Component report-Part B-10-Dec-01

The first issue relates to the balance between public and private control of the zones. Given Cambodia’s scarcity of experienced managers and limited financial resources the Government is likely to rely upon a private developer to create and manage one or more of the zones. This already appears to be the case in Koh Kong. Such developers would prefer themselves to undertake many of the policy and management decision making authorities that the law gives to the Council and Secretariat.

Ideally, these statutory authorities could be contracted to the developer. In particular, the developer could register investors and set fees. If the developer is providing the majority of services to the investor, including leasing land and then fees cannot be regulated by a sub-decree. An extremely important feature of any zone is the simplicity and efficiency of its administration. Private sector developers can more often provide these efficiencies than the public sector. Public interests are still protected through the provisions of government’s agreement with the developer and through legislation.

The Council and Secretariat would remain responsible for those one stop services that should be undertaken by an entity at arm’s length from the developer and investor. In particular, customs services and tax collection should be performed by an unrelated party.

The law envisages a very senior policy making body (Council) that delegates executive power to a mid-level executive body, that further appears to delegate management activities to representatives in each zone. This structure has the risk of blurring policy and management functions and becoming unnecessarily bureaucratic. Furthermore, accountability should be clarified. The Council should regularly report to Parliament. The Council appears to be represented by unnecessarily senior officials, that is, the Prime Minister.

However, the intended structure could function well if the Council acts as a policy making board of directors and the Secretariat as the executive management with line functions in each zone. Most public sector industrial estate development is undertaken by parastatals with adequate operational autonomy and their own board of directors to enable them to operate in a business like manner. The private sector is often represented on the board of directors reflecting the interests of enterprise investors and private sector developers.

The zone customs agency must demonstrate an ability to monitor imports and exports from the zone with minimum cost to the investor. Ideally, an entirely new agency could be established that is independent from the public sector — along the lines of a revenue authority discussed above. Or these functions could be contracted to the private sector.

Annex F 1 of the Kyoto Customs Convention sets out model provisions for the customs requirements of a duty free zone. The interesting customs feature of the proposed zones is the planned proximity of a duty free processing area with duty and tax paid accommodation and service areas. This suggests the following features could be appropriate:

For these reasons the RGC should not offer a reduced rate of profit tax to attract Thai investors. Further work should be undertaken to compare the cost structures of Thai manufacturers in their current location compared to their likely cost structure in a Cambodian zone. The additional value of GSP benefits and the current tax treatment of the Thai manufactures should be factored into the model.

Table 3.5 compares the net profit for an exporter selling a product with a CIF cost of $100 into three different markets. The selling price at which the exporter has contracted with the importer is $120. If the importer is located in the EU then the good enjoys completely duty free access and the profit is $20 or 16.7 per cent of the contract price. The EU average MFN tariff is 7.4 per cent therefore, the exporter’s profit rate falls to 10.5 per cent The profit spread on sales to the US and Japan are smaller than to the EU.

Table 3.5 shows that if production costs are conservatively assumed to be the same in Cambodian zone as they were in Thailand then if the exporter sells to the EU, they enjoy a gain equivalent to 7.4 per cent of their cost of sales. If their cost of sales is lower in Cambodia then their GSP gain would be greater than 7.4 per cent.

	Unit	EU GSP	EU MFN	US GSP	US MFN	Japan GSP	Japan MFN
CIF cost	$	100	100	100	100	100	100
Duty rate	%	0.0	7.4	1.8	5.0	1.7	4.3
Contract rice	$	120	120	120	120	120	120
Profit price	%	16.7	10.5	15.2	12.5	15.3	13.1

Calculated using 1999 average tariffs across all goods provided by Table 8, Market Access for Developing Countries’ Exports, WB and IMF, April 2001. Current GSP rate for LDCs used rather than 1999 average GSP rate.

The zone investment incentives should be aligned with the incentives offered by the Law on Investment. Consistent incentives will prevent investment locations within Cambodia competing with one another by offering lower rates of profit tax to foreign investors that have already chosen to locate in Cambodia. The current review of investment incentives should consider the tax treatment of Thai investors in the proposed zones.

The RGC should clarify that the US and EU will continue to permit GSP access of goods made in duty free zones in Cambodia. The US generally does not permit duty and tax free entry of goods made in foreign EPZs.

The requirement for a master plan is important. The plan should set out a design concept for the zone and the process by which investors comply with the design concept. Developers should be given the opportunity to participate in the formulation of the plan. The plan should be approved by the Council. The draft law does not require developers or investors to conduct an environmental impact assessment (EIA). The Government should ensure that minimum environmental standards are met by developers and investors. An environmental monitoring office could be established in the zones to evaluate EIAs and monitor compliance. Certificates of Origin could be issued by the zone customs office using the ELVIS and SEGAL links with the US and EU.

The linkages between trade policies and poverty have been addressed recently at length by several papers (World Bank 2001 and WTO 2001). There is a consensus that greater openness to trade can affect an economy in various ways, creating opportunities for the poor as well as risks: i) by affecting the prices of goods and services the poor consume and produce, changes in policy will benefit those who are net consumers of goods that become cheaper and those who can obtain higher prices for their products; ii) by affecting the demand for and returns to factors of production that the poor have to offer, such as unskilled labor; iii) it can affect government revenues and the resources available for antipoverty programs; iv) it can affect the potential for economic growth , which in turns affects poverty; v) due to likely transition costs and the possible increased volatility of growth coming from the opening up of markets, properly designed social protection mechanisms are recommended (G-8 Genoa Summit, July 2001). Following the scope of the document, this chapter will deal mostly with points i), ii), and iv), considering also scenarios where land is distributed to poor farmers and where transaction costs linked to the rice market are reduced. Besides the assessment of the impact of trade policy on poverty, this section intends to be a clearing house for the assessment of the expected impact of other, non-trade policy recommendation done in other parts of the study.

Lack of information for the Cambodian economy (for example, input-output table) forbids the use of any sophisticated economic model (for example, Computable General Equilibrium), and this poverty assessment is rather conducted in a partial equilibrium setting. It provides a proxy of the magnitude of the first order impact on poverty coming from changes in key prices and quantities. The setting of the model is flexible enough to capture impacts coming from:

This chapter presents in section 1 a summary of the poverty profile of Cambodia, highlighting some key issues related to the measurement of poverty in Cambodia. Section 2 identifies who the poor people are in term of consumption patterns and in term of sources of income. Section 3 presents the framework to be used in making the poverty assessment of different policies recommended in the other chapters of this document. Section 4 provides the results of the various simulations done, and Section 5 presents some conclusions.

Current measures of poverty in Cambodia are controversial. Changes in survey design and in interview practices make difficult to compare estimates of poverty rates from the 1999 Cambodia Socioeconomic Survey (CSES) with estimates from previous surveys. The 1999 CSES was done in two rounds of interviews, one in January–March and the other in June-August, with the intention of capturing the effect of seasonality in consumption and income. The detailed analysis of the data done by Gibson (1999) shows that there are important discrepancies in consumption, inequality and poverty estimates for the two round of the 1999 CSES that are not consistent with previous evidence of seasonality in Cambodia. For instance, the estimated headcount poverty rate for Cambodia in 1999 is 51 per cent if both rounds of interview data from the 1999 CSES are used, but it is 64.4 per cent if the first round is used and 35.9 per cent if the second round is used. Because of these discrepancies, it is difficult to see whether poverty increased since the previous survey in 1997. From Gibson’s report, (1999, page iii) we have:

If the data from Round 1 of the 1999 survey are used then poverty appears to have increased, regardless of the methods and adjustments used to ensure comparability with the 1997 data. If the data from Round 2 of 1999 are used, poverty appears to have decreased. If the discrepancies between survey rounds are ignored and all data from the 1999 survey are used poverty either increased or decreased since 1999 depending on the method used to ensure comparability with the 1997 data. A straight comparison of the poverty estimates from the unadjusted 1997 survey data with the combined-rounds 1999 survey data shows a small rise in the headcount index from 47.8 per cent to 51.1 per cent, with large proportionate increases in the depth and severity of poverty. An alternative comparison, which avoid using revised estimates for the headcount poverty rate in 1997, based on a consumption aggregate which exclude rent and medical expenses (the items adjusted in 1997) and used a lower poverty line. Under this comparison, and using the combined rounds data from 1999, there is a small reduction in the headcount index (from 36.1 to 35.4 per cent) but increases in the depth and severity of poverty. However, even this comparison does not remove the basic ambiguity caused by the variation across survey rounds in 1999; with this restricted consumption aggregate and lowered poverty line, there are significant increases in poverty recorded for all poverty measures and all regions from 1997 to Round 1 of 1999 and falls in poverty from 1997 to Round 2 of 1999.

Results for 1997 that have been incorporated into public policy in Cambodia (particularly into the poverty profile) are based on consumption data that were adjusted to correct for possible under-estimation of certain types of consumption, such as health, house rental values, and education. The adjustments made to the data had a big impact in raising the value of mean consumption by almost one-fifth (Gibson, 1999, p. 14).

In spite of its problems, the CSES survey done in 1999 was the first attempt to collect detailed information on household incomes in Cambodia. Households’ sources of income were collected with a detailed questionnaire, and 1999 data is the only available data that allow an assessment on the relative importance of each source of households’ income. Thus it is necessary to rely on the 1999 data to make the type of simulations this chapter on Poverty intends to make. As measures of poverty levels using the 1999 data are subject to controversies and this chapter tries to avoid making references to a particular measurement of poverty, and presents the information by deciles of the per capita adult equivalent expenditure instead. Following Deaton (1997), the deciles were formed by computing the per capita adult equivalent household consumption for each sample household as: total hh consumption/(0.5*# of children + # of adults). The reader should infer that improvements in the consumption level in the first five deciles as a result of any of the policies suggested by the IF analysis, has a pro-poor effect. To follow what seems to be the official data set for poverty analysis in Cambodia (see for instance, the Second Five Year Socioeconomic Development Plan 2001-2005), we present in this chapter results coming from using the second round from the CSES 1999 survey. All the computations were also done using the first round and the complete data set for 1999 and are available from the author upon request. Although to easy exposition those results are not shown in the text, an explanatory footnote indicates whenever these results differ qualitatively from those coming from the second round. The 1993–94 and 1997 poverty profiles identified many of the characteristics of the poor:

In turn, and although some of the cross-sectional poverty patterns are not robust across the two 1999 survey rounds, the 1999 poverty profile indicates that some of them are robust and in general corroborate the findings of previous poverty profiles. Amongst the latter:

Knowledge of how the poor obtain and spend their income is crucial in assessing the impact of economic policies on poor people. For the case of trade policies, it is expected that the impact on the sources of income of the poor will be more important that the impact on their consumption bundle. The reason is that trade reform will affect many relative prices, some of which will move in offsetting directions creating scope for adjustment of the consumption bundle. In contrast, as the poor people generally have limited assets, the most important of which is low skilled labor and/or small quantity of low quality agricultural land that do not allow for cushioning the effect of big changes, the impact on the sources of income (World Bank 2001). The approach used in this chapter — described below in section 3 — relies heavily in locating the poor people in terms of patterns of expenditures and of sources of income. The rest of this section presents first a description of what are the main expenditures of Cambodia’s people, ordered by deciles of adult equivalent per capita expenditures, and then a description of their main sources of income.

Table 4.1 shows by per capita adult equivalent consumption deciles the share of some key consumption aggregates in total expenditure (see annex G for more detail). In Cambodia, expenditures on food take about ¾ of total expenditures of the first five deciles, and are lower than 70 per cent only in the highest two deciles. As expected, expenditures on rice (all varieties) are by far the most important single item of all expenditures for the poorest deciles: this share is 28.4 in the poorest decile, about 23 per cent in the second and third deciles, and still above 20 per cent for the fourth and fifth deciles. Thus, any policy that has a high impact on rice prices is expected to have a high impact in poor households’ consumption. A simple exercise can show that, keeping the other prices and the sources of income constant, a reduction, say, of 10 per cent in the price of rice would imply an extra 3 per cent of disposable income for people in the poorest decile 10 per cent of 28.7 per cent) and 0.6 per cent for people in the richest decile (10 per cent of 5.7 per cent). Second in importance among food items is ‘Fish and fish products’ that involves between 9.2 and 11.1 per cent of total household expenditures in deciles 1 to 9 and only 5 per cent in the tenth decile. Regarding non-food expenditures, the mayor differences between deciles seem to be concentrated in ‘Housing, fuel and transportation’: this category of expenditure captures 46.7 per cent of the total expenditures in the richest decile and only 15.7 per cent in the poorest one.

Decile
Consumption item	1	2	3	4	5	6	7	8	9	10	avg
Food Total	75.7	76.3	76.0	73.7	73.7	72.6	72.0	70.3	64.4	38.6	63.5
Rice, all varieties	28.4	23.3	22.7	20.7	20.6	18.9	17.5	15.1	12.5	5.8	15.6
Fish & fish products	9.9	11.1	10.6	10.5	10.9	10.8	10.0	10.0	9.2	5.0	8.9
All other consumption items	37.4	41.9	42.7	42.5	42.2	43.0	44.5	45.2	42.7	27.8	39.0
Non Food Total	24.4	23.8	24.1	26.4	26.3	27.5	28.0	29.7	35.6	61.4	36.5
Housing, fuel and transportation a	15.7	15.5	15.2	17.1	17.3	17.4	17.5	19.4	24.1	46.7	25.5
Clothing b	2.9	2.8	3.0	3.0	2.8	3.4	3.4	3.2	2.9	2.3	2.9
Other expenditures c	5.8	5.4	5.9	6.2	6.1	6.6	7.1	7.1	8.6	12.4	8.2

a Includes house rent (rental value of subsidized housing, rental value of owner-occupied housing, hotel charges), house maintenance and repair, water and fuel, medical care, transportation and communication, and personal care

c Includes furniture and household equipment and operation, expenditures in recreation, education , personal effects and miscellaneous items.

Chart 4.2 shows the share of food expenditures in total expenditures by per capita consumption decile for households in Phnom Penh, in other urban areas and in rural areas. In rural as well as in other urban areas the share of food in total expenditures is higher than 70 per cent for the first 8 deciles, while in Phnom Penh this share is below 55 per cent for all deciles, and as low as 18 per cent for the 10th decile. These different patterns are just another way of observing the prevalence of relatively higher poverty levels — which are in general associated with a higher share of expenditures on food — in the rural areas of Cambodia. Any policy that affects food prices, everything else constant, will have a greater impact on households’ welfare in rural and in other urban areas, and a smaller impact — albeit still important for people in the poorest deciles — on households in Phnom Penh.

Table 4.3 shows by per capita adult equivalent consumption decile the contribution to total income made by different sources . Earnings from self employment are on average equal to 60.9 per cent of total income, 2/3 of which are coming from activities related to cultivation (22.4 per cent of total income), most of which is coming from rice cultivation (16.3 per cent), livestock (9.4 per cent), fish growing (5.6 per cent) and forestry and hunting (5.7 per cent). Income from wage employment was on average only 20.5 per cent, slightly higher than the 16.5 per cent coming from sources as rental income, interest received and imputed value of houses. Income from remittances represents on average only about 2.1 per cent of total income.

Decile
*Source of Income:*	1	2	3	4	5	6	7	8	9	10	*avg*
Self employment	67.8	71.4	71.7	70.3	73.6	68.5	71.2	70.7	63.3	36.3	60.9
from cultivation	27.8	29.7	31.1	30.3	31.9	32.1	30.6	32.3	19.8	3.2	22.4
from rice cultivation	21.4	24.3	25.3	23.6	25.7	25.0	22.7	20.4	9.7	2.1	16.3
from other crops	6.4	5.4	5.8	6.7	6.2	7.1	7.9	11.9	10.1	1.1	6.1
from livestock	16.6	14.3	14.2	13.0	12.4	11.9	11.6	12.0	8.3	1.5	9.4
from fish growing, etc	6.1	7.1	5.5	8.7	6.9	6.8	8.7	7.1	5.2	2.1	5.6
from forestry and hunting	8.3	10.2	9.7	7.3	7.7	7.9	8.1	6.1	4.9	0.6	5.7
from non farming activities	8.9	10.0	11.2	10.9	14.7	9.8	12.2	13.2	25.1	29.0	17.7
Other sources	32.2	28.6	28.3	29.7	26.4	31.5	28.8	29.3	36.8	63.9	39.1
from wages	19.0	15.0	15.9	14.5	14.3	18.5	17.6	16.5	21.0	30.2	20.5
from remittances	1.2	1.9	1.6	2.8	1.3	2.4	1.6	1.7	2.3	2.6	2.1
other (rents, dividends, etc)	12.1	11.7	10.9	12.4	10.8	10.6	9.6	11.1	13.4	31.0	16.5
Total	100	100	100	100	100	100	100	100	100	100	100

The table shows also that by decile of per capita adult equivalent consumption, a similar pattern to that of consumption emerges: the composition of income by sources is similar for deciles 1 to 8, and only for the ninth and tenth deciles the shares are different. The latter show a higher importance of non-farming activities in self employment sources of income (above 25 per cent), a higher importance of wage income, and, for the tenth decile, a higher proportion of income coming from other sources as rental income, interest received and imputed value of houses, as a reflect of a higher capital stock ownership.

Chart 4.4 shows the relative importance of wages as a source of income in different regions of the country: for deciles 1 to 8, wages are about 40 per cent of total income in Phnom Penh and about 20 per cent in the rest of the country.

Chart 4.5 shows the distribution of employment of the household head by industry and gender. 75 per cent of Cambodian household heads have ‘agriculture, hunting, and forestry’ as their main occupation, with a slightly higher proportion for female headed households (78 per cent). Employment in the public, education and health sectors is next in importance as a source of main income (9 per cent average for the whole country, 11 per cent for males head of households and 3.5 per cent for females) followed by employment in the wholesale and retail trade sectors (7.3 per cent average for the whole country). The ‘textile’ industry as well as the ‘hotel and restaurant’ industry is the main source of income for only 0.8 of Cambodia’s head of households.