The Development Academy of the Philippines (DAP), in its 1974/75 Social Indicators Project, made estimates of poverty incidence for 1961, 1965 and 1971, which were reference years of the most recent Family Income and Expenditures Surveys (FIES).¹ The DAP poverty line was based on the food menu recommended by the Food and Nutrition Institute, which the FNRI itself costed at 14 per day in December 1973 for a reference family of six. The DAP simply adjusted this “food threshold” for price differences across space and time using whatever price indexes were available. Its “total threshold” equals the food threshold divided by 0.6, on the assumption that the food: non-food distribution at the threshold of poverty would be 60:40. The food and total thresholds are given in Table 7.5.

¹ See A. S. Abrera, “Philippine Poverty Thresholds,” in Mahar Managahas (ed.) Measuring Philippine Development, Development Academy of the Philippines, Manila, 1976.

Table 7.2. Mean One-Day Per Capita Nutrient Intake and Per cent Adequacy, Philippines

Source: Food and Nutrition Research Institute, (FNRI). First Nationwide Nutrition Survey, Philippines 1978, Summary Report. FNRI Publication No. GP-11. Manila, 1980.

Table 7.3. Philippines, Comparison of Mean One-Day per Capita Nutrient Intake, Luzon-Visayas
1975-77 and 1978 FNRI Surveys

Table 7.4 Percentage Distribution of Households by Levels of Adequacy of Energy, Protein and Vitamin A, Philippines

Source: Same as for Table 7.2

The thresholds were then applied to the FIES distributions of both income and expenditures, with the finding that the poverty incidence was rising. The income distributions did not look very plausible, however, because their means were smaller than the means of the expenditure distributions, implying aggregate dissavings for all families. Income distributions were available for 1961, 1965 and 1971, while expenditure distributions were available only for 1965 and 1971. Using the latter, the DAP estimated that overall poverty, between 1965 and 1971, grew from 34 per cent to 41 per cent if the food threshold was used as the criterion, and stayed unchanged at 70 per cent if the total threshold was the criterion. For the rural sector in particular, it grew from 39 per cent to 48 per cent on the food threshold criterion and stayed unchanged at 76 per cent on the total threshold criterion.

Table 7.5 Philippines, Comparison of DAP and World Bank Poverty Lines for a Family of Six
(Pesos per Year)

Sources: Mahar Mangahas, Measuring Philippine Development, DAP, Manila 1976; Idem., “Measuring Poverty and Equity through Perception Variables”, The Philippine Economic Journal, Vol. XVI, No. 4, 1977; and The World Bank, Aspects of Poverty in the Philippines: A Review and Assessment, Vol. II. Report No. 2984-PH, 1980.

In mid-1979, the World Bank sent a Mission to appraise poverty in the Philippines, as similar Missions had earlier done in Thailand, Indonesia and Korea. The Poverty Mission’s procedure to obtain a poverty line was as follows, (a) The October-November 1970 food consumption survey of the National Food and Agriculture Council,¹ gave a per capita food mix of those in the per capita income class of 400 and below, which was then costed at 7977 prices, giving 361.7 on an annual basis. The average food basket contained 1,881 calories per capita on a daily basis. This gave 5.2 calories (= 1,881/361.7) per day per peso of annual food expenditure, (b) The daily energy allowance recommended by FNRI is 2,016 calories per capita, which thus requires an annual food expenditure of (2,016/5.2 =) 387.7 per capita per year. This implies 1.062 per capita per day or 6.37 for a family of six. (c) The proportion of the budget spent on food by ‘low-income’ families,² according to the 1971 Family Income and Expenditure Survey, is 69.3 per cent. Therefore 387.7/0.693 = 559.4 per capita per year is a threshold that allows for non-food expenditures of 30.7 per cent in the total expenditure budget. This threshold is then judgmentally rounded down to 500 per capita per year for 1971; this is the basic poverty line, (d) Judging that the urban poverty line is probably one-third higher than the rural poverty line, and using 1971 population weights, the urban poverty line becomes 605.0 and the rural line becomes 453.8 (working backwards, the implicit urban-rural weights are 31.6 per cent - 68.4 per cent.) (e) Lines for years aside from 1971 were obtained by adjustment with the Consumer Price Index (1957 = 72.0, 1961 = 79.8, 1965 (base year) = 100.0, 1971 = 160.2 and 1975 = 292.1). Adjustment factors for urban and rural areas were the same. For 1975, the urban line is [605.0 x (292.1/160.2) =] 1,103 and the rural line is [453.8 x (292.1/160.2),=] 827, both per capita per year.

¹ Apparently part of the series conducted by the Special Studies Division of the Ministry of Agriculture, and usually called the SSD series, not the NFAC series.

² Definition of ‘low income’ is unstated.

Thus the first essential difference in poverty lines is that the DAP food threshold derives from a basic level of 14 per day, for Metro Manila at the end of 1973, stemming from an FNRI norm, whereas the World Bank food threshold derives in effect from a basic level of 6.37 per day, at 1971 prices, for the country as a whole, stemming from Ministry of Agriculture data on actual consumption by low-income people. The difference cannot be accounted for merely by price changes, since the food component in the CPI only grew by about 50 per cent between mid 1971 and the end of 1973.³ In other words, after allowing for inflation, the World Bank line in 1973 prices would still have been only 9.55 per day, national basis, which further works out, following the World Bank procedure, to 8.64 per day for rural and 11.52 per day for urban areas.

³ From an index (1978 = 100) of 44.1 in 1971 to 55.6 in 1973 and 74.7 in 1974 (1983 Philippine Yearbook). The large jump at the end of 1973 and in 1974 was due to the first oil shock. The computation of 50 per cent is based on a comparison between 44.1 and the simple average of 55.6 and 74.7.

The second difference is the judgmental downward rounding in the World Bank line, which reduced it by a little more than 10 per cent. A third difference is the use of a food budget of 69 per cent compared to DAP’s 60 per cent. A fourth difference is the World Bank assumption that the urban: rural price ratio is 1.33:1 compared to the DAP assumption that the Metro Manila: other urban: rural ratio is 1.43:1.15:1. A final difference is that the World Bank actually used a per capita poverty line applied to a per capita per family distribution which was specially tabulated by the government at its request, whereas the DAP only had the published per family distributions to work with; this difference could lead to another reduction of about 6 percentage points in the poverty incidence rate.

For rural areas, the net result for 1971 (the base year used by both studies) is that the WB:DAP ratio of rural poverty lines is about 35 per cent. Obviously, given any common distribution to which the line is applied, the DAP line will give a higher poverty incidence rate, even allowing for the 6-point correction, as seen in Table 7.6

If the difference in results depends only on the poverty norm applied, who then is to judge as to which is the ‘better’ norm? My view on this is that normative matters are determined not by technicians’ opinion but by social opinion, and that the role of technicians is to carefully observe what the social opinion is. A particular technique for doing this, as well as recent empirical findings, is discussed in the next section.

However, the question of which purchasing power norm to use is less critical than the quality of the basic data on distribution to which the norm is to be applied. When the 1975 FIES data became available, it would have appeared - whichever norm one used - that the poverty incidence rate ‘exploded’ over a scant four years. The reason, according to the review of Mangahas and Barros,¹ was as follows.

The problem was obviously due to the FIES data. In the first place, after adjusting for inflation, the FIES data of 1971 and 1975 would assert that average real incomes of household had fallen, contrary to the growth trends in the National Income Accounts. Secondly, the FIES average income per household in 1975 is less than 6,000; but aggregate personal income (from the National Accounts) divided by the number of households yields an average of over 13,000 for the same year. (p. 90)

¹ See Mangahas and Barros, op. cit.

Table 7.6. Philippines, Comparison of DAP and World Bank Poverty Incidence Rate for 1971 (Percentage of all families)

Pointing out that the ratio of FIES aggregate household income to aggregate personal income had been 64.3 per cent in 1957, 61.2 per cent in 1961, 65.6 in 1965 and 65.5 per cent in 1971, but had somehow plummeted to only 44.0 per cent in 1975 (p. 67), they came to the conclusion that the 1975 survey was too unreliable to use in constructing a poverty trend.

The World Bank Poverty Mission was aware of this issue but was undeterred. Their report states (p. 8)¹:

“... the ratio of family expenditure estimated from the 1975 FIES to the one estimated from the national accounts data is 0.70 compared to 0.94 for 1971 (see Annex 1-C). Thus, the 1975 FIES appears to have underestimated family expenditures more than the 1971 FIES. This underestimation is probably largest at the higher income levels, but must also have occurred at lower income levels. Consequently, if the expenditures were underestimated by 5-10 per cent, as seems probable, the true’ poverty incidence would be 2-4 per cent lower than the tables show.”

¹ World Bank, Aspects of Poverty in the Philippines: A Review and Assessment, Vol. II. Report No. 2984-PH, 1980.

The World Bank’s poverty estimates have been fairly widely publicized, and, for whatever they are worth, are reproduced in Table 7.7. They would allege that, between 1971 and 1975: (a) poverty incidence rose by 9 points on the whole; (b) it rose by 16 points (or by 65 per cent!) in urban areas and by 6 points (or 15 per cent) in rural areas; and (c) in the rural areas of the following key agricultural regions, the poverty incidence rose tremendously: Central Luzon, from 15 per cent in 1971 to 27 per cent in 1975, Southern Tagalog, from 32 per cent to 47 per cent, Western Visayas, from 38 per cent to 49 per cent, and Central Mindanao, from 18 per cent to 28 per cent. In my opinion, such calculations are simply not credible.

Table 7.7. World Bank Estimates of Philippine Poverty Incidence
(percentages)

Sources: Same as for Table 7.5.

Several other income surveys were again conducted by the National Census and Statistics Office in 1978,¹ 1979, 1980 and 1981. All these surveys suffer from the same gross understatement as the 1975 PIES, and will, if taken at face value, imply huge poverty incidence rates. This may explain why no income survey has been published since the 1975 FIES in such standard references as the Philippine Yearbooks.

¹ The 1978 results were available by the time the World Bank Poverty Mission did its work; apparently they were ignored by the Mission.

One conclusion I would draw from this experience is that the present techniques for measuring income can by no means be taken for granted. In survey after survey, total expenditure exceeds total income, and as a result the income figures lose their plausibility. Perhaps some completely new approaches should be tried, e.g., measuring expenditures and savings separately and then adding the two to obtain income, instead of the present system of letting savings be the residual. Whatever the approach, clearly the need at this time is for innovation instead of complacency with the traditional data sets.