inequality, Distribution, and Growth...................................................................................................... 13 Vulnerability.................................................................................................................................................... 17 Poor Households and the Correlates of Poverty over Time........................................................... 18 V Chapter 2 : Sources of Household Growth........................................................................................... 25 Urban labor........................................................................................................................................................ 25 Labor Market and Poverty Reduction Trends...................................................................................... 25 Employment Composition: the Role of Gender and Education and Better Jobs................... 27 The Textile Sector.......................................................................................................................................... 32 Policy Recommendations........................................................................................................................... 35 The Role of the Rural Sector...................................................................................................................... 36 Rural Population............................................................................................................................................ 36 Evolution of Rural Poverty 2004 to 2011............................................................................................... 37 Consumption.................................................................................................................................................. 38 Housing Conditions and Consumer Durables in Rural Households........................................... 39 Rural Income................................................................................................................................................... 39 Income from Agricultural Crops............................................................................................................... 39 Income from Wage Labor........................................................................................................................... 42 Income from Non-Farm Self-Employment........................................................................................... 44 Income from Remittances and Transfers.............................................................................................. 44 Perceptions from Rural Households....................................................................................................... 44 Characteristics of the Rice Sector............................................................................................................ 45 Policies of the Royal Government of Cambodia................................................................................. 45 Remaining Challenges................................................................................................................................. 46 Drivers of Poverty Reduction.................................................................................................................... 47 Urban Households 2004-2007.................................................................................................................. 48 Rural Households Income 2004 to 2009............................................................................................... 48 Drivers of Poverty Reduction in 2004-2011......................................................................................... 49 Policy Recommendations........................................................................................................................... 50 Chapter 3 : Reaching The Poor and Vulnerable................................................................................. 53 Education............................................................................................................................................................ 53 Education Sector Background.................................................................................................................. 53 Private Sector Education in Cambodia.................................................................................................. 55 Education Outcomes.................................................................................................................................... 55 Education Characteristics........................................................................................................................... 61 Education Quality and Finance................................................................................................................ 65 Policy Recommendations........................................................................................................................... 66 Health..................................................................................................................................................................... 68 Child Mortality................................................................................................................................................ 71 Equity in the Provision and Use of Health Services........................................................................... 74 Equity of Health Spending......................................................................................................................... 77 Maternal and Child Nutrition.................................................................................................................... 79 Conclusions and Policy Recommendations......................................................................................... 82 VI Annex 1 : New Poverty Estimates in Cambodia 2009..................................................................... 85 New Consumption Aggregate....................................................................................................................... 85 New Poverty Lines.............................................................................................................................................. 91 2009 Poverty Results......................................................................................................................................... 97 Annex 2 : New Government and World Bank Poverty estimates 2004-2011.................... 105 Poverty Rates (Headcount Rate).................................................................................................................. 105 Consumption and Poverty Lines................................................................................................................. 106 New Royal Government of Cambodia and World Bank Poverty Comparisons........................... 108 Annex 3 : Simulation of Drivers of Poverty Reduction.................................................................. 111 2007-2009............................................................................................................................................................ 111 2004-2007............................................................................................................................................................ 112 2004-2009............................................................................................................................................................ 112 Bibliography.......................................................................................................................................................... 115 VII Acknowledgements This report was possible because of Ozaltin. Enrique Aldaz-Carroll provided overall fruitful cooperation not only from the Royal technical advice for the document and Government of Cambodia but also from other supported the interaction with the Royal development organizations, as well as assis- Government of Cambodia, international tance and hard work from our colleagues at institutions and other stakeholders. Lyden the World Bank. Kong, Linna Ky, and Phalline Sath provided useful logistic and administrative support. The report was prepared by a team led by The report was edited by Jeremy Gerardi Carlos Sobrado and Samsen Neak. Chapter 1 Mazzola. was written by Sodeth Ly and Enrique Aldaz- Carroll (Macroeconomic Conditions) and Carlos Useful guidance and comments were also Sobrado (Poverty and Well-being). Chapter 2 provided by other colleagues at the World was written by Elisa Gamberoni and Francisco Bank: Peter Lanjouw (reviewing the poverty Arias-Vazquez (Urban Labor), Samsen Neak methodology), as well as Andrew Mason and (The Role of Rural Sector), and Carlos Sobrado Jaime Saavedra (organizing and chairing a (Drivers of Poverty Reduction). Chapter 3 was workshop to review the initial poverty results). written by Tsuyoshi Fukao, Carlos Sobrado, The team worked under the guidance of and Simeth Beng (Education) and Timothy Annette Dixon (Country Director), Alassane Johnston and Miguel San Joaquin (Health). Sow (Country Manager), Sudhir Shetty Lucilla Bruni and Richard de Groot composed (Sector Director), and Mathew Verghis (Sector the portions on social protection and their Manager). The peer reviewers—Kenneth recommendations in Chapter 3. The report Simler and Salman Zaidi—provided extremely also benefited from contributions from Puja valuable feedback and comments. Dutta, Mariana Infante, Huot Chea, and Emre IX At the concept paper stage, the team (Secretary General, Ministry of Economy had opportunities for discussions with His and Finance), His Excellency Phan Phalla Excellency Hang Chuon Naron (Secretary of (Deputy Secretary General, Supreme National State, Ministry of Economy and Finance and Economic Council), His Excellency Tuon Permanent Vice-Chairman, Supreme National Thavrak (Under Secretary of State, Ministry Economic Council), His Excellency Phan Phalla of Planning) were sincerely appreciated. The (Deputy Secretary General, Supreme National authors are also grateful for the help from the Economic Council), His Excellency Chea Sophara Statistics Sweden team and from the members (Minister, Ministry of Rural Development), His of the Government’s Inter-ministerial Working Excellency Sao Chivoan (Secretary of State, Group on Redefining the Poverty Line, as well Ministry of Rural Development), His Excellency as Mr. Sarthi Acharya. The team would like to Theng Chan-Sangvar (Under Secretary of State, give special recognition to His Excellency San Ministry of Rural Development), Mr. Chrey Sy Than (Under Secretary of State, Ministry of Pom (Deputy Director of Rural Healthcare Planning) for promoting the advancement of Department, Ministry of Rural Development), statistics in Cambodia and for providing access His Excellency Ngan Chamroeun (Deputy Head, to the Cambodia Socio-Economic Survey, and NCDD Secretariat), His Excellency Pit Chamnan to His Excellency Hang Chuon Naron for his (Secretary of State, Ministry of Education Youth continual participation from the concept note and Sport), and Dr. Lo Veasnakiry (Director of to the draft report. Planning and Health Information Department, Ministry of Health). Funding from the United Kingdom’s Department for International Development The team benefited greatly from comments (DFID) through the Trust Fund for Poverty and interactions during the presentation on Reduction and Governance Analysis and poverty estimate results held at the Ministry of Coordination Support program (PRAGACS) in Economy and Finance. The participation and Cambodia is also gratefully acknowledged. assistance of His Excellency Vongsey Vissoth X Preface It is not possible to comprehend the soci- because of the various political groups backed oeconomic conditions of Cambodia today by their own military forces. Finally, in 1999, without mentioning the impact of not only the use of military force to challenge political the Vietnam War, but also the country’s rivals seemed to conclude, marking an end to nearly thirty years of concurrent internal con- decades of conflicts and wars. flicts and civil wars from 1970 to 1999. The war in Vietnam started in late 1955, eventu- It is difficult to imagine the extent ally spreading to Cambodia and Lao PDR— of these conflicts’ impacts in Cambodia. especially along the heavily bombed border Comprehensive estimates of the death toll with Vietnam, bombed heavily by the United from 1970 to 1987 range from 2.3 to 3.9 States. After a coup d’état in 1970, Cambodia million people, or almost one-third to one-half underwent multiple internal conflicts—most of the 7.1 million Cambodians alive in 1970 notably the revolution led by the Khmer Rouge (Rummel 1994). All urban centers were evac- from 1975-79 characterized by loss of all uated in 1975, and as of 2010, Cambodia has personal property, forced labor and migration, had the second lowest urbanization rate in imprisonment, and executions. An estimated all of Asia. Money was eliminated—as well as two million died—most owing to execution, most administrative records—including land famine and disease. The Vietnam occupation tenancy and birth certificates, which were at the end of 1978 did not stop the fighting. either lost or destroyed. Many of the land Their eventual withdrawal in 1990 led to the issues faced today can be directly traced back Paris Peace Accords signed in October 1991. to the massive displacement of people and the Fighting significantly reduced; nevertheless, loss of proper documentation that took place the 1990s was strife with political instability during this time. XI Fourteen years of peace have dramatically distant memory that belongs to the previous changed the country in almost every aspect, generation with either a limited or no rela- but the remnants of the armed conflicts and tions to their day-to-day life. Young adults in instability are still present in both Cambodian Cambodia are more concerned about their society and its institutions. For example, most future: employment opportunities, new tech- people in positions of power today—in both nology, new families, rice production, health, the private and the public sector—were born education, motorbikes, and so on and less either during or just before the Khmer Rouge about their past. years of 1975-1979, although many either suffered its consequences or participated in Over time, the Royal Government of the various armed conflicts. At a more personal Cambodia has been able to provide the level, in Cambodia today it is not uncommon basic living requirements for its citizens and to find people unsure about their birth date, enabled them to improve their socioeconomic and almost every Cambodian family has close status. With improved conditions and a better relatives—even members of their nuclear sense of security, its people’s expectations family—who were lost during those years. are changing more towards those of other advanced countries than that of other post- Institutions today are the product of a conflict countries. process of reconstruction during the 1990s, a period when the priorities were feeding the The positive changes are also accompanied people, reconstructing basic infrastructure, with new challenges. On one hand, people’s and establishing a minimum level of security expectations are no longer limited to having and the rule of law—all during a time of not food on the table and they are beginning to only political instability but continued civil demand access to better basic services— war. Under such extreme conditions, it is not including the rule of law, transparency from unusual for a country to underplay concepts their elected officials, and especially land such as inclusion, participation, and accounta- tenure security. They now have an earnest bility—even if such concepts would potential- desire to improve their living conditions. ly produce more benefits in the long term— Development also has created new problems; and to concentrate efforts on the more urgent for example, environmental concerns (pol- tasks of developing a functioning, lasting state. lution, land degradation, overuse of natural resources, and the like), the rising costs of The newer generations of Cambodians, living, as well as new health problems such with more distance to the past, are becoming as HIV/AIDS, avian flu, and so on. One of the more important and starting to take their biggest challenges for the Royal Government place in society. As of 2010, over two-thirds of of Cambodia is its ability to adapt to these new Cambodians were born after 1993. For most of conditions and to address problems that did them, the years of conflict are perceived as a not exist in the 1990s. XII Executive Summary Over the seven years from 2004 through conditions create new challenges for the Royal 2011, Cambodian economic growth was tre- Government of Cambodia. Furthermore, most mendous, ranking amid the best in the world. of the improvements in Cambodia originated Moreover, household consumption increased at very low values—thus leaving much work by nearly 40 percent. And this growth was to better the well-being of many Cambodian pro-poor—not only reducing inequality, but households. Outstanding gains have been also proportionally boosting poor people’s con- achieved, but it will take focus and further sumption further and faster than that of the actions to maintain Cambodia’s future growth. non-poor. As a result, the poverty rate dropped Most poverty in Cambodia is found in the coun- from 52.2 to 20.5 percent, surpassing all expec- tryside: about 90 percent of Cambodia’s poor live tations. However, the majority of these people in rural areas. To generate the maximum impact, escaped poverty only slightly: they remain highly government policies should concentrate on the vulnerable—even to small shocks—which could productivity of the rural poor’s major assets: their quickly bring them back into poverty. The main labor and their land. drivers of poverty reduction were high rice prices, higher rice production, higher revenue from Macro Economy non-farm businesses, higher wage rates of rural workers, and growth in salaried jobs in urban The end of internal conflicts in 1999 also areas. Improvements in health and education— marked Cambodia’s integration into inter- as well as government investment in infrastruc- national markets; more notably; it marked ture—provided a favorable environment for the the expansion of the industrial sector, fueled poor, allowing many of them to pull themselves by the apparel industry. With yearly growth out of poverty. Looking forward, some of these rates as high as 30 percent, industrial output drivers of poverty reduction are likely to stall. doubled by 2004 and represented 27 percent Coupled with increased vulnerability, the present of Gross Domestic Product (GDP). At the same XIII time, the agriculture share of GDP decreased the very low initial per capita Gross National from 44 percent in the mid-nineties to less Income (GNI) value in 2004 (US$400), the than 29 percent by 2004. impressive growth rate has not been enough to lift Cambodia from its status as a low income From 2004 to 2011, Cambodia’s per country; in 2011, the GNI remains at US$823, capita GDP (in constant 2000 US$) grew below the US$1,026 per person threshold for 54.5 percent, ranking fifteenth among middle income countries. But it is not that far 174 countries. Cambodia’s remarkable per- away from rising out of low income status— formance was possible because of a combi- sustained growth levels can bring the country nation of factors including sustained peace into middle income status by 2015. (Figure A) and the demographic dividend, free trade and open investment policies, and finally Consumption, Well-being, subsequent macroeconomic stability—all of Inequality, and Poverty which were conducive to increases in tourism, construction, the garment sector, agricul- The Royal Government of Cambodia and ture, and foreign direct investment (FDI). As the World Bank updated the way consump- in much of the world, the economy slowed tion and poverty was measured from 2004 in 2008 and contracted further in 2009 (1.4 to 2011 to better reflect the conditions of percent per capita constant 2000 US$), but Cambodia today. The first methodology to Cambodia was not as hard hit as most indus- measure poverty in Cambodia was developed trialized countries, and it recovered quickly. and applied to the 1993 Cambodia Socio- Economic Survey (CSES), but it is no longer If Cambodia continues to exhibit sus- valid for Cambodia. For example, the con- tained growth rates similar to those of sumption patterns of the poor have changed the last decade, it could become a middle significantly implying that the original poverty income country as early as 2015. Given lines are no longer a good reflection of the Figure A: Cambodia Per Capita GDP Growth Rate (Constant 2000 US$) Per Capita GDP Gowth Rate Source: National Accounts and population projections from National Institute of Statistics XIV poor today. As a result, the government and (quadruple)—all of which reaffirms consump- the World Bank—working independently but tion growth. coordinating actions—estimated poverty using the 2009 CSES. The World Bank team The average household in 2004 is com- benefited from the insights and preferences parable to the poorest household in 2011. made by the government and incorporated The aforementioned access to both housing many of the choices and decisions into its own services and durables by the poorest quintile estimate. The government benefited from the in 2011 is almost the same as the access to experience and previous work carried out by these same housing services and goods by the the World Bank and used some of the same middle quintile in 2004. In other words, these principles and results. Both teams used the indicators show that the poorest households new methodology to estimate poverty for of 2011 are similarly situated to the average 2004 through 2011, and the results were very households of 2004—again, a clear reaffirma- similar, especially from 2009 onwards.1 tion of growth. The average per capita consumption (in Consumption increases were pro-poor: constant 2009 CR) increased 37.8 percent the percentage increase in consumption from CR 6,399 in 2004 to CR 8,815 in 2011, was higher for the poor than for the rich. which was consistent with GDP growth. Although average consumption increased by Over the seven years from 2004, the average 37.8 percent, the increases were higher among per capita consumption has increased 4.7 the poor than for the rich. For households in percent for each year. This total increase of 37.8 the first quintile (the poorest 20 percent of the percent is well below the total per capita GDP population), the increase was 56.5 percent. growth of 54.5 percent. However, the variance The increase was lower for each subsequent is consistent with international experience: its quintile: 51.6 percent for the second quintile, explanation lies in the portion of the economy 45.0 percent for the third quintile, 38.4 percent included in GDP that has a limited impact on for the fourth quintile, and 26.8 percent for household consumption. the fifth quintile (the richest 20 percent of the population). Improved access to services and increased ownership of consumer goods The shared prosperity index improved. confirm consumption growth. The leading The share of consumption from the poorest 40 indicators in housing services and durables percent of the population increased from 20.7 illustrate that the improvements are as big as percent in 2004 to 23.1 percent in 2011. or greater than consumption increases. This is true not only for the following improved After an initial increase, inequality has access to housing services: electricity (triple), decreased every year since 2007. One of sewerage or septic tanks (double),and piped the most surprising results from the analysis water during the dry season (25 percent), but was the reductions in inequality after 2007. also for the following improvements in access Comprehensive measures of inequality like the to durables such as televisions (37 percent), Theil and Gini Index show an initial increase of motorbikes (double), and mobile telephones inequality followed by reductions until 2011. XV The Gini increased from 0.326 in 2004 to 0.374 Combined with pro-poor growth, the in 2007, but it decreased every subsequent consumption increase resulted in poverty year to 0.282 in 2011. decreasing from 53.2 to 20.5 percent. Unlike many other countries where inequali- But the actual gap between the rich and ty increases with growth, the pro-poor growth the poor has increased in absolute terms. in Cambodia enhanced poverty reduction— Because all of the inequality measures are surpassing all expectations and substantially relative, it is possible to reduce inequality—and exceeding the Millennium Development Goal at the same time, to increase the gap between (MDG) poverty target. the poor and the rich. This helps to explain why the perception of increased inequality—based But the majority of households that on absolute differences between the rich and escaped poverty did so by only a small the poor—do not match with the reported margin—they are highly vulnerable to decrease in inequality (based on relative falling back into poverty. For example, in changes). For example, having twice the income 2011 a small shock of CR 1,200 per day (US$0.30 is not that much of a difference if one person or the cost of two small water bottles from a has US$10 and the other has US$20; however, street vendor in Phnom Penh) would cause it is much more noticeable if one person has Cambodia’s poverty rate to double. (Figure B) US$100 and the other person has US$200. The measure of relative inequality is the same in Where did all the poor go? Not very both scenarios. But in the first scenario, the dif- far! Most moved from being poor to being ference is US$10, while the second scenario is vulnerable.2 The tremendous poverty US$100. This is why the measurement of ine- reduction was possible because many of the quality can decrease—even though the actual poor—who were just below the poverty line gap between the rich and the poor increases. in 2004—were able to move just above the poverty line in 2011. Figure B1: National Poverty Rate, Cambodia 2004-2011 Figure B2: Population Shares By Consumption, Cambodia Poverty Line = CR 5,326 Poverty Rate (%) Population % Consumption in 2009 Riels/day (000) Source: World Bank staff estimates based on CSES XVI However, conclusions from other inde- for increased production. Poverty reduction pendent international reports and indi- in rural areas was driven by the substantial cators support these poverty reduction increase in rice prices (24 percent), increased results. The MDG Progress Report ranks rice production (23 percent), better rural Cambodia’s improvements fifth out of 76 poor wages (16 percent), and improved income countries; the UNDP Human Development from non-farm self-employment (19 percent).3 Indicator ranks Cambodia as the country with (Figure C) the best improvement in the region from 2000 through 2010—above countries such as China, Poverty reduction in urban areas was Lao PDR, and Vietnam. driven by increased salaried employment (4 percent). The share of urban workers Drivers of Poverty Reduction engaged in salaried employment increased, reaching over 50 percent by 2011. Salaried The right government policies enhanced workers in urban areas have more years of opportunities for households to pull them- education than other workers and were able selves out of poverty. Government policies to take advantage of this education premium, in basic infrastructure improvement (roads, thereby increasing their income. communication, rural irrigation); in rice pro- duction (no price controls and no taxes); on Women have benefited from the garment minimum wage in apparel manufacturing; and industry, but wage inequalities remain in overall industrial policy provided the right important. Among salaried jobs, the garment environment for the poor to take full advan- industry is one of the main employers. About 85 tage of favorable conditions. percent of workers in the garment industry are female, and the higher wages paid by garment To identify the drivers of poverty reduction, factories have benefited women in Cambodia. a simulation was developed linking income Nevertheless, the overall gender wage gap was increase from several factors to their impact 30 percent—most of it from lesser returns to on poverty. Income and farmers information women with similar endowments to men: edu- derived from the Cambodian Socio Economic cation, age, and so forth. Survey (CSES) was used to compare changes in household behavior from 2004 through 2011. Figure C: Drivers of Poverty Reduction, Cambodia 2004-2011 Official government statistics for rice produc- tion, rice prices, and inflation were also used Non-Farm to estimate the impact of the changes in rice Business Urban Salaries Farm 4% prices. Wage 19% 16% Poverty reduction concentrated in rural Other factors 14% Cambodia (where 90 percent of the poor Rice Production reside) and was driven by rice farming. From 23% 2004 to 2009, the price of rice (in constant Rice Price 24% value) increased by 37.1 percent, boosting farmer’s income and providing incentives XVII Improving Human Development from 2005 to 2010, illustrated by the follow- ing indicators: decreasing maternal mortality Health and education improves people’s rate (from 437 to 288 per 100,000 live births); welfare while sharing important synergies. decreasing mortality rate for 0 -1 month old Healthier people are able to perform better in (from 30 to 25 per 1000 live births); decreas- almost any activity in which they engage. Good ing mortality rate for 1-12 months old (from health not only improves physical efficien- 66 to 45 per 1000 live births); decreasing mor- cy but also enhances concentration and the tality rate for 12- 60 months old (from 37 to 22 ability to process tasks that require planning, per 1000 live births); better measles immuni- organizing, and multitasking. Higher education zation coverage (from 79 to 93 percent); better opens opportunities to engage in salaried jobs DTP immunization coverage (from 82 to 92 that pay an important premium for education; percent); and more births attended by skilled it also increases the ability of self-employed health staff (from 44 to 71 percent). workers to manage their own businesses and to increase the returns to their investment and A noticeable exception is the lack of labor. Moreover, there are important syner- progress in combating malnutrition. gies between health and education from very Although substantial nutrition improve- early in life to later in life: an undernourished ments were observed from 2000 through child may not be able to walk to school and 2005, progress was limited, stalled, or even his ability to learn will be diminished; likewise, reversed from 2005 to 2010. The percent- educated mothers use the knowledge gained age of stunted children 0-60 months old from school to raise healthier children. decreased from 42 to 40 percent only, under- weight children stalled at 38 percent, and Because of better access to schools children with wasting actually increased from and better access to medical facilities and 8 to 11 percent. The results are especially sur- services, key education and health indi- prising given the improvements in consump- cators improved. Improvements from 2004 tion—including food intake. The reasons are through 2011 were observed in many edu- not fully understood, but there is a consen- cation outcome indicators: adults 25 years sus that the solution has to include several and older with primary education (10.0 per- actions targeting issues beyond the amount centage points increase), literacy among the of food consumed. This should include actions youth 15-24 years old (11.1 percentage points addressing the quality of the food, the envi- increase), literacy gender-gap for the youth ronment (open defecation), and hygiene prac- (reduced from 7.4 percent to less than one tices (washing hands and food, and other percent), the net enrollment rate in primary better hygiene practices). This study finds that school (18.2 percentage points increase), stunting is more common among the poor, and the gross enrollment rate in second- among children with mothers of lower educa- ary school (18.4 percentage points increase). tion, and among communities with over half Furthermore, there is no longer any gender of the households defecating in the open. gap among current students enrolled in either primary or secondary school. Health outcomes Health issues are more common among also showed remarkable improvements the poor. The following are some other XVIII health issues that hit the poor hardest: mor- overage is 3.1 years. Therefore, these students tality of 1-12 month old children, access to are finishing primary school at age 15 instead family planning, antenatal care (minimum of at age 12. This overage has two negative four visits), and the presence of skilled birth consequences: First, children are forgoing attendants. Coverage levels of targeted health education at an early age, missing a crucial programs based on Health Equity Funds (HEF) time for cognitive development, which makes and ID-poor cards are pro-poor, but coverage learning more difficult in years that follow. remains low. Among the poor seeking health Second, drop-out rates significantly increase treatment, only 20 percent reported using free with age, especially beginning at age 13. or subsidized treatment. Students entering primary school late create a situation in which more students will leave For the poor, the main source for health the education system earlier—thus having a care is private non-medical providers. The negative impact in secondary enrollment and main providers consulted for health issues retention. by the poor—55.1 percent of the time—are shops and markets selling drugs compared to Both economic reasons and access to only 17.9 percent for the richest Cambodians. secondary education are problematic for These providers are not well regulated: serious school enrollment, retention, and dropouts. problems in the quality of medications and in Secondary education coverage is expanding the quality of advice are reported every day. in Cambodia, but proximity to schools remains a problem. The number of students that make The improvements in education and it to secondary school is greatly reduced by health outcomes have not been enough to late entry into primary school. However, those compensate for the low values in 2004. This students that do enter secondary school report is true of almost every indicator; for example, high levels of dropouts in the initial grades. by 2011 only 29 percent of adults 25 years and The main reason is economic: it is either too older had finished primary school, illiteracy expensive or young people go to work—this is among 15-24 years old was 11.1 percent, the especially true for female students. Household net enrollment rate in primary school was 48.8 chores are reported by less than 14 percent of percent, and the gross enrollment rate in sec- dropouts as the main reason for leaving sec- ondary school was 54.2 percent. All the mor- ondary school (with no difference between tality rates in Cambodia are higher than the young women and young men). average for developing countries of East Asia and the Pacific (about double). And among the Overall education attainment remains poor, all education and health outcomes are significantly lower for the poor, for ethnic lower than the average. minorities, and in rural areas. Most of the problems described above are relevant to all The biggest problem in primary educa- Cambodians, but they are more prevalent tion is late entry into the first grade. Of all among the poor and ethnic minorities. The students entering first grade, 40 percent enter gap between the poor and the rich significant- 2 years late. The problem is then compound- ly broadens for enrollment and dropout rates ed over time with dropouts—by grade six, the in secondary education. XIX Beyond the Health Equity Funds (HEF) their main assets—land and labor in rural and scholarship programs, there are no areas, and labor in urban centers—thereby major government-funded social protec- lifting themselves out of poverty. Better tion programs in Cambodia. Social programs infrastructure and communications, favorable funded and administered by the government economic conditions, and government policies is a new concept in Cambodia. The tradition- allowing for the free movement of commod- al social safety nets came from family, friends, ity prices without tax burdens and business- and faith-based institutions, mostly from friendly industrial policies were also key factors pagodas. Social assistance is mainly food, for in helping the poor. These factors provided the distribution programs, and for assistance after poor with the flexibility to allocate their resourc- natural disasters. es to the most productive activities, as well as to capture an important share of benefits, thereby The National Social Protection Strategy allowing many to escape poverty. (NSPS) has yet to be implemented. In 2011, the government approved the first social pro- However, the key drivers of poverty tection strategy (NSPS), which “…sets the reduction are likely to stall. Increases in framework for sustainable and comprehen- the price of rice had a positive impact— sive social protection for all Cambodians over both in improving households’ welfare and the long term. This includes both contributo- in reducing poverty—but it is not likely to ry and non-contributory schemes.” While the continue. Therefore, public policies should not strategy was an important step to provide the rely on continued increases in rice prices for basic foundations and principles for an inte- further poverty reduction. grated social protection system in Cambodia, the implementation of the strategy remains Cambodia has not created artificial price slow with only limited new initiatives by 2013. control for commodities and other barriers that limit the free flow of resources. This is Lessons the right thing to do. International experience has shown that price control policies not only To reduce poverty, actions should be lessens the free flow of resources to more effi- focused where the poor live and endow- cient uses but also lessens the welfare of the ments of the poor. Two important lessons can people. International experience has proven be learned for poverty reduction in Cambodia that price controls usually distribute benefits from 2004-2011. First, improvements have from one group of people to another group. to reach rural Cambodia to reduce poverty Because most of the poor and vulnerable in (where nine out of ten poor persons live). Cambodia are involved in rice production, rice Second, actions should impact the activities in price intervention policies would only take which the poor are already engaged (includ- away some of the benefits from the group ing agriculture) and increase the returns to of people that need it the most. Moreover, their endowments—land and labor. creating artificial barriers would make it more difficult to freely move or switch activities and From 2004 through 2011, over half of the resources with changes in conditions over poor were able to maximize the return to time. XX Not all poverty reductions are the same. Investing in rural infrastructure needs to Poverty reduction can take many forms continue. The poor will benefit from further depending on income and consumption distri- investment in the upgrading of rural roads, bution. In Cambodia from 2004 through 2011, expanded irrigation facilities, increased access the reduction in poverty—while positive—was to electricity, cleaner water, and improved san- relatively small for the majority of households. itation. These households remain vulnerable to falling back into poverty—even from small shocks. Rural extension programs that promote better practices for improving productivity Implementation of the National Social need to be increased. Rice production needs Protection Strategy (NSPS) requires more to be intensified from one-crop to two-crops support to coordinate actions across gov- annually in areas that currently rely on only ernment agencies. It is important to recog- rain-fed, single season crops. Rice intensifica- nize that a social protection strategy requires tion requires a shift from subsistence farming several ministries and government agencies practices to commercial farming. This intensifi- to cooperate, not only on objectives but the cation should include giving farmers increased strategies to achieve those objectives. The access to improved rice seeds; likewise, it need for stronger support to the particular should promote crop diversification—espe- agency implementing the NSPS is necessary, cially in agro-ecological zones that are better especially at the early stages of social protec- suited for crops besides rice. This would facili- tion in Cambodia. New programs require a tate two important benefits: better productiv- stronger mandate to facilitate participation by ity and better diet diversity. all stakeholders. Activities that create more value added The endowments of the poor and vul- to agricultural products should be support- nerable are not fixed and can be improved ed. Value added includes all activities related over time. Another way to improve welfare for to agricultural products from after actual pro- the poor and vulnerable is to increase endow- duction to packages in shops or markets. ments. This would either create opportunities Activities that enhance the value chain from for increased returns from their existing activi- the farmers’ gate would keep much of the ties or allow them to move to more profitable value of these products within the country, activities. Among the more important endow- instead of allowing neighboring countries to ments for the poor are education and health— grasp most of this added value. This would of which the government is the main service include—but not be limited to—marketing, provider. Specific education and health rec- processing (such as rice milling), packaging, ommendations are put forward in the next and commercialization. In fact, it includes any section on policy recommendations. and all steps from the farmers’ gate to con- sumer’s hands. To achieve this goal, it is neces- Policy Recommendations sary to improve the connectivity among farms and farmers—as well as the rest of the supply Making a Robust Rural Economy chain—by using different means such as pro- moting cooperatives; improving the flow of information; creating credit instruments XXI suitable to small and middle-sized businesses; into the education system. Among rural knowledge sharing among existing, successful areas, education policies should concen- experiences in the country; and the like. trate on primary education—specifically by decreasing late entry and promoting enroll- Policies (namely commodity price ment by the remaining poor and minorities controls and unreasonable taxes) that who have never attended school. To counter distort prices in the agricultural sector these problems, scholarships and school should be avoided. Different to many coun- feeding programs have been found to be tries in the region, Cambodia’s agricultural effective—both internationally as well as in markets have been mostly free of government Cambodia. Therefore their coverage should be intervention, which has had two advantages: expanded. For rural households, access to sec- First, it has allowed farmers to take full advan- ondary education should increase gradually as tage of commodity price increases. Because the demand increases. most of poor people are linked to agricultur- al production, the benefits of increased prices Urban secondary dropout rates can be have reached the poor. Second, it did not reduced with an incentive program, and create artificial barriers or incentives tied to primary early entry for selected groups can specific activities or products, which created be improved. For the urban poor and vulner- opportunities for farmers and their family able, better education would increase pro- members to engage in the most efficient use ductivity in salaried jobs, which would result of their labor. Maintaining the same policies in increased incomes. The main recommen- in the future would allow for not only a better dation is to keep increasing the education flow of resources in and out of the agricultur- level of future workers to increase productiv- al sector, but also improved allocation of the ity, which will improve incomes. Programs like assets of rural households. cash transfers and other targeted schemes are well-suited to improve secondary education in Although agriculture has lifted people Cambodia. out of poverty, it probably cannot be sus- tained alone. In the middle and over the Health Equity Funds (HEF) coverage long run, non-agricultural activities are the and utilization needs to be increased most important factor for sustained economic for the poor and the vulnerable, while growth and well-being improvements in maintaining low user fees for public Cambodia. The country should be ready to health care. A large proportion of HEF step-up efforts to promote industry and members did not use their benefits, especially services sectors in both urban and rural areas. for primary care. The reason for this needs to be better understood. Moreover, over- Continue Investing in Human Development sight mechanisms need to be strengthened to ensure that all eligible poor are not only Rural children need to be brought into covered, but aware of their benefits. A robust the education system earlier; moreover, the impact evaluation study is recommended to remaining poor and minorities that have evaluate utilization changes of HEFs services never been to school need to be brought among the poor. XXII Controls must be established to ensure to facilitate its role as coordinator of several the quality of health care (medicines and government ministries and institutions. Such advice) of private providers outside of the a mandate has to be accompanied with public health system. Strengthening over- increased resources to initiate some of the sight and regulating the private and informal NSPS programs by providing seed money and sector is crucial. Likewise, it is vital to develop by attracting other resources from the govern- mechanisms of accreditation to combat coun- ment, as well as from other stakeholders. terfeit drugs. In the future, it is recommend- ed that the standards of pre-service training Social protection programs should (medicine and nursing degrees) are thoroughly target not only the poor—but the vulner- evaluated. Such an evaluation should start with able as well. The reported poverty reduction an independent needs assessment to identify from 2004-2011 should not be a reason to areas for support and for future investment. reduce coverage of targeted programs. Many of the formally poor remain very vulnerable, A multi-sector approach needs to be and the NSPS should include such households adopted to reduce maternal and child mal- in their planning to prevent them from falling nutrition. Support for community-based back into poverty. programs needs to be increased while the elimination of open defecation, better feeding Use the same labor standards applied in practices, food fortification, and micronutri- the garment factories to other industries ents supplements (targeting the poor and and sectors. Labor laws protecting workers children) requires substantial improvement. and enrollment into the social security system Currently, a large proportion of Cambodia’s are good tools to reduce vulnerability and to children do not receive a minimum accepta- improve household well-being. Building on ble diet, and this has not changed since 2005. the experience it has gained from the garment The health sector needs to strengthen the sector, the government has the option to monitoring of child growth in health facilities expand coverage into other industries to (similar to observation of any other disease); protect vulnerable people in urban areas of it also needs to link the identification of mal- the country. nutrition to appropriate response mecha- nisms (this may include the prescription of for- tified foods, micronutrient supplements, and END NOTE thorough counseling on feeding practices). 1. The World Bank estimates are used for this Enhancing Social Protection System study. Comparisons of the estimates could be found in the annex 2. Stronger institutional support must be provided to the agency implementing the 2. Brett Ballard from the Australian Agency National Social Protection Strategy (NSPS); for International Development, during the furthermore, the resources necessary for decision review meeting. its implementation should be appropriate- ly allocated. To accelerate the implementa- 3. The percentages corresponds to the impact tion of the NSPS, a stronger mandate should in national poverty reduction: 100% = 32.7 be given to the central agency responsible percentage XXIII What Happened With The Poor? Chapter 1 Cambodia experienced surprisingly The aim is to provide a clear, concise analysis impressive economic growth from 2004 for continued poverty reduction in Cambodia. through 2011. The per capita Gross Domestic Product (GDP) increased by 50.9 percent, and Macroeconomic Conditions the average per capita consumption increased by 38 percent. At the same time—and very After several years of stagnant per capita unexpectedly—inequality began to fall after Gross National Income (GNI) during the 2007, further increasing the poor’s income. This nineties and the beginning of the new mil- reduction in poverty was beyond all expecta- lennium, Cambodia GNI grew at a surprisingly tions—including the Cambodia Millennium fast pace of 11.5 percent for seven consecu- Development Goal target for poverty (CMDG). tive years, increasing its value to US$830, more The striking growth certainly moved people than double the 2004 GNI (Figure 1). Given the out of poverty—but for a large majority, it was very low initial value in 2004 ($400), the impres- not large enough. Much of the poor moved sive growth rate has not been enough to lift just above the poverty line, remaining highly Cambodia out of its Low Income status and vulnerable to falling back into poverty. the GNI remains below $1,026 per person in 2011. But it is not that far from rising out of Low This chapter describes both the main Income status: if Cambodia has similar growth improvements and the main weaknesses of rates to the preceding years, it would become a the Cambodian economy; concurrently, it middle income country sometime between 2014 performs a detailed exploration of not only and 2015. Gross Domestic Product (GDP) per poverty conditions but also the characteristics capita figures from the Royal Government of of the poor. At the same time, it gives special Cambodia are similar to the GNI values with attention to qualifying these improvements for almost the same growth rates at any period of a better understanding of their implications. time. 1 The composition of Cambodian economy It relied on four sources of growth: rapid has changed dramatically since 1995, garment exports, strong tourism receipts, a reducing agriculture’s share and increas- continuing construction boom, and increas- ing industry’s role. Agricultural share in GDP ing yields in agriculture. Sustained growth has steadily decreased from 45 percent in was possible because four key factors: First, 1995 down to 26 percent in 2011 with most Cambodia achieved peace and stability and of the gains captured by the industrial sector derived a growth dividend as evidenced by (Figure 2). Almost all the observed changes significant inflows of foreign direct investment in GDP shares were during the second half (FDI) and official development assistance of the 1990’s and the first part of the millen- (ODA). Second, the global and regional envi- nium. From 2004 to 2011, shares of agriculture, ronment during this period was very favorable, industry, and services in GDP experienced only and Cambodia was able to tap this potential small changes. through open trade and investment policies. Third, macroeconomic policies—essentially At almost 10 percent for each year from through fiscal policy and given the high 1998 to 2008, GDP growth was very strong. actual dollarization of the economy—were conductive to stability. Fourth, the growth of Figure 1: Cambodia Per Capita GNI, 1995-2020 the garment sector through bilateral trade agreements with the United States and the European Union under the Everything But Current/Nominal US $ Arms agreement. Both agreements allowed $ 1,026 Middle Income in 2011 for better access to markets against higher GNI Per Capita labor standards. This sustained growth was interrupted in 2008 and 2009 because of both internal Note: The World Bank middle income threshold is updated every year to account and external reasons. Overheating of the for inflation. Dotted lines are projections based on constant growth rate equal to the average of the last four years. economy in 2007 and 2008 was evidenced Source: GNI from World Bank, World Development Indicators. GDP figures from by very rapid growth in credit to the private the RGC Macroeconomic Report. sector and a bubble in real estate prices. When Figure 2: Cambodia GDP Composition By Sector, 1995-2011 combined with the successive shocks of food and fuel price increases, resulting high infla- Argriculture tion peaked at 35.8 percent in May of 2008. Services After this, the global financial crisis led to a sharp deceleration in growth. The crisis hit % of total GDP Cambodia severely in 2009 and flattened its real growth rate. The government adjusted Industry its policy mix in response to external shocks Taxes, others by increasing reserve requirements and then decreasing the same requirement in response to the global financial crisis. This led to a gov- Source: National Accounts, National Institute of Statistics ernment-led a fiscal stimulus in 2009. 2 Supported by an enabling economic However, envisioning a scenario of limited policy and an improved external environ- progress reform, the indicator would breach ment, Cambodia’s economy has recovered the threshold and would not be able to absorb well since 2010. Real GDP growth reached significant risk. This underscores the need 6.0 percent and 7.1 percent in 2010 and 2011, to continue reforms for sustainable strong respectively. The growth was supported by a growth that should include not only pursuing resilient agriculture sector, by a recovering fiscal consolidation and mobilizing revenue, garment and construction industry, and by but also improving debt and contingent liabil- strong tourism. The Government has recently ity management. revised the 2012 real GDP growth upward to 7.3 percent from the former estimate of 7.1 The overall macroeconomic outlook percent—this reflects the better performance remains positive. Real GDP growth is pro- of not only the agricultural sector and but also jected to reach about 7 percent in 2013. The the tourism industry. real GDP growth is underpinned by a contin- ued strong performance of the tourism and Mainly because of petroleum and con- agriculture sectors, and the garment industry struction material imports for building hydro- is likely to benefit from improved external power plants under “Build, Operate, and demand. The growth in the construction sector Transfer” (BOT) contracts, the trade deficit is is projected to continue, but at a slower pace. estimated to have grown to US$ 2.0 billion: it has remained at about 18 percent of GDP in Inflation continues to wane, and it is pro- 2012 (Figure 3). Because of continued, sub- jected to remain at about 3 percent in 2013. stantial foreign direct investment inflows— In the meantime, the nominal exchange rate which reached US$1.5 billion in 2012—the remains stable. Fiscal consolidation continues balance of payments remains positive and the and overall fiscal management remains sound. current account deficit (excluding transfers) is The overall debt sustainability outlook remains projected to stay at about 12 percent of GDP. favorable owing in part to the policy of avoiding The recent debt sustainability analysis (DSA) non-concessional borrowing. Moreover, most shows that Cambodia’s debt distress rating borrowing is being used for productive sectors. remains low under the baseline scenario. This is vital to sustainable debt management. Figure 3 a: Trade Balance (Million US$) Figure 3b: Current Accounts And FDI (Million US$) Other exports (Left) Garment export (Left) Trade deficit (Right) Direct investment, net Current account, excl. official transfers Source: Balance of Payments, National Bank of Cambodia Source: Balance of Payments, National Bank of Cambodia 3 Cambodia’s external public debt and pub- Prospects for fiscal balance are antici- licly-guaranteed debt stock is estimated pated to improve in 2013 and 2014 because to reach 30.1 percent of GDP in nominal revenue collection has improved, increasing terms in 2012. However, fiscal risks from con- by 29 percent in the first half of 2012 compared tingent liability are gradually building up; to the same periond in 2011. This is attributed at the same time, power purchasing agree- to growth of direct and inditect taxes. While ments will become effective and require a rel- the fiscal management remains under control, atively large portion (up to 15 percent) of the it is increasingly vulnerable to unpredictable domestic revenue (if the risks fully materialize). external financing and increased pressure of Therefore, the financing of the power sector, operational and maintenance budgets. specifically, needs to be carefully managed. Financial deepening continues and credit Cambodia’s economic outlook is subject growth—although eased—continues to raise to considerable risks, arising from a fragile concerns as a financial risk. The credit growth global economic outlook and from domestic does not appear to be concentrated in either a factors like rapid credit growth, contingent single or a variety of sectors. liabilities and potential weather shocks (IMF 2013). A deepening of the euro area crisis Foreign direct investment (FDI) inflow and global financial turbulence could end up is expected to account for 9.3 percent of affecting Cambodia's garment exports and its GDP for 2012. In the first half of 2012, 72 financial sector. Credit growth—though eased new projects were approved led by Korean, (29.2 percent in January 2013 year on year Chinese, and Japanese investors (compared down from 34.6 percent in January 2012)— to 57 projects approved over the same period continues to raise concerns as a financial risk in 2011. The number of new firms registered because of weak banking supervisory capacity. at the Ministry of Commerce also increased The public-private power sector projects could by 10 percent to 1,712 during the first half of also potentially result in contingent liabili- 2012 (compared to 1,563 at the same time ties for the government if they are not imple- in 2011). Gross foreign reserves continue to mented according to plan, which could nega- rise, reaching the US$3.2 billion mark by June tively affect the budget. Cambodia could also 2012; it is projected to reach US$3.5 billion by be affected by weather shocks and natural dis- the end of 2012 (representing 4.5 months of asters, similar to the recent Mekong flooding, imports). which could impact agricultural production and individual incomes. Revenue collection did not increase in 2011 (13.2 percent of GDP in 2011—similar If these risks materialize, the policy to collection in 2010) and there were record space of the government to mitigate their high general government outlays of 22.2 impact would be limited. Persistent high percent of GDP in 2011. The fiscal deficit dollarization limits the country’s ability to became high again in 2011 reaching an esti- use monetary policy to mitigate shocks and mated 9.0 percent of GDP (compared to 7.5 implies that the government needs to rely percent of GDP in 2010 when the govern- on fiscal policy as its main tool for macroe- ment exercised an aggressive cut of non- conomic stability. However, the present low essential current expenditures). levels of government deposits limit the fiscal 4 buffer that can be used to mitigate shocks. This of electricity, corruption, and transport are also underscores the need to continue reforms that identified as three of the four largest obsta- enhance the country’s resilience to shocks, cles by firms in the latest Investmnet Climate like pursuing fiscal consolidation through Assessment 2012 survey (ICA). The high cost improved efficiency, mobilizing revenue, and of doing business in Cambodia particularly improving debt and contingent liability man- hurts small and micro enterprises, which rep- agement. resent 99.6 percent of firms in Cambodia (2011 Cambodia Economic Census). The development of the private sector in Cambodia is constrained by a poor business Poverty and Well-being environment. Cambodia ranks 133rd out of 185 economies in the overall Ease of Doing After discounting for inflation, the yearly Business index (World Bank series). This is well average GDP per capita for 2004-2011 was below the average for East Asia and Pacific 6.1 percent (50.9 percent for the entire Region, which ranks 86th. For example, starting period). Because inflation reduces a house- a business in Cambodia requires 9 proce- hold’s ability to acquire goods and services, dures, takes 85 days, and costs 100.5 percent GDP per capita in constant dollars is a better of income per capita, which gives it the poor measure of well-being. After three years with ranking of 175th in the world. The following growth rates close to 10 percent, Cambodian are other aspects in which the business envi- GDP growth decelerated in 2008 and was ronment fairs badly: resolving insolvency (152), negative in 2009. It started to improve, reaching dealing with construction permits (149), enforc- 5.5 percent in 2011 (Figure 4). Individual ing contracts (142), getting electricity (132), sectors do not have a uniform tendency; the and trading across borders (118). The high cost industrial sector shows the biggest yearly Figure 4: Constant GDP Growth (Per Capita) By Sector, Cambodia 2004-2011 Total GDP Agriculture Industry GDP Per Capita Growth Services Source: World Bank staff estimates based on National Accounts 5 increase in 2006 (17.3 percent) and the lowest It was not until 2004 that a more standard- and only negative growth (-10.9 percent in ized methodology (including the question- 2009). Agricultural growth has become more naire) was adopted in Cambodia. Since 2004, stable in the last seven years with only small LSMS type surveys have been collected every changes over time. In 2009, agricultural growth year from 2007 through 2013. The sample size did not decrease. for the 2004 and 2009 surveys were close to 12,000 households; in other years, the sample Socio-Economic Surveys size was about 3,600 households. All of the surveys were collected over the entire year Cambodia estimated poverty for and are representative for Phnom Penh, other the first time in 1997 based on the first urban areas, and rural areas, as well as for the Cambodia Socio-Economic Survey 1993/94. entire country.5 Because of a lack of reliable census informa- tion, the sampling frame used the village pop- After the 2004 questionnaire, some ulation data prepared by the United Nations sections have been added, others removed Transitional Authority (UNTAC). The survey or modified, but most of the questions used included 5,578 households and was carried to estimate poverty remain the same, espe- out by the National Institute of Statistics (NIS) cially since 2007. The Swedish International with support from the Asian Development Development Cooperation Agency (Sida) Bank (ADB) and the UNDP. The survey included provided permanent advice and financial 15 out of 21 provinces (59 percent of the support since the 2004 CSES—promoting best villages) excluding sparsely populated areas, practices, standardization, and limiting com- areas that were expensive to reach, or unsafe parability problems over time. to survey at that time. The survey included most of the elements of a Living Standard Measuring Poverty Measurement Survey (LSMS) with the infor- mation necessary to estimate both per capita Poverty was first estimated in Cambodia consumption and poverty lines (World Bank using the CSES 1993/94. Consumption per 1997). capita was selected as the preferred indi- cator of household well-being. Extreme During the second part of the nineties, (food) poverty lines were estimated based three other household surveys were imple- on the food consumed by poor households mented, but all presented serious limi- in 1993/94, and total poverty lines were esti- tations: (i) the 1996 survey included only a mated based on poor households’ consump- reduced version of the consumption module; tion of non-food items. The main source of (ii) the 1997 CSES was collected over a period information was the recall data; if no recall of only two months and it coincided with a data was available, the information reported deep political crisis; and (iii) the 1999 CSES on the diary section was used.6 presented inconsistencies arising from inad- equate training, which resulted in not only Over time, several improvements have under recorded consumption but also overes- been added to the questionnaire, but it was timated poverty.4 decided not to incorporate such changes into 6 the poverty estimate. The main reason for was provided to improve government staff not changing the way consumption is esti- members’ ability to measure poverty and to mated was to ensure comparability over time understand poverty results better. In 2011 and to ensure that such comparisons were the Government appointed a team of special- fair. Nevertheless, the preferences and habits ists to measure poverty. The work was carried of the population change as time passes. out in parallel with the World Bank, allowing This is especially true for countries similar to both teams to provide inputs, insights, and Cambodia that have experienced important assistance to each other. The World Bank team changes since 1993/94 making an “update” benefited from the insights and preferences necessary to obtain results that are relevant made by the government and incorporated for Cambodia today.7 many of the choices and decisions into its own estimate. The government benefited from the From the first time poverty was estimated experience and previous work carried out by through 2010, the World Bank was leading the World Bank and used some of the same the work with limited participation from the principles and results. (Box 1) Royal Government of Cambodia. Training Box 1: Measuring Poverty In Cambodia: The World Bank Updated Methodology In 2009 The recall section of the CSES 2009 contains all Consumption Aggregate In Cambodia 2009 the necessary information to create the per capita consumption for each household. The components • All food consumed at home or outside (20) include all food consumed at home or outside the • Monthly value of the home (2) home (purchased, produced, received as gifts, or • Housing services (electricity, water, etc.) (10) otherwise), housing (rented or owned), housing • Transport and communication (2) services (firewood, electricity, gas, water, and so • Purchase value of selected durable goods (16) forth), transport and communication, purchase • Personal goods (3) values of selected durable goods, personal use • Spending on recreation and entertainment (3) goods, recreation and entertainment, outlays on • Education expenditures (7 x member) education and health, and the like. Total monthly household consumption was estimated by adding • Health related expenditures (2 x member) all the individual components: this was converted • Others (including goods received in kind) (4) into a measure of individual (per capita) welfare per day using household size. all; for most of the households, the imputed values represented less than 5 percent of their own con- The consumption aggregate is computed using sumption with only 8 cases reporting too many a total of 60 household questions and 9 questions imputations to be used. For 2009 the average asked to all household members. Missing infor- per capita daily consumption was CR 15,709 for mation and outliers were identified and imputed Phnom Penh, CR 10,184 for other urban areas and based on values of similar households. Only 3.3 CR 6,013 for rural households (in per capita, daily percent of households required any imputation at nominal Cambodian Riels). 7 The poverty line is equal to the cost of a basic in the other two regions: non-food consumption food basket plus an allowance for a minimum represented 36 and 34 percent, respectively. The amount of non-food goods and services (housing, same shares were applied to the non-food part clothes, personal goods, entertainment, and so for a total poverty line of CR 5,326 for Phnom forth). The basic food basket composition is drawn Penh, CR 4,273 for other urban areas, and 3,914 from the information provided in the dairy section for rural households (in per capita, daily nominal of the household survey and reported by the Cambodian Riels). poorest 5-25 percent of households. The total food quantities are the necessary amount to provide Households with per capita daily consumption the minimum average caloric requirement of 2,200 below the poverty line were classified as poor and Kcal per day per person: the cost was estimated for non-poor otherwise. The national poverty rate Phnom Penh, other urban areas, and rural house- in 2009 was 23.9 percent with poverty being the holds. The basic non-food part of the poverty line lowest in Phnom Penh (4.3 percent), the highest in is estimated from a group of households with rural areas (27.5 percent) with urban households total consumption that hover around the food outside Phnom Penh nearly midway between poverty line; in the CSES 2009, 498 households the former two (12.7 percent). Poverty levels with have total consumption within 10 percent of the the “old” methodology were substantially lower food poverty line value. In these households, food with a national average of 20.7 percent in 2009. A represented 64 percent of total consumption in complete comparison for the entire period is pre- Phnom Penh and 66 percent of total consumption sented in Annex 2. The 2009 Cambodia Socio-Economic In November 2012, the government Survey was chosen to “update” and to and the World Bank compared poverty esti- improve the way poverty is measured in mates and agreed that the results were Cambodia. The update follows the same prin- very similar. After estimating poverty in 2009, ciples with the same objective of measuring both teams repeated the exercise going back well-being using household consumption and in time for 2004, 2007, and 2008 and forward poverty lines. It includes several changes and for 2010 and 2011. Poverty estimates were improvements: (i) the use of all information almost the same for 2009, 2010, and 2011, available since 2004; (ii) new caloric require- with differences of one percentage point or ments and population weights; (iii) a new food less. Headcount rates in 2007 were also very basket to represent conditions in 2009; (iv) similar (a 2.3 percentage point differential). In an update of non-food value for the poverty 2008, inflation was very high in Cambodia, and line to represent preferences of the near the World Bank team used an alternative index poor in 2009; and (v) the use of an expanded that accounted for most of differences. Annex Consumer Price Index (CPI) that includes 2 includes detailed results from the World Bank information from several provinces.8 Annex update and compares the poverty rates with 1 includes a full explanation of how the con- the government’s new numbers. Because the sumption aggregate and poverty lines were World Bank uses the results for other analysis measured. beyond headcount rates requiring specific 8 characteristics, it was agreed to use both esti- Phnom Penh, most poverty reduction took mates and to recognize each other’s work. The place from 2004-2007, and poverty actually World Bank results are used for this study. increased until 2010. In other urban areas, the distinction occurred after 2009, and a poverty Poverty Levels and Trends increase in 2011 (Figure 6). In seven years, Cambodia has reduced Any successful poverty reduction poverty by over half—reaching 20.5 program has to target rural households and percent in 2011. After a modest poverty direct actions to promote human capital and reduction from 2004 to 2007, poverty was income generation activities in these rural reduced drastically in Cambodia for the next areas, including health, education, agri- two years. Indeed, the drop from 50.1 percent culture and fisheries, and agricultural and to 23.9 percent in 2008 and 2009 was one fisheries related industries. By 2011 poverty of the most impressive poverty reductions remains the highest in rural areas (23.7 percent) observed anywhere in the world. After 2009, and the lowest in Phnom Penh (1.5 percent) poverty reduction has been more modest— with poverty in other urban areas falling nearly averaging slightly more than 1.5 percentage halfway between the two (16.1) (Figure 6). points per year (Figure 5). Figure 5: National Poverty Rate With Confidence Intervals, Cambodia The size of the 2007-2009 poverty reduc- tion was unexpected. It prompted a series of reviews comparing results with independent Poverty Rate poverty estimates from a consultant working out of Thailand. This independent study cross- checked other well-being indicators based on the same survey, and conducted internal reviews with World Bank experts. Finally, com- parisons were made to other international indicators—the results were also compared Source: World Bank staff estimates based on CSES to the government estimates: it was found that all of the overall trends were the same. Figure 6: Poverty Rate By Region, Cambodia 2004-2011 The high inflation levels of 2008 increased the non-sampling error for that year but there is Phnom Penh general agreement for the total inflation from Other Urban 2007-2009. Other Rural Poverty Rate Poverty has been reduced in the three main areas of the country, but with some distinctions in time. About 80 percent of Cambodians live in rural areas; it is therefore not surprising that rural poverty has evolved at the same pace as the national trend. In Source: World Bank staff estimates based on CSES 9 Even after decreases between rural and Poverty has not only decreased in urban poverty (from 27.6 percentage points Cambodia but the remaining poor are in in 2004 to 14.9 in 20119), the higher poverty better condition than in the past, including incidence in rural areas, combined with a very the extreme poor. The Poverty Gap Index—a low urbanization rate (21.2 percent urban, measure of how poor the poor actually are— 78.8 percent rural), concentrates more than has also significantly decreased from 16.2 nine out of ten poor Cambodians in rural percent in 2004 to 4.2 percent in 2011. Even households with no significant change over the Depth Index, a measure that highlights time (Table 1). the conditions for the poorest of the poor has decreased during the same period of time Poverty is not higher in female headed from 6.6 percent to 1.3 percent. Both measures households. There are only small differences show the same tendencies for urban and for and no discernible trend: in some years, female rural households.11 households have higher poverty and in other years they exhibit lower poverty (Table 1). Total per capita consumption increased Non-Khmer households are a reduced group by 37.8 percent from 2004-2011 (4.7 per- representing 2.9 percent of the population cent per year) compared to 50.9 percent (National Institute of Statistics 2011), and the increase in GDP per capita. Since the poverty information for them is only reliable during line is fixed in real terms (by definition), the years with the expanded CSES sample. poverty reduction is determined by increases Poverty rates for non-Khmer headed households in consumption and by changes in distribu- (minorities) were about six and half percentage tion (inequality). In order to put the poverty points higher than Khmer-headed households reductions in context, the total and yearly con- from 2004 to 2009. But this should be taken sumption growth from 2004 was estimated. more as a tendency because the differences Consumption grew at a yearly average of 4.7 are not statistically significant.10 percent from 2004 to 2011, for a total increase of 37.8 percent. At the same time, per capita Poverty remains significantly higher Gross Domestic Product grew a total of 50.9 for children, yet lower for the elderly in percent. Cambodia. Poverty incidence for children six years and younger is 27.2 percent, over ten Other Well-being Indicators percentage points above that of working age people (Table 1). Poverty incidence decreases This report develops a wealth index with age, and it is the lowest for the elderly. The calculated as the first principal component surprisingly low poverty rate of 15.4 percent using durables ownership (44 items) and for Cambodians 60 years and older indicates housing characteristics (34 indicators) from a strong cultural practice of support for the the Cambodia Socio-Economic Survey (CSES) elderly. High levels of support for the elderly and presents results for 2004 and 2011.12 are also easier because there are more than 15 (A wealth index is a summary measure of durable working age adults for each elderly person. goods and housing characteristics that is 10 Table 1: Poverty Rate, Poverty Gap, Depth Index And Contribution To Poverty, Cambodia   Year   2004 2007 2008 2009 2010 2011 Poverty Rate TOTAL 53.2% 50.1% 38.8% 23.9% 22.1% 20.5% Male household head 53.6% 51.1% 38.7% 24.2% 22.6% 20.1% Female household head 51.3% 45.5% 39.0% 22.4% 19.3% 22.5% Khmer household head (HH) 53.0% s.s. s.s. 23.6% s.s. s.s. Non-Khmer HH (Minorities) 59.2% s.s. s.s. 30.3% s.s. s.s. 0-6 years old 62.5% 57.6% 47.7% 30.4% 30.6% 27.2% 7-20 years old 57.3% 54.7% 43.5% 27.0% 25.4% 23.9% 21-59 years old 48.2% 45.6% 34.2% 20.9% 18.6% 17.3% 60 years and older 42.9% 41.2% 30.1% 17.0% 13.0% 15.4% Poverty Gap Index TOTAL 16.2% 15.6% 10.4% 5.3% 4.7% 4.2% Phnom Penh 4.0% 0.4% 0.6% 0.9% 0.8% 0.2% Other Urban 11.6% 10.4% 6.5% 2.5% 2.4% 3.5% Rural 18.1% 18.2% 12.1% 6.1% 5.4% 4.8% Depth Index (Squared Poverty Gap Index) TOTAL 6.6% 6.5% 3.9% 1.7% 1.5% 1.3% Contribution To Poverty TOTAL 100% 100% 100% 100% 100% 100% Phnom Penh 2.5% 0.5% 0.6% 1.7% 2.0% 0.8% Other Urban 8.1% 7.2% 7.1% 5.1% 5.9% 8.2% Rural 89.4% 92.3% 92.3% 93.2% 92.1% 91.0% Consumption Aggregate in per capita/day 2009 Phnom Penh Riels (Poverty Line = 5,326.36) Cambodia 6,399 7,176 7,580 9,325 9,105 8,815 Yearly growth from 2004 3.9% 4.3% 7.8% 6.1% 4.7% Total growth from 2004   12.1% 18.5% 45.7% 42.3% 37.8% s.s. Small sample. Source: World Bank staff estimates based on CSES 11 increasingly used as a proxy for longer-term set target, a score of one was provided. Those household income.13) Two indicators derived that were at least 50 percent on track received from the wealth index are used and compared: a score of 0.5; those below received a score First, the percentage change of the wealth of 0. With a score of 5.5, Cambodia shares the index is compared to the percent change in per fifth best performance with Nepal—just below capita consumption. Second, a proxy poverty Vietnam and Lao PDR (Barmeier 2010). rate based on the wealth index is compared to the actual poverty rate for 2004 through 2011. Cambodia’s ranking in the world in terms of human development remains very low, The wealth index results correspond to but changes in the last ten years are the the changes in consumption and in poverty best in the region. The Human Development results over time. The overall increase of 45.6 Indicator (HDI) is tracked by the UNDP all over percent in the wealth index in seven years the globe. The HDI combines several indica- is only slightly lower than the 56.5 percent tors to capture health, education, and living increase in average per capita consumption standards by normalizing individual country (Table 2). Furthermore, from 2004 and 2011, values within the range of observed values. the proxy poverty estimate derived from the From 2000 through 2010, Cambodia shows wealth index improved at almost the same rate the best improvement in the region14 at 18 (63.3 percent) as the actual poverty decrease percent—above countries like China and Lao (61.4 percent). PDR (16 percent) or Vietnam (12 percent). But Cambodia’s 2011 HDI of 0.523 ranks it among Impressive well-being improvements the worst in the region and the 139th worst in are also reported by other international in- the world (UNDP 2000 to 2011). (Figure 7) dependent studies. In 2010 an MDG Progress Index was estimated by Leo and Barmeier, Housing and Durables which included 76 countries classified as the poorest countries by the International Improvement in housing and housing Development Association at the World Bank services are observed in every category (IDA). The Index compares country perfor- including construction materials (floor, walls, mance against the core MDG targets of pov- and roofs), access to water, electricity, and erty, hunger, gender equality, education, child sanitation. For example, Cambodian house- mortality, health, and water. For each individ- holds showed the following housing improve- ual component that was on track or above the ments in the seven years from 2004 to 2011: Table 2: Wealth Index And Consumption In Cambodia Actual value Poverty rate Measure 2004 2011 change 2004 2011 change Wealth Index 1.6 2.3 45.6% 41.6% 15.3% -63.3% Consumption 2,693 4,214 56.5% 53.2% 20.5% -61.4% 12 Figure 7: Change In The Human Development Indicator By Country 2000-2010 HD % Change from 2000-2010 World Vietnam EAP China Laos Cambodia Source: World Bank staff estimates based on UNDP yearly reports (website database) metal roofs increased from one-third to almost Cambodians are less hungry, feel safer in one-half; access to electricity from a grid more their neighborhoods, and experience less than tripled, reaching 37.5 percent in 2011; the crime. Another way to measure well-being is use of kerosene lamps for lighting decreased to ask people directly questions about their from over one-half to less than one-quar- well-being and then compare those answers ter; and access to sewerage with septic tanks over time. In 2004 the average household more than doubled. Moreover, improved experienced food shortages almost two weeks water source availability increased nearly each year, compared to about half a week in ten percentage points (Table 3). But the use 2011. Additionally, neighborhood safety has of firewood as cooking fuel showed a small improved by more than one-third, reaching decrease with almost no improvement in rural nearly 80 percent in 2011, while victims of rob- areas. For selected indicators, the poorest house- beries has decreased from 4 percent in 2004 to holds in 2011 are better off than the average 2.6 percent in 2011. households in 2004. Households in the poorest quintiles in 2011 show better access to elec- Inequality, Distribution, and Growth tricity, improved water sources, sanitation, and roofing than households in the third quintile Since 2007, inequality has declined for seven years earlier (Figure in Table 3). four consecutive years. From 2004 to 2007, inequality was increasing, becoming a chal- Durable goods ownership has also lenging welfare issue for Cambodia. It was increased dramatically from 2004 and 2011. expected that inequality would continue The selected durable goods in Table 4 illus- increasing, becoming a major impediment for trates that ownership in 2011 was twice what poverty reduction—even with high economic it was seven years earlier. Similar to housing growth. But after 2007, the sources of house- indicators, the poorest households in 2011 are hold growth not only shifted to those favoring better off than the average households were in the poor but augmented the overall impact of 2004. With the exception of television sets, economic growth by increasing the share of more households in the lower quintile in 2011 consumption of the poorest households. The own durables than households in the middle GINI coefficient, an overall measure of inequal- quintile in 2004 (Figure in Table 4).15 ity, increased from 0.326 in 2004 to 0.374 in 13 Table 3: Selected Housing Conditions (% Of Households) By Year And Quintile Cambodia 2004-2011 year 2004 2011 Roof Metal 34.5% 49.1% Connected to grid 12.7% 37.5% Light source Battery 24.4% 35.0% Kerosene lamp 55.0% 23.4% Piped, well or borehole 37.6% 46.2% Dry seas. water Pond, river, rainwater 35.7% 35.9% Sewerage/septic tank 20.8% 42.1% Metal roof Electricity Piped water Sew./septic Sanitation tank None 74.2% 54.0% Liquefied gas LPG 5.9% 12.2% Cooking fuel Firewood 84.6% 79.8% Yellow indicators are positive for household wellbeing. Q3 in 2004 is the middle quintile, Q1 in 2011 is the poorest quintile Source: World Bank staff estimates based on CSES Table 4: Selected Durables Own (% Of Households) By Year And Quintile, Cambodia 2004-2011 2004 2011 Television 46.1% 63.0% Electric Fan 12.4% 28.6% Stove 2.9% 24.6% Bed set 30.4% 44.3% Dining set 6.8% 13.2% Water pump 8.3% 14.0% Television Electric Stove Bed set Motobike Mobile Motorbike 28.6% 56.5% Fan Phone Mobile Phone 12.7% 63.0% Q3 in 2004 is the middle quintile, Q1 in 2011 is the poorest quintile Source: World Bank staff estimates based on CSES 14 200716. Since 2007 consumption has become the increase of inequality detected that year). more equitable in Cambodia, and by 2011 the Consumption also increased in 2008 and 2009 GINI coefficient was 0.282. The same tendency for the entire population—with the exception can be observed when comparing the con- of the richest quintile in 2009 (showing the sumption of the poor and the rich: in 2004 the impact of the financial crisis). In 2010 and 2011, richest quintile consumed almost five folds as only the two lowest quintiles had increased much as the poorest quintile, this then increased consumption (fueling a small decrease in to over six fold in 2007. It has subsequently poverty), and consumption for the upper two decreased, reaching four folds in 2011 (Table 5). quintiles actually decreased (Figure 9). Inequality within each of the three main Both of the aforementioned factors (distribu- regions in Cambodia is decreasing, while tion and growth) play a role in poverty changes inequality between the regions is almost over time. To provide an idea of how each factor unchanged. The Theil Index, another measure influenced poverty, a series of simulations was of inequality, can be decomposed between carried out. The results show a very dynamic and within regions and help to understand composition of growth and distribution as inequality and inequality changes over time. they relate to poverty reduction, with both The overall Theil Index shows the same evo- components helping to reduce poverty some lution as the GINI coefficient: first increasing years and hindering improvements in others. in 2007 and then decreasing in the follow- In 2004-2007, growth was responsible for all ing years (Table 5). While internal differences the poverty reduction rates while inequality within regions have decreased over time, actually increased poverty. In 2007-2008 both reductions in differences between each of the factors helped to reduce poverty. But in 2008- regions in 2011 (0.032) is almost the same as it 2009, growth was again responsible for poverty was in 2004 (0.030) (Figure 8). reduction while inequality played no significant role whatsoever. Finally, in the last two years Average growth depends on individ- (2009-2011) distribution was responsible for all ual consumption status. All Cambodians’ poverty reduction while growth was negative consumption increased in 2007—with the and actually increased poverty. (Table 6) exception of the poorest quintile (prompting Table 5: Consumption Shares And GINI Values, Cambodia   2004 2007 2008 2009 2010 2011 Share of poorest 20% 8.4% 7.5% 8.4% 8.5% 9.0% 9.6% Share richest 20% 42% 46% 41% 41% 40% 38% Share richest/poorest 4.9 6.1 4.9 4.8 4.4 4.0 GINI * 100 32.6 37.4 32.3 31.9 30.2 28.2 Theil Index * 100 18.6 27.0 18.2 17.9 16.0 13.7 Source: World Bank staff estimates based on CSES 15 Table 6: Simulated Poverty Impact Of Growth And Distribution (% Points), Cambodia 2004-2011 Period Year to Year 04-07 07-08 08-09 09-10 10-11 Observed poverty -3.2% -11.3% -14.8% -1.9% -1.5% Growth (Consumption) -8.8% -4.0% -13.7% 1.9% 2.2% Distribution (Consumption) 4.8% -6.3% -0.2% -3.2% -3.9% Interaction (Growth * Distribution) -0.9% 0.9% 0.9% 0.5% -0.2% Growth = poverty with distribution in year 1 and average consumption in year 2. Distribution = poverty with average consumption in the year 1 and distribution in year 2 Source: World Bank staff estimates based on CSES Figure 8: Population, Theil Index Value And , Theil Index Change, Cambodia 2004-2011 Between regions Rural Other Urban Phnom Penh Theil Index Source: World Bank staff estimates based on CSES Figure 9: Consumption Average By Quintile, Cambodia 2004-2011 Per Capita/day Constant thousahand of Riels Phnom Penh 2009 Poverty Line Source: World Bank staff estimates based on CSES 16 Vulnerability 2011. The increased consumption moved the highest number of Cambodians possible out The impressive poverty reduction from of poverty. Plotting the population shares for 2004 through 2011 was a much welcome consumption levels show the highest con- improvement in people’s well-being. However, centration of people in 2004 just to the left of it is important to understand that while the the poverty line (poor), whereas in 2011, the absolute improvements for those escaping highest concentration of people is just to the poverty were positive, they were not enough right of the poverty line (CR 5,326 per capita to compensate for unexpected or undesirable per day in 2009 Phnom Penh Riels), and there- shocks that would cause many people to fall fore not poor (Figure 10). The estimated back into poverty. Poverty classifications divide poverty rate was very sensitive to relatively the population according to their consump- small changes in consumption. tion at one specific point, called the poverty line. However, it is vital to understand that how The impressive poverty reduction was close or how far away a household is from that possible because of how consumption was specific point is not reflected by the poverty distributed from 2004 through 2009. rate. To understand poverty and vulnerabil- To illustrate this point, the distance to the ity in Cambodia, this concept is essential. It is poverty line was estimated for the group of explained further in the following paragraph. households between the 20.5 percent mark (the poverty rate in 2011) and 53.2 percent Poverty reduction was augmented sub- mark (the poverty rate in 2004).17 All of these stantially by the high concentration of households were poor in 2004, yet none were people just below the poverty line in 2004 poor in 2011. This significant group of house- and people just above the poverty line in holds’ average distance to the poverty line in Figure 10: Population Share By Consumption Per Capita, Cambodia 2004-2011 2009 Poverty Line = Riel 5,326 Population % Consumption in Per Capita 2009 Riels/day (000) Source: World Bank staff estimates based on CSES 17 2004 was CR 1,072 per day per capita (in 2011 population will then outpace the reduction of Phnom Penh Riels). In other words, in 2004 children in Cambodia (Figure 11). As a result, these households needed an average of just the number of potential workers for each child above one thousand Riels to escape poverty. will keep increasing from about two in 2010 to In 2011 households within the same percen- nearly four in 2055, easing burdens on raising tile (20.5-53.2) were no longer poor with an children. But the number of potential workers average of CR 1,202 above the poverty line.18 for each elderly person will experience a dramatic decrease at the same time from sev- Vulnerability of falling back into poverty enteen in 2010 to four in 2055, thus increasing has substantially increased in Cambodia. the burden on caring for older people. Given the high concentration of non-poor households close to the poverty line, small Poor Households and the Correlates reductions in consumption would bring them of Poverty over Time back into poverty. For example, a loss of CR 1,192 per capita per day (2011 nominal Riels) The typical poor Cambodian household would double the poverty rate to 41.0 percent is rural, has 5.7 family members, uses almost in 2011.19 In 2011 the implicit poverty elastic- two-thirds of its budget for food, has a house ity to the poverty line is close to 3.5, compared with hard roof material, does not have elec- to 1.1 in 2004. tricity, owns almost one hectare of land and a bicycle, is headed by a male over 45 years old Cambodia’s demographic evolution is with a spouse or companion, and the working expected to continue for an overall positive age adults have an average of 3.3 years of edu- effect on the economy, but it also threatens cation. Among the poor, one in five live in a to undermine its traditional social protec- household headed by a female and 41 percent tion system. On the one hand, the low fertility receives remittances (Table 7). rate will further reduce the dependency ratio over the next three decades—contributing From 2004 through 2011, access to to GDP growth, but with lower impetus than housing services and ownership of durable in the past 15 years. On the other hand, the goods by the poor improved although years lower fertility rate coupled with the longer life of education, access to clean water, and expectancy will place an increasing economic household size have remained almost the burden on families, eroding the traditional same. Important increases were reported by protection system for the elderly. This high- the poor in access to sanitation (6.2 to 13.2 lights not only the need to focus on worker percent), and electricity (5.4 to 8.5 percent). productivity improvements as an engine of Similarly important increases were reported growth, but also the need to promote a formal in ownership of mobile telephones (1.4 to 39.1 savings plan for retirement for more workers. percent), motorbikes (12.9 to 29.7 percent), and bed sets (17.5 to 26.1 percent). At the In the long run, taking care of children same time, poor people reported little change will become easier and caring for the elderly in average years of education for working age much harder. The overall dependency ratio is adults (3.1 to 3.3 years), access to piped water expected to decrease until 2030—the elderly 18 Figure 11: Population And Dependency Ratio, Cambodia 1995-2055 Working age: 15-16 Population share Dependency ratio Children: 0-14 Elderly: 65+ Workers per elder # of worker Workers per child Dependency ratio = (Total Children + Elderly)/ Working Age. Values are the 3 period moving average. Source: Estimates based on United Nations 2011 or a protected well in both the wet and dry between the average and poor Cambodian seasons (42.7 to 43.5 percent), and household households are as follows: they have one less size (5.59 to 5.67 members). (Table 7) person than the poor (from 0 to 20 years old), they use 56.1 percent of their budget for food The poor and the average house- (compared to 63.8 percent for the poor), they holds share both similarities and impor- have more than triple the access to sanitation tant differences. The poor and the average (42.6 percent) and electricity (37.7 percent), Cambodian households have very similar they have almost two more years of education household head characteristics (age, gender, (for working age adults), they are better con- and companion), agricultural land size, access nected by mobile telephones (63.0 percent) to piped water or protected wells in both the and motorbikes (56.5 percent), and they had wet and dry seasons, remittances (frequency), remittance values that were more than double and bicycle ownership. The differences those of poor households (Table 7). 19 Table 7: Selected Indicators Of The Poor And Average Households, Cambodia 2004, 201120 Poor Average Indicator 2004 2011 2004 2011 Rural HH 89.8% 92.2% 81.4% 79.3% Household size 5.59 5.67 4.98 4.53 # 0-6 years old 0.93 1.05 0.70 0.63 # of 7-20 years old 2.19 1.98 1.82 1.36 # of 21-59 years old 2.19 2.31 2.15 2.19 # of 60 years & older 0.28 0.33 0.31 0.35 HH head age 43.7 45.5 44.9 46.6 Female HH head 20.1% 22.0% 21.6% 22.7% HH head married or living together 80.9% 78.8% 79.4% 77.9% Food/Total Consumption * 63.3% 63.8% 59.5% 56.1% House + services/Total Consumption * 21.1% 16.6% 19.5% 18.7% Transport & Communication/Total Consumption * 1.3% 4.0% 3.0% 6.4% Roof of hard material: tiles, metal 59.1% 73.1% 71.0% 88.5% Piped water or protected well in wet & dry season 42.7% 43.5% 46.9% 45.9% Sewer or septic tank 6.2% 13.2% 20.8% 42.1% Electricity 5.4% 8.5% 19.7% 37.4% Average years of education for 20-60 years old 3.1 3.3 3.9 5.1 Agricultural land: hectares 1.04 .96 .98 .97 Rice producing household 75% 69% 65% 59% Received remittances 40.6% 43.4% Average remittance value/day/per capita 118 321 Refrigerator 0.0% 0.0% 1.9% 4.7% Wardrobe/cabinets 5.3% 14.2% 18.2% 44.1% Mobile 1.4% 39.1% 12.7% 63.0% Bed sets 17.5% 26.1% 30.4% 44.3% Motorbike 12.9% 29.7% 28.6% 56.5% Bicycle 64.1% 67.4% 64.0% 68.0% * Values are the average of individual household percentages. Note: numbers are household averages and not population averages Source: World Bank staff estimates based on CSES 20 To identify the impact of household char- Cambodia. (This was true for all of the three acteristics on poverty, a multivariate regres- years analyzed: 2004, 2009, and 2011.) But sion was estimated linking per capita con- households headed by a non-Khmer have a sumption of each household with a series higher probability of being poor. Also, consump- of indicators reflecting geography, house- tion improves if the household heads speaks hold head attributes, household size, educa- more than one language; it also improves as tion, land ownership, rice production, hours both male and female heads of households of work, and durables ownership.21 By con- become older. sidering all variables at the same time, the results help identify the real impact of a single Overall consumption is reduced as the variable after discounting for differences of size of the household increases, but the other variables. The results do not capture negative impact is lower for older house- the dynamic impact of economic growth hold members. Indeed, the negative impact and other poverty related changes; they of extra household members is highest for are limited to the variables available for the children 6 years and younger, followed by analysis as well as the range values for each children 7 to 20. Negative consumption year. Moreover, although several variables can impacts are less for working adults and even be theoretically linked to poverty changes, the less for the elderly. results only ensure the link between variables and consumption (correlation). Also while the Each year of education for the working selected indicators can impact consumption, age adults is related to higher consumption. it is possible for consumption to impact the For each additional year of education for those indicators (causality can run in both ways), for household members able to work increases example, more land can increase household average consumption in the household. income, but also, households with more income can purchase more land. Small land holdings are becoming vital for poverty reduction. In 2004, owning up to Geographic location is one of the most two hectares of land was not correlated with important factors explaining poverty in household consumption. As net income from Cambodia. Even after taking into considera- agriculture improved, owners of small plots of tion disparities in assets and household com- lands became better off. By 2011, owning even position, households in rural areas have lower half of a hectare of land improved household consumption levels, followed by urban house- consumption. Owning a motorbike, a proxy holds outside Phnom Penh. Households in for assets, is associated with better consump- the plateau and mountain areas have a higher tion and a lower probability of being poor.23 probability of being poor compared with the rest of the country22 (Table 8). Farmers have lower consumption levels than other households. Consumption Female headed households do not improves with the size of the agricultural land; have a higher (or lower) probability of however, the analysis did not find any differ- being poor. Male and female-headed house- ence among households planting rice or other holds have similar levels of consumption in crops (other crops excluded from the table). 21 Table 8: Correlates Of Poverty In Cambodia, 2004, 2009 And 2011 2004 2009 2011 Intercept 8.8 ** 8.9 ** 9.2 ** Rural Plateau/Mountain -0.33 ** -0.26 ** -0.34 ** Geography (Phnom Rural non Plateau/Mountain -0.17 ** -0.21 ** -0.30 ** Penh excluded) Other Urban Plateau/Mountain -0.20 ** -0.16 * -0.29 ** Other Urban non Plateau/Mountain -0.07 * -0.04 ns -0.19 ** Female HHH -0.02 ns 0.03 ns 0.02 ns Non Khmer HHH -0.05 * 0.05 ns -0.14 ** Household Head HHH speaks > 1 language 0.16 ** 0.16 ** 0.10 * (HHH) HHH is with a partner 0.06 ** 0.07 ** 0.03 ns HHH age 0.01 ** 0.01 ** 0.01 ** HHH age squared (*100) -0.01 ** -0.01 ** -0.01 ** 0 to 6 years old -0.17 ** -0.14 ** -0.17 ** Household Size (# 7 to 20 years old -0.10 ** -0.10 ** -0.11 ** of members) 21 to 59 years old -0.06 ** -0.07 ** -0.08 ** 60 and older -0.04 ** -0.06 ** -0.07 ** Average years of educ. for 20-60 years old 0.05 ** 0.05 ** 0.05 ** Rice producer -0.14 ** -0.14 ** -0.13 ** > 50 m2 and up to 1/2 ha. -0.002 ns 0.01 ns 0.04 ns > 0.5 ha and up to 1 ha. -0.01 ns 0.04 * 0.07 ** Size of agricultural > 1 ha and up to 2 ha. 0.02 ns 0.04 ns 0.09 ** land in hectare > 2 ha and up to 4 ha. 0.06 ** 0.06 ** 0.12 ** > 4 hectares 0.20 ** 0.18 ** 0.22 ** Per capita hours in paid employment 0.003 ** 0.002 ** 0.001 ns Did the HH have a Motorbike last year 0.29 ** 0.18 ** 0.19 ** R2 .46 .40 .507 ** significant at p ≤ 0.05; * significant at p ≤ 0.15 Source: World Bank staff based on CSES 22 After discounting by location and land own- 10. Due to large confidence intervals for the ership, the rice producer variable is more of a non-Khmer estimates: 52.1%-66.3% in proxy for farmers than anything else, and the 2004 and 19.3-41.3% in 2010. results shows that being involved in farming is related with lower consumption levels in 11. The Poverty Gap Index and the Depth Cambodia .24 Index measure how far away each indi- vidual household’s consumption is from END NOTE the poverty line. Both measures cover only poor households. 4. The 1993/94 and 1996 surveys were called Socio-Economic Survey of Cambodia 12. For a detail explanation of the wealth (SESC), all others are called Cambodia index, see (World Bank 2013) Socio-Economic Survey (CSES). 13. See, for example, Deon Filmer and Lant 5. The larger samples (2004 and 2009) are H. Pritchett, “Estimating Wealth Effects also representative for five agro-ecologi- without Expenditure Data--or Tears: An cal zones: Phnom Penh, Plains, Tonle Sap, Application to Educational Enrollments Costal and Plateau/Mountains. Another in States of India.” Demography 38(1): large sample is planned for 2014 (every 115-132 (February 2001). five years). 14. Excluding some small islands in the pacific 6. Recall data is collected by directly asking the households how much have they bought, 15. Other durables were owned by too few received, or consumed during the last households (namely, washing machines), days. Diary data is a daily written record or were too small (namely, cameras), of actual expenditures and consumption or were highly related to electric fans over one month. (namely, radios or video and CD/DVD players), or were used for farming (plows, 7. This concept is applied in other exercises tools, and the like). like the Consumer Price Index when the list of products and their relative weight is 16. GINI coefficient can range from a value of reviewed and changed to reflect new pref- 0 (total equality, everybody with the same erences by Cambodian households over consumption) to 1 (total inequality with time. one person reporting all the consumption in the country). As the GINI coefficient 8. The official CPI used in the past was based increases, so does inequality. on information from the capital city Phnom Penh only. 17. There is no panel data to estimate the actual value of the households that moved 9. Total urban poverty headcount rate (Phnom out of poverty. Penh and Other Urban areas) was 29.9% in 2004 and 8.7% in 2011. 18. 2011 Phnom Penh per capita Riels were 23 used in this instance to have a better ref- the center-south of Cambodia (Kampong erence to Cambodia purchasing power Speu) and seven provinces in the North today. Most of the document uses nominal and North East part of the country: Oddar or 2009 Phnom Penh Riels. Meanchey, Preah Vihear, Stung Streng, Ratana Kiri, Mondul Kiri, and Kratie 19. 2011 Phnom Penh per capita Riels were used for a better reference to current 23. The actual variable use was ownership of a Cambodia purchasing power. motorbike for more than one year. 20. All statistics presented in this table are 24. The regression was also run with a dummy expanded to the entire population (not for any crop (instead of only rice) and the the households). results were the same. 21. The dependent variable was the natural logarithm of per capita consumption. 22. Plateau/Mountain includes one province in 24 Sources of Household Growth Chapter 2 After reviewing the impressive household is based on micro-data from both household consumption increase, the inequality reduc- surveys and government statistics. The sim- tion, and the poverty reduction, it is only ulation identifies the main drivers of poverty natural to pose the question: Where did it all reduction and their impact on poverty. come from? This chapter attempts to answer this question not only by identifying the Urban labor sources of income growth, but by examining its contributing factors. It first examines the Labor Market and Poverty Reduction Trends evolution of the salaried urban labor market (including labor force composition, female Although the changes in poverty rates labor, returns to education, and the like) and involve both rural and urban areas, there identifies main areas for further action. Second, have been marked differences in the timing the chapters explores Cambodia’s rural sector, of these changes. Poverty rates in Cambodia starting with the changes in consumption fell from 53.3 percent in 2004 to 20.5 percent over time and income sources (including agri- in 2011. While poverty reduction in rural areas cultural crops, wage labor, self-employment, concentrated during 2007-2009 (86 percent), and so forth). It then identifies priority areas most of the poverty reduction in Phnom Penh for government interventions. poverty took place earlier, from 2004-2007 (92 percent). Changes in other urban areas show The chapter concludes with a simulation similarities to both Phnom Penh and rural built from the results of previous sections and areas, but variations over time are closer to from the impact of rice price estimates, which that of rural areas (Figure 12 and Table 9). 25 Salaried employment represents From 2008 to 2011, the increase in salaried almost one-third of the labor force and over employment in urban areas was more half of the labor force in urban Cambodia. modest (4.1 percentage points). Own account Generally, available employment opportuni- workers remained almost unchanged in urban ties are vital determinants of poverty reduc- areas from 2004 to 2011 (0.3 percentage points tion because labor and land represent the increase), but increased 7.4 percentage points main assets of the poor. Cambodia is not an in rural areas (see the Role of the Rural Sector exception to this rule. The labor market in for the impact on rural income). In both urban Cambodia underwent a striking transforma- and rural areas, unpaid family workers have tion in 2004-2011: salaried employment— significantly decreased (Figure 13). which generally provides more stable earnings compared to other occupational categories— One of the main factors driving poverty increased one-third from 22.6 percent in 2004 reduction is the increased share of paid to 30.3 percent in 2011. In urban areas, salaried employment; this is illustrated by the dif- employment reached 50.5 percent in 2011 ferences in consumption expenditure (Figure 13). across employment categories of the household head. The average per capita con- The difference in timing suggests that sumption for households of which the head the determinants of poverty reduction in of the household is employed in a salaried rural and urban areas differ. Indeed, the job was computed; the average consump- rapid expansion of salaried employment has tion for households of which the head is a occurred mainly in urban areas whereas own self-employed worker was also computed. account workers increased in rural areas. From The ratio of the per capita consumption 2004 to 2008, the share of salaried employ- between these two types of households is ment increased 9.8 percentage points in used to obtain a measure of the “consump- urban areas, but remained almost unchanged tion premium” for salaried employment. This in rural areas (0.9 percentage point increase). consumption premium was calculated for Figure 12: Poverty By Region (2004=100%), Cambodia 2004-2011 Table 9: Poverty Rate Changes Phnom Penh Normalized poverty reduction 2004-2011 Other Urban change = 100% Other Rural Phnom Other Period Rural Penh Urban Poverty Rate Actual % points 14.3% 23.6% 35.3% reduction 2004-2011 100% 100% 100% 2004-2007 92% 20% 3% 2007-2009 -11% 95% 86% 2009-2011 19% -15% 11% Source: World Bank Staff estimates based on CSES 26 Figure 13: Employment Share By Occupational Categories In Urban And Rural Areas Urban Rural Cambodia Unpaid family worker Own account worker Paid employee Note: Spouses classified as unpaid family workers were reclassified to own account workers. Employers representing less than 0.2% of the labor force were classified as paid employees. Others, representing less than 0.5% of the labor force were classified as unpaid family workers. Source: World Bank estimates based on CSES Figure 14: Consumption Premium For Salaried 2004 through 2011 for all years for which there Emplyment, Cambodia, 2004-2011 was available data (Figure 14). The findings state that salaried employment is systemati- cally associated with higher welfare than self- employment. For example, in 2004, an average household involved in wage employment experienced 16 percent higher consump- tion per capita than an average household of self-employment. The consumption premium exceeds 10 percent throughout the entire Source: World Bank estimates based on CSES period (with the exception of 2010, dropping to 4 percent). In 2011, the trend reversed, and Employment participation is lower for the consumption premium reached 10 percent women, young adults, mature workers, and for households involved in wage employment. people with more education. The employ- ment- to-population ratio is computed by Before moving to the next section, which gender, age, and education categories (Table looks at the labor market from a supply side, 10). The employment-to-population ratio is it should be noted that the labor market faces a common indicator for developing coun- important challenges in future job creation. tries and has several advantages over unem- The workforce of Cambodia is remarkably ployment rates, which tend to be both very young—as of 2010, more than two-thirds of all low and unrepresentative of actual employ- Cambodians were born after 1993. ment conditions. Employment participation is higher for males and individuals 25-44 years Employment Composition: the Role of old; moreover, it is higher at the extremes Gender and Education and Better Jobs on either end of the educational attainment 27 continuum. Lack of employment is not the sector, which attracts women and contributes main concern in labor markets—policy efforts to rising female wages. In this process, labor should also focus on the working poor, rather market incentives are accompanied by higher than focusing on only unemployed individu- investments in human capital of women, with als. (Table 10) a consequent increase in female labor force participation. Cambodian female labor force Although men are more likely to par- participation is now above average relative to ticipate in the labor market, female labor other countries with similar per capita Gross force participation is high compared to Domestic Product (GDP). (Figure 15) other countries at similar levels of devel- opment (Figure 15). There is a well-known In urban areas, the increase in paid U-shaped relationship between a country’s employment has benefited women and level of development and its female labor par- men equally. For example, unpaid employ- ticipation rate. A country’s initial economic ment decreased from about 25 percent in development is generally associated with a 2004 to almost 11 percent in 2011 for females transition of the economy from agriculture and males; self- employment did not change to manufacturing—this reallocation results much for females (-0.7 percentage points) nor in men earning higher wages than women. for males (1.1 percentage points). The overall Moreover, this economic shift is usually not reduction in unpaid labor was accompanied accompanied by the concurrent development by an increase in the share of salaried jobs for of universal childcare services or changes in women and men in urban areas. The share of social norms that would contribute to the full female salaried employment increased from participation of women in the labor market. 28.9 percent in 2004, to 38.3 percent in 2008, Consequently, the increase in household and to 43.5 percent in 2011. This remarkable income goes hand-in-hand with a reduction in increase in female salaried employment had female labor force participation: women stay important effects on household standards of at home (unpaid) and are in charge of house- living. Same as women, a greater percentage of hold chores and child rearing. Later economic men found salaried employment: an increase development usually expands the service from 44.1 percent in 2004, to 53.2 percent in 2008, and 57.4 percent in 2011 (Figure 16). Table 10: Employment To Population Ratio Education 2004 2007 2008 2009 2010 2011 Age 2004 2007 2008 2009 2010 2011 None 73% 95% 92% 83% 89% 89% 15–24 76% 78% 76% 76% 78% 78% Primary 82% 86% 85% 86% 87% 87% 25–34 87% 90% 90% 90% 91% 94% Lower Sec. 76% 77% 75% 75% 82% 82% 35–44 88% 92% 91% 92% 94% 94% Higher Sec. 70% 87% 82% 79% 68% 68% 45–54 85% 89% 89% 90% 92% 90% Male 85% 90% 89% 87% 88% 89% 55–64 72% 80% 78% 78% 81% 83% Female 74% 77% 75% 77% 80% 81% 65+ 43% 45% 42% 43% 46% 50% Source: World Bank estimates based on CSES 28 Figure 15: Female Labor Force Participation And GDP, Latest Year Cambodia Vietnam Thailand Female L. Force Participation LN (GDP per capita) Source: The World Bank, World Development Report Figure 16: Urban Male And Female Employment Share Female Male Unpaid family worker Own account worker Paid employee Note: Spouses were reclassified from unpaid family worker to own account worker. Employers were classified as Paid employee. Others was classified as unpaid family worker. Source: World Bank estimates based on CSES 29 Self-Employment is more common percent. Likewise, the share of urban salaried among low-skilled workers (primary or no workers with complete secondary education education), especially in the urban areas. increased to 16.4 percent (over double that Although low-skilled workers represent two- the self-employed) and university graduates thirds of the self-employed in urban areas, represent almost one-fifth of the urban paid their share drops to only 41.5 percent among employees (compared to only 1.7 for the self- all paid employees. In rural Cambodia, low- employed) (Table 11). skilled workers have also higher concentra- tion of the self-employed (84.4 percent) than The population pyramids from Cambodia as paid employees (75.9 percent) (Table 11). represents a challenge for the creation of The workforce of Cambodia is remarkably salaried employment and suggest that self- young, and the labor market might not be able employment and household enterprises are to absorb all new entrants into the salaried bound to continue as the main source of employment labor force. employment in the medium term. For instance, Fox and Pave (2012) find that for Sub-Saharan Compared to 2004, educational levels Africa (SSA), household enterprises generate improved for all urban workers, but there the majority of new non-farm jobs. The authors was a limited change for any rural workers. find that these enterprises are fundamental From 2004 to 2011, the share of self-employed for inclusive growth because high economic people in rural areas with lower-secondary growth rates will not be able to absorb the education increased from 20.0 to 24.3 percent, new entrants that are leaving the farm sector, and the share of those with secondary edu- who also do not possess education levels for a cation doubled, increasing from 3.0 to 7.7 salaried job. The authors also show that a house- hold (controlling for education) can perform Table 11: Employment Share By Education, Residence, just as well by adding a household enterprise And Salary, Cambodia 2004-2011 as primary employment as by adding a wage     Salaried Self-Employed income. The authors conclude that “With 40-50     2004 2011 2004 2011 percent of households engaged in non-farm No grade 0.1% 0.0% 0.5% 0.0% enterprises on average, and the share increasing Primary 47.9% 41.5% 76.0% 66.3% in many countries, any investments which result Lower Secondary 28.2% 24.2% 20.0% 24.3% in more household having a viable HE (house- Urban Secondary 13.9% 16.4% 3.0% 7.7% hold enterprise) or higher incomes for even half University 9.9% 17.9% 0.5% 1.7% of the HEs would have a substantial impact on Total 100% 100% 100% 100% Gross Domestic Product (GDP) and poverty.“ To estimate the returns to education for No grade 0.3% 0.0% 0.5% 0.4% salaried workers 15 to 65 years old, two mul- Primary 75.6% 75.9% 87.8% 84.4% tivariate regressions were constructed. Both Lower Secondary 18.4% 16.9% 11.0% 12.1% Rural regressions used a logarithm of monthly Secondary 4.7% 5.9% 0.7% 3.0% wages in constant Riels as the depend- University 1.1% 1.3% 0.1% 0.1% ent variable. The independent variables— Total 100% 100% 100% 100% factors that explain increases or decreases in Source: World Bank estimates based on CSES 30 wages—were educational completion: for the Returns to education have been declin- first regression, completed years of education ing in Cambodia since 2008. Results from were used. For the second regression, comple- the first regression show a clear decline to tion of primary, lower secondary, secondary or returns to education after 2008. Returns to tertiary education were used (each person was each year of education declined a total of 1.8 assigned a value of one for the highest com- percentage points from 7.4 percent in 2008 to pleted level of education and a 0 for all other 5.6 percent in 2011. Compared to completed education levels). Also, to make sure that the primary school, the returns to secondary edu- results are not combining the importance of cation (lower and upper) were reduced from education with other household characteris- one-third to almost one-half from 2009-2011. tics, an extra set of variables were used. Reductions in returns to tertiary education have been more modest. Workers with more education earn higher wages—with the biggest premium Investing in education pays off in the for university graduates. By 2011, each year Cambodia labor market, but there is a of education increases wages by 5.6 percent. mismatch in skills. The observed decline Even compared to workers with completed in the returns to education, together with primary education, finishing lower-second- the findings from a recent World Bank ary school increases wages by 16.1 percent, study (World Bank, 2012), suggests that the and completing upper-secondary school adds Cambodia labor market might be experienc- another 8.4 percent to wages. Finishing uni- ing a mismatch between the demand and versity is associated with wages 84.5 percent supply of skills. Despite large returns to educa- higher than those with only primary education tion, the majority of the employed population (Table 12). has a only primary education. Table 12: Returns To Education By Year And Level Completed, Cambodia 2004-2011 2004 2007 2008 2009 2010 2011 Regression 1: returns to each year of education completed Years of education 0.069 0.070 0.074 0.062 0.053 0.056 Regression 2: returns to completed education level: primary excluded (Raw coefficients) No grade 0.451 -0.591 -0.689 -0.316 * * Lower secondary 0.282 0.192 0.220 0.239 0.132 0.161 Higher secondary 0.598 0.438 0.402 0.447 0.226 0.245 Tertiary 0.992 0.958 1.147 0.877 0.867 0.846 In 2010 and 2011 there were no individuals in the "No grade" educational category. 31 The percentage of firms reporting skills training graduates compared to university as a major constraint to growth in the 2007 graduates, as well as difficulties in accessing is about 15.5 percent. Similarly, 22 percent of good training providers for formal training to Cambodia’s foreign firms identified skills as a address skills gaps. “severe” or a “very severe” constraint to their businesses (World Bank, 2012). An example of There is a substantial gender wage the skills mismatch has also been found in one gap among salaried workers. In 2009, the of the main exporting sectors: the garment average salary of men was 30 percent higher sector. Apparel technicians and supervisors than the average salary of women. The Oaxaca from China are dispatched to apparel factories decomposition was applied to understand in Cambodia through Chinese human resource this raw wage differential: it attributed the agencies (Natusuda , Toto, and Thoburn, 2009). gender wage differential to the following two Even among graduates, there appears to be rationales: (i) differences in endowments (for a disconnection between fields of study and example, gender disparities in education); market needs, with an oversupply of college and (ii) an unexplained component, which graduates that have difficulty acquiring a includes distinctions in the returns to the job in their field of study because too many endowments (for example, variances in the students are enrolled in business-related dis- monetary returns to education). The findings ciplines and law, while too few are enrolled in suggest that the gender distinctions in returns engineering (World Bank, 2012). to endowments explain the largest part of the unconditional gap (Table 13). In particular, Cambodia has a shortage of senior man- women, compared to men, face a penalty not agement with the appropriate skills and only from marriage but from living in urban soft skills, which needs to be improved areas. Examining the dissimilarities in endow- among all employees. According to the World ments (which are more prone to be affected Bank 2012 skills report, “…the HRINC 2011 by policies), it was found that disparities in survey, over 70 percent of employers reported the average years of education are particularly a major shortage in management skills, 36 important—and increases inequality. percent in middle management and supervi- sory skills, and 34 percent in professional staff The Textile Sector skills. Among the most deficient soft skills, 52 percent of employers identified work atti- Compared to other sectors, the textile tudes in unskilled workers; 45 percent cited sector has a much larger share of female decision-making skills in semi-skilled workers; workers. Because more than 80 percent of and 64 percent mentioned analytical skills in textile workers are female, this sector could rep- skilled workers. Furthermore, employers com- resent a laboratory for expanding economic plained about difficulties in finding employees opportunity to women (Table 14). Policies with not only specific vocational skills but also aimed at expanding benefits and working con- basic skills such as literacy and numeracy.” ditions in industries with a large prevalence of female labor participation are more likely to Other findings from the same report transmit improvements into several welfare include both a relative shortage of vocational dimensions: children, education, and health. 32 Table 13: Oaxaca Decomposition Of Gender Wage Differential, Cambodia 2009 Endowments (Inequality Of Opportunities) Returns to Endowment (Penalty) University /0.025*** Married /0.064*** Lower secondary /0.013*** Against Women Higher secondary /0.007*** Urban /0.052*** Urban /0.001 Lower secondary /0.001 Age /0.000 Higher secondary /0.001 Neutral University /-0.001 Constant: 0.088 Sector (all) /-0.067 Favor Women Sector (all) /-0.091 Age /-0.019 Total effect Slightly in favor / -0.040 Slightly against / 0.120 Source: World Bank Staff Estimates based on CSES Table 14: Female Labor Participation Across Sectors, Cambodia 2004-2011 Food Con- Social Agricul- For- Textiles Other Trans- Public Other Ser- Wood struc- Sales Ser- ture estry Apparel Manuf. port Admin services vices tion vices 2004 50.0% 48.6% 45.8% 81.3% 61.8% 28.9% 12.4% 68.1% 7.1% 13.3% 36.9% 44.0% 2011 50.8% 66.6% 45.1% 84.2% 25.8% 33.4% 8.8% 68.8% 4.4% 11.8% 44.6% 50.1% Source: World Bank Staff Estimates based on CSES Other sectors with high female participation by 2011. This gender wage gap is consistent with 2011 are sales (68.8 percent) and forestry (66.6 previous studies on Cambodia’s labor market percent), and female labor participation in agri- (Savchenko, et al. 2011). culture and other services is close to 50 percent. The apparel sectors benefits women. Women in the garment industry are com- The interaction term of the wage regres- paratively better off than women employed sions shows a positive premium for women in other industries. Within the entire employed in the apparel sector (Figure 17). economy, the gender wage gap has remained After 2009, a positive trend begins in the mag- constant over the period with a slight widening nitude of this gender-industry premium—this after 2009. Controlling for education and other might reflect the favorable conditions imple- demographic characteristics, women earned mented through the Better Work Program on an average of 20 percent less than men in 2009, female employment in the garment industry and this gap widened to almost 30 percent in overall. Similarly, the minimum wage laws 33 implemented in the garment industry might working conditions in Cambodia’s garment fac- also be partly responsible for creating the afore- tories according to national and international mentioned favorable conditions. (Figure 17) standards but also helps factories to improve working conditions and productivity. Several The garment industry experienced studies evaluated the impact of the Better Work major changes after the end of the Multi- Program on factories—with a special emphasis fiber Agreement in 2004. The major achieve- on how working conditions change over time. ment in proactive policies in Cambodia’s The studies consider factors affecting the deci- garment factories in the post-Multi-fiber sions of firms to improve working conditions Agreement (MFA) period was the Better Work or to regress and retrogress (to pre-program program, which grew out of a trade agreement working conditions). The studies also explore between Cambodia and the United States. the link between working conditions and plant Under this agreement, the United States closures and whether or not a business reason allowed Cambodia better access to United exists for improving working conditions. The States markets in exchange for improved main conclusion from the empirical studies working conditions in the garment sector. demonstrates that improving working con- This program has put in place an institutional ditions and labor compliance did not under- structure that creates opportunities for collab- mine the competitiveness of Cambodian facto- oration amid the government, industry asso- ries; on the contrary, the improvements helped ciations, firms, and trade unions. The Better Cambodian factories to maintain a position in Factories Cambodia program is managed by the global economy. The following summarizes the International Labor Organization (ILO) and the main findings: supported by the government, by the Garment Manufactures Association of Cambodia GMAC, • Improvements in working conditions do and by the unions. not increase the probability of plant closure; moreover, in several cases, improvements The program works closely with may actually increase chances that a plant other stakeholders, including international will not close. buyers. It not only monitors and reports on • Reputation sensitivity is vital. Thus ending Figure 17: Coefficients On Wage Regressions the public disclosure requirements of the Better Work Program lowered compliance with labor standards. Female (Garment) • Once firms switch to better working condi- tions, they rarely go back and reverse the gains entirely. The pattern is similar to the forward direction of technological adoption. This finding supports the idea that these Female improvements do not undermine the man- agement and productivity of factories. All coefficients are statistically significant at 1%. Source: CSES 2004 - 2011 34 An argument against attempts to improve Future growth will not come from only working conditions in factories states the fol- agriculture. While Cambodia remains predom- lowing: improvements raise costs and is there- inately rural, sustained medium and long term fore non-competitive. The actual results of income growth has to be supported by non- studies, however, find little support for this agricultural activities. For those that do not argument. On the contrary, these studies’ have the option to start their own businesses, results suggest that improving working con- salaried labor is the best option. Therefore, ditions may improve factory competitiveness the government should be ready to step-up (including profitability). Currently, research efforts to promote not only industry but other is being conducted to formally evaluate this non-agricultural sectors that provide salaried hypothesis by comparing improvements in jobs. This study is not designed to provide rec- working conditions with productivity. ommendations related to economic develop- ment and industrial policy, but it recognizes There is an unfriendly business environ- their importance in improving well-being and ment for small firms in Cambodia. While reducing vulnerability. salaried employment dominates the urban labor force, a full third of it is engaged in self- The skill mismatch between supply and employment, mostly in small businesses. Small demand must be reduced to increase labor and micro enterprises represent 99.6 percent productivity within existing salaried jobs. Cambodian firms (2011 Cambodia Economic Overall improvement of the formal education Census). Still, Cambodia ranks 133 out of 185 sector is one way to help, but more specific countries in the Ease of Doing Business Index actions should include the following: (i) main- (World Bank Series). Most of the worst indica- stream good technical and vocational edu- tors for Cambodia are related to the rule of law cation and training programs; (ii) upgrade a (enforcement and application): they include small number of skills providers, including resolving insolvency, dealing with construc- non-formal training centers, in collaboration tion permits, enforcing contracts, and trading with local industries; (iii) consider methods across borders. Amid the infrastructure bottle- that encourage employers to invest in training necks, the cost of electricity and transport are their workforce; and (iv) explore a variety of highlighted as two of the four main obstacles financing tools to promote incentives for skills by firms (ICA 2012 survey). providers that deliver good results, including higher education institutions. (World Bank Policy Recommendations 2012 Matching Aspirations) Salaried jobs are better suited to take Best practices already used in Cambodia full advantage of many Cambodians labor should be promoted to improve employee by increasing their productivity and by pro- working environments and stability. viding a more stable source of income. To Policies and practices already under imple- get the most of salaried jobs, the following mentation in the apparel sector are the best four recommendations are needed to increase example for a set of conditions that benefit the number of salaried jobs, to increase pro- the working environment. Consequently, the ductivity, and to promote best practices from government can take actions on two fronts: some sectors into other sectors. (i) create opportunities for knowledge sharing 35 among companies that have already adopted The Role of the Rural Sector good practices for improving working condi- tions. These companies can then showcase Rural Population these strategies, and (ii) expand labor regula- tions into other sectors of the economy such Cambodia remains predominately a as food, entertainment, and tourism related rural country. Rural Cambodia makes up 90 businesses (the minimum wage and other percent of the country’s 1,621 communes. benefits associated with the Better Factories Moreover, the bulk of the population Cambodia Strategy can be used as model for lives within this 90 percent, and this has these other sectors). changed very little over time. According to the Cambodia Socio-Economic Survey, the The government should also consider rural population has decreased only two per- interventions to help household enter- centage points from 81 percent in 2004 to prises and micro-firms to expand their busi- 79 percent in 2011. Looking farther back and nesses. Both of these self-employment activ- using results from the population census, the ities should be viewed as a major source of rural population decreased by only 1.2 per- employment, but they also need to be restruc- centage points in ten years (from 81.7 percent tured to give workers an eventual, viable entry in 1998 to 80.5 percent in 2008). By 2011, 21 point into the labor market. Recent research on percent of all Cambodians (or about 3 million self-employment recognizes that “developing people) live in urban areas. Phnom Penh and promoting self-employment is therefore accounts for about half of this amount, while not a coping strategy, but a growth strategy” the other half lives spread across 25 cities and (Fox and Pave-Sohnesen, 2011). Improvement district towns throughout the country. in laws commonly regulating small and micro enterprises is among the most needed actions. There are two main reasons why urban- Distribution, reduced electricity costs, and ization remains low in Cambodia. First, improved transport infrastructure also need to although 3.5 million Cambodians were be prioritized to improve Cambodia’s business migrants in 2008, most were moving from rural environment. areas to other rural areas. Moreover, much of this took place during the eighties and first part of the nineties. After discounting for dif- Figure 18: Share Of Rural And Urban Population ferent population growth rates, the urban share increased from 18.3 percent in 1998 to only 19.5 percent in 2008. (Figure 18) Second, it is still easy for urban migrants— Urban either short term or long-term migrants—to return to their rural homes if a shock occurs, Rural suggesting that rural areas provide a safety net during crises. This was clearly evident in CSES Census CSES CSES 2009 when the global financial crisis affected 2004 2008 2009 2011 the garment and footwear, construction, Sources: NIS(2009) and WB Staff Estimates using CSES 36 and service industries. Many of these laid-off poverty reduction in other urban areas was workers returned to their families in rural 1.5 percentage points for every year in 2004- areas, waiting for an opportunity to return to 2007; 11.2 percentage points for every year in the cities. This caused a temporary increase of 2007-2009, and then 1.7 percentage points for 1 percent in the rural population in 2009. every year in 2009-2011. However, the pace in Phnom Penh was different. Poverty reduced Evolution of Rural Poverty 2004 to 2011 4.4 percentage points for every year in 2004- 2007; 0.8 percentage points for every year Rural poverty remained high in 2007, in 2007-2009; and 1.4 percentage points for but fell sharply in 2009 before leveling off every year in 2009-2011. (Figure 20) again in 2011. Poverty estimates using the Cambodia Socio-Economic Survey showed Because about 80 percent of the population that the proportion of people living below the and 90 percent of the poor live in Cambodia’s rural poverty line fell from 59 percent in 2004 rural areas, rural poverty reduction has driven to 58 percent in 2007. The proportion dropped most of the overall poverty reduction in sharply to 28 percent in 2009 before leveling off to 24 percent in 2011. Figure 19: Poverty In Rural Cambodia Poverty reduction in rural areas exhib- ited three clear trends from 2004 to 2011. At first, it declined slowly at 0.4 percentage points each year from 2004-2007, but then began to accelerate substantially at 15.2 percentage points every year from 2007-2009. The third trend slowed to 1.9 percentage points every year for 2009-2011. (Figure 19) In comparison, trends for other urban areas were similar, if not as extreme. The pace of Source: WB Staff Estimates using CSES Figure 20: Pace Of Poverty Reduction And Per Capita Daily Consumption Changes 2004-2011 (a) Poverty reduction per year (b) Per capita daily consumption changes per year គិតជាភាគរយ គិតជាភាគរយ Phnom Penh Other Urban Other Rural Phnom Penh Other Urban Other Rural Source: WB Staff Estimates using CSES 37 Cambodia. Total poverty reduction for all of year. In comparison, per capita daily consump- Cambodia parallels poverty reduction for its tion of people living in Phnom Penh and other rural areas: slowly in 2004-2007, very quickly urban areas during the same period increased in 2007-2009, and again slowly for 2009-2011. by 40 percent and 24 percent, respectively— Cambodia’s overall poverty reduction pace from CR 11,330 to CR 15,886 and from CR 8,183 was 1.1 percentage points for 2004-2007; 13.1 to CR 10,163, respectively. percentage points for 2007-2009; and 1.7 per- centages points for 2009-2011. By 2009, the average per capita daily consumption of people living in rural areas Migration had almost no effect in the (adjusted for inflation) increased dramatically poverty reduction. To account for the impact to Riel 8,183—an increase of 39 percent in two of net rural to urban migration, the Ravallion- years, or 19.5 percentage points each year. This Huppi decomposition was estimated using is very impressive compared to the changes in poverty results from 2004 and 2011. The per capita daily consumption of people living results of the decomposition analysis indicates not only in Phnom Penh—where consumption that population shifts account for only 0.5 decreased by 1 percent in two years, but also percent of poverty reduction. of people living in other urban areas—where the consumption increased by 25 percent or Consumption 12.5 percentage each year. Poverty reduction trends are reflected By 2011, the average per capita daily by per capita consumption. Comparable consumption of people living in rural areas estimates from the Cambodia Socio-Economic (adjusted for inflation) declined to CR 8,815—a Survey (CSES) indicated that the average per drop of 6 percent in two years, or 3 percent- capita daily consumption of people living in age points each year. Despite these declines rural areas (adjusted for inflation) increased in average consumption, the drop in the rural from CR 5,641 in 2004 to CR 5,875 in 2007, rep- poverty level was possible because of redistri- resenting an increase of 4 percent in five years, bution of consumption that favored poorer or by a mere 1.3 percentage points for each segments of the population. (Figure 21) Figure 21: Per Capita Daily Consumption Of People Living In Three Main Regions 2004-2011 Other Rural Phnom Penh Other Urban Riel/day (thousand) Riel/day (thousand) Riel/day (thousand) Source: WB Staff Estimates using CSES 38 Housing Conditions and Consumer households typically make their living from Durables in Rural Households multiple sources; this could help to identify main income sources that rural people use to The changes in the living standards of cover parts of their consumption. Estimates the rural population were reflected through of household income from CSES suggested a not only the increase in consumption, but trend broadly consistent with consumption also the improvements in housing and trends. The estimates showed that the average ownership of consumer durables. Estimates per capita daily income (adjusted for infla- of housing ownership from the CSES showed tion) of people living in rural areas rose from that the number of rural households with CR 4,100 in 2004 to CR 5,884 in 2009—a rise thatched-roofs declined from 24 percent in of 44 percent in five years or 8.7 percent for 2004 to 14 percent in 2011. Many rural house- each year. By 2011, the income continued to holds now have permanent roofs made of gal- increase to CR 6,144, although the growth rate vanized-iron and concrete or fibrous cement. was 4 percent over the same two years. The number of households with galvanized- iron roofs increased from 31 percent in 2004 to Income from Agricultural Crops 48 percent in 2011. Households with concrete or fibrous cement roofs increased from 5 Agricultural crops were the main contrib- percent in 2004 to 10 percent in 2011. utor to the sharp rise in the total household income. Economic activities and livelihoods Ownership of key durables such as tele- in rural Cambodia rely heavily on agricultural visions, mobile phones, and motorbikes also activities. Estimates from CSES indicated that increased considerably. In 2011, 61 percent real average per capita daily income of people of rural households reported that they had living in rural areas from agricultural crops a television; 85 percent a cellphone; and more than doubled within five years—from 64 percent a motorbike. This represents an CR 507 in 2004 to CR 1,101 in 2009. This repre- increase of 16 percentage points from 2004 sents an increase of 117 percent from 2004 to for televisions; 77 percentage points for cell- 2009, or 23 percentage points each year. Since phones; and 37 percentage points for motor- 2009, the per capita daily income from agricul- bikes—a substantial increase in the owner- tural crops continued to rise by 12 percent to ship of consumer durables; thus, an important CR 1,232 in 2011. (Box 2) change in the living standards of the poor The rise in income from agricul- Rural Income tural crops is mostly from rice. Cambodia is a net exporter of rice, and rice is grown on In rural areas, the changes in living 84 percent of the country’s nearly 3 million standards and poverty reduction can hectares of cultivated land. Of all people living be explained by household income and in rural areas, 85 percent are rice farmers. With sources of income. An alternative way of this dominance of rice, the country is esti- measuring living standards is by looking at mated to have a paddy surplus of 3-4 million not only the incomes of households but at tons, beginning in 2008 (1.5-2 million tons the varied sources of such incomes. Rural in milled rice equivalent). Prior to 2008, the 39 Box 2: Household Income vs. Household Consumption In many developing countries, including The second reason is that a large number of the Cambodia, it is common for estimates of house- population is self-employed on farms or in small hold income to be lower than consumption. This family business—many of them do not have paradox stems from a number of reasons: First, regular income, nor do they have proper records parts of household consumption, particularly food, of income, which makes data gathering very dif- are produced at home or by in-kind transactions. ficult. Third, people generally underreported their income because of privacy concerns. Per capita daily consumption and income 2004-2011 Comparisons of income with consumption Per capita can create paradoxical results, and such com- consumption parisons should treated with caution. Because Riel per day income tends to fluctuate substantially—even within a short time period—for comparison Per capita purposes is better to use longer periods of time. income In this report, comparisons of income and con- sumption trends were done for the periods 2004- Source: CDRI (2013) using CSES 2009 and 2009-2011. paddy surplus was estimated at 2 million, 2.2 farm somewhere between 0.5 to 1 hectare of million, and 2.6 million tons in 2005, 2006, farm land. (Table 16) and 2007, respectively. At an average price of US$250 per ton of paddy, the paddy surplus In 2009, estimates from CSES showed that translates into US$750 million to 1 billion for the about 37 percent of rural households operated rural economy beginning in 2008. (Table 15) agricultural land in excess of one hectare; hence, they have a rice surplus. The survey Of these rural households, 37 percent are also found that 21 percent of rural households net rice producers; 17-41 percent are net con- have agricultural land ranging from 0.51-1 sumers; and the remaining is neutral. To feed hectare, which may produce enough rice for human consumption for a household of five consumption only. The net consumers make (consuming approximately 709 kilograms of up between 17-41 percent, which also com- milled rice or 1.4 tons in paddy equivalent prises the 17 percent of rural people that do a year according to the estimates from the not operate any agricultural land, and the 24 CSES’s 2009), calculations suggest that house- percent that operate less than half of a hectare holds must harvest slightly over 0.5 hectares of land. Therefore, rising rice prices benefited of farm land at an average yield of 2.7 tons per roughly over one-third of rural households, hectare. But, to meet all consumption needs— about another one-third had no effect, and including human and animal consumption, the remaining one-third experienced negative seeds, and crop losses—the household must effects. 40 Table 15: Income Composition And Changes Income (Riel per day) % Changes 2004 2009 2011 04-'09 09-'11 Wage labor 918 1,355 1,835 48% 35% Agricultural crops 507 1,101 1,232 117% 12% Livestock 377 295 317 -22% 8% Fishing 201 219 151 9% -31% Forestry & hunting 286 319 324 12% 2% Non-farm self-emp. 856 1,398 1,211 63% -13% Remittances & transfers 957 1,197 1,073 25% -10% Total 4,100 5,884 6,144 44% 4% Source: CDRI (2013) using CSES Table 16: Agricultural Land (% Rural Households) CSES Hectares CDRI 2002 2004 2009 2011 None 20 21 17 16 0.01-0.5 25 25 24 26 0.51-1.0 20 21 21 22 ha 1.0 > 35 33 37 36 1.5 1.01- - 11 11 11 1.51-2.0 - 8 9 9 2.0 > - 13 17 16 Total 100 100 100 100 Source: Helmers et. al. (2004) and WB Staff CSES The gains in rice are mainly from higher expanding and because large farms typically prices and from increased yield. Although belong to commercial farmers. Estimates from the country’s total cultivated rice area has CSES showed that the average farm size for expanded from 2.3 million in 2004 to 2.6 in rural households remained at 1.2-1.4 hectares. 2007, and to 3 million in 2011 for both wet and (Figure 22) dry seasons (resulting from land that has been cleared from degraded forests or demining), But the rise in prices allowed farmers to the average farm size in rural areas remains boost income substantially. From 2004 to 2009, generally unchanged. This situation stems, international prices of rice increased by 134 in part, because new households have been percent in Thailand and 88 percent in Vietnam. 41 Similarly, during the same period, local prices Income from Wage Labor of rice in Phnom Penh’s retail markets increased by 119 percent. The main input of rice farming Wage labor is the second source of in Cambodia is labor; therefore, the increase in incomes’ increases in rural households. accompanying fertilizer and fuel prices had a The increase in labor wages is because of the minimal effect on farmer’s profit margins. Thus upsurge in the number of working people from farmers began to gain substantially from the each household, because of working longer rise in prices beginning in early 2008 because hours, and because of increasing wage rates. farmers began to harvest 2006/2007 crops in The average per capita daily income of people December. The increase in yield has also con- living in rural areas from wage labor rose by tributed to this gain. For wet season rice, the 48 percent from 2004 to 2009, or 9.5 percent yield increased from an average of 1.7 tons per each year—from CR 918 in 2004 to CR 1,355 in hectare in 2004 to an average of 2.6 tons per 2009—it continued to increase to CR 1,835 in hectare in 2009—rising by 52 percent. For dry 2011. The rural people of Cambodia are now season rice, the yield increased from 3.5 tons working longer hours, as well as working more per hectare in 2004 to 5 tons per hectare in than one job. A decomposition analysis found 2009—rising by 44 percent. (Box 3) that the bottom 20 percent of the rural popu- lation had an increase of 69 percent in wages. These increases in household earnings were Figure 22: Average Paddy Yield Per Hectare 54 percent from the number of paid employ- Wet season Dry season ees and 9 percent from the escalation of the number of hours worked. The remaining 6 percent stems from factors such as house- hold size, hourly rate, and the interaction amid these factors. (Figure 23) Daily wage rates in rural areas more than doubled from 2007 to 2012. From 2007 to 2008, the daily nominal wage rate increased sharply. Source: MAFF’s Annual Reports In some cases, these increases exceeded Figure 23: Increases In Average Daily Wage Rates In 2007 And 2008 Riel/day Male Riel/day Female Ploughing Transplanting Harvesting Construction Ploughing Transplanting Harvesting Construction Source: Knowles (2010) using CSES 42 Box 3: Two Stories From The Ground On The Effects Of High Rice Prices High rice prices helped Battambang to In Kandal Province, high rice prices and reclaim its renown as Cambodia’s rice basket developments of infrastructure created incen- tives for people to invest in their farms Before the 1970s, Battambang province was known as Cambodia’s rice basket. However, after This story comes from Rokarchunloeung the Khmer Rouge lost power in the capital, civil commune in Kandal province. Before 2011, the war continued to rage in the 1980s and 1990s, only access to the commune was crossing the preventing Battambang from reclaiming its legit- Mekong River with ferries. And it was expensive, imate place in the Cambodian economy. Instead, costing more than US$1 (CR 5000); moreover, Kampong Cham province was much more pros- the trip took over an hour. In 2011, a new bridge perous due to its role as a trade gateway with was built crossing the river, making the trip much Vietnam. faster, cheaper, and easier. Furthermore, roads that connect the commune—as well as road within the Visits to Battambang from 2003 to 2005 gave commune—were improved significantly with the the impression of a quiet town, with old French rehabilitation of National Road 8 and the building style buildings. But this changed with the increases of new commune and village roads. in the prices of rice in 2007—which then doubled in 2008. The Cambodian Rice Millers Association Before the doubling of rice prices, farmers did and Vietnamese rice traders have become com- not seek extension services and focused only on petitors with Thai rice traders, benefiting farmers local rice varieties. But when they began to get with multiple buyers and subsequent higher higher prices, they began competing for water, prices. Farmers, formerly disconnected with the seeking extension services, putting in more efforts world market, have now shifted from subsistence to double their yields, and choosing rice varieties farming to commercial farming. in demand by the market. Source: Extract from Hang Chuon Naron’s Written Comments for Decision Meeting of Cambodia Poverty Assessment 2013 on May 2, 2013 adjustments for inflation, rising by 57 percent 41 percent to CR 10,000; and construction by from 2007 to 2008 in rural Cambodia. This 37 percent to CR 10,000. Female workers daily has helped many workers—particularly the wage rates for plowing surged by 120 percent landless in rural areas—to partly offset the to CR 6,000 a day; transplanting by 41 percent impact of high food prices. Estimates from to CR 10,000; harvesting by 44 percent to CR CSES showed that daily wage rates for many 10,000; and construction by 23 percent to CR types of work in rural areas rose suddenly; 8,000. moreover, the rise was observed for both male and female workers. From 2007 to 2008, male In 2012, the daily wage rates continued to workers daily wage rates for plowing surged rise dramatically, not only because of rising by 44 percent to CR 18,000 a day; transplant- costs of living in the country, but also because ing by 45 percent to CR 10,000; harvesting by of costs of labor in neighboring countries. For 43 instance, in rural Cambodia, the daily wage interviewees in household surveys—the rates for plowing increased to CR 20,000- recipients of remittances—tend to underre- 25,000 a day; transplanting to CR 12,000- port cash received from relatives; in addition, 15,000; and construction to CR 20,000-25,000. Cambodia does not have a sizeable internal (Figure 24) cash transfer program. Estimates from CSES showed that average per capita daily income Income from Non-Farm Self-Employment from remittances and transfers increased by 25 percent over five years, or 5 percent each year The third contributor was income from from 2004 to 2009. Compared to the growth of non-farm self-employment. One-third of other income sources from agricultural crops, rural households engaged in non-farm busi- wage labor, and non-farm self-employment, nesses in 2009—either as their primary or the contribution of remittances and transfers their secondary job. Estimates from CSES to household income is small. showed that average per capita daily income from non-farm, self-employment rose by 63 Perceptions from Rural Households percent over five years, or 12.7 percent each year—from CR 856 in 2004 to CR 1,398 in 2009. Rural people perceive that they can However, in 2011, the income from non-farm, attain higher income from land and labor self-employment declined to CR 1,211, or by than other assets. A rural survey conducted 13 percent from 2009. in October of 2012 (interviewing 1,560 house- holds in 78 villages in 14 provinces) revealed Income from Remittances and Transfers that people were generally optimistic about the rise in income sources from farms and Income from remittances and transfers wages. Of 1,535 households (98 percent of all is by far the smallest contributor to the people interviewed) that engaged primarily increase in household income from 2004 in agricultural activities, 30 percent thought to 2009. International experience shows that their income from agricultural activities Figure 24: Decomposition Of Factors Contributing To Wage Labor Income, Cambodia 2009 Contribution to wage income # of paid employees Paid employees hours worked Household size Wage of paid employees Interaction Poorest Source: Knowles (2012) using CSES 44 would increase in next three years because standard by international buyers. As a result, they thought productivity and prices would many informal rice traders buy paddy rice from increase. About 37 percent thought that it Cambodian farmers and sell it to Vietnamese would stay the same, and 24 percent thought and Thai traders. They then add value by further their income would decline. The perception processing—including milling, packaging, and of declining income stems from concerns commercialization. (Table 17) about bad weather and crop diseases, as well as increases in costs. The perceptions of agri- Policies of the Royal Government culture income appear to concur with both of Cambodia net rice producer and consumer estimates: 37 percent of rural households are net rice pro- The remarkable gain in rural areas ducers; 17-24 percent net consumers; and the is due, in part, to the effects of agricul- remaining the neutral. tural prices—and rice prices in particular; however, policies of the Government have Wages from labor showed a somewhat also played an important role. During the similar trend: of 1,126 respondents (72 percent 2008 global food crisis, the government did of all people interviewed) who engaged in not make any price interventions. Instead, waged employment, 40 percent thought their it took an open approach to price and trade income from would rise; 34 percent thought control, allowing for greater competition. it would remain the same; and 16 percent Since 2008, there has been strong competition thought it would decline. among local traders and international traders in buying paddy at the farm gates from neigh- Characteristics of the Rice Sector boring countries. As a result, many farmers do not have to transport their paddy to Most Cambodian farmers plant rice only markets—this has proven to be more helpful once a year. In many farm plots across various than harmful. parts of the country, rice is grown once a year— mostly during rainy season. This is because of Moreover, the government’s devel- a lack of water during the dry season and the opment priorities—focusing on infra- continued use of traditional technology. As structure, agriculture, and rural develop- a result, rice production and yield is far from ment, which are articulated in the National its full potential. Neighboring Thailand and Strategic Development Plan—have been Vietnam grow more than once per year and maintained. Government capital spending their yields are much higher. Table 17: Rice Milling Capacity (Tons Of Paddy Per Hour), Cambodia Rice milling capacity is improving, but remains limited in Cambodia. Most of   Milling Polishing Cambodian rice flows to neighboring countries Mid 2009 96 72 in paddy form. Despite rapid modernization in recent years, local mills are unable to meet Mid 2010 245 201 demand because of cash-flow limitations, lack of storage capacity, and inability to meet quality Mid 2011 322 305 Source: Tom Slayton and Sok Moniroth (2012) 45 from the Ministry of Agriculture, Fisheries, for rural people to basic services and markets. and Forestry (MAFF), the Ministry of Water Irrigation facilities were reconstructed, result- Resources and Meteorology (MOWRAM), and ing in 42 percent of rice crops with pumped the Ministry of Rural Development (MRD) water in 2010, compared with only 25 percent in focused on developing agriculture and 2005. The percentage of rural households with rural infrastructure. Their capital spending access to improved drinking water increased increased from CR 88 billion in 2004 to CR 228 from 40 percent in 2000 to 58 percent in 2010; billion in 2008 and to CR 612 billion in 2010, the percentage of rural households with elec- before decreasing to CR 216 billion in 2011. tricity increased from 2 percent in 2004 to 24 This represents an increase from 4.1 percent percent in 2011. (Figure 25) of the government’s total spending in 2004 to 4.7 percent in 2008 and to 9.5 percent in 2010 Remaining Challenges before decreasing to 2.8 percent in 2011. Despite remarkable gains, the living The Government’s investment has not conditions of people in rural areas still only shown results, but also improved living require attention—and the task may be conditions. By 2011, about 76 percent of the more demanding. This dramatic improve- country’s rural roads were rehabilitated from 69 ment stems from a relatively low level—the percent in 2005, which helped improve access task of poverty reduction is not over. In some Figure 25: Improvements Of Key Rural Infrastructure In Rural Areas (a) % of rural roads rehabilitated (b) % of rice crop covered by irrigation facilities Source: RGC (2012) Source: NGO Forum on Cambodia (2011) (c) % of households with improved drinking water (d) % of rural households use electricity for lightning Source: WHO/UNICEF’s JMP Source: WB Staff Estimates using CSES 46 respects, the task has become even more diffi- is also limited. But its results are good indica- cult. Poverty is still mostly a rural phenomenon tors of changes over time although it should with 90 percent of the poor living in rural areas. be interpreted with care. To understand the But the poor today are different from the poor drivers of poverty reduction, a simulation was in 1990s: they seem to have not only higher created using many of the parameters from expectations, but also greater aspirations. Ten the survey (micro-data) and results from the years ago, the sole priority was enough food labor and rural analysis. This simulation iden- for the family. Currently, priorities—in addition tifies characteristics that are linked to poverty to food—consist of motorbikes, TVs, mobile reduction over specific time periods and phones, children’s education, and quality regions (Phnom Penh, other urban, and rural). healthcare. Special care was given to identify character- istics that were exclusive to each time-region The sluggish pace of poverty reduction combination. in 2011 is a good reflection of the fact that poverty reduction is far from complete. The From 2004-2011, most of the poverty fast poverty reduction in 2009 could have been reduction in Cambodia came from improv- a one-time event, based, in part, on taking ing rural household conditions, as well as advantage of serendipitous factors. Recently, it changes that were time-specific from 2007 has been proven difficult to endure. Estimates through 2009. From all poverty reduction show that the pace of poverty reduction in rural over the seven year period, 3.6 percent was areas has slowed to an annualized rate of less from Phnom Penh, 8.1 percent was from other than one percentage point. When compared to urban areas, and the remaining 88.3 percent 15.2 percentage points for each year from 2007 was from rural Cambodia (Figure 26a). to 2009, it is clear that more needs to be done. Furthermore, most of the poverty reduction (80.0 percent) occurred from 2007 through Drivers of Poverty Reduction 2009. The remaining reductions were evenly divided from 2004-2007 (9.7 percent) and from Based on results illustrated in this chapter, 2009-2011 (10.4 percent) (Figure 26b). on results from the Cambodia Socio-Economic Surveys (CSES), and on results from gov- Households from ‘other urban’ areas ernment statistics, this section provides a present characteristics that are similar to link between poverty reduction and house- both Phnom Penh and the rural areas of hold characteristics (labor force, commod- Cambodia. For example, in many of the other ity market, agricultural production, and self- urban areas, rice fields and other agricul- employment) from 2004 through 2011. ture exist within a very short distance from the center of town (often less than a kilom- The CSES household survey (Living eter). At the same time, many of the indica- Standards Measurement Survey or LSMS type) tors from other urban households are closer is a cross sectional survey: it does not have to those in Phnom Penh than to those in rural panel data. Because of this limitation, it is diffi- households. This includes access to services, cult to establish a direct relationship between education levels, income per capita, and the poverty reduction and changes in household like. Disentangling the drivers of growth characteristics. Moreover, income information between urban and rural for the “other urban” 47 Figure 26: Contribution To Poverty Reduction In Cambodia 2004-2011 (a) By region (b) By time period P.P. O. Urban Total poverty reduction = 32.7 percentage points (first number) = 100% (second number) Source: World Bank staff estimates based on CSES households was not possible. Therefore, this When examining variations of consump- analysis treats “other urban” households as a tion expenditures across employment cate- combination of the two other regions: Phnom gories of household heads, the increase in Penh and rural. the share of paid employment is the main factor driving poverty reduction (Urban Urban Households 2004-2007 Labor Section). Changes in urban poverty in 2004-2007 contributed to a national poverty Industrial activity is linked mainly to urban reduction of 1.4 percentage points. households in Cambodia. From 2004 to 2007, per capita industrial Gross Domestic Product Rural Households Income 2004 to 2009 (GDP) grew 40.8 percent. This was followed by a decrease in 2008 through 2009, but it began The first factor explaining the increase recuperating in 2010, again reaching the levels in consumption for rural households from of 2007. In contrast, per capita agricultural GDP 2004 to 2009 is the price of rice. Households growth from 2007 to 2008 was 6.1 percent. improved their income because of improved The increased availability of salaried jobs in rice prices from 2004 to 2009. Better prices had urban areas has fueled the dramatic drop in two direct effects on household income: first, poverty rates in the past decade. Almost all the value per unit of production increased, and of the poverty reduction in Phnom Penh was second, total production increased owing to observed from 2004 to 2007, a period with a more planted area and increased use of inputs. fast increase in salaried jobs (10 percentage points). This was in stark contrast to rural areas: To estimate the impact of the rice price poverty decreased only 1.1 percentage points, increase, a simulation based on information and the share of salary earners increased by from the CSES and other government statis- less than one percentage point. tics was performed (Annex 3). The simulation’s 48 main results are based on an examination of a households (not only the reference group) reference group of households that were poor yields an estimate of general changes in the in 2007 but not poor in 2009. The simulation entire rural sector associated with overall presented in Annex 3 estimates the income growth. Income from non-agricultural, self- effect of the increased rice prices to be CR employment increased CR 542 per day per 683.72; it estimates the impact of increased person from 2004 to 2009 (but 2009 2011 productivity and area harvested at CR 661.94 showed a decrease). The rural section illus- per day per person at 2009 Phnom Penh CR. trates this explanation of income changes over time for rural households. (Figure 27) The second factor explaining the increase in consumption is income from Drivers of Poverty Reduction in 2004-2011 wages in agriculture. Because of data limi- tations, it was not possible to differentiate To estimate the impact of the drivers of the sources of wages between different agri- poverty reduction in 2004 through 2011 in cultural activities, but there is no doubt that rural Cambodia, the increases in sources of an important share of the wages was related income identified in Figure 27 were equated to rice. For example, rice represents about 80 to poverty reduction. Poverty reduction in percent of the total planted area in Cambodia, Phnom Penh in 2004 through 2007 was driven and about 80 percent of rural households by increase of salaried jobs. reported growing rice. Moreover, rice experi- enced the highest price increase from 2007 to Poverty reduction in urban households 2009 compared to all other locally produced outside of Phnom Penh is a combination of food products (reported in the Consumption urban and rural characteristics, and the drivers Price Index Basket). As rice prices increased of poverty reduction are divided between them. by 39 percent, other product price increases Half the poverty reduction in urban households were as follows: beef 25 percent, chicken 24 outside Phnom Penh was due to the same drivers percent, fish and seafood 24 percent, fresh of growth identified for rural households; the eggs 14 percent, fresh fruits 27 percent, and other half was due to increase in salaried jobs. vegetables 22 percent. Figure 27: Income Increases In Reference Group 2004-2009 From 2004 to 2009, agricultural income from wage labor increased 49.0 percent, Unidentified, equivalent to CR 450. However, from 2009 to Non-Farm Business, 2011, wages from agricultural activities in the reference group did not increase (Figure 27). Agric. Wage, The third factor that explains the increase in consumption is higher income Rice from small businesses. Non-agricultural, Production, self-employment income was available in the surveys for all rural households in the CSES Rice Price, from 2004 to 2009 and from 2009 to 2011. Using the average income increase for all rural 49 The main drivers of poverty reduction in in commodity markets. Some of the most 2004 to 2011 were higher rice prices (7.8 per- important reasons farmers were able to take centage points), increased rice production advantage of increased rice prices was the lack (7.6 percentage points), higher agricultural of policies that distort prices (most notably wages (5.2 percentage points), and non-farm price controls or price subsides). Moreover, businesses (6.2 percentage points). No drivers the lack of any tax on rice production benefit- were assigned to the remaining 4.4 percent- ted farmers greatly. Maintaining these same age points in poverty reduction (Figure 28). policies in the future would allow for a better flow of resources—both in and out of the agri- The direct impact of higher rice prices cultural sector—at the same time, it would and the increase in rice production was allow for improved allocation of farmers’ assets. responsible for almost half the reduction in poverty. Agricultural wages—also closely Continue improvement to rural infra- related to rice—contributed to 16 percent of structure, while using poverty to help set the total reduction in poverty; at the same time, priorities for future projects. Overall rural improved income from non-farm business was infrastructure projects, roads, communica- responsible for one-fifth of all poverty reduc- tions, and electrification allowed the poor to tion. Increases in salaried employment in urban take advantage of improved economic con- areas accounted for 4 percent of the decline in ditions from 2004 through 2011. It is recom- poverty. Finally, the simulation does not explain mended not only to continue expanding rural the remaining 14% percent of the reduction, infrastructure, but to give priority to regions but it occurred in 2009-2011. (Figure 28) like the Plateau-Mountains, which are associ- ated with higher poverty rates. Similar appli- Policy Recommendations cation of poverty considerations should be used to select the type of projects that would Maintain policies of low intervention have a greatest impact on the poor. Figure 28: Drivers Of Poverty Reduction In Cambodia 2004-2011 Non-Farm Business, Urban salaries, Farm Wage, 6.2, 19% 1.4, 4% 5.2, 16% Other factors Rice Production 4.4, 14% 7.6, 23% Rice Price, 7.8, 24% Note: First number is the percentage points reduction (total 32.7 percentage points), second number is the percent of total poverty reduction (total = 100 percent) 50 Increased profit margins from rice pro- include processing, packaging, distribution, duction were possible due to improved and marketing. Increasing the milling capacity productivity. Because a big share of the poor of the country would help, but it should go and vulnerable are involved in rice production, much further. For example, black pepper policies that increase productivity would be from Kampot province is produced, pro- pro-poor. This recommendation’s objective is cessed, packaged and commercialized within not to increase total production alone, but to Cambodia, and it has international recogni- so in a way that improves farmers’ net income tion as “Kampot Pepper”. Each step in the value without increasing risks. Thus the following chain ladder provides jobs and investment three recommendations should be under- opportunities for Cambodians, as well as for taken: (i) improve access to higher quality rice foreign direct investment. seeds, (ii) continue investing in water manage- ment projects to promote more than one crop Agriculture has lifted many rural people a year and to reduce the risk of flooding, and out of poverty, but relying solely on agricul- (iv) augment extension services and research ture for future improvements is not realis- efforts to improve agricultural practices for tic. For medium and long-term growth, non- farmers with less than two hectares of land. agricultural activities are vital for sustained The recommendations should be accompa- economic and wellbeing improvements in nied by improved access to credit programs. Cambodia. The country should be ready to boost efforts to promote both industrial Promote diversification of the agricul- development in urban areas and other non- tural sector. While rice production has been agricultural sectors in rural areas. To achieve one of the major drivers of poverty reduction, these economic and social goals, the future rural households—and especially the poor— policies of Cambodia must expand agricul- will benefit from a more diverse range of agri- ture for short and medium growth; it then cultural products. This would first require must shift to industry and service growth for finding good matches between the agricultural medium and long-term growth. activity (crops, animals, and the like) and the agro-ecological conditions in different parts of END NOTE the country (for example maize and soya in the uplands). Second, a wider selection of foods for 25. The consumption premium presented local consumption will improve both nutrition in Figure 13 should be interpreted with and health. Third, diversification provides some caution. The premium is not only gener- protection against product-specific shocks. ated for being salaried employed but also by several workers’ characteristics that Increase the value added to agricul- also explain the earnings profile. Yet the tural outputs already being produced in welfare premium (in Figure 13) suggests the country. Wage labor has been the second that salaried employment is an efficient most important driver of poverty reduction. vehicle to lift households out of poverty. Rural communities have the potential to retain more value from their agricultural produc- 26. For example, several developing coun- tion by engaging in other parts of the com- tries report official unemployment rates modities value chain. Value added activities of 1 and 2 percent (See Jobs Trend, Winter 51 2012. The World Bank.) hectares (WB, 2009). For rural Cambodia as a whole, the average farm size is 1.4 hectares. 27. We do not include the category “University and more”. The sample size for individu- 34. The 2008 was a rapid inflation year and it als in this category is small; in 2009, this was the year of global food and fuel crisis. category included only 1.5 percent of indi- viduals between 15 and 65 years old. 35. Wage data collected by World Food Program in 12 provinces since 2010, where it has its 28. Completion of primary education was regional offices also suggested a trend of excluded from the second regression every rapid rising in wage rates. For example, year and no education was excluded in 2010 agricultural wage in rural areas increased and 2011. This is necessary to avoid multi- by 15 percent from 2010 to 2011 and con- collinearity. Because primary was excluded, tinued to increase by 16 percent from 2011 the results obtained for other levels are in ref- to 2012; non-agricultural wage in rural erence to primary education students. For areas increased by 16 percent from 2010 to example, wages for people with lower 2011 and by 13 percent from 2011 to 2012. secondary are 16.1 percent higher than those In urban areas, agricultural wage increased with only primary education in 2011. by 14 percent from 2010 to 2011 and by 18 percent from 2011 to 2012; non-agricultural 29. This extra set is also known as the covar- wage increased by 8 percent from 2010 to iates and included gender, economic 2011 and by 12 percent from 2011 to 2012. sector, region, marital status, age, and age squared. All the covariates from the extra 36. It is common that rural household have set were highly significant at 1% (p ≤ 0.01). multiple jobs because relying on one job is not sufficient to support living standard. 30. All regressions control for education, sector of employment, region, marital status, age, 37. In addition to this, other priorities in and age square. The coefficients presented National Strategic Development Plan in Figure 16 are statistically significant at 1 2006-2010 are health and education; the percent. plan was subsequently updated to 2009- 2013 to align with national election cycle, 31. World Bank (2011) the Poverty and Social but priorities were maintained. Impacts of the Global Economic Crises. 38. This figure includes all the poverty reduc- 32. Additional explanations of redistribution tion in Phnom Penh and half the reduction that make poverty continue declining in other urban areas. despite a drop in average consumption could be found in chapter 2. 39. Wage includes any income from agri- cultural paid jobs (daily wages or salary 33. This is also true for poor households. The earnings). Actual earnings were in refer- average farm size among rural poor was 1.5 ence to last month. 52 Reaching The Poor And Vulnerable Chapter 3 Cambodia has made important strides both administered or funded by the government in primary education investment and in access. (See Box 5). In order to meet the new demands Primary education is now nearly universal— and challengers, the government has to even in the more remote areas of the country. continue investing in the three main social But major challenges persist. These problems pillars: health, education, and social protection. include late entry into school, dropouts (espe- cially at the secondary level), and poor quality Education of education—all of which are new challenges for the government. Secondary education and Education Sector Background health facilities are not nearly as accessible as primary education. Basic coverage (prox- Cambodia’s education system is mostly imity) and access (cost) reduce the usage of public, regulated by the following six govern- many of these services, explaining some of ment institutions: The first is the Ministry of the low outcomes indicators. Nevertheless, Education, Youth and Sport (MoEYS)—which is improvements have been detected through- the overarching national institution for educa- out all sectors—with the noticeable exception tion. It provides education services, monitors of nutrition, which has shown no observed quality and access at all educational levels, progress over the last five years. and ensures that basic conditions are met. The second is the Public Higher Education Social protection—a relatively new concept Institutes (HEIs), which are under the jurisdic- in Cambodia—lags even further behind. With tion of 14 Ministries and entities.40 It provides the exception of scholarship programs target- tertiary education services offering bachelor, ing women and subsidized healthcare target- master, and doctor degrees. The third are the ing the poor, only small programs like food various private institutions that receive no state distribution and flood assistance are either subsidies. The fourth is the Ministry of Labour 53 and Vocational Training, charged with provid- Government Education Policy ing vocational training services at the second- ary and tertiary levels. The fifth is the National Since the mid-1990s, development Early Child Development Committee (ECD); policy in Cambodia has been guided by five it coordinates public and community-based, year plans, originally named the Economic early childhood education in conjunction with Development Plans and later, the National MoEYS, the Ministry of Interior, the Ministry Strategic Development Plans (NSDP). In 2001 of Health, and the Ministry of Women Affairs. in response to these national plans, MoEYS The sixth is the Accreditation Committee in developed the first, five-year Education Cambodia (ACC), which operates under the Strategic Plan (ESP). To date, this plan has been Council of Ministers and is directly supervised substantially revised twice (ESP 2006-2010 by the Prime Minister’s Office: it is responsible and ESP 2009-2013).41 for the accreditation of both public and private higher education institutions. The policy development process began in 2001 with the adoption of the first Education The various levels of Cambodian education Strategic Plan 2001-2005, the Education Sector are organized into the following series of cycles: Support Program 2001-2005, and the develop- (i) home-based child care for children under six ment of the National Education for All Action years old, (ii) preschool for children from three to Plan in 2003. These three programs outlined a five years old, (iii) primary education for children number of educational goals for all Cambodian six to eleven years old, (iv) secondary education children by 2015. The first Education Strategic that is divided into lower-secondary and upper- Plan (ESP) set the government’s goal and secondary for 12 to 17 year-olds, and (v) tertiary mid-term vision for 2001-2005 as follows: “… or higher education for pupils above 17 years to develop an inclusive, easily accessible and old. In parallel to the general stream, students high quality service which is available to all, can select technical or vocational training at the independent of wealth, gender, ethnicity, and upper-secondary level. This is available for 15 to physical and mental well-being” (page 7, MoEYS 17 year olds, tertiary students, and all students 2001). In addition, a work plan was created to older than 17 years (Table 18). The Ministry of outline specific areas of engagement. Education, Youth, and Sport (MOEYS) is also responsible for not only non-formal education The second version (Education Strategic (equivalency programs, literacy, and commu- Plan 2006-2010) set out to achieve its goals nity learning centers), but also teacher training. through implementing the following three Table 18: Formal Education Ages, Levels, And Grades In Cambodia Children age 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 + Secondary: Secondary: Upper & Tertiary or Education level Preschool Primary Technical/ lower Higher educ. Vocational Education grade Child care 1 2 3 4 5 6 1 2 3 4 5 6 1 2 3+ 54 national policies: (i) achieving equitable students in 2011 (Figure 29). Over the past access to education services, (ii) improving the decade, enrollment of private tertiary students quality and efficiency of education, and (iii) increased by a factor of ten from 8,419 students strengthening institutional capacity to deliver in 2001-2002 to 146,834 in 2011-2012. Primary education. Subsequently, the third version and secondary education (accounting for (Education Strategic Plan 2009-2013) aimed to about 92 percent of all Cambodian students in accelerate efforts within the sector to meet the 2011) remains almost entirely public with only Cambodian Millennium Development Goals marginal increases in private enrollment. and the Education for All National Plan 2003- 2015. The three priority areas of the second Education Outcomes education plan remained intact, for the post part in the Education Strategic Plan 2009-2013, Literacy but it placed a special emphasis on expanding early childhood education; secondary educa- Literacy has increased, regardless tion; and non-formal, technical and vocational of age, gender, location, or residence education. of Cambodians. Literacy rates have been steadily improving from 64.2 percent in 2004, to Private Sector Education in Cambodia 70.6 percent in 2011. Improvements in literacy rates from 10 to 12 percent were reported in The private sector is increasing its role all age groups, urban, and rural households. in preschool while dominating tertiary This further includes men and women, as well education. Traditionally, the public educa- as both the richest and the poorest quintile of tion sector played the lead role in Cambodia’s the population (Table 19). The largest literacy education system. However, private education increases were in those of the elderly in urban is increasingly growing in importance, espe- areas and in the richest quintile (25.6 and 17.1 cially in pre-primary (it has more than doubled percentage points, respectively). For young from 7.7 percent in 2004 to 19.1 percent in people, the largest improvements were for 2011). Currently, private tertiary education women and for the poorest quintile (about dominates, accounting for 71.9 percent of all 15 percentage points each). There is a strong Figure 29: Public Private School Enrolment Distribution, Cambodia 2004-2011 7.7% 19.1% 1.3% 1.8% 1.1% 1.3% 74.8% 71.9% Private Public Preschool Primary Secondary Tertiary Source: CSES 2004 and 2011 55 correlation between increasing levels of improved for Cambodians who are 25 education (including literacy) and decreasing years and older, significant education dis- levels of poverty (World Bank 2007). parities remain an issue. Adults’ education level is a product of past performance and By 2011, there was no longer a literacy improvements—both of which require many gender gap for people from 15 to 24 years years to have any impact on the entire group. old. Indeed, female gender literacy has not Nevertheless, today’s labor productivity is only improved from 76.8 percent in 2004 to linked to the human capital stock and should 91.3 in 2011, but reduced the gap to less than be considered when evaluating Cambodian’s 1 percentage point from the original 7.4 per- potential. Therefore, it is important to note the centage points. The gap for young people— lower education attainment by women, rural the group reflecting more recent education residents, and the poor (Figure 30). Among policies— has also reduced amid both urban rural and poor households, disparities increase and rural and the richest and the poorest after grade three. For example, the completion (Table 19). of grade six by rural people is less than half that of urban people; among the poorest quintile, Education Attainment completion of grade six is less than one-third of that of the richest quintile.42 Although education attainment has Table 19: Literacy Rate By Age Groups, Urban-Rural, Gender, And Poverty, Cambodia 2004, 2011 Youth (15-24 years old) Adults (15 years and older) Elderly (65 years and older) Groups 2004 2011 Change 2004 2011 Change 2004 2011 Change Urban 89.7% 96.9% 7.2% 81.7% 92.3% 10.6% 45.5% 71.0% 25.6% Rural 78.5% 90.2% 11.6% 66.5% 75.7% 9.2% 31.7% 41.8% 10.0% GAP 11.2% 6.8% -4.4% 15.2% 16.5% 1.4% 13.7% 29.2% 15.5% Male 84.2% 91.9% 7.7% 80.2% 87.6% 7.4% 63.7% 75.8% 12.1% Female 76.8% 91.3% 14.5% 59.4% 72.1% 12.7% 13.1% 26.0% 12.9% GAP 7.4% 0.6% -6.8% 20.9% 15.5% -5.3% 50.6% 49.8% -0.8% Richest 93.8% 97.5% 3.7% 84.5% 90.3% 5.8% 44.4% 61.5% 17.1% Poorest 65.5% 80.6% 15.2% 53.6% 64.6% 11.0% 22.4% 32.2% 9.8% GAP 28.3% 16.9% -11.4% 30.8% 25.7% -5.1% 22.0% 29.3% 7.3% All 80.5% 91.6% 11.1% 69.0% 79.3% 10.3% 33.8% 45.9% 12.1% Source: World Bank staff estimates based on 2004 and 2011 CSES 56 Government policies to promote girls’ 96 percent have completed first grade (76 education levels successfully achieved percent for older adults), 75 percent have parity with that of boys’ levels. Scholarships completed sixth grade (38 percent for older and other education programs targeting girls adults), and 36 percent have completed lower- over the years have improved girls’ school secondary school (17 percent for older adults). enrollment and completion to the point that Differences according to place of residence (for by 2011,no difference in the level of education the most part, urban and rural) have reduced remains amid young men and women from 15 for the first seven years of education for an and 20 years old (Figure 31). average of 17 percentage points, but there has been no change amid later grades. The educa- Education attainment for young adults tion gap between the poorest and richest has has improved for all groups and dispari- also been reduced by 17 percentage points— ties have been reduced. Young adults from but only for the first five years of education. 15 and 20 years old have shown remarka- The richest students have increased their par- ble improvements in education; for example, ticipation in grades 6 to 12 much more than Figure 30: Education Attainment By 25 Years And Older a. By Year b. By Place of Residence in 2011 Urban Rural School year completed School year completed c. By Gender in 2011 d. By Wealth in 2011 Male Poorest Female Richest School year completed School year completed Source: World Bank staff estimates based on CSES 57 the poorest students. In fact, the gap has the poorest children in Cambodia). Children actually increased by an average of 14 per- from other ethnic groups show no difference centage points (Figure 31). than average Cambodian children in primary education attainment, but they show much Young children from local ethnic groups lower secondary attainment. have exhibited lower education attain- ment at all grades from grades 1 through 12. Enrollment Rates There is a distinct disadvantage for kids from local ethnic minorities in Cambodia. This starts The two most common measures of educa- from first grade with attainment levels of only tion enrollment are gross enrollment and net 75.5 percent, which is substantially lower than enrollment rates. The gross enrollment rate is children from even the poorest households (90.8 defined as the proportion of students attend- percent attainment). The problem extends to all ing school (of any age) divided by those that levels of primary education for ethnic minority should be in school—according to age. The children; they have lower education attainment net enrolment rate is defined as the propor- than the poorest children in all six grades (an tion of students of the appropriate age (not average of 14.2 percentage points lower than all students) divided by those that should be Figure 31: Education Attainment By 15-20 Years Old, Cambodia 2011 a. By Residence in 2011 b. By Gender in 2011 Male Urban Rural Female School year completed School year completed d. By Ethnicity in 2011 c. By Wealth in 2011 Khmer Poorest Local Ethnic Richest Others non-Khmer School year completed School year completed Source: World Bank staff estimates based on CSES 58 in school—according to age. Gross enrollment below 40 percent after grade three and below rates can be interpreted as a measure of the 25 percent after grade six (Figure 32). total number of people going to school; the net enrollment rate is a a measure of student going Over 30 percent of children do not enter to school at their appropriate grade level.43 grade 1 at the appropriate age; furthermore, Moreover, there are two possible ways to define significantly low promotion rates prevail net enrollment rates for an entire primary or at the lower primary level—combined, secondary level. For this study, the strict defini- this leads to very low net enrolment rates ton of individual grade net enrollment rates (37.4 percent) at grade 6 (Figure 32). The (Box 4) was chosen because it is more sensitive passage from grade 6 to grade 7 creates two to not only late entry and, but also repetition— challenges for students: the first is the need to both problematic in Cambodian education. pass the national exam at the end of grade 6; the second is that education facilities for sec- Net enrollment rates in Cambodia (using ondary schools are more limited than those the strict definition) have substantially of primary schools. The enrollment reduc- improved for all grades. From 2004 to 2011, tion between grade 6 and 7 (one out of three the net enrollment rate for primary students students) is thus a consequence of supply and increased by over half (from 30.6 to 48.8 demand factors. But this is not the case within percent); secondary net enrollment rates have primary education: there is no specific require- more than doubled (from 8.1 to 20.0 percent). ment to move from one grade to the next— While improvement has been observed in all and normally, access to all grades of primary grades, dropout from grades four to five and five school is the same. Nevertheless, in 2011, four to six have nearly been eliminated. Similalr to all out of ten students in grade 1 do not attend countries expanding their education coverage, grade 4. Factors contributing to these statistics Cambodia’s gross enrollment rates are diffi- include repetition and dropouts after achiev- cult to interpret. For example, more children in ing basic reading and writing skills. However, a school would increase the enrollment rates, but clear reason for this attrition rate is not clear.44 grade repetion and late entry would also increase Dropout rates and repetition trends are not gross enrollment rates. On the other hand, more captured in the Education Management infor- children starting school at the appropriate age mation System (EMIS).45 (Figure 32) would decrease gross enrollment for first grade. Primary and secondary net enroll- Because of low rates in the past, the ment rates for all grades have improved— above-mentioned improvements are not regardless of place of residence, gender, or enough and net enrolment rates remain wealth. Also females have had a slightly higher low. Cambodia is improving net enrolment net enrollment rate than males since 2007— rates quickly—in both primary and second- both in primary and in secondary education. ary education over the last seven years (an average of 2.7 percentage points each year for Children from the poorest households primary and 1.7 percentage points each year are being left behind at the beginning of for secondary). However, net enrollment rates their education. There is a clear relationship remain low in Cambodia in 2011—with values between wealth and education—in Cambodia, 59 Box 4: Measuring Net Enrollment Rates Net enrollment rates can vary significantly depending on the definition used. The strict definition of net enrollment rate is defined as follows: the proportion of students attending the correct education grade according to their age.a # of students 6 years old and attending first grade 1st. grade Net enrollment rate (NER1st) = *100 # of children 6 years old at the begining of the year To estimate the primary net enrollment rate, the weighted average of grades one through six is computed.b Primary Net Enrollment rate (strict) = Average(NER1st, NER2nd, NER3rd, NER4th, NER5th, NER6th) A more lax definition—using the education level (primary or secondary) as the unit of measurement— is sometimes used: # of students 6 to 11 years old and attending primary Primary Net enrollment rate (lax) = # of children 6 to 11 at the begining of the year *100 The strict net enrollment rate using the Primary And Secondary Net Enrollment Rates Using The grade as the measurement unit captures rep- “Strict” And “Lax” Definitions, Cambodia, 2004-2011 etition rates and late enrollment rates more accurately. Net enrolment rates using primary or secondary levels as the unit of measurement do not accurately capture repetition rates until Strict Lax the student turns 12 (for primary). Moreover, it fails to identify late enrollment once a child has enrolled in school. Some governments use the more lax def- inition to report net enrollment rates. Rates reported by international institutions suggest the use of the education level—and not the grade— Primary Secondary as the unit of measurement. Both measures are reported for comparison and to relate them a. Age at the beginning of the school year as defined by the Ministry of Education. to official government statistics. All other net enrollment rates reported in the document are b. Weighted by the number of kids derived using the “strict” definition. the biggest difference in primary net enroll- enrollment rates above the national average ment rate is among the poorest children. With (Table 20). only a 34.4 percent net enrollment rate, the poorest children are the only group below the With the exception of gender, second- national average. Even children in the next ary net enrollment rates show impor- wealthiest group (quintile 2) have primary net tant disparities. Attrition and repetition are 60 Figure 32: Net And Gross Enrollment Rates In Cambodia 2004 And 2011 2004 Gross 2011 Gross 2004 Net 2011 Net 1 2 3 4 5 6 1 2 3 4 5 6 Primary Secondary Source: World Bank staff estimates based on CSES Table 20: Net Enrollment Rates For Primary And Secondary, Cambodia 2011 Place of residence Gender Wealth by quintile National P.P. O. Urban Rural Male Female Poorest Q2 Q3 Q4 Richest Primary 67.3% 58.0% 48.8% 47.3% 50.4% 34.4% 49.6% 50.4% 58.3% 62.9% 48.8% Secondary (all) 50.2% 28.3% 15.4% 18.8% 21.2% 5.6% 13.7% 14.8% 27.7% 45.4% 20.0% Grade 9 60.0% 30.6% 17.4% 17.7% 25.5% 4.2% 15.7% 20.1% 26.7% 51.5% 21.5% Grade 12 43.6% 16.6% 9.1% 13.9% 14.0% 5.3% 5.7% 9.4% 11.1% 35.5% 14.0% Source: World Bank staff estimates based on CSES much greater amid the rural, the poor, and the Education Characteristics middle class. In 2011, secondary net enroll- ment rates in Phnom Penh were over three Enrollment by Age times higher than those of rural households (compared to only one-third for primary). There is a strong relationship between Moreover, children in the richest quintile had enrollment and wealth—starting at age secondary enrollment rates at over 8 times of eleven. Although the enrollment gap between those of the poorest quintile (but less than 2 the poorest and richest children from 7-11 times in primary) and over 3 times of those years old is less than 10 percentage points, in the middle quintile (but only one-fourth in at age 12 the gap increases to 18 percentage primary). points. At 18 years of age, the gap widens to 61 57 percentage points (Figure 33d). (Similar Dropouts and Children Who tendencies are found throughout all wealth Have Never Gone to School quintiles.) There is no difference in attendance rates among males and females. Differences The rate of children that have never between urban and rural households are rel- gone to school is improving in Cambodia. atively small—until age 13 (less than 8 per- Children from 11 to 18 years old that have centage points). The difference increases sig- never attended school has reduced from 8.5 nificantly at age 15 to almost 20 percentage percent in 2004 to 3.6 percent in 2011, with points—with only small changes for older Reasons For Not Entering School children (Figure 33b and c). Reason 2004 2011 Improvements in attendance have been mostly from young children. Overall attend- Cost/work 31.6% 57.1% ance has improved for children 6-7 years old by Household Chores 35.4% 13.5% around 13 percentage points and for children 8 10 by an average of four percentage points. But Motivation/grades 26.1% 22.2% no real improvement in attendance was reported for children from 11 through15 years old. Other 6.8% 7.2% Figure 33: School Attendance (Any Grade) By Age, Cambodia a. By year b. By Place of Resident in 2011 Urban Rural Age in years Age in years c. By Gender in 2011 d. By Wealth in 2011 Poorest Male Female Richest Age in years Age in years Source: World Bank staff estimates based on CSES 62 almost no incidence in Phnom Penh (0.8 or secondary school. Dropping out of school percent) or other urban areas (1.7 percent). because of household chores reduced from Nevertheless, 9.7 percent of children from the the mid to high thirties in 2004 to about 13 poorest households never go to school—a percent in 2011. Lack of interest and low number very close to the national average grades are more commonly cited as a reason seven year ago (Table 21).46 Economic consid- for dropping out of primary education (28.1 erations (the need to work or education costs) percent) than dropping out of secondary edu- have become the main reasons for children cation (18.0 percent) in 2011 (Table 22). never enrolling in school. This was mentioned in over half of such cases (from less than one- As a reason to drop out of school, house- third in 2004). Household chores are no longer hold chores are as important for males as the main reason for never attending school, they are for females. Household chores were dropping significantly from 35.4 percent in mentioned almost 14 percent of the time as the 2004 to only 13.5 percent in 2011. reason not to return to school by both males and females. Because household chores are The reasons for dropping out of school not restricted to traditionally housekeeping are very similar to the reasons for never activities, males and females report activities entering school. This is true for both primary associated with helping the household at the and secondary education. Among primary same rate. The current primary factor affecting and secondary education students, economic dropout rates—regardless of location, gender, reasons have increased from about one- or wealth—is economic reasons. This is followed third to more than one-half of all dropouts. by a lack of motivation, low grades, and house- Secondary education requires more expenses hold chores. Other reasons, including supply for the household: transportation, materials, side factors, are always a minor factor related to and so forth; at the same time, secondary age dropping out of school. (Table 22) students are older and able to earn money. It is therefore not surprising that economic reasons Overage are mentioned more frequently for dropping out of secondary school (61.1 percent) than Overage is defined as attending a lower from primary school (53.3 percent). education grade according to official age for first grade and assuming no repetition. Late Household chores are no longer the entry and repetition are the two major factors main reason for dropping out of primary contributing to the overage in school enrollment. Table 21: Percentage Of Children 12-17 Years Old Never In School, Cambodia 2014-2011 Place of residence Gender Wealth by quintile National Year P.P. O. Urban Rural Male Female Poorest Q2 Q3 Q4 Richest 2004 1.8% 6.1% 8.3% 7.2% 7.9% 14.9% 8.6% 6.8% 4.1% 2.0% 7.5% 2011 0.4% 1.3% 3.6% 3.7% 2.4% 7.7% 1.7% 3.1% 1.2% 0.4% 3.1% Source: World Bank staff estimates based on CSES 63 The impact of late entry is twofold: The first 0.4 years in 2011. But it is important to under- stems from forgoing education at a young stand that total overage rates increase with age—a critical period in child cognitive devel- repetition of school grades. In 2011, it was 0.8 opment is missed, making learning more dif- years for third graders and 1.1 years for sixth ficult in the future. The second is late entry graders (compared to 2.2 years in 2004). The into the first grade, as well as repetition. This average reduction for overage for all six primary increases the age of children in latter grades— school grades was one full year. But overage in pressuring them to leave school to work, either Cambodia varies significantly: a large propor- doing household chores or working outside tion of students report much greater overage the home for money. Regardless of the reason, values. Of the 40 percent of students with the overage results in a diminished capacity for highest overage, late entry in 2011 was a full two learning and a higher dropout rate in second- years; overage was 2.7 years for third graders ary education. and 3.1 years for six graders (Figure 34).48 On average, late entry (overage in first Overage can be directly linked to the grade) seems to be a much smaller problem dropout rate of students in Cambodia— in 2011 than it was seven years earlier. especially in the latter grades. For example, Estimates of late entry into school have sub- four out of ten students would finish sixth stantially decreased from 1.3 years in 2004 to grade at age 15 instead of at age 12, thus Table 22: Dropout Of 12-17 Year Olds, By Year And Main Groups, 2011 Cambodia Year In 201147   Reason 2004 2011 Rural Male Female Q1 and Q2 Cost/work 30.2% 53.3% 52.8% 48.8% 58.0% 60.4% Primary Household Chores 34.5% 12.5% 12.2% 11.4% 13.5% 13.2% Motivation/grades 27.7% 28.1% 29.0% 33.6% 22.4% 22.5% Other 7.7% 6.1% 6.0% 6.2% 6.1% 3.8% Cost/work 34.8% 61.1% 61.3% 48.7% 68.8% 64.9% Secondary Household Chores 39.5% 13.2% 13.1% 17.4% 10.6% 12.6% Motivation/grades 22.7% 18.0% 16.9% 23.4% 14.7% 13.9% Other 3.1% 7.7% 8.7% 10.5% 5.9% 8.6% Cost/work 31.4% 56.4% 56.2% 49.2% 62.4% 61.1% Household Chores 35.6% 13.7% 13.4% 13.8% 13.6% 14.1% Total Motivation/grades 26.0% 23.0% 23.1% 29.1% 17.9% 19.5% Other 6.9% 6.9% 7.3% 7.9% 6.1% 5.2% Source: World Bank staff estimates based on CSES 64 increasing the pressure to drop out to engage Figure 34: Overage By Grade Over Time, Cambodia 2004-2011 in labor or family activities. Currently, a full 39.5 percent of students are not receiving formal a. Overage for all students education until eight years of age (instead of six years). (Figure 34) Overage Year The official education policy in Grade 6, 1.1 Cambodia—the Education Strategic Plan— Grade 3, 0.8 has been silent on the issue of overage enrollment although Cambodia faces Grade 1, 0.4 serious overage rates for at least 40 percent of students from grade 1 to grade 6. Even though the causes of overage enrollment rates are late school entry or repetition or both, b. Overage by students above median by year grade (39.5%) policy discourse has focused, almost exclu- sively, on repetition and dropout only. The analysis of this study suggests that proper-age Overage Year enrollment has real effects on future attend- Grade 6,3.1 ance and progression into later grades. Grade 3, 2.7 Education Quality and Finance Grade 1, 2.0 To get a full picture of the Cambodian public education system and its ability to provide citizens with the skills they need to Source: World Bank staff estimates based on CSES secure professional jobs and to attract foreign investment, it is necessary to look beyond with students with higher cognitive skills have simply getting children into the classroom and more consolidated democracies (Barro, 1999; keeping them there. The quality of teaching Campante & Glaeser, 2009). These previous must be addressed. However, at present, such studies clearly show the significance of paying indicators to measure the progress of student attention to students’ learning outcomes as a learning are not available. key indicator on education quality. In practice, standardized national assessment is an effec- Previous research identified improved tive policy and practical tool to gather and learning achievements as the essential evaluate information on student learning and factor for measuring the quality of educa- their cognitive development. It is also neces- tion. Several recent studies show countries sary to make informed decisions about the next that have higher student achievement on steps in the educational process (Clarke, 2012). international exams also have higher rates of economic growth (E. Hanushek & Woessmann, Almost 90 percent of MoEYS budget is allo- 2009; Pritchett & Viarengo, 2009). Other studies cated for primary and secondary education, have extended this by showing that countries while the remaining 10 percent is budgeted 65 across the remaining sub-sectors: Early Child which have been growing by about 10 percent Education (ECE), tertiary, and Non-Formal every year. In contrast to this, the growth rate Education (NFE) (Figure 35). of the Program Budget for each student has been smaller. (Table 23) The education budget trends have been increasing for some time, reaching a peak Budget formulation and execution is of 19.2 percent of MoEYS budget in 2007. a serious issue that impacts all education Recently, however, there has been a downward sector activities. A greater cause for concern decline following the 2007 apex (Figure 36). is that the budget has been significantly Although the percentage of recurrent budget underspent over the past few years. In 2010, it share shows a declining trend, the amount was underspent by more than US$30 million— actually allocated has been increasing. In a figure as large as the total program budget fact, the unit expensed cost for education has for that year. Although low teacher salary has increased from $46.10 for each student in 2008 been a recurrent issue, the largest under-spent to $72.43 for each student in 2012. However, item is remuneration (budget items under most of the increase in cost for each student Chapter 64). In 2010, it was underspent by is the result of increases in teachers’ salaries, over US$20 million, still just for budget items under Chapter 64. This suggests that although Figure 35: Total And Programed Education Budget By Sub-Sector, 2010 the education budget’s absolute amount increased, these increases were not spent Total Budget effectively—resulting in a smaller budget ratio for education amid the total national budget Programed Budget for the following fiscal year. Policy Recommendations Early Child Primary Secondary Tertiary Non-formal Further expansion of non-formal education Source: World Bank Staff Estimated (alphabetization campaigns) targeting poor Non-program program budget figures based on salary allocations for public employed teachers/staff and their different salary level across each sub-sector. Program budget is based on actual budgeted amount excluding Youth and Sports. Table 23: MoEYS Yearly Cost Per Student Figure 36: Budgeted And Actual Recurrent Expenditures, Cambodia Program Non-Pro- Year Total (US$) Budget gram Budget 2008 $ 7.07 $ 39.03 $ 46.10 Recurrent expenditures in total budget 2009 $ 8.59 $ 46.06 $ 54.65 2010 $ 8.93 $ 47.41 $ 56.34 Budgeted Actual 2011 $ 9.89 $ 52.94 $ 62.83 2012 $ 9.82 $ 62.61 $ 72.43 Source: World Bank Staff based on figures provided by MoEYS and European Union Source: WB Staff based on MoEYS’s EMIS data 66 adults is necessary to cover the remaining grade late or falling grades and providing 8.4 percent of illiterate youth and the 20.7 the right incentives to change both atti- percent of illiterate adults. Better coverage tudes and practices. Scholarships and school of primary education and the promotion of feeding programs have been found to be an non-formal education have decreased illit- effective means to counter this phenome- eracy rates in Cambodia. But illiteracy rates non—not only in Cambodia but also interna- remain high for the poorest Cambodians (19.4 tionally. Their coverage should be expanded. percent) and for adults 15-64 years old (20.7 percent). Alphabetization campaigns can Programs like targeted cash programs be an option to reach them. Such programs and other targeted transfer schemes are should concentrate efforts in rural areas and well suited to improve secondary education in the poorest households and toward adult in Cambodia. Improvements of secondary females. education outcomes require improvements in primary completion rates. The most important The most important recommendation reason for dropout rates is economic; thus any to improve primary education outcomes program that aims at keeping more students is promoting enrollment in first grade for in secondary school must take this into con- all children older than 5 years, especially sideration. Economic reasons for repetition among the rural poor. While the average and dropouts are more commonly cited from overage enrollment for first graders (late young women and from the poorest 40 percent entry) stands at a reasonable 0.4 years in 2011, of the students. Rigorous evaluations of the a full 38 percent of students enters school— Cambodia Education Sector Support Project on average—two years late. The loss of this (CESSP) scholarship program have shown this critical time in children’s development is an program to significantly increase enrollment impediment for future learning that decreases rates among children from poor households, the benefits of education and increases the especially girls (Filmer and Schady, 2009). Such chances of dropouts in later years. coverage should be expanded. The second most important option to International experience strongly suggests improve primary education outcomes is to that standardized national assessments are reduce repetition. Repetition from grade one effective tools to make an informed decision to grade six adds another 0.7 years overage by about the educational process. The Ministry’s grade six. Similar to late entry, repetition adds Quality Assurance Department has been con- 1.1 years of overage for an important share of ducting official national assessment tests for students in grade six (four out of ten). When grade 3 for the first time, which is a positive combined with late entry, it adds a total 3.1 trend, but it is likely to improve quality only if years of overage by grade six. it is made systematic and routine. Any effort to remedy these two problems As a portion of total budget share in must address two issues: identifying the recent years, education expenditure has group of children that are entering first declined; it has also been significantly 67 under-spent for many years. The educa- There are currently 6 national hospitals, 83 tion budget showed a progressively upward referral hospitals, and 1024 health centers trend throughout much of the past decade or and 121 health posts throughout the country. more, reaching 19.2 percent in 2007. Yet the The Ministry of Health employs 19,700 staff; more immediate trend has been downward including 3,200 doctors; 9,000 nurses; and since 2007, reaching 15.9 percent in 2012. 4,600 midwives. It has achieved its objective of However, the greater concern is that these ensuring at least one primary midwife in each budget outlays have been significantly under- health center, and it is now seeking to increase spent for many years. Budget formulation and coverage of secondary midwives. execution thus becomes an urgent issue nec- essary for improving overall financing and Health is a priority sector for the govern- capacity across all education sector activities. ment. The government’s Rectangular Strategy prioritizes the construction of referral hospitals More resources should be allocated and health centers, the provision of free health to early child education to meet rising care for the poor, and the continued preven- demand. However, any increases in the Early tion and treatment programs for communi- Child Education budget should be achieved cable diseases. Other interventions include by increasing the overall education budget, the promotion of maternal and child health, rather than by reducing allocations in other and the adoption and enforcement of health sub-sectors. Almost 90 percent of MoEYS laws and regulations. The Strategy’s policy budget is allocated for primary and second- statements are further elaborated upon in the ary education; the remaining 10 percent is Second Health Strategic Plan (HSP2) for 2008- budgeted across the following sub-sectors: 2015. The Plan focuses on the development of Early Child Education (ECE), tertiary, and Non- health systems to improve health outcomes; it Formal Education (NFE) (Figure 35). (Box 5) has established the five following sector-wide strategies to achieve these goals: (i) health Health service delivery; (ii) health care financing; (iii) human resources for health; (iv) health infor- Cambodia has made significant progress mation systems; and (v) health system govern- in reestablishing its health system in the ance. National health financing strategies and decades following genocide and civil policies are further specified in the National conflict. Cambodia lost up to 90 percent of Charter on Health Financing (1996) and the their trained health staff during the Khmer Strategic Framework for Health Financing Rouge regime; moreover, the subsequent (2008-2015). civil conflict that lasted until 1998 made pro- vision of health service difficult. Since then, Cambodia has made remarkable Cambodia has made substantial progress progress toward the health Millennium in rebuilding its health system. The public Development Goal. Concurrent with health system currently consists of the central progress in economic conditions, key health Ministry of Health (MOH), 26 Provincial Health indicators have also improved. From 2000 Departments, and 77 Health Operational to 2010, the maternal mortality rate (MMR)50 Districts, which typically have one district or dropped from 437 to 288 (per 100,000 live provincial hospital, and 10-15 health centers. births) and the under-five (U5) mortality rate1 68 Box 5: Social Protection System In Cambodia Cambodia’s recent achievements in poverty plishing the following: (i) protect the poorest reduction are remarkable. Nevertheless, one in and most disadvantaged who cannot help them- five people remain poor, and an additional one selves, (ii) mitigate risks that could lead to negative in five people risk falling into poverty from a coping strategies and further impoverishment, and very small shock. The fast pace of poverty reduc- (iii) promote the poor to move out of poverty by tion experienced thus far might not continue in building human capital and expanding opportuni- the future because the smaller proportion of the ties. From 2012-2013, the government worked with population living below the poverty line is more development organizations on the preparation of difficult to reach and economic growth could slow three pilot programs to experiment with program down as a byproduct of increased development. design and delivery mechanisms for safety nets. Implementation of these pilots is starting in 2013. Poverty and vulnerability to shocks call for a social protection system that helps households Remarkably, since 2007, the government exit poverty, cope with shocks, and invest in has developed and implemented (with financial human development. Development and urban- and technical support from GIZ and AusAID) an ization have brought about an erosion of tradi- official poverty targeting system—the IDPoor— tional family based safety net systems, which have which is available to any program. This target- historically helped the population cope with des- ing mechanism is based on household assess- titution and risk49. This erosion of former safety ments through a proxy means test from a survey nets leaves households vulnerable to shock; in questionnaire combined with village-level discus- addition, it leaves them with a limited ability to exit sions. Households identified as extreme poor or poverty. In the 2011 CSES, 98 percent of the popu- poor are classified as IDPoor 1 and 2, respectively, lation reported not receiving any form of pension and receive an IDPoor card. In 2012, the IDPoor or scholarship from the government. Moreover, 80 program covered almost all rural areas with at least percent of the poor does not have access to free one survey round; survey waves are repeated in or subsidized health treatment even at primary each province every three years. The development level, implying that they need to borrow (28%) or of an urban poverty targeting system is planned sell assets (9%) to pay for health care. High levels of for 2013. The IDPoor system unfortunately does child malnutrition among the poor are especially not allow for post-identification, thus requiring worrying because of the negative effects malnu- programs to set up their own redress mechanism trition has on cognitive development, as well as its for targeting (notably, the health equity funds. large human and economic costs at both the indi- vidual and aggregate levels. Despite government commitment and avail- ability of a targeting mechanism, the coverage The Royal Government of Cambodia has of safety nets remains at only 2 percent of the recently recognized the need to introduce poorest quintile of the population, compared to an integrated social protection system; it has an average of 53 percent in East Asia and 49.3 created the National Social Protection Strategy percent among other developing countries. (NSPS) for 2011-2015, signed by the Prime Until 2013, the government has been funding Minister in 2011. This strategy aims at accom- three main safety net programs: welfare payments 69 for specific vulnerable groups (elderly people, of coverage, the following are key recommendations orphans, veterans) through the Ministry of Social to reduce poverty and vulnerability, and to foster Affairs; scholarships for poor children in lower-sec- investment in human development among the poor: ondary schools through the Ministry of Education, and the Health Equity Funds (HEFs) through the 1) Increase the amount of government Ministry of Health—all with co-financing from resources allocated to social protection. This development partners. Unfortunately, no infor- will ensure the financial and institutional sus- mation is available on the beneficiaries, coverage, tainability for these programs. As Cambodia or transfer amounts of the welfare payments by develops toward middle income status, donor the Ministry of Social Affairs; thus no benefit inci- funding will decrease. Government resources dence analysis can be done (ILO 2012). An impact are thus a paramount necessity if the country is evaluation by the World Bank has found that the to establish a social protection system. scholarship program significantly increases enroll- ment among beneficiaries, though no information 2) Strengthening the IDPoor targeting mech- is available on the accuracy of the poverty target- anism. This can be achieved by improv- ing system employed (Filmer and Schady, 2009). ing accuracy and allowing re-redress and The health section in this chapter finds that tar- post-identification, as well as by developing geting accuracy and usage among the poor of methods to identify those at risk of falling into HEF in accessing medical treatment are limited, poverty (vulnerability targeting). with about 80 percent of the poor without access to an HEF or IDPoor card. The other existing safety 3) Expanding coverage and improving tar- net programs are funded and often directly imple- geting of demand-side instruments. This mented by development partners and NGOs, and will incentivize the use of key services among their coverage is limited (ILO, 2012 and RGC, 2010). the poor. This includes expanding scholarship The most notable interventions among these programs for children from poor households are labor-intensive public work programs, imple- and improving not only targeting but also pro- mented in a few provinces under separate programs moting usage of Health Equity Funds (HEFs) by the Ministry of Economy and Finance and the among the poor, as well as testing new inter- Ministry of Interior—although financed by the Asian ventions to cover demand side gaps in key Development Bank and the World Food Program/ services (such as in nutrition and sanitation). AusAID. With the exception of Health Equity Funds, social insurance mechanisms do not cover the poor Promote programs open to the near poor. because they are limited to civil servants, veterans, Programs cannot be open to only the poor: vulner- and formal sector employees through the National ability needs to be decreased across the popula- Social Security Fund for Civil Servants (NSSFC), the tion. Such programs could consist of an expanded National Fund for Veterans (NFV), and the National and self-targeted public works program open to Social Security Fund for private sector employees anyone to get an income case of unemployment, (NSSF), respectively (ILO, 2012). as well as community-based social health insur- ance for which the near poor pay contributions. Given the nascent nature of the social protec- tion sector in Cambodia, and its relatively low level 70 dropped from 124 to 54 (per 1000 live births) independent predictor for under-five mortal- (Figure 37). Overall, the under-five morality ity. Children born to mothers with no edu- rate decreased to a point in 2000 when 1 in cation are twice as likely to die compared to 8 children born in Cambodia did not survive those whose mothers received secondary edu- their fifth birthday. But by 2010, the rate cation or higher. Yet the poorest girls are twice improved to just 1 in 19 children. Cambodia is and five times less likely to attend primary and now on target to meet these two Millennium secondary education, respectively, compared Development Goals (MDG). to the wealthiest quintile. (Figure 37) Child Mortality Inequities have increased for mortal- ity of children aged one month to one year Child mortality is highly correlated with and neonatal mortality has shown limited wealth, especially for children 1-12 months old. improvement, particularly for the poor and Inequality is an intrinsic problem of most devel- rural populations. Underfive mortality can be oping countries and Cambodia is no exception. seperated into several components: neonatal Aside from the moral dilemma that this poses, mortality (deaths of children from birth to one inequality generates inefficiency and delays month), post-neonatal mortality (deaths from in development. Despite significant declines 1 – 12 months), and child mortality (death from in child mortality (children under 5 years), the 1 – 5 years). A review of components of child reduction in child mortality since 200551 was mortality reveals several trends: First, deaths twice as high in urban areas compared to rural, of children aged 1-5 have declined sharply and higher for the richest income quintiles across all quintiles, and declined by almost half compared to the poorest. Child mortality has for the poorest from 2005-2010—although remained unchanged at 3.3 times higher for poor children of this age are still three times the poorest quintile compared to the wealth- more likely to die than the highest quintile. iest since 2005 (Figure 38), and three times Second, post-neonatal mortality shows the higher for rural children compared to urban highest levels of inequities. While post-neona- children. Education of the mother is also an tal mortality for the poor has declined by half Figure 37: Cambodia Is On Target To Meet Under-Five And Maternal Mortality MDGs Deaths per 100,000 births Deaths per 1,000 birtths Poorest Maternal Overall mortality MDG target Source: CDHS 2000, 2005, 2010 71 in the past decade, mortality rates are six times tality is strongly influenced by the quality of that of the poorest compared to the wealthiest maternal and child care—including prenatal, quintile. Third, neonatal mortality has seen the delivery, and post-delivery. (Figure 38) least impovement across all income groups— dropping from 37 for each 1,000 live births in Insufficient coverage and equity of 2000 to 28 in 2005, but declining only to 27 in primary health interventions account for 2010: neonatal mortality is now the predomi- much of Cambodia’s child mortality. In nant source of child mortality for both the rich 2010, the ten most prevalent causes of under- and the poor (Figure 38). While neonatal mor- five death were from complications during tality declined in urban areas from 2005-2010, birth (asphyxia, prematurity, and sepsis), from it showed no improvement either in rural areas diseases that can be prevented by vaccines or among the poorest people (Figure 42) (pneumonia and measles), or from a lack of access to water and good sanitation (diarrhea) Reductions in child mortality reflect (Figure 39). Insufficient coverage and equity improvements in preventive services and disparities along the continuum of care improved socioeconomic status, while still exist (for example, antenatal care and neonatal mortality is strongly influenced quality of birth delivery and post-natal care); by quality of care. The significant reductions moreover, a lack of vaccinations and access in child mortality likely reflect improvements to potable water and good sanitation are still in overall socioeconomic status and increased accountable for a large fraction of the under- coverage of preventive interventions, such as five death toll. immunizations. The growing disparity in post- neonatal mortality likely reflects a range of Communicable diseases such as malaria factors including continued disparities in soci- have experienced a substantial reduction, oeconomic status, maternal education, envi- however, mainly because of the up-scaling of ronmental health and sanitation, malnutrition, not only insecticide-treated nets but also the and inadequate access to quality preventive promotion of prompt access to treatment. The and curative health services. Neonatal mor- country now approaches the pre-elimination Figure 38: Mortality Rates For Children Aged 0 5 Years By Year, Cambodia Child (12-60 mo) Postneonatal (1-12 mo) Neonatal (0-1 mo) Deaths per 1,000 children Source: CDHS 2000, 2005, 2010 72 stages, although the threat of drug resistance sion analysis controlling by relevant covariates is still a concern. showed a greater chance of death if birth took place in a public hospital compared to a birth at There are indications of increased risk home and that the risk was reduced for deliver- of death when births were attended by ies in private hospitals. These results were not unskilled personnel. What explains the slow statistically significant, however, due to the progress on neonatal mortality? The fact that small sample size. Higher risks at public hospi- 23 percent and 17 percent of all under-five tals could be partly attributable to more high- deaths (Figure 39) are due to prematurity and risk deliveries, but shortcomings in quality of neonatal sepsis or birth asphyxia, respectively, care may also be a factor. is an indication of health system failures along the continuum of care (antenatal care, delivery, The growing burden of non-communi- and post-natal care). Neonatal mortality has cable diseases will increasingly affect the also dropped substantially less in rural areas, poor as well as the better off. Cambodia which accounts for the largest segments of faces a dual burden of both communicable and the population—as well as among the poorest non-com municable diseases. The burden of people (Figure 40). Regression analysis of non-communicable diseases already exceeds the Demographic and Health Survey (DHS) that of communicable diseases. According to data supports this suggestion, indicating the World Health Organization, 832 deaths for that the risk of death increased when births each 100,000 people in 2004 were due to non- were attended by traditional birth attend- communicable diseases whereas 660 deaths ants, nurses, and midwives compared to births were due to communicable diseases. As the attended by qualified doctors. Similarly, regres- country industrializes and life expectancy Figure 39: Most Prevalent Causes Of Under Five Mortality, Cambodia 2010 Birth asphyxia, Pneumonia, Other Diarrhea, diseases, Injury, Prematurity, Congenital abnormality, Neonatal sepsis, 6% Note: Malaria HIV and Measles were ≤2 % each. Source: Child Health Epidemiology Reference Group (CHERG) 2012 73 Figure 40: Neonatal Mortality By Residence And Wealth, require follow-up treatment and continuity of Cambodia 2000-2010 services. For example, while 89 percent of the poorest women had at least one antenatal visit in 2010, only 40 percent of poor mothers (poorest) attended the recommended four antena- tal care visits compared to 80 percent amid Deaths per 1,000 births Rural the richest. This leads to having less access to essential nutritional support (for example, folic Overall acid and iron supplementation) and subse- quent quality of birth delivery. Consequently, an increased risk of neonatal and maternal mortality exists. Skilled birth attendance has (richest) improved dramatically in the past decade, but still shows high gaps between the wealthi- est and the poorest. Women in rural areas are more than three times as likely to give birth Source: CDHS 2000, 2005, 2010 at home compared to urban women (CDHS 2010). The poor are more likely to live farther prolongs, it is expected that non-communica- from a public health center and to face greater ble diseases will dominate the overall burden financial and opportunity costs for follow up of disease in the forthcoming years. Chronic visits or transportation for delivery. In contrast, diseases and conditions such as hypertension, service provided through outreach (such as diabetes, and cancer will increasingly affect vitamin A distribution) has achieved relatively the poor, with major consequences for health equitable coverage, but it could be improved. outcomes—furthermore, it will put more people at risk of impoverishment because of Despite significant improvements in health spending. A focus on primary preven- immunization, the poorest still suffer tion and treatment will be necessary to reduce from substantially lower coverage rates. the disease burden and to reduce costs for not Vaccination coverage for most diseases has dra- only the people, but also the health system— matically increased over the last two decades with appropriate referral services and financial from 40 percent of children vaccinated for protection for more severe cases. measles, diphtheria, whooping cough and tetanus (DTP), and Bacillus Calmette–Guerin Equity in the Provision and (BCG) in 1990 to over 90 percent in 2010. Use of Health Services These relatively high rates of coverage have contributed to reductions in child illnesses Although coverage of some preventive and deaths. But only 65 percent of the poor interventions and health behaviors has are fully immunized, compared to 88 percent achieved good levels of equity, coverage of the wealthiest quintile. Ensuring high vac- remains highly inequitable for others cination coverage for the poor and remote (Figure 41). Overall, equity is better for services communities could further reduce sickness that require a single contact than for those that and mortality from diseases like pneumonia 74 Figure 41: Access To Key Health Interventions, Q1 And Gap With Q5, Cambodia 2010 Gap Q1-Q5 Q1(poorest) ng s . d 3 ) ng re tn es o. sit fe P ca 6m at ni i DT sl st ed vi an ia ea ea h + fe (> rt on pl br M 4 bi + A m C ily r ly T d AN in eu OR m ille Ea m Pn Fa ta Sk Vi Pre-natal Birth Post Natal Note: ORT + feeding refer to treatment of diarrhea Source: DHS 2010. and measles, ranking third and tenth among were unlicensed drug shops and markets (47 the top causes of under-five mortality, respec- percent), health centers (13 percent), private tively. The ultimate consequence of differen- pharmacies (13 percent), and other private tial coverage in preventive disease measures medical providers (9 percent). In addition, the and health care practices is disparities in mor- use of provincial and district hospitals by the tality across wealth quintiles. poor declined by half from 2007 to 2011 (to only 1.4 percent and 0.6 percent, respectively). Over 90 percent of the poor seek medical The wealthiest sought care nearly three-quar- care when sick, but the majority seeks ters of the time in the private sector, but they care through unlicensed drug shops and were also more than twice as likely as the poor markets. The percentage of the poor seeking to use district, provincial, or national hospitals some sort of health care when ill increased (Table 24). Residents of Phnom Penh are the from 62 percent in 2004 to 92 percent in 2011. most frequent users of national hospitals (5 This is an encouraging trend. The percent- percent of the time), but nearly 70 percent of age of the poor seeking care in the public Phnom Penh residents purchased medicines sector increased steadily from 2004 to 2009, directly through private pharmacies when sick. but subsequently declined from 26 percent The percentage of rural residents seeking care in 2009 to only 16 percent in 2011—with 35 through home visits to or by trained health percent seeking care in the private sector, and workers is the same as those visiting health 48 percent seeking care through drug shops centers (9 percent)—indicating a continued and the nonmedical sector. In 2011, the most prevalence of dual practice in rural areas. frequent sources of curative care for the poor 75 Health seeking behavior is influenced drugs shops was about US$2, while private by cost relative to wealth, perceptions of pharmacies were only slightly more expensive quality, accessibility, and perceived severity at an average of US$5 (Table 24). But drug of the illness. It is important to differentiate shops and pharmacies tended to be closer between seeking healthcare for primary and with less waiting time. However, for second- secondary care. The poor have a preference ary care (in-patient), where costs are substan- for informal drug sellers and private providers tially higher, the poor—as well as the average at the primary care level, where prices across population—preferred the public sector and sectors are comparatively low: this appears to social schemes aimed at reducing costs, such be driven by convenience (for example, prox- as Health Equity Funds (HEFs), which have imity, waiting time, and so forth). From 2009 increased the demand for public facilities to 2011, the average spending by the poor at among the poor. Table 24: Health Delivery Providers Used And Average Costs, Cambodia, 2007-2011 % seeking Percentage seeking care in 2011 Average costs ‘09-‘11   2007 Wealth- Phnom Poorest Poorest Rural All Poorest iest Penh National hospital (NH) 0.7% 0.9% 2.2% 1.3% 5.0% $84 $4 Provincial hospital (RH) 3.4% 1.4% 2.9% 1.6% 0.8% $63 $9 District hospital (DH) 1.9% 0.6% 3.3% 2.5% 0.0% $34 $8 Health center/Health Post 19.6% 13.4% 3.5% 8.8% 0.5% $71 $11 TOTAL PUBLIC 25.6% 16.3% 11.9% 14.2% 6.3% $39 $6 Private hospital 1.5% 0.7% 2.8% 1.6% 1.5% $51 $18 Private clinic 9.0% 6.9% 19.4% 13.3% 11.1% $43 $10 Trained health worker anywhere 19.9% 5.8% 8.4% 9.1% 4.3% $18 $17 Private pharmacy 6.9% 12.7% 39.2% 16.2% 68.2% $8 $5 TOTAL private medical 37.3% 26.1% 69.8% 40.2% 85.1% $20 $7 Shop selling drugs/market 37.4% 46.9% 14.1% 35.9% 7.6% $4 $2 Other private 2.6% 9.3% 3.0% 7.9% 0.0% $16 $5 Kru Khmer/Monk 0.4% 0.9% 0.8% 1.0% 0.6% $26 $24 TOTAL private non-medical 40.4% 57.1% 17.9% 44.8% 8.2% $7 $3 Grand Total 100% 99% 100% 99% 99% $18 $5 * Notes: values correspond only for those seeking care in each provider type. Totals are weighted by frequency use. Table reports the first place care was sought, among those who sought care. Average costs reflect total health costs for the past 30 days, which may include visits to other facilities and are expressed in constant 2009 Phnom Penh dollars (US$1=4,044 Cambodian Riels) Source: World Bank Staff Estimates based on CSES 2007, 2009, 2010, 2011. 76 Equity of Health Spending …but health spending remains an important source of debt and impoverish- Health spending (as a percentage of ment for the poor and near-poor, and for the income) and catastrophic health spending chronically ill. Despite these improvements, has declined since 2009, owing to rising an estimated 2 percent of Cambodians fell into incomes and the increased use of low-cost poverty in 2011 because of health costs, only services. Health spending can be an important a slight decline from 3 percent in 2004. Health source of impoverishment. As a percentage of spending remains a significant burden on the total income, health spending has declined poor, with 18 percent of the poor incurring debt from 3.3 percent in 2007 to 2.5 percent in 2011 because of health expenses. Health spending is for all Cambodians and from 2 percent to 1.5 also a significant burden for the 9.4 percent of percent for the poor (Figure 42). Rural people Cambodians households that reported at least currently spend 2.8 percent of their income on one family member being ill for over a year in health, however, compared to only 1 percent 2011 (owing to chronic illness or injury). Poor for residents of Phnom Penh. Catastrophic families spent 25 percent of per capita monthly health expenditures (defined as a household income for long-term illness. But in rural areas, spending more than 40 percent of their annual the average health spending for chronically ill nonfood expenditure on health) averaged persons was 125 percent of per capita income, almost 5 percent in 2004, 2007, and 2009. It suggesting a significant burden and incurred fell to 3.6 percent in 2010 and fell again to 2.8 debt for the entire household. percent in 2011. This overall decline in health spending as a percentage of income and cata- Health Equity Funds (HEFs) and other strophic spending appears to be due to a com- exemption schemes for the poor reduce bination of rising incomes and a shift to lower health expenditure and debt, while increas- costs services through private pharmacies and ing the likelihood of using public services. drug shops. In Cambodia, like in most coun- Health Equity Funds (HEFs) were first estab- tries, health spending tends to increase dispro- lished by NGOs in the late 1990s to pay user portionately with income, reflecting greater fees for inpatient services on behalf of the disposable incomes and access to savings. poor. They have since expanded to 44 of Figure 42: Health Spending As % Of Total Income All Health Expenditure Catastrofic Health Expenditure Catastrophic spending is defined as spending more than 40 percent of nonfood consumption on health. Source: CSES 2004, 2007, 2009, 2011 and World Bank calculations. 77 Cambodia’s 77 operational districts in 2013. An but 4 percent had an HEF card but did not use estimated two-thirds of the poor are covered it. Thus about 80 percent of the poor did not geographically, and the government has com- access free or subsidized treatment. About mitted to achieving national coverage by 5 percent of households in the 3rd and 4th 2015. Coverage is also expanding from inpa- wealthiest quintiles report using ID Poor or tient care to outpatient services, with over other subsidized cards to get treatment, sug- 300 of the Cambodia’s 1,100 health centers gesting inclusion errors in the ID Poor system eligible for HEFs. These funds have emerged (Figure 43). as an important source of flexible income for public facilities, with 60 percent of revenue The 2010 Cambodian Demographic and used for staff incentives, and the balance for Health Survey (CDHS) also found low usage operating costs. The poor are identified by rates of not only HEF but also Community Based the national ID Poor system, managed by the Health Insurance by the poor: only 4 percent of Ministry of Planning who issues HEF cards. But the poorest quintile reported having services Health Equity Fund Operators (HEFOs, usually paid for by HEFs (and 1 percent by Community NGOs) also use a “post-ID” system to provide Based Health Insurance). Moreover, 18 percent free services to poor people that do not have of the poor borrowed with interest; 10 percent cards. These fund operators are also responsi- borrowed without interest; and 9 percent ble for monitoring quality of care for the poor sold assets to pay for healthcare (Figure 44). and ensuring that no additional payments are In contrast, the wealthiest quintile rarely made. A recent impact evaluation found that borrowed or sold assets to finance health care. HEF schemes managed by third party opera- In addition, a recent study (GIZ, 2011 and Ir, tors reduced, but did not eliminate out-of- 2008) showed that between 30 and 50 percent pocket health expenditures by the poor and of the poor or ultra-poor families continue reduced indebtedness. The evaluation also facing financial barriers to curative care. found positive synergies between HEFs and supply-side interventions such as health Government health spending is gener- service contracting (Flores, et.al, 2011). ally pro-poor, but the equity and efficiency of public spending could be increased, and Coverage and use of Health Equity the reliance on out-of-pocket payments Funds remains low, however, because of reduced. Out of pocket spending is the major incomplete coverage of the ID Poor system, source of health financing in Cambodia. and possibly gaps in distributing cards, Government spending on health rose from as well as the low use of public services. about 0.8 percent of GDP in 2000 to about 1.4 Despite the high theoretical coverage of HEFs percent of GDP in 2010 (about 11 percent of the and other voucher schemes, a majority of national budget), which accounts for one-fifth the poor do not have HEF cards, and those of overall health outlays in Cambodia. External with cards sometimes did not use them. In donor financing represents about 0.8 percent the 2011 CSES, about only 10 percent of the of GDP, but provides the majority of support poor reported using an HEF or other cards to for many critical health programs and interven- receive free or subsidized treatment, another tions for the poor, including the Health Equity 10 percent reported using their ID Poor card, Funds. A recent public expenditure review 78 Figure 43: Use Of Subsidized Treatment, 2011 Figure 44: Source Of Health Care Financing, 2011 % of households % of households poor rich poor rich Saving Borrow with interest Subsidy, has card No need, but has card Borrow no interest Sold assets Subsidy has ID-poor Subsidy, other reason Health E. Funds Community insured Source: CSES 2011 Source: CDHS 2010 highlighted several relevant findings (World Maternal and Child Nutrition Bank 2011). First, government spending on primary care and preventive services is pro- Women continue to have high rates poor, since the poor benefit disproportion- of both malnutrition and anemia. Among ately from these services. Conversely, the women aged 15–49 years, one of the remain- better off are more likely to benefit from gov- ing challenges is the high prevalence of a low ernment subsidies to national or provincial body mass index (BMI), which has remained hospitals. Second, the equity and efficiency of unchanged from 2000 to 2010 at about 20 resource allocations among provinces could percent (Figure 45). This is of great concern be made more transparent, including using given the adverse consequences for their poverty as allocation criteria. Third, 70 percent children—such as low birth weight. On the of all resources of the Ministry of Health are positive side, considerable improvements managed centrally (including the purchasing were observed in maternal anemia prevalence. of drugs and equipment), and more resources However, anemia is still more prevalent among need to be delegated to health centers and women who are of high parity (more than hospitals. Finally, the Ministry of Health four children), have little or no education, are spends over half its budget on pharmaceuti- pregnant, and live in poor households and in cals—significant savings could be obtained rural areas. through more efficient purchasing of pharma- ceuticals and equipment. These savings could Adequate maternal support through be used to scale up and deepen coverage of nutrition and education has impor- other priority equity enhancing interventions, tant benefits for both mothers and their including health equity funds. children. For example, iron supplements for women during pregnancy protect not only 79 mothers but infants against anemia, and it is In 2010, the prevalence of stunting, under- estimated that one-fifth of perinatal mortality weight children, and wasting was 40 percent, and one-tenth of maternal mortality are attrib- 28 percent, and 11 percent, respectively, utable world-wide to iron deficiency anemia and anemia is at epidemic proportions at 55 (Brabin, 2001). Moreover, anemia results in percent for children under 5. Child malnutrition an increased risk of premature delivery and contributes to an estimated 6400 child deaths low birth weight (Figure 45). In 2010, just annually. Cambodia is estimated to lose over over 45 percent of women reported that they US$146 million in GDP to vitamin and mineral had received vitamin A and iron and folic acid deficiencies every year. The long-term con- supplements in the six-week period follow- sequences of child malnutrition are severe— ing the delivery of their last-born child, which leading to poorer cognitive development— may reflect the fact that only over 60 percent which affects both human capital formation of women received comprehensive antena- and economic growth. Adults who were mal- tal care or post-natal care. There is thus the nourished when they were children not only die potential to reduce maternal (and childhood) younger but also have higher rates of chronic malnutrition by strengthening these existing diseases. Moreover, girls who do not receive delivery mechanisms and to incorporate addi- adequate nutrition in the first few years of their tional strategies. Such could include cash lives have children who are more likely to die, transfers to poor, pregnant women encour- thus creating a vicious cycle (Barker, 1993). aged to earlier antenatal and postnatal care, and the promotion of education or availability Child malnutrition is correlated with of fortified foods. socioeconomic status, but other key deter- minants are maternal malnutrition, breast- Despite some gains from 2000, improve- feeding and child feeding practice, and ments in child malnutrition has stagnated access to sanitation. Regression analysis and wasting has increased from 2005, with using the Cambodian Demographic and significant implications for human capital Health Survey (CDHS) data shows that children development and well-being (Figure 45). are more likely to be stunted or underweight Figure 45: Maternal And Child Nutrition Issues, Cambodia, 2000, 2005 And 2010 Stunted Underweight Wasted Anaemia Anaemia Undermutrit. Overmutrition Children 0-60 months old Women 15-49 years old Source: CDHS 2000, 2005, 2010 80 if their families are poor, if open defecation communities still defecated in the open and is common practice in their communities, if only 50 percent of the households in Cambodia their mothers are uneducated, and if they had access to potable water. In the same year, were not breastfed (Figure 46). This analysis only 1 in 4 girls enrolled in secondary educa- further shows that stunted women are more tion, which is a risk factor for future genera- likely to have malnourished children after con- tions born to these future mothers, perpetu- trolling by socioeconomic status, education ating the problem. Thirty percent of mothers of the mother, open defecation, and feeding did not practice exclusive feeding, 20 percent practices. This indicates that malnutrition has were underweight, and 45 were percent inter-generational effects, and emphasizes anemic, which highlights the need to improve the importance of investing in early childhood maternal health to reduce low birth weight nutrition, especially in girls. Complementary (Table 25). In addition, 76 percent of children and enhanced feeding after the age of six did not receive a minimum acceptable diet, months reduces the prevalence of under- and on the therapeutic side, only 1 percent of weight children. This highlights the impor- an estimated 80,000 severely malnourished tance of not only food diversification but also children received specialized treatment with the correct timing of weaning.52 (Caulfield, emergency food (UNICEF). A number of inter- 1996) The results are consistent with the inter- ventions to reduce malnutrition have success- national literature (Bhutta, 2008) and empha- fully been tested elsewhere, and there is inter- size the importance of treating mother and national recognition that interventions such child health in an integrated manner. as food fortification can tackle micronutrient deficiency. Community identification of severe Cambodia has experienced important malnutrition or promoting supplementary progress in promoting exclusive breast- and diversified feeding through multi-sector feeding and reducing open defecation, community-based programs represents cost- among other advancements—but further effective strategies to reduce malnutrition. Yet attention is needed. In 2010, 70 percent of in Cambodia, iodized salt (that now reaches Figure 46: Children Stunted by Wealth, Mother Education and Breastfeeding And Open Defecation % of stunnted Children None Never Ever poorest years years HH's HH's Wealth quintile Mother's education Breastfeeding Open defecat. Source: DHS 2000, 2005, 2010 81 83 percent of houses) is the only fortified food the private sector probably indicates a per- produced at scale (UNICEF). ceived lower quality of the public sector. The quality of care in the public sector needs to be Conclusions and Policy Recommendations assessed and consequently improved, both in terms of clinical quality and responsiveness Health outcomes have improved dra- to the needs of poor clients. This will require matically, but a significant gap between the a combination of investments in training and poor and the better off still exists—reflect- equipment; strengthening, monitoring, and ing not only gaps in coverage of lifesaving oversight of clinical quality by the Ministry of interventions but also gaps in quality of Health (MOH); strengthened oversight and care. A fundamental barrier for reaching the regulation of private and informal health pro- most vulnerable is social exclusion, so much viders; and better social accountability to so that the needier are systematically missed citizens and local governments in the context from most interventions. of the ongoing decentralization process. The government should give priority to Financial protection for the poor and closing the equity gap in life-saving coverage near-poor must be further enhanced indicators including antenatal and post- through improved coverage of ID Poor and natal care, vaccination, and facility deliver- Health Equity Funds (HEFs), while main- ies. This will mean shifting more resources to taining low user-fees in the public sector primary care, ensuring adequate financing for to avoid any excess burden on the near- outreach services to remote communities, and poor. The low reported use of HEFs and other improving service coverage in remote areas. subsidized schemes is concerning, as is the Demand side interventions such as Health apparent exclusion and inclusion errors in the Equity Funds, vouchers, or cash transfers could ID Poor system. be used to further incentivize access to these key interventions. The implementation and monitoring of the ID Poor system and the HEF post-ID poor That the large majority of the poor opt for system must be strengthened to ensure that Table 25: Priority Interventions That Can Potentially Reduce Malnutrition Further Risk factor for childhood malnutrition The gap Open defecation 70% population defecates in the open No access potable water 50% Households with no access to potable water Low education of mothers 30% mothers with no education No exclusive breastfeeding first 6 mo. 30% not exclusively breastfed Low BMI of mothers 20% mothers with low BMI Anemia in mother 45 % mothers anemic Source: CHDS 2010 82 all eligible poor are covered. Furthermore, ture is prevention and promotion programs— identified poor receive their HEF cards and including anti-smoking campaigns and taxes, are made aware of its benefits. There also and healthy lifestyle programs—coupled with requires a scaling up of HEFs nationally with strengthened primary care to manage chronic expanded coverage to health centers to allow conditions at relatively low cost. HEFs to finance outpatient as well as inpatient services. There will also need a strengthening The provision of non-communicable of HEF oversight mechanisms to ensure service disease services should be at public health quality and to eliminate additional payments. centers and hospitals with appropriate copay- ment policies for the better off—coverage for Low user fees must be maintained in the HEFs and social health insurance should be public sector concurrent with the government’s assessed based on affordability. commitment to universal health coverage, par- ticularly at the primary care level. Also, the A multi-sector approach needs to be establishment of stronger oversight and the adopted to reduce maternal and child mal- regulation of fees for non-insured patients at nutrition, with a focus on aggressively tar- public hospitals must be done to avoid over- geting the poor and rural areas. A stunted, charging and catastrophic expenditures. Costs underweight, and anemic mother tends to and options must be considered to expand deliver a weaker child, who is more likely to be social health protection to the near-poor and malnourished and at greater risk of dying. the informal sector in the coming years, with at least partial subsidy from government. In the health sector, the monitoring of child growth in health facilities needs to be strength- The prevention and health promotion ened; nutritional counseling for pregnant for chronic, non-communicable diseases women improved, breastfeeding and feeding needs to be strengthened, and social pro- practices for children promoted; treatment tection mechanisms for chronic illness of severely malnourished children in public need to begin developing. With the increase health facilities scaled up; and micronutrient in life expectancy and better control of infec- distribution and coverage to pregnant women tious diseases, the burden of disease will shift and children improved. toward non-communicable diseases in the forthcoming years. Chronic diseases such as Community-based programs should be hypertension and diabetes treatment will established not only to encourage diet diver- become an increased burden on the poor as sity and good child feeding practices, but also well as the better-off, cardiovascular disease to improve sanitation and the elimination of and cancers will increasingly contribute to open defecation. This needs to be comple- mortality and to impoverishing expenditures. mented by food fortification and micronutri- ent supplementation campaigns that target The first priority for government expendi- children early in life. 83 END NOTE 40. These fourteen ministries/entities are: enrolment (6-11 ages for primary is their Ministry of Education, Ministry of Health, net enrolment rate). Ministry of Labor and Vocational Training, Ministry of Agriculture Forestry and 46. Children 12 to 17 years old were selected Fisheries, Ministry of National Defense, as they provide a good representation or Ministry of Cults Religion, Ministry of recent education trends. Lack of attend- Culture and Fine Arts, Ministry of Economy ance and dropout rates for younger and Finance, Ministry of Interior, Ministry children are too small in 2011 to have of Public work and Transport, Ministry of statistically significant results. For older Mine and Energy, Ministry of Social Affairs, people, the dropout reason was not asked Council of Ministers, and National Bank of before 2009. Cambodia. 47. Other groups did not have enough obser- 41. ESP was revised in 2004 and ESP 2004- vations to be included. 2008 was developed. There were no major revisions for ESP 2001 to 2005. 48. Students in the 40 percent group corre- spond to those with overage values greater 42. Wealth is defined by quintiles according to than the median overage for the corre- consumption with 20% of the population sponding year and grade combination. in each quintile. 49. Royal Government of Cambodia, 2009. 43. Because gross enrollment rates will count even adults going to school, it is possible 50. Calculated for 6 years prior to survey to have values above 100 precent. But net enrollment rates are always between 0 and 51. Child mortality is calculated for the 10 year 100 percent. period prior to each survey (i.e. 2005 is 1996-2005) 44. Increasing net enrolment rates over time does not explain the high dropout rate. 52. Regression analysis shows a positive and Comparing net enrolment rates between significant relationship between reducing grade one any given year to grade four underweight children and the times a three years later yields an average dropout week a child was given meat, poultry, rate of 4 out of 10. fish, shellfish, or eggs. The Underweight Z-score increased by 0.04 for each time a 45. EMIS doesn’t use age specific data for week the food items were provided. 84 New Poverty Estimates Annex 1 in Cambodia 2009 New Consumption Aggregate The main source of information was the recall data; when no recall data was available (trans- Socio-Economic Surveys in Cambodia portation, communications, personal care, and hotel accommodations53), the informa- Beginning in 1993/94, Cambodia has con- tion reported from the diaries was used.54 ducted several Socio Economic Surveys, collect- ing information from households and individual Over time, several improvements have household members. Coverage of the surveys been added to the questionnaire, but it was has increased from 59 percent of the villages in decided not to incorporate such changes 1993/94 to 100 percent in the 2004 Cambodia into the poverty estimate. This was done to Socio Economic Survey (CSES). Starting in 2004, maintain the way consumption was estimated, the survey design, the questionnaire, the field not only to ensure comparability over time but work, and the overall methodology in collect- also to ensure that such comparisons were ing and processing the information were stand- “fair”. Nevertheless, as times change, so do ardized and applied to the 2007—2011 surveys, the preferences and habits of the people. This as well as the ongoing 2012 CSES. is especially true for countries like Cambodia that have experienced important changes Poverty was first estimated using the since first estimating poverty after decades CSES 1993/94. Consumption per capita was of conflict. Therefore the above-mentioned selected as the preferred indicator of a house- updates are necessary to obtain results that hold’s wellbeing. Extreme (food) poverty lines are accurate for Cambodia today.55 were estimated based on the food consumed by poor households in 1993/94, and total The CSES 2009 was chosen as the year to poverty lines were estimated based on poor update and to improve the way poverty is households’ consumption of non-food items. measured in Cambodia (“new” will refer to 85 these results).56 This document includes a component is included in Box 1.) A price index description of the “new” methodology and will was created for each of the main three regions point out the main differences with the old (Phnom Penh, other urban, and other rural) estimates (“old” will refer to the old method- to account for different costs in various parts ology).57 It is important to point out that not of the country. Total monthly household con- only do both methods follow the same prin- sumption was estimated by adding all of these ciples, but that both methods have the same individual components: this was converted into objective: measuring well-being using per a measure of individual (per capita) welfare for capita consumption.58 each day using household size. (Box 6) Components of Total Consumption Missing Values and Outliers with the New Method Similar to other household surveys, the The recall section of the 2009 Cambodia CSES 2009 database included cases with Socio Economic Survey contains all necessary missing information, or information provided components to create both the total consump- that was considered an outlier. Classifying tion and the per capita consumption for every an observation as an outlier is not a precise household. These components include costs process and special care was given following for all food consumed at home or outside the the same criteria for all cases: (i) outlier iden- home (purchased, produced, received as gifts, tification took place at the individual variable or otherwise), housing (rented or owned), level before any grouping of values; (ii) in case housing services (firewood, electricity, gas, of doubt, an observation was not classified as water, and so forth), transportation and com- an outlier; (iii) a combination of local knowl- munication, purchase values of selected edge, comparisons with other households and durable goods, personal use goods, recrea- statistical properties59 was used to classified tion and entertainment, education and health, an outlier; (iv) a record of each case was kept and the like. (A detailed description of each throughout the entire process, providing infor- mation at the aggregate level for each house- hold; and (v) outliers and missing cases were Box 6: Components Used For The New Consumption Aggregate imputed using averages reported by house- In Cambodia 2009 holds with similar characteristics. All food consumed at home or outside, purchased, produced, or otherwise acquired/received Food Consumption Monthly value of the home Housing services (electricity, water, gas, etc.) Food consumption was estimated using Transport and communication section 0.1.B of the questionnaire. The variable Purchase value of selected durable goods including total consumption for each item Personal goods (Q01BC05) was used. The total value of food Spending on recreation and entertainment consumed was reported over the last 7 days Education expenditures from the date when the survey was con- Health related expenditures ducted. To estimate monthly food consump- Others (including goods received in kind) tion, reported values were divided by 7, multi- plied by 365, and then divided by 12. 86 Monthly Value of House own imputed rent value (the natural logarithm) to seven household characteristics: household The monthly use-value of the house is part size, Phnom Penh households, other urban of total household consumption; therefore, it households, roof quality, water source, access must be included in the consumption aggre- to toilets, and access to electricity (Table 26). gate. This information was included in section 04 of the questionnaire. For rented houses, Housing Services the rental price is considered a good estimate of the value of using the house. For renting Housing services information is included households, the monthly amount reported in in section 04 of the questionnaire. All informa- question Q04_25A was used. For households tion was provided on a monthly basis and no that own their home or otherwise do not pay transformation was necessary. Items included rent, the monthly use value provided by the expenditures in water (Q04_16), sewage or owners in question Q04_25B was used. waste water (Q04_20), garbage collection (Q04_21), electricity (Q04_23A), gas or liquid Among 25 households (0.21 percent of petroleum gas (LPG ) (Q04_23B), kerosene the sample), no information was provided or (Q04_23C), firewood (Q04_23D), charcoal the reported value was classified as an outlier. (Q04_23E), batteries (Q04_23F), and other Since the value of the house is an important energy sources (Q04_23G). share of the household budget, special atten- tion was given to impute the value of these 25 Transportation and Communication households. A multivariate regression was esti- mated: the input for the calculation came from Total transportation and communication using the information from households that expenditures are reported in section 01.C own their home. This was done by relating their of the questionnaire. Total transportation Table 26: Linear Regression Of Owner Estimated Use Value Of Housing60  Dependent Variable: LN (estimated rent) Estimated β t value p value Constant 9.938 479.9 .000 House size: 1: 0-20 m2; 2: 21-40 m2.; 3: > 40 m2 .386 44.6 .000 Phnom Penh: Region = 1 1.243 46.6 .000 Other Urban : Region = 2 .471 21.2 .000 Good Roof: Q04_05 = 2, 3, 4,6 or 8 .353 18.9 .000 Good Water: Q04_12 = 1, 2, 34 or 5 .004 .3 .761 Good Toilet: Q04_19a = 1, 2 or 3 .399 24.3 .000 With Electricity: Q04_22a = 5 .568 4.5 .000 Adjusted R2 = 0.517 and F probability (ANOVA) < 0.001. Source: World Bank estimates based on CSES 2009, NIS, Cambodia 87 expenditures (Q01CC06 & Q01C01=2) and com- these values were used. Clothing and footwear munication expenditures (Q01CC06 & Q01C01=3) questions were reported over the last 6 months; were reported from the last month of the survey; the values were divided by 6. Personal effects thus no transformation was necessary. were reported over the last 12 months; the reported values were divided by 12. Monthly Use Value of Durable Goods Spending on Recreation Durable goods information is included and Entertainment in section 09 of the questionnaire. The exact age for items more than one year old was not Recreation and entertainment informa- asked, and the standard methodology used in tion is reported in section 01.C of the ques- other studies could not be applied. Originally, tionnaire. Total values—including cash and in an attempt to estimate the use value of all kind expenditures—are captured in question durables was made, but later consultations Q01CC06 and divided into several groups with the government resulted in removing according to the variable Q01C01. These such imputations. The purchase value was groupings include recreation within Cambodia used for selected items bought during the (Q01C01=8); recreation abroad (Q01C01=9); last 12 months. The selected items were as and gambling (Q01C01=12). All respondents follows: radios, televisions, bicycles, sewing were asked questions about such spending machines, video tape players, stereos, electric over the last 12 months; the reported values fans, kitchens and stoves, dining sets, mobile were divided by 12. phones, electric irons, suitcases, batteries, bed sets, wardrobes, and cabinets. In addition, Education Expenditures because motorcycles have become an impor- tant item—even in poor households—their Section 02 of the household questionnaire expenditure was used. These selected pur- includes all information for education expen- chased items or their imputed value was used ditures. Total education expenditures for each only for items acquired over the last 12 months person included formal, non-formal, public, (Q09_C07) and divided by 12 for the monthly and private education. Estimated total educa- value.61 tion outlays were derived from six questions: school fees (Q02C16a), tuition (Q02C16b), Personal Goods text books (Q02C16c), other school supplies (Q02C16d), transportation (Q02C16f ),62 gifts Personal goods information is reported in to teachers, school building funds, and the like section 01.C of the questionnaire. Total values— (Q02C16g). Five households did not provide including cash and in kind expenditures—are information for individual expenditures, and included in question Q01CC06 and grouped the reported total was used (Q02C16h). All according to variable Q01C01. These groupings respondents were asked for expenditures for included personal care (Q01C01=4); clothing the past school year; the values were divided and footwear (Q01C01=5); and personal by 12. Total education expenditures for each effects (Q01C01=11). Personal care items were household were equal to the sum of all house- reported for the last month of the survey, and hold members’ expenditures. 88 Health Related Expenditures Weighing the per Capita Consumption Aggregate by Regional Price Index Health expenditure information was reported in section 13 of the questionnaire. To compare the welfare conditions of house- Each household member was asked individ- holds living in different parts of the country, ually, which is believed to be more accurate a regional price index was estimated. This than a single question for the entire house- regional price index is based on the value of the hold. The question for health expenditures total poverty lines for each of the primary three asked respondents for the total amount regions: Phnom Penh, other urban, and other spent for treatment from any type of health- rural households.63 Phnom Penh was chosen care provider (Q13BC11); each household as the base for the price Index and assigned member was asked for expenditures over the a value of 1. For urban households outside last 30 days. The total health expenditure for Phnom Penh, the estimated price index was each household was the sum of all household 0.8022; for rural households outside Phnom members’ expenditures. Penh, the estimated price index was 0.7348. Other Consumption Results The final group included information from Imputed Values section 01.C and 07 of the questionnaire. From section 01.C, question Q01CC06 was used for Imputing missing values and extreme domestic salaries (Q01CC01=7) and miscel- values (outliers) is always a delicate process. To laneous items (Q01CC01=13). From section have a good measure of the potential impact 07, total imputed values by the informant of such estimations, a record of the value of were used (Q07_C05) for goods received from each estimate was kept and compared to the Cambodia or abroad as gifts (Q07_C01=10) total consumption aggregate. Out of 11,970 or as barter (Q07_C01=11). All four ques- households in the sample, 11,572 did not tions were asked over the last 12 months; the require any imputation on any variable. In only reported values were divided by 12 for the 398 households (3.3 percent of the sample), monthly estimate. values were imputed: for 323 households, the total imputed value represented 5 percent or Total Consumption Aggregate less of total consumption; for the remaining 67 and Daily per Capita Value households, the total imputed value ranged from 5.1 and 14.9 percent. Only 8 cases had The total consumption aggregate was imputed values that represented 15 percent or estimated by adding all individual compo- more of the consumption aggregate. nents for each household. The daily per capita consumption aggregate was calcu- The 8 households with a total imputed value lated by dividing each household’s consump- of 15 percent or more of total consumption tion aggregate by the household size and by were dropped from the sample. The informa- 30.4167. tion provided by these 8 households was not reliable enough to construct the consumption 89 aggregate; therefore, their elimination from To compare the level of consumption the sample was considered to be the best among different regions, it is recommended option. To compensate for the dropped house- to understand the differences in prices among holds, their weighted values were distributed regions. This is especially true for welfare proportionally to other households within the analysis—and any other type of compari- same Primary Sampling Unit.64 In other words, sons—that requires ranking households on households physically close to the eliminated their potential to consume goods and services. households were used to represent these unre- Using the price index created with the CSES liable figures. The “old” method did not make 2009, the average per capita consumption in the same effort to identify and replace outliers. Cambodia is estimated at CR 9,325 per day— ranging from CR 15,709 in Phnom Penh to CR Comparing the “Old” and the “New” 8,183 in rural areas (Table 28). Thus it is not per Capita Consumption Aggregates surprising that the price adjusted values are higher than the current values: this is because A good overall measure of the impact of average goods and services outside Phnom updating the manner in which per capita con- Penh are cheaper. In other words, with the sumption aggregate is estimated is to compare same amount of money, an individual or house- average values for each region. In current hold can buy more goods or services outside of Cambodian Riels (CR),65 the “new” per capita Phnom Penh than within Phnom Penh. consumption in Cambodia is CR 7,325 per day: a value very close to the results obtained Another method to understand how per applying the “old” method (CR 7,212). Table 27 capita consumption is distributed in Cambodia shows that the difference between both esti- is by estimating averages per quintile.66 Per mates is larger in Phnom Penh (3.2 percent) capita consumption at 2009 constant prices and smaller in both other urban areas (0.9 ranges from CR 3,953 for each day for the percent) and rural areas (1.2 percent). poorest 20 percent of the population. It then increases modestly—reaching CR 10,236 per Levels of Consumption in Constant day in the fourth group. In the last quintile, per 2009 Cambodian Riels capita consumption increases substantially, reaching CR 19,009 per day (Table 29). Table 27: Average Per Capita Daily Consumption, Cambodia 2009 Current Riels/day Increase Old New Riels/day % Phnom Penh 15,225 15,709 484 3.2% Other urban 10,096 10,184 88 0.9% Other rural 5,940 6,013 73 1.2% CAMBODIA 7,212 7,325 113 1.6% Source: World Bank estimates based on CSES 2009, NIS, Cambodia 90 Share of Consumption by Component budget than the poor: in 2009, housing repre- sented 24.1 percent of the budget for house- The relative importance of each compo- holds in Phnom Penh and only 9.5 percent nent of total consumption is a useful tool. The for rural households. Households in Phnom average per capita nominal consumption was Penh also allocate a higher percentage of computed not only for each region but also their budget to recreation and entertainment, for the entire country. The percentage of each transportation and communication, and edu- component was estimated, and it is presented cation. But they allocate smaller shares to in Table 30. At 53.2 percent, food represents health related expenditures. the biggest share of all components within Cambodia. The second expenditure type is New Poverty Lines related to the house and its services (electricity, water, and so forth), representing 19.2 percent The Concept of Poverty of the average Cambodian’s budget. The third largest share is transportation and commu- The concept of poverty has been recog- nication, which represents 6.7 percent of this nized as a multi-dimensional condition. At the budget. Smaller shares are allocated to health core of these dimensions, poverty is defined as (3.8 percent) and education (2.4 percent). the lack of something for which a minimum, decent standard of living is required. As such, Food shares are a smaller part of total con- many factors contribute to poverty—if not sumption for Phnom Penh—40.2 percent— poverty by definition, these factors can consid- because the consumption levels are higher erably lower living standards. Some examples (compared to rural households at 58.1 percent) of these factors are as follows: limited access to (Table 30). On the other hand, the use value of justice; gender discrimination; ethnic discrim- the house behaves inversely, with the richest ination; vulnerability; insecurity and safety households allocating a higher share of their issues; pollution and environment issues; child Table 28: Constant Riels (2009 Phnom Penh) Per Capita Consumption Per Region, Cambodia 2009   Phnom Penh O. Urban O. Rural CAMBODIA Per Capita Consumption/day in constant Riels 15,709 12,694 8,183 9,325 Source: World Bank estimates based on CSES 2009, NIS, Cambodia Table 29: Per Capita Daily Consumption By Quintile, Cambodia, 2009 Decile Avg.   1 2 3 4 5 Constant Riels 3,953 5,818 7,610 10,236 19,009 9,325 Current Riels 2,937 4,342 5,733 7,921 15,691 7,325 91 Table 30: Shares Of Total Consumption By Components And Regions In Cambodia 2009 Phnom Other Other  Consumption aggregate components CAMBODIA Penh Urban Rural All food consumed at home or outside 40.2% 48.7% 58.1% 53.2% Monthly use value of house 24.1% 16.6% 9.5% 13.4% Housing services (electricity, water, gas, etc.) 6.5% 6.0% 5.5% 5.8% Transport and Communication 9.0% 7.3% 5.9% 6.7% Monthly use value of durable goods 1.8% 2.0% 1.9% 1.9% Personal goods 4.1% 4.9% 5.6% 5.2% Spending on recreation and entertainment 2.2% 1.8% 0.9% 1.3% Education expenditures 5.2% 3.0% 1.4% 2.4% Health related expenditures 1.8% 2.8% 4.5% 3.8% Others (including goods received in kind) 5.1% 7.0% 6.7% 6.4% Total 100% 100% 100% 100% Daily per capita nominal consumption (CR) 15,709 10,184 6,013 7,325 The values are the percentages of the average nominal values reported in 2009 Source: World Bank estimates based on CSES 2009, NIS, Cambodia labor; domestic abuse; physical and mental example, it is not clear what is more important handicaps; hunger and malnutrition; lack of in the manifestation of poverty: Is being hand- adequate housing; restricted access to health icapped or being abused a more significant and education; and the like. The number factor? Is not having enough food to eat or not of dimensions varies greatly; their relative having access to basic medical attention more importance changes from country to country. likely to create poverty? Another example Moreover, these dimensions can vary within can be stated in the following question: Is it the same country, not only across regions but more dangerous to live near a river prone to also over time. flooding or to live in a neighborhood with a high crime rate? There have been several attempts to develop measurements of poverty by combin- In this work, consumption reported by ing several dimensions simultaneously. Most households is used as the measurement of of these measurements take the form of an welfare: it is compared to minimum amounts index; they have met with limited and varied (poverty lines), which classify either a household degrees of success. The root of the problem or a person as poor or not poor. Households are stems from putting all dimensions onto one classified as “extremely poor” if their daily per measurement so that they can be compared capita consumption is below the “food poverty (economists call this the “unit problem”). For line”. Households are classified as “poor” if their 92 daily per capita is below the “total poverty line”. and services that are also considered basic for a Thus a household that is classified as “extremely human being. Each poverty line is drawn from poor” is also by definition “poor”.67 different segments of the population to make them relevant to the resulting classifications. By using the local currency to value con- sumption, one can add different types or Food (Extreme) Poverty Line dimensions related to poverty within a common unit: Cambodian Riels (CR). Therefore, The food poverty line is the cost of a food this alleviates some of the ‘unit problems’ men- basket that provides a specific amount of tioned in the aforementioned paragraph. (This calories. There are three elements necessary to was done for food, education, health, enter- estimate the food poverty line: the minimum tainment, and so on.) But this approach does caloric requirement of an individual, the quan- not incorporate or attempt to incorporate all tities or components of the food basket, and the dimensions of poverty. To address other the prices used to estimate the cost of the food dimensions of poverty, it is recommended to basket. do complementary analyses—placing direct attention on specific issues. For example, if one Minimum Caloric Requirements believes the elderly are more vulnerable than other population groups, it is recommended The total calories for the food basket are to explore the conditions of people 60 years based on minimum caloric requirements esti- and older. It is also necessary to explore the mated for Cambodia—taking into consid- characteristics, relationships to other varia- eration both age and gender. This estimate bles (including poverty), sources of income, follows the recommendations of the FAO, support, and so forth. WHO, and UN in 200468 (which were adapted to Cambodia). Because Cambodia is a devel- This document uses a ‘new’ approach to oping country with almost 80 percent of the measure consumption and poverty lines: it is population engaged in agricultural activities, based on the conditions in the Cambodia Socio the requirements are estimated for medium Economic Survey (CSES) 2009. Comparisons to medium-high levels of activity. For each will be made with the previous method devel- age and gender group, the minimum recom- oped in the 1993/94 CSES; they will be marked mended kilocalorie intake for one person for as the ‘old’ method. one day was estimated. Finally, an allowance was added for pregnant women. Table 35 Poverty Lines includes the requirements according to gender and age, and the population share for each Two poverty lines have been estimated: The combination. First, the average kilo-caloric first corresponds to the cost of a food basket (Kcal) requirement was estimated (weighted that contains a minimum amount of calories by the population share) and 2,194.75 Kcal/ for the proper biological function of a human day, and second the allowance for pregnant being. It is called the Food (Extreme) Poverty women (5.25 Kcal/day) was added for a total Line. The second poverty line is equivalent to per capita minimum requirement of 2,200 Kcal/ the food poverty line plus an allowance cor- day. Originally, estimates with higher levels of responding to a minimum of non-food goods activity and higher caloric requirements were 93 modified to reflect the Royal Government of by multiplying the quantities by the prices for Cambodia’s preferences. each of the three regions of the country. An allowance of 10 percent to the total cost was Food Basket Composition originally added to compensate for wastage, but later dropped after conversations with To decide what items to include in the food government officials. The price, the cost of basket and in what amounts, data from a diary each food item, and the total cost of each food survey was used. Diary information is detailed basket is included in (Table 37). Total values of enough to allow for weight, unit price, and the food basket in current Cambodian Riels caloric content estimation. To make the food per capita are CR 3,411 for Phnom Penh, CR basket appropriate for the poor, households 2,791 for households in other urban areas, within the 5 percent to 25 percent lower con- and CR 2,570 for rural households (Table 31). sumption group were selected.69 Food used for production, charities, or payments of loans was Reduced Food Basket Costs excluded—only food items used for household consumption were used. The diary data col- In the future, estimating the value of the lected information on 193 different food items. food basket will require pricing the same food basket with new prices. The food basket quan- The food basket contained 111 products— tities do not change, but the prices change after eliminating products without caloric over time. However, obtaining reliable prices content information, products reported by less is not easy for several reasons: First at 111 than 5 percent of the households, or products items, the number of individual items in the reported in ambiguous units. Of these 111 food basket is large. Second, in periods of products, the reported consumption was high inflation, monthly prices fluctuate, there- transformed to weight and the median values fore, requiring the need for better information. in kilograms were computed. Also, for these Third, some of the categories in the basket same 111 products, the median price per include several food items; thus, updating kilogram for each of the major three regions of these prices over time poses difficulty. Fourth, the country was computed. future budget restrictions can limit the ability to collect all the necessary information for an The total caloric content of the selected accurate assessment. food basket was estimated at 1,484 Kcal per day.70 The amount of food required to satisfy For these reasons, a second food basket the minimum caloric requirement of 2,200 was was created: the reduced food basket. It uses computed by multiplying the reported quanti- the following principles: maintaining total ties by 1.4824 (2,200/1,484). The list products, food basket costs as close as possible to the the quantities necessary for the 2,200 Kcal per original estimates, maintaining products with day requirement, and the individual Kcal for a clear definition and a clear unit measure- these products are in Table 35. ment, and maintaining products that provide the most calories and that are most commonly Food Basket Cost or frequently consumed by households. After applying these principles, the resulting food The cost of the food basket was estimated basket includes only 29 food items (compared 94 to 111); at the same time, it includes more than allowance starts with the concept that the 96 percent of the long (111 items) food basket’s food poverty line provides the minimum food calories. The cost of the reduced food baskets requirements and that consumption is labeled is about 91.5 percent of the long food basket. as “essential” or “basic”. The second concept is the principle that a household may substitute To keep costs as similar as possible to the some of its “basic” food consumption for other long food basket, the total quantity of the goods and services only if these other goods 29 food items was increased. After the final and services are as important as the food not adjustment, the reduced food basket costs are consumed. Said another way, a household can as follows: CR 3,408.87 for Phnom Penh, CR go hungry; at the same time, they may use 2,820.08 for households in other urban areas, part of their budget for non-food goods and and CR 2,583.10 for rural households (Table services because those goods and services are 31). Cost differences between the two sets of as important or as necessary as food.71 food baskets are 1.1 percent for Phnom Penh and almost 0 for both other urban households Households with total consumption that outside Phnom Penh and for rural house- hover around the food poverty line do not holds. Given the advantages of tracking the meet or barely meet the minimum food reduced food basket’s cost over time and the requirements. Nevertheless, these house- limited negative impacts, it is recommended holds still allocate part of their budget to to use the reduced food basket. The complete non-food consumption. This non-food con- reduced basket is in Table 37. sumption has to be as important as the food not consumed; therefore it is called “basic”. In Total Poverty Line the Cambodian Socio Economic Survey (CSES) 2009, 498 households had total consump- The total poverty line is equal to the food tion within 10 percent of the food poverty line (extreme) poverty line plus an allowance for value.72 In these households, food represented a minimum amount of non-food goods and 64 percent of total consumption in Phnom services (housing, clothes, personal goods, Penh and 66 percent of total consumption in entertainment, and so forth). Because there the other two regions: non-food consumption is not an accepted minimum amount of represented 36 and 34 percent, respectively.73 non-food goods and services, determining the minimum amount is done indirectly. The food poverty line should represent 64 or 66 percent of the total poverty line, and the The technique to estimate the non-food non food allowance should be the remaining Table 31: Long (111 Items) And Reduced (29 Items) Food Basket Values, Cambodia 2009 Phnom Penh Other Urban Other Rural Cost of 111 items Food Basket in current Riels/day 3,408.87 2,820.08 2,583.10 Reduced Food Basket Cost in current Riels/day 3,448.05 2,815.70 2,583.10 Difference in Riels per day / percentage 39.18/1.1% -4.38/-0.2% 0.00/0.0% Source: Poverty group estimates with CSES2009, NIS, Cambodia 95 36 or 34 percent of the total poverty line. half times as much as the “old” values (Table Given the food poverty line values reported in 33). Both estimates are based on how much Table 1, the corresponding total poverty lines food and non-food was consumed by groups are as follows: CR 5,326.36 (Phnom Penh), CR or households within the samples (the Engels 4,272.85 (other urban), and CR 3,913.79 (rural) coefficient). Why has the Engels coefficient (Table 32). changed so drastically from 84 to 65 percent after using the “new” method? Poverty Lines Comparisons between the “Old” and the “New” Method There are two reasons why the food allow- ance in the “new” method is much larger than The “old” and the “new” food poverty lines that of the “old” estimate: the base year and values are very similar: the “new” method price evolution. The base year came from a estimates are CR 14 higher than the “old” for time when the “old” method was first devel- Phnom Penh (+0.4 percent), CR 87 lower for oped. It used information from the 1993/94 other urban households ( 3.0 percent), and CR household survey: a time in Cambodia when 138 lower for rural households (-5.1 percent) poverty was much higher and households ded- (Table 33). The difference between the “old” icated a higher share of their budget to food.74 and the “new” food poverty line is a product For example, rural households’ non-food share of two factors: First, by redefining the food in 1993/94 was estimated at 21.1 percent basket in 2009, flexibility was introduced that compared to 34 percent in 2009.75 With fun- has allowed households to switch food con- damental changes in Cambodian society, this sumption into cheaper products. Second, with change in budget allocation is not surprising: it higher incomes, households have increased has been 16 years since the 1993/93 estimate, food expenditures with more expensive and income levels have more than doubled. calories. The net effect is mostly positive because the efficiency gains were higher than The second reason is the evolution of prices the more expensive calories being consumed. over the 16 years from 1993 to 2009—measur- ing price changes over a long period of time is However, the two methods reached very always difficult. The unit value from the diary different values for the non-food allowance: questionnaires and the prices from the price the “new” values were more than two and a questionnaire both indicate that inflation in Table 32: Poverty Lines Per Region In Current Riels Per Day In Cambodia, 2009   Total Poverty line a Food allowance b Non Food Allowance c Phnom Penh 5,326.36 3,408.87 1,917.49 Other urban 4,272.85 2,820.08 1,452.77 Other rural 3,913.79 2,583.10 1,330.69 CAMBODIA 4,080.75 2,683.33 1,397.42 a Food poverty line/0.64 or 0.66; b Same as the Food Poverty Line value; c Total Poverty Line - Food allowance. The total for Cambodia is the average of the expanded sample. Source: Poverty group estimates based on CSES 2009 96 non-food prices was lower than inflation in 2009 Poverty Results food prices in all three regions throughout the entire period. By 2009, the “old” method was With the “new” method, extreme poverty producing even lower shares of non-food con- in Cambodia is estimated at 5.3 percent, with sumption than it had previously; for example, relatively lower values in the urban areas rural households’ non-food share was esti- (0.6 percent in Phnom Penh and 2.0 percent mated at 15.8 percent. in other urban areas) and somewhat higher in rural households at 6.3 percent. The “old” The reasons for the differences are clear: the method estimates have similar extreme (food) important question is which method to use? poverty rates—all of which are within 2.4 per- The answer is explained as follows. The closer centage points or each other (Table 34). the year in question is to the 1993/94, the use of the “old” method is more accurate. The closer Total poverty with the “new” method is the year in question is to the 2009, the use of substantially higher due to the much larger the “new” method is more accurate. In addition, non-food allowance. The total rate is estimated by having estimates using both methods over at 23.9 percent, with values of 4.3 percent in a period of time, comparisons of tendencies, Phnom Penh, 12.7 percent in other urban areas, trends, and factors related to poverty can be and 27.5 percent in rural areas (Table 34). With made. If the differences are a matter of only the “old” method, total poverty in Cambodia levels, most of the poverty analysis and gener- was estimated at 14.6 percent. ated knowledge will not change. Table 33: Poverty Line Comparisons In Current Riels Per Day In Cambodia, 2009 Food (Extreme) Poverty Line Non-Food Allowance Region Old New Difference Old New Difference Phnom Penh 3,395 3,409 14 790 1,917 1,127 Other urban 2,907 2,820 -87 531 1,453 922 Other rural 2,721 2,583 -138 492 1,331 839 Source: Poverty group estimates based on CSES 2009, NIS, Cambodia Table 34: Poverty Headcount In Cambodia 2009 Food (Extreme) Poverty Headcount Total Poverty Headcount Region Old New Difference Old New Difference Phnom Penh 0.8% 0.6% -0.2% 1.8% 4.3% 2.5% Other urban 2.8% 2.0% -0.8% 6.9% 12.7% 5.8% Other rural 8.7% 6.3% -2.4% 17.0% 27.5% 10.5% Cambodia 7.4% 5.3% -2.1% 14.6% 23.9% 9.3% Source: World Bank staff estimates using the 2009 CSESS, NIS, Cambodia 97 Table 35: Caloric Requirements And Population Weight By Age And Gender In Cambodia 2009 MALES FEMALES Age (years) Kcal/day Pop. Share Kcal*share Age (years) Kcal/day Pop. Share Kcal*share <1 671 1.2% 8.0 <1 615 1.1% 6.5 1 a 1.9 942 1.0% 9.2 1 a 1.9 843 0.9% 7.9 2 a 2.9 1,115 1.0% 10.9 2 a 2.9 1,041 1.0% 10.0 3 a 3.9 1,239 1.1% 13.1 3 a 3.9 1,140 0.9% 10.3 4 a 4.9 1,338 1.0% 13.0 4 a 4.9 1,239 0.9% 10.9 5 a 5.9 1,462 1.0% 15.3 5 a 5.9 1,313 1.1% 14.2 6 a 6.9 1,561 1.2% 18.7 6 a 6.9 1,413 1.1% 15.2 7 a 7.9 1,685 1.1% 18.8 7 a 7.9 1,536 1.0% 15.9 8 a 8.9 1,809 1.1% 19.3 8 a 8.9 1,685 1.0% 17.6 9 a 9.9 1,958 1.2% 22.5 9 a 9.9 1,834 1.2% 21.2 10 a 10.9 2,131 1.2% 25.1 10 a 10.9 1,983 1.1% 21.9 11 a 11.9 2,329 1.1% 24.7 11 a 11.9 2,131 1.1% 22.5 12 a 12.9 2,528 1.1% 28.2 12 a 12.9 2,255 1.1% 24.3 13 a 13.9 2,751 1.3% 34.5 13 a 13.9 2,354 1.1% 26.9 14 a 14.9 2,974 1.2% 36.2 14 a 14.9 2,429 1.1% 26.8 15 a 15.9 3,147 1.4% 42.7 15 a 15.9 2,478 1.1% 27.8 16 a 16.9 3,296 1.3% 42.4 16 a 16.9 2,478 1.2% 29.1 17 a 17.9 3,370 1.2% 40.2 17 a 17.9 2,478 1.1% 28.3 18 a 18.9 2,681 1.3% 33.9 18 a 18.9 2,020 1.2% 24.5 19 a 19.9 2,664 1.1% 28.5 19 a 19.9 2,004 1.1% 22.1 20 a 24.9 2,731 5.1% 138.7 20 a 24.9 2,039 5.2% 106.1 25 a 29.9 2,733 4.6% 126.2 25 a 29.9 2,040 5.1% 103.5 30 a 34.9 2,737 2.6% 71.5 30 a 34.9 2,209 2.8% 62.8 35 a 39.9 2,742 2.9% 79.9 35 a 39.9 2,212 3.2% 69.8 40 a 44.9 2,726 2.4% 65.4 40 a 44.9 2,202 2.8% 61.4 45 a 49.9 2,711 2.4% 64.8 45 a 49.9 2,193 2.7% 60.1 50 a 54.9 2,718 1.6% 43.1 50 a 54.9 2,197 2.4% 52.0 55 a 59.9 2,730 1.3% 35.8 55 a 59.9 2,205 1.9% 42.2 60 a 64.9 2,234 1.1% 24.4 60 a 64.9 1,938 1.3% 26.1 65 a 69.9 2,241 0.7% 15.6 65 a 69.9 1,942 0.9% 18.1 70+ 2,241 1.1% 25.5 70+ 1,942 1.7% 32.6 48.6% 1,176.14 51.4% 1,018.61 Weighted average = Sum (Kcal/day * Population Share) = 1,176.14 + 1,018.61 = 2,194.75 Plus Pregnancy Allowance: Birth ration of 0.025 * Extra caloric requirement of 210 = 5.25. Total = 2,194.75 + 5.25 = 2,200.00 98 Table 36: Food Items, Calorie Content, Prices And Costs Of Food Basket, Cambodia 2009 Kcal in Prices per Kilogram Cost in Current Riels/day   Refuse gr./day Kcal/day .100 gr PhnomPenh Other Urban Rural PhnomPenh Other Urban Rural Rice quality 2 (bad quality) 356.3 0% 428.5 1526.60 2,000 1,800 1,500 856.91 771.22 642.69 Other rice 353.5 0% 3.8 13.60 2,200 2,200 2,117 8.45 8.45 8.13 Corn on the cob 159.9 60% 2.1 1.30 3,333 2,500 1,500 6.91 5.18 3.11 Pnum pan 263.8 0% 0.4 1.10 7,500 7,500 6,000 3.02 3.02 2.42 Yellow noodles 373.7 0% 0.5 1.70 10,000 7,692 7,692 4.49 3.45 3.45 Others noodles 343.6 0% 1.3 4.50 8,000 6,667 6,667 10.59 8.82 8.82 Cakes, tarts, pies, quiches 370.2 0% 0.4 1.60 10,000 8,333 10,000 4.19 3.49 4.19 Other biscuit/cookies 460.2 0% 9.5 43.80 5,000 3,333 3,500 47.58 31.72 33.30 Rice cakes 234.1 0% 0.8 1.80 2,633 3,000 2,500 2.06 2.34 1.95 Other traditional cakes 376.7 0% 1.8 6.60 3,333 3,333 3,000 5.82 5.82 5.24 Other cereal preparations 362.8 0% 0.4 1.60 5,000 3,333 3,167 2.17 1.44 1.37 Pork without fat 374.9 18% 5.5 17.00 15,000 14,000 15,000 83.00 77.46 83.00 Pork with fat 456.8 12% 13.0 52.40 15,000 12,000 12,000 195.38 156.30 156.30 Bone of pig 0.0 85% 4.6 0.00 14,000 10,000 10,000 63.74 45.53 45.53 Beef 132.4 20% 4.3 4.50 20,000 15,000 15,000 85.25 63.94 63.94 Bone of cow 0.0 85% 1.6 0.00 12,000 7,000 7,143 19.54 11.40 11.63 Chicken 152.6 45% 5.9 4.90 16,000 10,000 12,000 93.90 58.69 70.43 Duck 233.4 0% 0.9 2.10 12,000 8,833 8,000 10.57 7.78 7.05 Other meat 176.0 25% 6.1 8.00 7,200 6,000 5,000 43.53 36.27 30.23 Other locally proc.meat 204.8 0% 0.4 0.90 16,000 12,000 12,000 6.89 5.16 5.16 Mud fish (large) 99.2 50% 2.2 1.10 10,000 8,000 8,000 21.55 17.24 17.24 Mud fish (small) 99.2 50% 15.5 7.70 10,000 8,000 8,000 154.49 123.59 123.59 Snake fish 87.5 43% 0.7 0.30 7,000 7,000 6,667 4.56 4.56 4.34 Cat fish 92.3 51% 9.9 4.50 9,000 7,000 7,000 89.25 69.41 69.41 Sea fish (small) 79.8 63% 2.6 0.80 8,000 5,000 4,500 20.98 13.11 11.80 Other fish 136.8 31% 58.8 55.50 8,000 6,000 5,000 470.02 352.51 293.76 Crabs 104.4 62% 6.0 2.40 3,500 2,000 1,500 21.13 12.08 9.06 Other seafood 94.3 45% 6.6 3.40 5,000 4,000 3,000 32.97 26.38 19.78 Smoked fish 212.3 0% 1.2 2.50 10,000 10,000 10,000 11.73 11.73 11.73 Fermented /cheese fish 102.1 0% 9.9 10.10 7,000 6,000 6,000 69.33 59.42 59.42 Dried fish 192.8 24% 3.8 5.50 13,333 9,000 8,000 50.36 33.99 30.22 Canned fish 138.1 0% 1.9 2.60 9,286 6,667 8,500 17.42 12.51 15.95 Dried prawns or shrimps 316.4 0% 0.2 0.70 10,000 15,000 16,667 2.16 3.24 3.60 Duck eggs 182.8 12% 10.6 17.00 5,882 5,882 5,882 62.34 62.34 62.34 Other fresh eggs 128.1 9% 0.2 0.20 10,000 10,000 10,000 1.55 1.55 1.55 Fermented/salted eggs 189.7 12% 0.2 0.40 7,407 7,407 7,407 1.78 1.78 1.78 Vegetable oil / soybean 900.0 0% 0.8 7.00 6,000 5,000 6,000 4.69 3.91 4.69 Other cooking oils 900.0 0% 2.5 22.70 5,495 5,495 6,000 13.85 13.85 15.12 Pork fat 427.5 0% 0.9 4.00 6,000 6,000 6,000 5.60 5.60 5.60 99 Table 36: Food Items, Calorie Content, Prices And Costs Of Food Basket, Cambodia 2009 Kcal in Prices per Kilogram Cost in Current Riels/day   Refuse gr./day Kcal/day .100 gr PhnomPenh Other Urban Rural PhnomPenh Other Urban Rural Banana 93.6 34% 9.6 5.90 1,600 1,500 1,000 15.28 14.32 9.55 Pineapple 48.1 45% 1.4 0.40 2,000 1,429 1,600 2.83 2.02 2.27 Other mangoes 79.2 28% 5.0 2.80 2,000 2,000 1,500 9.92 9.92 7.44 Lime 46.8 23% 0.7 0.30 3,000 5,000 2,500 2.21 3.68 1.84 Rambutan 68.5 56% 1.8 0.50 2,500 2,500 1,500 4.53 4.53 2.72 Longan (mien) 70.8 50% 1.1 0.40 4,500 4,000 4,000 5.08 4.52 4.52 Papaya 39.5 28% 8.9 2.50 1,500 1,000 1,000 13.41 8.94 8.94 Water melon 24.3 37% 5.8 0.90 1,500 1,000 1,000 8.73 5.82 5.82 Other melons 35.9 49% 1.1 0.20 1,000 1,000 1,000 1.14 1.14 1.14 Calamansi 47.8 62% 0.8 0.20 3,000 3,000 3,000 2.53 2.53 2.53 Chico 68.9 20% 0.5 0.30 2,000 2,000 2,000 0.93 0.93 0.93 Jackfruit 78.2 61% 1.0 0.30 1,000 1,000 1,000 1.04 1.04 1.04 Apples 53.4 18% 0.6 0.30 4,000 4,667 5,000 2.40 2.80 3.00 Other fresh fruits 67.5 12% 3.8 2.20 3,500 2,000 1,667 13.15 7.52 6.26 Tamarind 302.9 62% 4.7 5.40 3,000 2,000 2,000 13.98 9.32 9.32 Other preserved fruits 364.6 0% 0.1 0.50 6,000 6,000 4,000 0.74 0.74 0.49 Coconut 362.3 54% 2.8 4.70 4,250 2,000 1,667 11.90 5.60 4.67 Peanuts 617.1 29% 0.3 1.20 6,000 5,000 5,000 1.70 1.42 1.42 Peanut Preparation 602.8 0% 0.6 3.40 6,000 5,000 5,000 3.38 2.82 2.82 Trakun (watercress marsh 17.9 16% 24.1 3.60 1,667 1,200 1,000 40.21 28.95 24.12 cabbage) Garlic 50.9 10% 3.8 1.70 3,000 3,000 3,600 11.34 11.34 13.61 Spring Onion / leeks leaves 32.1 9% 1.0 0.30 5,000 5,000 4,000 5.20 5.20 4.16 Cabbage leaves 25.3 14% 4.7 1.00 2,571 2,000 2,000 11.97 9.31 9.31 Leaf and stem vegetables 14.9 26% 24.0 2.60 3,000 2,000 2,000 72.10 48.07 48.07 Cauliflower 30.3 40% 0.6 0.10 3,500 3,000 2,733 2.14 1.84 1.67 Other leaf and stalk veg- 28.1 0% 20.3 5.70 2,500 2,000 1,400 50.67 40.54 28.38 etables Tomatoes 22.7 6% 3.8 0.80 2,500 2,000 2,000 9.49 7.59 7.59 Bell peppers, sweet 32.3 13% 0.7 0.20 3,292 3,000 3,000 2.27 2.07 2.07 Ridge gourd 20.7 26% 10.7 1.60 2,000 1,333 1,000 21.45 14.30 10.72 Bitter gourd 16.1 20% 1.1 0.10 2,667 2,000 2,000 2.82 2.12 2.12 White/yellow/green gourd 20.0 15% 18.7 3.20 1,714 1,200 1,000 32.07 22.45 18.71 Cucumbers 15.9 28% 16.3 1.90 2,000 1,500 1,250 32.52 24.39 20.33 Brinjals/eggplant 31.4 9% 7.0 2.00 2,000 1,667 1,500 13.95 11.63 10.46 Other fruit vegetables 54.7 13% 19.8 9.40 2,000 1,200 1,000 39.51 23.71 19.76 Onions/shallot 46.8 10% 1.4 0.60 3,000 3,000 3,000 4.17 4.17 4.17 Turnip 39.6 21% 1.6 0.50 1,800 1,500 1,500 2.91 2.42 2.42 Carrots 35.4 17% 0.5 0.10 3,000 2,667 3,000 1.42 1.26 1.42 Other root vegetables 42.4 7% 4.0 1.60 4,000 2,000 2,000 15.81 7.90 7.90 100 Table 36: Food Items, Calorie Content, Prices And Costs Of Food Basket, Cambodia 2009 Kcal in Prices per Kilogram Cost in Current Riels/day   Refuse gr./day Kcal/day .100 gr PhnomPenh Other Urban Rural PhnomPenh Other Urban Rural Sweet Potatoes 121.5 13% 1.4 1.50 1,200 1,100 1,200 1.67 1.53 1.67 Cassava 146.4 27% 0.9 1.00 1,500 2,000 1,500 1.37 1.82 1.37 Traov 116.9 16% 0.7 0.70 2,000 2,000 1,600 1.45 1.45 1.16 Mushrooms 31.0 3% 1.2 0.40 8,000 5,000 4,000 9.47 5.92 4.74 Other tubers and products 347.3 0% 0.6 2.20 1,500 1,000 1,500 0.94 0.63 0.94 of tuber Green gram/Mung beans 300.9 0% 0.8 2.30 4,000 4,000 4,000 3.05 3.05 3.05 Bean sprouts 40.6 7% 0.9 0.30 2,000 2,000 2,000 1.77 1.77 1.77 Long green beans 34.8 0% 3.9 1.40 2,500 2,000 2,000 9.75 7.80 7.80 Other pulses/legumes 345.2 0% 0.7 2.20 4,000 3,000 4,000 2.60 1.95 2.60 Cucumber pickles 17.8 0% 1.5 0.30 4,000 3,000 2,500 6.07 4.55 3.79 Cabbage pickles 41.5 0% 5.4 2.20 3,000 2,500 2,500 16.10 13.42 13.42 Other prepared & pre- 26.6 0% 1.9 0.50 5,000 2,500 2,500 9.68 4.84 4.84 served veg. Granulated (refined) 397.6 0% 4.6 18.30 2,400 2,400 2,600 11.04 11.04 11.96 Other sugar 374.4 0% 3.4 12.60 3,000 3,000 3,000 10.11 10.11 10.11 Confectionary products 534.7 0% 0.1 0.40 12,500 14,286 12,000 0.92 1.05 0.88 Other sugar products 268.0 0% 3.4 9.10 3,333 3,000 4,000 11.30 10.17 13.57 Ice 0.0 0% 8.8 0.00 500 500 500 4.42 4.42 4.42 Ice cream 157.1 0% 0.4 0.60 5,000 6,000 5,000 2.06 2.47 2.06 Other edible ices 51.9 0% 0.6 0.30 3,333 2,041 2,041 2.03 1.24 1.24 Soy sauce 178.8 0% 0.9 1.60 2,823 2,822 2,646 2.45 2.45 2.30 Fish sauce 47.9 0% 6.2 3.00 2,352 1,764 1,764 14.67 11.00 11.00 Other sauces 115.6 0% 2.2 2.50 4,000 3,000 3,000 8.67 6.50 6.50 Salt 0.0 0% 7.4 0.00 1,400 1,100 1,100 10.30 8.09 8.09 Black or white pepper 364.1 0% 0.4 1.60 12,000 10,000 10,000 5.11 4.26 4.26 Red pepper spice 274.6 0% 0.2 0.50 10,000 10,000 5,000 1.74 1.74 0.87 Monosodium glutamate 0.0 0% 3.8 0.00 10,000 10,000 10,000 37.66 37.66 37.66 (MSG) Ginger 317.9 0% 1.2 3.80 5,000 5,000 4,000 5.93 5.93 4.74 Other spices and season- 95.8 28% 4.3 3.00 4,000 3,000 3,000 17.38 13.03 13.03 ings Canned soft drinks 38.7 0% 0.9 0.30 4,735 3,519 3,472 4.03 3.00 2.96 Bottled soft drinks 38.7 0% 1.3 0.50 2,604 2,604 2,604 3.28 3.28 3.28 Fruit drinks 45.5 0% 2.2 1.00 2,041 2,020 2,041 4.39 4.34 4.39 Locally produced spirits/ 245.4 0% 6.9 16.90 2,116 2,116 2,116 14.54 14.54 14.54 liqueurs Locally produced wine 162.7 0% 3.1 5.00 2,116 2,116 2,116 6.51 6.51 6.51 TOTAL 976.94 2200 3,448 2,816 2,583 101 Table 37: Food Items, Calories Content And Costs Of Reduced Food Basket, Cambodia 2009 Kcal in Quantity/day Prices per Kilogram Cost in Current Riels/day   Refuse 100 gr. Value Unit PhnomPenh Other Urban Rural PhnomPenh Other Urban Rural Rice quality 1 (good 353.5 0% 61.65 grams 2,000 2,000 2,000 123.31 123.31 123.31 quality) Rice quality 2 (bad quality) 356.3 0% 484.74 grams 2,000 1,800 1,500 969.49 872.54 727.11 Pork without fat 374.9 18% 6.26 grams 15,000 14,000 15,000 93.90 87.64 93.90 Pork with fat 456.8 12% 21.58 grams 15,000 12,000 12,000 323.64 258.91 258.91 Bone of pig 0 85% 5.15 grams 14,000 10,000 10,000 72.11 51.51 51.51 Beef 132.4 20% 4.82 grams 20,000 15,000 15,000 96.45 72.34 72.34 Chicken 152.6 45% 6.64 grams 16,000 10,000 12,000 106.24 66.40 79.68 Mud fish (large) 99.2 50% 2.44 grams 10,000 8,000 8,000 24.38 19.50 19.50 Mud fish (small) 99.2 50% 17.48 grams 10,000 8,000 8,000 174.78 139.83 139.83 Cat fish 92.3 51% 11.22 grams 9,000 7,000 7,000 100.97 78.53 78.53 Other fish 136.8 31% 73.93 grams 8,000 6,000 5,000 591.44 443.58 369.65 Fermented /cheese fish 102.1 0% 11.21 grams 7,000 6,000 6,000 78.44 67.23 67.23 Dried fish 192.8 24% 4.27 grams 13,333 9,000 8,000 56.98 38.46 34.19 Duck eggs 182.8 12% 0.140 units 5,882 5,882 5,882 70.52 70.52 70.52 Other cooking oils 900 0% 13.62 grams 5,495 5,495 6,000 74.81 74.81 81.70 Banana 93.6 34% 33.16 grams 1,600 1,500 1,000 53.05 49.73 33.16 Papaya 39.5 28% 10.11 grams 1,500 1,000 1,000 15.17 10.11 10.11 Trakun (watercress marsh 17.9 16% 50.22 grams 1,667 1,200 1,000 83.71 60.27 50.23 cabbage) Leaf and stem vegetables 14.9 26% 27.19 grams 3,000 2,000 2,000 81.57 54.38 54.38 Ridge gourd 20.7 26% 12.13 grams 2,000 1,333 1,000 24.27 16.18 12.13 White/yellow/green gourd 20 15% 21.16 grams 1,714 1,200 1,000 36.28 25.40 21.16 Cucumbers 15.9 28% 18.40 grams 2,000 1,500 1,250 36.79 27.59 23.00 Brinjals/eggplant 31.4 9% 7.89 grams 2,000 1,667 1,500 15.79 13.15 11.84 Granulated (refined) 397.6 0% 5.20 grams 2,400 2,400 2,600 12.49 12.49 13.53 Ice 0 0% 10.00 grams 500 500 500 5.00 5.00 5.00 Fish sauce 47.9 0% 6.22 mm 2,352 1,764 1,764 16.60 12.45 12.45 Salt 0 0% 8.32 grams 1,400 1,100 1,100 11.65 9.15 9.15 Monosodium glutamate 0 0% 4.26 grams 10,000 10,000 10,000 42.60 42.60 42.60 (MSG) Locally produced spirits 245.4 0% 7.77 grams 2,116 2,116 2,116 16.45 16.45 16.45 and liqueurs Total             3,408.87 2,820.08 2,583.10 Note: 1 liter = 1,000 milliliter and 1 Kg. = 1,000 grams. Unit conversions for selected products: 1221-Duck eggs: 1Krg. =11.765 units; 1262-Other cooking oils: 1 Lt. =0.91 Kg.; 1280Banana: 1 hand=0.7 Kg.; 1480-Ice: 1 Lt. =0.98 Kg.; 1491Fish sauce: 1 Kg. =0.8818 Lt.; and 2080-Locally produced distilled liqueurs: 1 Kg. =1.0582 Lt. With the conversions provided here, more than 99% of all products reported in 2009 were transformed into one single unit (the same unit listed in the table). For 1221-Duck eggs: 1 unit = 85 grams, and for 1491Fish sauce: 1 liter =1,134 grams (1 milliliter= 1.134 grams) 102 END NOTE 53. Hotel accommodation expenditures were 25 households without information. later estimated using the recall data. 61. Because motorcycle purchases are not as 54. Recall data is collected by directly asking frequent as the other items, the purchase households how much they bought, value was first divided by 5. received, or consumed during the days preceding the survey. Diary data is a 62. School transportation could have been written record of actual expenditures and classified in the Transportation and consumption. Communication grouping, but it was decided that it was more important to 55. This concept is fully understood and applied have an accurate measure of all expendi- in other exercises like the Consumer Price tures for education. Index when at different time intervals, the list of products and their relative weight is 63. A detailed explanation of the poverty reviewed and changed to reflect new pref- line estimates is provided in a similar erences by Cambodian households. document. 56. 2009 was used to “update” the methodol- 64. Primary Sampling Units are groups of ten ogy because the expanded sample was to twenty households sampled as a unit close to 12,000 households. and in proximity, typically selected from the same groups of households identified 57. For a detailed description of the “old” during the population census work. method, see Knowles, J. Poverty Estimates for Cambodia 2009. February 2012. 65. Current Riels are the values as reported by the households before correcting for 58. If no mention of “old” or “new” is made, the regional price differences. “new” methodology and results should be assumed. 66. Each quintile includes exactly twenty percent of the population, ordered from 59. The number of standard deviations from lowest to highest consumption. the mean was estimated for all variables. Use of subpopulations was customary to 67. For daily per capita consumption estimates, take into consideration differences among see World Bank, Measuring Consumption regions, or for example, public and private Using the 2009 Cambodia Socio-Economic health services, or school level (for educa- Survey (CSES2009), December 2012. tion expenditures). 68. FAO/WHO/UNU in 2004. Human Energy 60. The exponential of the equation’s predicted Requirements. Report of a Joint FAO/ value was used as the imputed rent for the WHO/UNU Expert Consultation. Rome. 103 69. The poorest 5 percent of the population 72. Given the small number of poor house- is excluded because their low budget holds in the urban areas, the range was severely restricts their decisions to select increased to 40% for Phnom Penh and to food. For Phnom Penh and Other Urban, 15% for Other Urban households. The ref- the number of observations was too small, erence group included 23 households so the 5-25 internal percentiles for each of from Phnom Penh, 35 from other urban the two urban regions was estimated and areas, and 440 rural households. used. 73. This share is also known as the Engels coef- 70. Observed calorie consumption does ficient. not include all dropped products due to limited use, lack of appropriate calorie 74. As poverty decreases, the share of the content, difficult measurement issues (i.e. budget dedicated to food also decreases bowls of rice), or meals outside home. (Engels’s Law). 71. The idea is related to the concept of 75. This is the estimated share for “households “revealed preferences” when households whose total expenditures equal the food “show” their preferences. In this case, poverty line” from The World Bank 1997, households’ non-food consumption is A Poverty Profile of Cambodia 1993/1994 “revealed” to be as important as the “basic” discussion paper No. 373, p18. food not consumed. 104 New Government and World Bank Annex 2 Poverty estimates 2004-2011 In 2011 a team of analysts from the National in 2004 to 5.3 percent in 2009; in 2011, it fell Institute of Statistics (NIS); the General moderately to 3.8 percent (Figure 47). The Directorate of Planning; the Economic, Social, Royal Government of Cambodia poverty rates and Cultural Council (ECOSOCC); and the for 2009, 2010, and 2011 are very similar to the World Bank estimated Cambodian poverty. World Bank estimates–with differences of less This new poverty estimate was developed for the than one percentage point (Figure 49). 2009 Cambodia Socio Economic Survey (CSES).76 This new estimate not only considers changes Food Poverty and Total Poverty were esti- in Cambodia, but also reflects current con- mated for three regions of the country: Phnom ditions within the country. These same prin- Penh, other urban, and rural. Each of these ciples were applied to the 2004, 2007, 2008, regional averages show very similar changes 2010, and 2011 data sets to estimate con- to the national poverty rate—with important sumption, poverty lines, and poverty rates. reductions from 2004 to 2009 and a slight The Royal Government of Cambodia con- reduction in 2010 and 2011.77 ducted a similar exercise using the 2009 CSES household survey. The overwhelming majority of Cam- bodia’s poor live in the rural regions: this Poverty Rates (Headcount Rate) general trend has not changed. Poverty is much higher in the rural areas of Cambodia; it is Total poverty has significantly decreased lowest in Phnom Penh (Table 38). In 2004, about since 2004. Total poverty fell from 53.3 percent 89 percent of the poor lived in rural regions. By in 2004 to 23.9 percent in 2009; over the next 2011, 91 percent of the poor lived in rural regions. two years, it fell again reaching 20.5 percent in About 90 out of 100 food-poor persons lived in 2011. Food poverty has also decreased sub- rural households—both in 2004 and in 2011. stantially. Food poverty fell from 16.0 percent 105 Figure 47: Poverty Rates In Cambodia 2004, 2007-2011 Total Poverty Food Poverty Source: World Bank estimates using CSES, provided by NIS, Cambodia In constant 2009 Phnom Penh Riels, Consumption and Poverty Lines average per capita consumption has increased from CR 6,399 in 2004 to CR 9,325 First, to measure poverty, the total value in 2009; average per capita consumption of all goods and services consumed by each then decreased to CR 8,815 in 2011. The household was computed. Then, the per cost of the basic food basket has increased capita consumption per day was estimated from CR 3,153 in 2004 to CR 3,543 in 2011; the using household size. Second, the cost of a total poverty line has remained constant at CR food basket containing 2,200 calories per day 5,326 per day per person (Table 40). was computed. (This is the food poverty line.) Finally, the total poverty line was estimated Since 1993, Cambodians have decreased by adding a quantity for the consumption of the share of their household budget for food. goods and services distinct from foods consti- In 1993, the average household in Cambodia tuting the food poverty line. Households with allocated 68 percent of their total budget for consumption values below the food poverty food items. With higher incomes and better line are considered ‘Food Poor’ and house- socioeconomic conditions, the share has holds with consumption values below the steadily decreased over time–reaching about total poverty line are considered ‘Poor‘ 55 percent in 2004 and nearly 51 percent in 2011 (Figure 48). In 2011, per capita consumption was CR 17,340 in Phnom Penh; CR 9,430 in other In 2011, the average household in urban areas; and CR 6,398 for each day in Cambodia dedicated 51.1 percent of its budget rural households. Consumption aggregates for food, 14.8 percent for housing, 6 percent and poverty lines for each region for 2004, for house utilities, and 7.6 percent for commu- and from 2007 through 2011 are presented in nication and transportation. Personal expen- Table 39 (in current Cambodian Riels [CR]). ditures accounted for 5.3 percent, followed 106 សូចនាករប៉ាន់ស្មានថ្មីនៃភាពក្រីក្រ ចេញដ�ោយរដ្ឋាភិបាល និងធនាគារពិភពល�ោក សម្រាប់ឆ្នាំ២០០៤-២០១១ ​​​​​​​​​ Table 38: "New" Headcount Rates By Regions In Cambodia 2004, 2007-2011 Food Poverty Total Poverty Year Phnom Penh Other Urban Rural Phnom Penh Other Urban Rural 2004 3.81% 11.43% 17.89% 15.83% 39.67% 58.97% 2007 0.06% 8.70% 15.26% 2.66% 35.04% 57.86% 2008 0.62% 4.54% 7.17% 2.54% 26.83% 44.60% 2009 0.56% 1.98% 6.28% 4.27% 12.67% 27.53% 2010 0.74% 2.84% 5.86% 4.54% 12.61% 25.42% 2011 0.00% 3.75% 4.38% 1.53% 16.10% 23.72% 2004 Contribution 2.0% 7.8% 90.2% 2.5% 8.1% 89.4% 2011 Contribution 0.0% 10.1% 89.9% 0.8% 8.2% 91.0% Source: World Bank estimates using CSES ​​​​​​​​​ Table 39: Daily Consumption And Poverty Lines, Cambodia 2004, 2007-2011 Per Capita consumption Food Poverty Line Total poverty line   Year Phnom Other Other Phnom Other Other Phnom Other Other Penh Urban Rural Penh Urban Rural Penh Urban Rural 2004 7,148 4,030 2,549 1,980 1,573 1,423 3,361 2,623 2,407 2007 13,104 6,544 3,482 2,499 1,907 1,765 4,394 3,430 3,157 2008 16,902 7,728 5,571 3,654 2,744 2,540 5,928 4,594 4,597 2009 15,709 10,184 6,013 3,409 2,820 2,583 5,326 4,273 3,914 2010 16,888 10,614 6,045 3,606 2,984 2,733 5,609 4,501 4,122 2011 17,340 9,430 6,398 3,894 3,221 2,951 6,014 4,828 4,422 Source: World Bank estimates using CSES Table 40: Daily Consumption And Poverty Lines ( 2009 Phnom Penh Constant Riels), Cambodia  Year Per Capita consumption Food Poverty Line Total poverty line 2004 6,399 3,153 5,326 2007 7,243 2,981 5,326 2008 7,580 2,997 5,326 2009 9,325 3,505 5,326 2010 9,105 3,521 5,326 2011 8,815 3,543 5,326 Source: World Bank estimates using CSES 107 Figure 48: Food Share In Household Budget, Cambodia 2004, 2007-2010 Source: World Bank estimates using CSES by health and education (2.7 and 2.6 percent, New Royal Government of respectively). Small durables and recreation Cambodia and World Bank costs accounted for 1.7 and 1.2 percent of the Poverty Comparisons budget. Richer households spent a smaller budget share on food and a larger share on Food Poverty Lines housing, communication and transporta- tion, durables, entertainment, education, and In 2004, 2007, and 2009, the unit values health. reported by households were used to calcu- late the cost of the reduced food basket. In The Royal Government of Cambodia esti- 2008, the 2007 values were updated with mated poverty using the 2009 Cambodia the monthly food index, created by James Socio Economic Survey (CSES) by comput- Knowles. In 2010 and 2011, the food poverty ing per capita expenditure and poverty lines. lines were calculated using the values of the Households with expenditure values below previous year updated with the food CPI (the the poverty line were considered poor. In “Combined” series, using both Phnom Penh 2004, 2007, 2008, 2010, and 2011, per capita and Cambodia’s provinces). expenditure was estimated using the CSES household surveys. Poverty lines were esti- Non-food Component of mated by adjusting the 2009 estimated value Total Poverty Lines with the Combined Consumer Price Index (CPI).78 The resulting poverty rates are very In 2004, 2007, and 2008, the values for similar to World Bank estimates, especially for 2009 were updated or modified with James 2009, 2010, and 2011 (Figure 49). Knowles’ non-food price index. In 2009, the 108 សូចនាករប៉ាន់ស្មានថ្មីនៃភាពក្រីក្រ ចេញដ�ោយរដ្ឋាភិបាល និងធនាគារពិភពល�ោក សម្រាប់ឆ្នាំ២០០៤-២០១១ Figure 49: New Poverty Estimates From World Bank And Royal Government Of Cambodia WB RGC Source: World Bank and Royal Government of Cambodia estimates using CSES shares were estimated by observing people with consumption values hovering near the END NOTE poverty line. In 2010 and 2011, the 2009 values were updated with the non-food CPI (the 76. The “old” poverty estimates were devel- “Combined” series, using Phnom Penh and oped based on the 1993-94 CSES house- Cambodia’s provinces).79 hold survey. They reflect conditions from almost twenty years ago. The “new” and the “old” poverty estimates show almost the same evolution over time: 77. The exception is for “Other Urban” house- holds in 2010 and 2011 in which poverty • Food Poverty, The “new” estimates are increased. slightly lower than the “old” estimates. 78. The Combined CPI includes prices from • Total Poverty. The “new” estimates are sig- Phnom Penh and the provinces. nificantly higher than the “old” estimates. 79. Food and total poverty line values in 2010 Figure 50 shows estimates of both Food and 2011were updated using the same Poverty (a) and Total Poverty (b); it further illus- method as the government of Cambodia trates the similarity of the changes from the “old” and the “new” methodologies. 109 Figure 50: Headcount Rates In Cambodia With “Old” And “New” Methodologies a. Food Poverty Rate New Old b. Overall (Total) Poverty Rate New Old Source: World Bank estimates using CSES 110 សូចនាករប៉ាន់ស្មានថ្មីនៃភាពក្រីក្រ ចេញដ�ោយរដ្ឋាភិបាល និងធនាគារពិភពល�ោក សម្រាប់ឆ្នាំ២០០៤-២០១១ Simulation of Drivers of Annex 3 Poverty Reduction 2007-2009 To explain the increase in poverty reduc- change. Since traditional savings are limited tion in rural households from 2007 to 2009, in rural Cambodia, this simulation uses income an examination is required into the increase changes for the reference group households in consumption for the group of rural house- as the direct source of increased consumption. holds that escaped poverty.80 This “Reference Group” includes households within the 27.5 From 2007-2009 the price of rice percentile (poor in 2009) and the 57.9 per- increased considerable changing the centile (poor in 2007). The average daily per behavior of some farmers. According to capita consumption for the Reference Group the Consumer Price Index, the price of rice in 2007 was 4,415 in 2009 Phnom Penh Riels, increased 37.1 percent from 2007 to 2009 (row and 6,465 in 2009. In other words, the average J, Table 41), compared to an actual decrease per capita increase for the reference group of 6.1 percent in 2009-2011.82 The impact of was 2,050 per day.81 Less than 40 percent of the price increase affected household income the increase was in 2008 and over 60 percent in three different ways: first by increasing the in 2009. In 2009, per capita agricultural GDP— actual value of the harvest, second by increasing mostly linked to rural households—increased the planted area (bringing into production oth- 3.8 percent in Cambodia, compared to a 10.9 erwise not used land or switching from other percent decrease for industrial GDP —mostly less profitable crops), and third by allowing linked to urban households—. households to increase the use of inputs and to boost productivity (specifically, an increase in Household consumption increase is access to credit owing to higher than expected possible when household income and savings income and better returns on investment). 111 From 2007 to 2009, total agricultural land impact on farmers’ behavior and the observed use increased by 7.6 percent in Cambodia; national averages from government statistics households in the Reference group reported are reasonable indicators of what happened an impressive increase of 40.4 percent (row K, for poor and non-poor households alike. Table 41); the remaining households reported a decrease of 3.6 percent.83 From 2009 to 2011, From 2004 to 2007 the price of rice households in the reference group actually increased, in constant Riels, 4.8 percent exhibited a decrease in rice production by (Cambodia Consumer Price Index: CPI). 8.5 percent (harvested area). Productivity According to government statistics, during increases from 2007 to 2009 were estimated the same three year period, total rice pro- from government statistics at 11.8 percent for duction in Cambodia increased 33.0% due wet season rice and 2.3 percent for dry season to an impressive productivity increased of rice (row L, Table 41).84 After examining all the 18 percent combined with a 12.2 percent changes attributed to the increase in the price increased area harvested.85 The production of rice, the net gain was 1,393 Riels per person increase was possible because increased input per day for all rice farmers (Table 41). use estimated at a cost of 20% the increased production. Because not everybody is growing rice, total income increase has to be adjusted. Using the estimated value of rice for 2007 After adjusting for the 20% of households not of CR 957 (Table 41), the total impact of the growing rice in the Reference group, the net price increase in 2004-2007 was CR 44 per increases are: 429 Riels from the price increase capita per day, and the production increase of original production, 471 Riels from produc- contributed another CR 191 per capita per day. tion increase (with original price) and 211 from the price increase of extra production (last 2004-2009 column, Table 41). The total average for the entire group was an increase of 1,111 per day The total impact is the sum of the two per person in 2009 Phnom Penh Riels. period estimates. For 2004-2009 the impact of the increased rice price was: CR. 429 + CR. 2004-2007 211 + CR. 44 = CR. 683.72 and the impact of increased productivity and area harvested was: From 2004 and 2007 rice price increases CR. 471 + CR. 191 = CR. 661.94. The combined were moderate. Planted area as well as har- income increase was CR. 1,345.66 per day per vested Tons of rice showed important person 2009 Phnom Penh Riels. increases in part due to very low yields in 2004. Unfortunately in 2004 the CSES survey Agricultural Wages 2004-2009 did not include questions regarding planted or harvested area for agricultural crops and Using the CSES 2004 and 2009, the increase no estimate for the impact of rice from 2004- income from agricultural wages was 49%. 2007 can be derived from the micro data. In Given a wage labor income of CR. 918 (Table the other hand, the small price changes during 15), the corresponding income increase is CR. the period are not expected to have major 450 (day/person, 2009 Phnom Penh Riels). 112 Table 41: Estimated Income Increase From Rice In 2008 And 2009, Cambodia     Wet Rice Dry Rice Total 2009 Phnom Penh Riels Initial conditions in 2007 Farmers Ref. group A Hectares 0.97 0.22 1.19 B Yield per Kilogram 2.20 4.50 C=A*B Tons of paddy rice 2.13 0.98 3.11 D=C*0.55 Milled (55% of paddy) 1.17 0.54 1.71 E=D*1,000 Milled in Kilograms 1,172 537 1,708 F Production value Riels/Kg in 2007 910 910 910 Nominal Riels/year/HH (thou- G=E*F (,000) sand) 1,066 488 1,554 2,126 1,696 H=G/365 Value per day/HH 2,920 1,337 4,257 5,826 4,647 I=H/4.86 Nominal Riels/day per capita 601 275 877 1,200 957 Changes between 2007 and 2009 in “Reference Group” J Price Increase 37.1% 37.1% 37.1%     K Area increased 40.4% 40.4% 40.4%     L Productivity increase 11.8% 2.3%       Impact of changes in per capita consumption per day by different factors M=I*(K+L) Production increase * 2007 price 314 118 432 591 471 N=I*J Price increase * old production 269 123 393 538 429 O=N*J Price increase * production increase 141 53 193 265 211 M+N+O Total Increase 724 294 1,018 1,393 1,111 113 END NOTE 80. Because the CSES is not a panel survey, 84. Since consumption in 2008 and 2009 is the corresponding households are not the affected by the yields from the previous ones that escaped poverty but the group years, the values for 2007 and 2008 were of household representing those that used. Information from the CSES is also actually escaped poverty. from the previous year since agricultural harvest questions are from the previous 81. Unless otherwise stated, Riels quantities season. The 2006 CSES rice productivity reported here are in 2009 Phnom Penh estimate was 2.20 tons of paddy rice per Riels. hectare in the wet season and 4.50 tons per hectare in the dry season, compared to the 82. A single price increase was used to either wet Royal Government of Cambodia 2.27 Tons or dry season rice because there is no infor- (wet) and 3.94 (dry). The estimates are con- mation related to the time farmers sold their sidered close enough to use the productiv- product. Total rice price inflation is the differ- ity growth reported by the government. ence in the average constant price of rice in 2009 and 2007 divided by the 2007 price. 85. Since consumption in 2004 and 2007 is affected by the yields from the previous 83. Households in the reference group include years, the RGC official productivity values 30.4 percent of rural households in Cambodia for 2003 and 2006 were used. Information and their behavior is not representative of from the CSES is also from the previous the entire country or all rural households. year since agricultural harvest questions are from the previous season. 114 Bibliography Alderman, H., and R. Yemtsov. “Productive Role Group. What Works? Interventions for of Social Protection.” Background Paper for Maternal and Child Under-nutrition and the World Bank 2012-2022 Social Protection Survival.” The Lancet, 371, Issue 9610, Pages and Labor Strategy, Washington D.C., 2012. 417 – 440, 2008. Barro, R. J. “Determinants of Democracy.” The Brabin BJ, Hakimi M, Pelletier D. “An Analysis of Journal of Political Economy, 107 (S6), Anemia and Pregnancy-Related Maternal S158–S183, 1999. Mortality.” J. Nutr., 131 no. 2 604S-615S, 2001. Barker DJ, Gluckman PD, Godfrey KM, Harding JE, Owens JA, Robinson JS. “Fetal Nutrition Cambodia Development Resource Institute and Cardiovascular Disease in Adult Life.” (CDRI). “Studies on Levels and Sources of The Lancet, 341, Issue 8850: 938–941, 1993. Rural Income. Background Report for World Bank's Cambodia Poverty Assessment.” Barmeier, Benjamin Leo and Lulia. “Who are the Phnom Penh, 2013. MDG Trailblazers?” A New MDG Progress Index. Center for Global Development, Campante, F., & Glaeser, E. L. “Yet Another Tale Working Paper 222, 2010. of Two Cities: Buenos Aires and Chicago.” Cambridge, MA, 2009. Bhutta ZA, Ahmed T, Black RE, Cousens S, Dewey K, Giugliani E, Haider BA, Kirkwood Clarke, M. “What Matters Most for Student B, Morris SS, Sachdev HP, Shekar M; Assessment Systems: A Framework Paper.” “Maternal and Child Undernutrition Study Washington D.C., 2012. 115 Caulfield LE, Bentley ME, Ahmed S. “Is Hanushek, E. A., & Woessmann, L. “Do Better Prolonged Breastfeeding Associated with Schools Lead to More Growth? Cognitive Malnutrition? Evidence from Nineteen Skills, Economic Outcomes, and Causation.” Demographic and Health Surveys.” Int. J. Cambridge, MA., 2009. Epidemiol 25 (4): 693-703., 1996. Helmers, K., Gibson, J., and Wallgren, P. “Final Ferreira, F., D. Filmer, and N. Schady. “Own and Report of the Rural Sources of Income Sibling Effects of Conditional Cash Transfer and Livelihood Strategies Study”. Report Programs. Theory and Evidence from prepared for the World Bank, 2011. Cambodia.” Washington DC: World Bank Policy Research Working Paper, 2009. Hennicot, J-C. “Cambodia - Social Protection Expenditure and Performance Review.” Filmer, D. and Schady, N. “School Enrollment, Geneva: International Labour Office, 2012. Selection and Test Scores.” Washington DC: World Bank Policy Research Working Paper HRINC. “Higher Education and Skills for the 4998, 2009. Labor Market in Cambodia.”, 2010. Filmer, D. and Schady, N. “Getting Girls into International Labor Organization (ILO). “Social School: Evidence from a Scholarship Protection Expenditure and Performance Program in Cambodia.” World Bank Policy Review.”, 2012. Research Working Paper 3910., 2009. IMF. “Cambodia: 2012 Article IV Consultation”. Gladys Lopez-Acevedo and Raymond Country Report No. 13/2, 2013. Robsertson (Editors). “Sewing Success? Employment, Wages, and Poverty following Ir Por et al. “Challenges in Identifying the Poor: the End of the Multi-fibre Arrangements.” An Assessment of Household Eligibility for The World Bank, 2012. HEF after Four Years of pre-identification in Oddar Meanchey.” HSO&P 23:385-407-; BTC Flores, Gabriela, Ir Por, Chean Men and others. Project Evaluation Surveys, 2008. “Financial Protection of Patients through Compensation of Providers: The Impact Knowles, James. “Poverty Estimates for of Health Equity Funds in Cambodia.” Cambodia, 2009.” Unpublished, 2012. Tinbergen Institute Discussion Paper, TI 2011 – 169/3, University of Amsterdam, ———. “Poverty Estimates for Cambodia, 2011. 2008.” Unpublished, August 2010. GIZ, “Determinants of Non-Utilization of Public Louise Fox, and Thomas Pave-Sohnsen. Health Services among Poor Households “Household Enterprises in Sub-Saharan Covered by a Social Protection Scheme: An Africa.” Policy Research Working Paper Evaluation in Kampot Operational District.” 6184. The World Bank, 2012. Cambodia, 2011. 116 National Institute of Statistics. Ministry ———. Council for Agricultural and Rural of Planning. “Consumer Price Index”. Development. “Towards a Social Protection Cambodia, 2011. Strategy for the Poor and Vulnerable.” Phnom Penh, Cambodia, 2010. ———. “Statistical Yearbook of Cambodia 2011”. Phnom Penh, Cambodia, 2011. Rummel, R. J. “Death by Government”. New Brunswick, N.J.: Transaction Publishers, ———. “Cambodia Socio-Economic Survey 1994. 2009”. Phnom Penh: Ministry of Planning, 2010. Savchenko, Yevgeniya and Gladys Lopez- Acevedo. “Female Wages in the Apparel ———. “National Report on Final Census Industry post-MFA: the Cases of Cambodia Results”. Phnom Penh, Cambodia: National and Sri-Lanka.” Policy Research Working Institute of Statistics, Ministry of Planning, Paper, 3649. The World Bank, 2012. 2009. UNDP. “Human Development Indicator”, 2000 Natsuda, Kaoru, Kenta Goto and John to 2011. Thoburn. “Challenges to the Cambodian Garment Industry in the Global Garment United Nations. “World Population Prospects Value Chain.” Working Paper 03. Center for Revision by the Population Division of Asia Pacific Studies. Asia Pacific University, the Department of Economic and Social Beppu, Japan, 2010. Affairs”., 2011. NGO Forum on Cambodia. “Learning from World Bank. “A Poverty Profile of Cambodia.” Irrigation Projects in Cambodia: Benefits Discussion Paper No. 373, October. and Challenges”. Phnom Penh, Cambodia: Washington DC, 1997. NGO Forum on Cambodia, October 2011. ———. Cambodia: “Halving Poverty by 2015? Pritchett, L., & Viarengo, M. “Producing Poverty Assessment 2006.” The World Bank, Superstars for Economic Mundial: The 2006. Mexican Predicament with Quality of Education.” Cambridge, MA., 2009. ———. “World Development Indicators (WDI)”, 2010. Royal Government of Cambodia (RGC). “Mid- term Review 2011 on National Strategic ———. “Matching Aspirations: Skills for Development Plan Update 2009-2013.” Implementing Cambodia’s Growth Phnom Penh: Royal Government of Strategy.” Report Number 67349-KH, 2012. Cambodia, April 2012. ———. “Analysis of a Comparable Wealth ———. “National Social Protection Strategy Index over Time in Cambodia 2004-2011.” for the Poor and Vulnerable (2011-2015).” Technical Report to the World Bank by Phnom Penh, Cambodia, 2011. James Knowles, 2013. 117 ———. “Commodity Prices Pink Sheet”. http:// Water and Sanitation Program (WSP). econ.worldbank.org/WBSITE/EXTERNAL/ “Economic Assessments of Sanitation EXTDEC/EXTDECPROSPECTS/0,,conten Interventions in Cambodia.” Jakarta: WSP tMDK:21574907~menuPK:7859231~pa East-Asia and the Pacific Regional Office, gePK:64165401~piPK:64165026~theSit 2012. ePK:476883,00.html, n.d. ———. “Economic Assessments of Sanitation ———. “The World Bank (2011) Social Interventions in Cambodia.” Jakarta: WSP Protection Atlas.” Cambodia National Risk East-Asia and the Pacific Regional Office, and Vulnerability Assessment, 2008. 2012. World Food Programme. “WFP Cambodia School Feeding 2000-2010: A Mixed Method Impact Evaluation.” Rome, 2010. 118 THE WORLD BANK