Chapter 3. The pro-poor potential of social protection and fiscal policy

This chapter focuses on the impact of existing social protection programmes on poverty and inequality. It finds that social assistance in Cambodia is well targeted through the IDPoor system but has little impact on poverty and inequality because of low levels of coverage and low benefit levels. Health Equity Funds, however, do ease the burden of health spending for a significant number of poor households. Pension benefits for retired civil servants and military veterans dominate social protection spending and are unsustainable. Social protection spending in Cambodia is low by regional standards. The tax system is progressive but increases poverty, an impact that public transfers are too small to offset.

  

The previous chapters examined the needs for social protection in Cambodia’s fast-changing society (Chapter 1) and discussed the extent to which the country’s current social protection system is aligned to the needs of the population of today and tomorrow (Chapter 2). This chapter focuses on the impact of the existing social protection programmes on poverty and inequality. In doing so, it will not only examine the main social protection programmes but also analyse the way these policies are financed, since both aspects are likely to affect social policy outcomes.

Social assistance is undermined by low coverage

The low coverage of social assistance programmes means that the sector currently has little impact in terms of reducing poverty or inequality at a national level. Figure 3.1 shows coverage rates for scholarship programmes acquired from CSES data (distinguishing between those that are administered by the Royal Government of Cambodia (RGC) and those which are not). No quintile has a coverage rate higher than 2%.

Figure 3.1. Scholarship coverage is lowest for students from the poorest households
Incidence of scholarship beneficiaries among urban and rural households by income quintile in Cambodia (2014)
picture

Note: The sample used for the government scholarships and other scholarships are households with school-attending children. Therefore, quintiles of household consumption in 2014 Cambodian riel (KHR) are defined according to the following cut-offs: 698 791 / 938 771 / 1 218 666 / 1 688 181.

Source: Authors’ calculations based on NIS (2014), Cambodia Socio-Economic Survey 2014.

The level of benefits is also not high: the value of the scholarship, USD 60 per year, equates to just 10.8% of the national poverty gap. However, it should be noted that this benefit level would be more than enough to close the poverty gap if received universally, reflecting the diminished severity of poverty between 2004 and 2014.

Technical vocational education and training (TVET), meanwhile, accounted for 16% of individuals currently attending an educational facility in 2014. There are no data for estimating the incidence of TVET coverage, though it is important to note that its targeting criteria relate to educational attainment and sector of employment rather than income. Figure 3.2 estimates TVET coverage across the income distribution based on microsimulation analysis. Given the size of the informal sector in Cambodia, there is no marked difference in coverage rates across deciles.

Figure 3.2. Individuals across the income distribution make use of TVET programmes
TVET coverage by household consumption decile in Cambodia (2014)
picture

Note: The deciles of household consumption in 2014 Cambodian riel (KHR) are defined according to the following cut-offs: 350 282 / 456 077 / 549 250 / 636 399 / 729 553 / 829 604 / 961 286 / 1 137 181 / 1 495 614.

Source: Authors’ calculations based on NIS (2014), Cambodia Socio-Economic Survey 2014.

Health spending is pro-poor and coverage is growing

In 2014, 8.3% of the population reported having benefited from free or subsidised health care (Figure 3.3) (NIS, 2014). In practical terms, this indicates the extent of coverage by Health Equity Funds (HEF), although a small number of those surveyed might have benefitted from fee exemptions or Kanta Bopha hospitals.1 This coverage level corresponds with the findings of Flores et al. (2013), who state: “In areas in which a HEF was operating, 7.1% of households reported receipt of free or subsidised health care in the last 12 months, and a further 2.2% reported entitlement without receiving treatment. In areas with no HEF, the corresponding figures are 2.0% and 0.01%.”

Figure 3.3. A majority of households do not have access to the HEF
Share of households with access to free or subsidised health care in Cambodia (2014)
picture

Note: MHH = male-headed household; FHH = female-headed household.

Source: Authors’ calculations based on NIS (2014), Cambodia Socio-Economic Survey 2014.

Nonetheless, the proportion of respondents who accessed free or subsidised healthcare is lower than HEF data indicate. This is likely to reflect the fact that people covered by HEF do not always use the public health facilities with whom HEF contracts (see Chapter 2).

In rural areas, the percentage of persons living in a household with access to subsidised health care is 9.2%, compared with 4.9% in urban areas. This finding is consistent with the fact that the Identification of Poor Households Programme (IDPoor) system targets rural areas. CSES data suggest a greater proportion of those with access to subsidised healthcare are women (52.7% versus 47.3% for men). Taking into account the gender of the household head, however, male-headed households report much greater access to free or subsidised health care: a rate of 68.8%, compared with 31.2% for female-headed households, which make up only 21.1% of all households.

The proportion of households reporting access to free or subsidised health care is highest among the first consumption quintile and decreases gradually to under 10% for the richest quintile (Figure 3.4). The first and second quintiles make up more than half of those who reported benefiting from free or subsidised health care. This finding is consistent with the good targeting performance of the IDPoor system. The CSES questionnaire specifies access at point-of-service during the previous 12 months alone, which may explain why self-reported coverage rates are not higher.

Figure 3.4. Access to subsidised health care is highest for the poor
Access to free or subsidised health care at point of service by consumption quintile in the last 12 months in Cambodia (2014)
picture

Note: The quintiles of household consumption in 2014 Cambodian riel (KHR) are defined according to the following cut-offs: 674 952 / 902 023 / 1 154 459 / 1 572 958 / 18 680 054.

Source: Authors’ calculations based on NIS (2014), Cambodia Socio-Economic Survey 2014.

According to 2014 CSES data, out-of-pocket health expenditure rises with income. In the poorest consumption quintile, individual health expenditure ranged from minimum KHR 600 (Cambodian riel) (0.32% of the 2014 national poverty line) to maximum KHR 800 000 (428.71% of the 2014 national poverty line) (Figure 3.5). The weighted mean of out-of-pocket health expenditure for the first quintile in the 30 days preceding the survey was KHR 31 468 (16.8% of the 2014 national poverty line).

Figure 3.5. Out-of-pocket health spending is highest for the richest households
Out-of-pocket health expenditure by consumption quintile in the last 30 days in Cambodia, in KHR (2014)
picture

Note: The quintiles of household consumption in 2014 Cambodian riel (KHR) are defined according to the following cut-offs: 674 952 / 902 023 / 1 154 459 / 1 572 958 / 18 680 054.

Source: Authors’ calculations based on NIS (2014), Cambodia Socio-Economic Survey 2014.

Catastrophic out-of-pocket health expenditure is much higher in rural areas. As captured by the 2014 CSES, 8.96% of households reported catastrophic health expenditure2 in the 30 days preceding the survey, over 90% of which were in rural areas (Figure 3.6). This underlines the importance of social health protection in protecting all people against health shocks.

Figure 3.6. Catastrophic out-of-pocket health expenditure affects a majority of households
picture

Source: Authors’ calculations based on NIS (2014), Cambodia Socio-Economic Survey 2014, available at nada-nis.gov.kh/index.php/catalog/CSES.

Box 3.1. Impact evaluation of health care micro-insurance in rural Cambodia

The French NGO Groupe de recherches et d’échanges technologiques (GRET) launched a health care micro-insurance scheme, Sokapheap Krousat Yeugn (SKY), in 245 villages in the provinces of Takeo, Kandal and Kampot. From November 2007 to December 2008, the Agence française de développement (AFD) conducted an impact evaluation of the scheme which found that insurance was successful in helping families decrease their overall health expenditure (Levine et al., 2010).

SKY was estimated to decrease total health care costs of serious health shocks by over 40%. Households with coverage had more than one-third less debt and more than 75% less health-related debt, compared to households who did not have coverage. The scheme was also found to have changed health-seeking behaviour, increasing utilisation of insurance-covered public facilities and decreasing use of unregulated care, which was not covered by the scheme.

SKY had no detectable impact on preventive care, and there was little evidence of improvements in health outcomes. However, perceived or actual health care quality issues may have been at the root of that. The results remain important in developing more and better micro-insurance pilot programmes and methods of evaluating their impacts in Cambodia.

Public spending on social protection is low and not pro-poor

As Figure 3.7 shows, the RGC prioritised expenditure on the social sector3 between 2009 and 2015. This sector received the largest allocation (4.5% of GDP in 2015) and showed the strongest growth over the survey period. The Ministries of Health and Education (MoH and MoEYS, respectively) combined accounted for just under 80% of total social spending on average between 2009 and 2015, while the Ministry of Social Affairs, Veterans and Youth Rehabilitation’s (MoSVY) contribution increased from 13% to 19% over the same period. Economic sector4 spending rose slightly between 2011 and 2015 but was only the fourth greatest expenditure after defence and general administration, which amounted to 3.2% and 1.9% of GDP in 2015, respectively.

Figure 3.7. Social spending is high and rising
Government spending by sector in Cambodia (2009-15)
picture

Source: Ministry of Economy and Finance (2016).

In 2016, transfers to the National Social Security Fund for Civil Servants (NSSF-C) and National Fund for Veterans (NFV) accounted for 88.8% of MoSVY’s budget (Table 3.1). Of these two funds, the NFV absorbs the greater proportion despite the military being the smaller group. Both social insurance schemes are presently non-contributory (with no plan to introduce contributions for the NFV) and represent a direct (and rapidly increasing) burden on the fiscus, crowding out other social protection initiatives that are targeted at lower-income households.

Table 3.1. MoSVY budget breakdown (2016)

Programmes

Spending, KHR (million)

Spending, USD (million)

% of MoSVY expenditure

National

Provincial

Total

NSSF-C

5 900

265 666

271 566

67.9

38.0

NFV

5 352

355 068

360 420

90.1

50.5

Emergency support to vulnerable groups

7 779

6 374

14 153

3.5

2.0

Child welfare

5 236

7 062

12 298

3.1

1.7

Disabled people

10 943

634

11 577

2.9

1.6

Elderly people

327

335

662

0.2

0.1

Others (institutional development, etc.)

205

43 044

43 249

10.8

6.1

Total MoSVY budget

35 742

678 183

713 925

178.8

100

Source: MoSVY (2016).

Pension payments are received predominantly by wealthier households in urban areas: while 17.5% of households with at least one senior family member received a pension in the highest consumption quintile in urban areas, only 1.4% did so in the poorest quintile in 2014 (Figure 3.8). By contrast, pension benefit incidence was only about 3% in rural areas, with even lower coverage in the poorest rural quintile.

Figure 3.8. Pension coverage is highest in urban areas
Pension coverage by consumption quintile and region in Cambodia (2014)
picture

Note: The quintiles of household consumption in 2014 Cambodian riel (KHR) are defined according to the following cut-offs: 674 952 / 902 023 / 1 154 459 / 1 572 958 / 18 680 054.

Source: Authors’ calculations based on NIS (2014), Cambodia Socio-Economic Survey 2014, available at nada-nis.gov.kh/index.php/catalog/CSES.

Between 2012 and 2016, spending on retirement benefits by the NSSF-C and the NFV increased approximately sixfold and fourfold respectively in nominal terms (Figure 3.9). This increase reflects growth in the number of civil servants, which has been reinforced by wage increases long considered necessary to improve conditions for state employees (World Bank, 2013). According to an actuarial analysis carried out by the ILO, civil service wages have increased by an average of 18% a year over the past ten years (ILO, forthcoming).

Figure 3.9. Pension coverage and spending are rising
NSSF-C and NFV beneficiaries and expenditure (2012-16)
picture

Source: Ministry of Economy and Finance (2016).

The minimum monthly salary for members of the NSSF-C and NFV is currently around USD 150. However, the RGC expects to raise it to around USD 250 as of 2018. Because replacement rates and minimum benefit levels are linked to current wages, salary increases automatically increase pension expenditures and liabilities.

The NSSF-C and the NFV are accruing long-term obligations that will constrain the RGC’s spending decisions in the future even if contributions are enforced for one or even both arrangements. It is imperative that this debt be accurately quantified and recognised as part of the process of establishing a new pension system envisaged by the Social Protection Policy Framework (SPPF).

Cambodia’s social protection spending is low by regional standards

According to the Asian Development Bank (ADB) Social Protection Index (SPI), only Indonesia spends a lower proportion of GDP on social protection than Cambodia at 0.5% of GDP, compared with Cambodia’s 0.6% (Figure 3.10) (ADB, 2013). Viet Nam spent the most on social protection in Southeast Asia (SEA) in 2010 at over 5% of GDP. In 2009, Cambodia spent 0.2% of GDP on social insurance, 0.3% on social assistance and 0.1% on labour market programmes.

Figure 3.10. Social protection expenditure in Cambodia is among the lowest in the region
Social protection expenditure in SEA, % of GDP (2008-10)
picture

Note: Lao PDR = Lao People’s Democratic Republic.

Source: ADB (2013), Social Protection Index (database), hdl.handle.net/11540/79, accessed September 2016.

According to the ADB’s SPI, Cambodia performed better than Lao PDR and Indonesia in average expenditure per beneficiary5 but was much weaker than the rest of the comparison Association of Southeast Asian Nations (ASEAN) countries in terms of the percentage of GDP allocated to social protection (Figure 3.11). Cambodia’s expenditures per intended beneficiary amounted to 2% of the poverty line (set at 25% of GDP per capita), compared with 9.5% for SEA as a whole.

Figure 3.11. Cambodia’s depth and breadth of social protection are low by regional standards
Depth and breadth of social protection in SEA, ADB SPI score (2013)
picture

Note: In the Social Protection Index, depth refers to the ratio of total expenditures to total beneficiaries (normalised by the value of the relative poverty line), while breadth refers to the ratio of total beneficiaries to the total reference population. Depth can thus be interpreted as the average expenditure per actual beneficiary as a percentage of poverty line expenditures.

Source: ADB (2013), Social Protection Index (database), hdl.handle.net/11540/79, accessed September 2016.

Taxes and transfers are failing to reduce poverty

Countries are increasingly assessing the coherence of fiscal and social policies to ensure complementarity. Revenue and expenditure measures are mutually reinforcing; people pay the taxes that fund public services and social protection transfers. Considering only the revenue or expenditure side of the fiscal framework when analysing government efforts to reduce poverty and inequality provides only a partial picture (Bastagli, 2015).

In 2015, value added tax (VAT) on imported goods accounted for 2.0% of Cambodia’s gross domestic product (GDP), VAT on domestic goods accounted for 1.2% of GDP, and excise duties on imported and domestic goods accounted for 2.2% of GDP. Payroll taxes have been on an overall upward trajectory, from 0.3% of GDP in 2010 to 0.6% in 2014 and 0.4% in 2015. The tax on company profits was a much more important source of revenue at 1.9% of GDP. Those revenues are expected to increase from 2016 onwards as a result of a push for companies to register on the “real” rather than “estimated” tax regime.

Modelling based on 2014 CSES data suggests that the quantum of tax paid rises with consumption; so too does the average tax rate (Figure 3.12). The average tax rate is notably lower for the poorest decile. It jumps for the second decile and thereafter increases relatively mildly until the seventh decile, at which it levels off until the tenth. The increase in both quantum of tax and tax rate for the tenth decile reflects both increased consumption at the top end of the distribution and the fact that only the very top earners are likely to pay the salary tax.6

Figure 3.12. Cambodia’s tax system is progressive
Tax quantum and average tax rate by consumption decile per capita in Cambodia (2014)
picture

Note: The deciles of household consumption per capita in 2014 Cambodian riel (KHR) are defined according to the following cut-offs: 131 675/ 157 611 / 181 599 / 203 509 / 228 697 / 258 686 / 294 930 / 352 555 / 456 316.

Source: Authors’ calculations based on NIS (2014), Cambodia Socio-Economic Survey 2014.

Figure 3.13 shows the net effect of taxes and transfers on the income distribution using counterfactual analysis. It models four scenarios. The baseline scenario models the status quo, i.e. current household consumption after the receipt of social protection benefits and payment of taxes. The first counterfactual scenario models household consumption in the absence of public transfers. The second counterfactual scenario models household consumption in the absence of taxes. The third counterfactual scenario models household consumption after subtracting both transfers and taxes.

Figure 3.13. Taxes and transfers are failing to reduce poverty
Mean annual per capita consumption by decile in Cambodia, in USD: baseline and counterfactual scenarios (2014)
picture

Notes: *USD = United States dollar.

Baseline (status quo) = transfers and taxes; counterfactual 1 = taxes, no transfers; counterfactual 2 = transfers, no taxes; counterfactual 3 = no transfers, no taxes. The deciles of household consumption per capita in 2014 Cambodian riel (KHR) are defined according to the following cut-offs: 131 675/ 157 611 / 181 599 / 203 509 / 228 697 / 258 686 / 294 930 / 352 555 / 456 316.

Source: Authors’ calculations based on NIS (2014), Cambodia Socio-Economic Survey 2014.

The model highlights two important trends. First, inequality (as evident from the disparity in consumption levels across the distribution) is lower after taxes but the receipt of transfers plays a minimal role in equalising consumption levels because the coverage and level of benefits is low. Second, all groups along the income distribution, including the poorest, would have higher levels of consumption in the absence of taxes, with or without transfers.

The average total tax burden for poor households is 14.9% of the average total consumption and 67.8% of the national poverty line. For the many households close to the poverty line, taxation can mean the difference between being above or below the threshold. Calculations based on the 2014 CSES show that taxation pushes about 5.1% of the population (166 322 households or 774 389 individuals) into poverty.

The impact of social protection transfers in alleviating poverty is modest due to their low coverage and benefit levels. Based on microsimulations of existing social protection programmes, 0.6% of the population (19 567 households or 91 105 individuals) who have post-tax per capita consumption above the poverty line would be poor without these transfers. Figure 3.14 shows estimates for the headcount poverty rate in the absence of transfers and taxes; it indicates that, while progressive, the fiscal framework’s net impact is impoverishing.

Figure 3.14. Transfers are not sufficient to offset the impact of taxes
Poverty headcount rate according to taxes and transfers in Cambodia: baseline and counterfactual scenarios (2014)
picture

Notes: Baseline (status quo) = transfers and taxes; counterfactual 1 = taxes, no transfers; counterfactual 2 = transfers, no taxes.

Source: Authors’ calculations based on NIS (2014), Cambodia Socio-Economic Survey 2014.

These calculations confirm the importance of assessing the combined effect of taxes and transfers. If the SPPF is to achieve the poverty reduction objectives, transfer amounts will have to increase to offset the impact of taxation on the poor. This is especially the case should official development assistance be declining (as discussed in Chapter 4). Thinking ahead to the implementation of the SPPF, it will be important to assess the impact of different financing strategies, especially on poor and vulnerable households.

References

ADB (2013), The Social Protection Index Assessing Results for Asia and the Pacific, Asian Development Bank, Manila, www.adb.org/sites/default/files/publication/30293/social-protection-index.pdf.

Bastagli, F. (2015), “Bringing taxation into social protection analysis and planning”, ODI Working Paper, No. 421, The Overseas Development Institute, London, odi.org/publications/9671-bringing-taxation-social-protection-analysis-and-planning-working-paper.

Flores, G. et al. (2013), “Financial protection of patients through compensation of providers: The impact of health equity funds in Cambodia”, Journal of Health Economics, Vol. 32, Issue 6, pp. 1180-1193, http://dx.doi.org/10.1016/j.jhealeco.2013.09.012.

ILO (forthcoming), Actuarial valuation of the National Social Security Fund for the Civil Servants and actuarial study on the implementation of a new pension scheme in the private sector, International Labour Organization, Geneva.

Levine, D. et al. (2010), Assessing the Effects of Health Insurance: The SKY Micro-Insurance Program in Rural Cambodia, Impact Analyses Series, No. 4, Agence française de développement, Paris, librairie.afd.fr/en/assessing-the-effects-of-health-insurance-the-sky-microinsurance-program-in-rural-cambodia/.

NIS (2014), Cambodia Socio-Economic Survey 2014, National Institute of Statistics, Ministry of Planning, Royal Government of Cambodia, Phnom Penh.

NIS (2009), Cambodia Socio-Economic Survey 2009, Household Survey 2009, National Institute of Statistics, Ministry of Planning, Royal Government of Cambodia, Phnom Penh.

NIS (2004), Cambodia Socio-Economic Survey 2004, Household Survey 2004, National Institute of Statistics, Ministry of Planning, Royal Government of Cambodia, Phnom Penh.

World Bank (2013), Public service pay in Cambodia: the challenges of salary reform, World Bank, Washington, DC, documents.worldbank.org/curated/en/913481468214829704/Public-service-pay-in-Cambodia-the-challenges-of-salary-reform.

Notes

← 1. The first Kanta Bopha paediatric hospital was established in 1992 and since they have grown to a network of hospitals which are funded mainly through private donations and support from the Swiss Agency for Development and Cooperation.

← 2. According to the World Health Organization’s definition of catastrophic health expenditure.

← 3. The social sector contains the Ministries of Information; Health; Education, Youth, and Sport; Culture and Fine Arts; Environment; Social and Veteran Affairs and Youth Rehabilitation; Cults and Religion; Women Affairs; and Labour and Vocational Training.

← 4. The economic sector comprises the Ministries of Agriculture, Forestry and Fisheries; Rural Development; Public Works and Transport; and Water Resources and Meteorology.

← 5. Divided by the national poverty line to adjust for differences in GDP per capita (ADB, 2013).

← 6. These estimates are based on microsimulation of seven major taxes, including taxes on profit, salaries, property, VAT, excise tax, and tax on petroleum imports and other import duties.