Income and consumption distribution trends in the Philippines, 1950-1970

INCOME AND CONSUMPTION DISTRIBUTION TRENDS IN THE PHILIPPINES, 1950-1970. by A. Berry Series: Studies in Employment and Rural Development No.29 Division: Employment and Rural Development Department: Development Economics Development Policy Staff International Bank for Reconstruction and Development The views expressed represent those of the author alone and not necessarily those of the Bank. This report may not be published nor may it be quoted as representing the views of the Bank and its affiliated organizations. Washington,. D. C., September 1975 INCOME AND CONSUMPTION DISTRIBUTION TRENDS IN THE PHILIPPINES, 1950-1 970 Table of Contents Page No. A. EVIDENCE ON DISTRIBUTIONAL TRENDS: THE FIES 1 Demographic Aspects of Family Income 9 and Consumption Distribution Trends Age of Household Head anc Income 9 Family Size and Family Income 10 Price Differences 16 B. WAGE RATES AND OTHER MICRO-TYPE EVIDENCE ON 18 DISTRIBUTION TRENDS C. CONCLUSIONS 28 List of Tables Table No. Title Page No. 1 Comparison of Levels of Personal Income, 2 Personal Direct Taxes, and Personal Savings: CY 1956, 1961 & 1965 2 Indicators of Income Distribution: Total, rural 4 and urban, 1956, 1961, 1965 and 1970/71 3 Consumption Distribution, 1956-1970/71 6 4 Consumption Shares by Income Class, 1956, 8 1961; 1965, 1970/71 5 Poverty and Age of Household Head All Families 11 6 Family Income and Income Per Capita, by Family 12 Size, 1971 7 A Comparison of Distribution of Income by Families 13 and by Persons, 1970/71 8 A Partial Profile of Family Characteristics by 15 Family Income Class: Rural, 1970/71 9 Agricultural Real Daily Wage (Prices of 1965) 20 10 Index of Real Wages: Salaried Employees and Wage 21 Earners in Selected Non-Agricultural Industries in the Philippines, by Industry Division, 1952-1972 (1965=100) 11 Evidence on the Wage Share in Manufacturing 22 12 Evidence on Aggregate Income Consumption Trends, 214 1965 to 1970/71 National Accounts and FIES 13 Family Income and Consumption Trends, 1965-1970/71 25 by Occupation of Household Head (Fies Statistics) 14 Evidence on Income Trends by Groups, 1965 to 26 1970/71: FIES Family Income Data and Selected Series from Other Sources Table No. Title Page No. A-1 Family Consumption Pattern: Comparison 29 Time A-2 Consumption Shares by Income Categories, 30 Manila A-3 Distribution of Families by Age of Household head 31 and Family Income, 1971, Rural A-4 Number of Families, Median Size of Family, Average 32 and Median Number of Children Under 18 Years of Age Living with the Family, Median Number of Family Members Employed by Income Class 1957, 1961, 1965, and 1971 A-5 Family Size, Income, and Consumption: Rural and 33 Urban Areas Compared 1970/71 Income and Consumtion Distribution Trends in the Philippines, 1950-1970 Like moat devzloping countries, the Philippines has relatively uequal distributions of income and of expenditures as far as can be ascertained from available data. According to 1971 information (and converting at the 1971 exchange rate), it appears that about 45% of its population is found in families with incomes of less than US$100 per capita. As relevant as the absolute levels of income for the population and its various subgroups are the trends for the relatively less well off groups. A major source of worry with respect to the development of the Philippine economy in the latter part of the 1960s and the early seventies was the declining real wages recorded for many occupations, despite increases in per capita income for the population as a whole. These data suggest a worsening of the income distribution in an already poor country, leading to a bleak prospect for many Filipinos at present and in the near future. This essay reviews same'of the relevant information on income distribution and tries to draw tentative conclusions about trends over the last fifteen or twenty years.2/ Basic sources of information are the Family Income and Expenditure Surveys (FIRS) of the Bureau of the Census, the national accounts, and data on wage trends. Since the FIES are the richest and most complete source of information, but are at the same time subject to familiar biases related to the difficulty of getting complete reporting, it is important first to check consistency between this and other sources. A. Evidence on Distributional Trends: The FIES As Table 1 indicates, there is a substantial discrepancy between the .FIES and the national accounts.3/ A substantial shortfall between personal income as reported in the FIES and that calculated in the national accounts occurs for all four of the FIES, and ranges around 30% in each case. 1/ Assuming underreporting of income of about 25% in the 1970/71 Family Income and E=penditure Surveys. (See below). 2/ The ILO mission concluded that in certain important respects, income dis- tribution worsened between 1956 and 1971, based on the FIES surveys. While the GINI coefficient rose only marginally and the index of quintile inequal- ity fell, the share of the lowest 20% fell from 4.5 to 3.7% and from 7.0 to 4.4% in the rural areas, a really serious decline. At the same time the top 5% and 10% lost somewhat according to the ILO figures. (See below). 3/ They are interdependent in the sense that some of the assumptions about consumption behavior used in the national accounts calculations came from the FIES; but the estimates of income per capita from the two sources are essentially independent. l 3 The personal consumption recorded in the FIBS, on the other hand, is much closer to the national accounts estimates in recent years,, being onlY 13-14% lower in 1970/71 whereas it had been around 35% lower in 1956/57. Since the difference between personal income and personal consumption is personal savings, the two sources are particularly divergent with respect to their estimation of that variable. Without a more in depth analysis, one. cannot judge how these discrepancies affect the usefulness of the FIES surveys in tracing out income distribution trends. It is often argued that national accounts tend to underestimate the variables they seek to measure; it seems a reasonably safe presumption that they do not overestimate the true values. ' But the change in level of accuracy over ti-me is much harder to guess. Tentatively one might conclude that the FIES studies have had increasing accuracy and coverage of the expenditures they report ever the period in question, but perhaps not with respect to income. If, say, 65% of consumption was reported in 1956/57 and 87,cQ" in 1970/71., it is important to know for which income cate- gories the completeness of reporting increased over the period; conclusions with respect to 'trends in con.-mTtion distribution could be very sensitive to this. And with presumed underreporting of at least .10% for personal income, it would be hard to sa7 much about income distribution trends without some evidence on how relative underreporting by groups changed over the years.l/ With these caveats in mind, we turn to the FIES themselves, and what they:- suggest with respect to income distribution trends over time. For all families together, the four surveys indicate virtually no change in the Gini coefficient for family income and a decrease in the Index of Quintile Inequality, occurring between 1965 and 1971; over thatperiod, the recorded share of the top 5% defined substantially; for most ot-kier groups no dramatic change-swere noted. (See Table 2.) But the lowest and second lowest quintiles of rural families lost substantially (and continously for the lowest one) between 1956 and 1970/71 ; the sharp relative decline for the poorest quintile -- from 7.0% to 4.4% of rural income 7- indicates a decline in absolute terms as well; it was the source of a decline in the recorded income share of the bottom 20% (rural and urban together) from 4.5 to 3.7% over this 15 year period.2/ 1/ Meanwhile it would 'be desirable to dig deeper into the possible under- estimation of consumption in the FIES by comparing its figures at the product level with independent (national accounts) estimates of output, adjusted for imports and exports. In interpreting the 1965-1970/71 movements of real variables, especially of consumption but also of income, it is important to note that the last sample, sometimes referred to as that of 1971, corresponds in fact to the period May 1970-April 1971. The estimates of food, alcoholic beve- rages and tobacco were, however, based on the week previous to the sample visit, i.e. to May 1971. It was true also of earlier samples that these expenditures were blown up from those recorded in the week before the 'visit. In the 1965 FIRS, for example, the visit occurred in May 1 66. This creates problems which become particularly severe when there is inflation. If the appropriate dating point for the expenditures was January 1971, then the Consumer Price Index in Manila would 4.1% less than in 1971 overall, and in regions outside Manila, it would be 8% less, and overall about 7% less. . .. ...... . Table 2 Indicators of Income Distributions Total, rural and urban, 1956, 1961, 1965 and 1970/71 1956 1961 1965 1970/1 Total Rural Urban Total Rural Urban Total Rural Urban Total Rural Urban Families Grouped by Family Income Lowest 20 percent 4.5 7.0 4.5 4.2 5.9 3.8 3.5 5.0 3.8 3.7 4.4 4.6 Second 20 percent 8.1 11.1 8.0 7.9 11.8 7.5 8.0 9.5 8.0 8.2 8.9 9.4 Third 20 percent 12.4 14.7 12.2 12.1 13.5 12.5 12.8 15.3 12.0 13.2 13.9 13.4 Fourth 20 percent 19.8 21.1 20.0 19.3 21.9 19.5 20.2 23.0 18.7 21.0 21.8 21.9 Top 20 percent 55.1 46.1 55.3 56.4 46.9 57.1 55.4 47.2 57.5 53.9 51.0 5D.7 Top 10 percent 39.4 30.1 39.6 41.0 31.1 40.9 4D.0 30.0 41.7 36.9 34.4 33.4 Top 5 percent 27.7 . . 29.0 . . 28.7 . . 24.3 22.6 22.6 Index of quintile inequality 0.44 0.34 0.44 0.46 0.36 0.46 0.45 0.38 0.47 0.40 0.1 0.41 Gini coefficient 0.48 0.38 0.49 0.90 0.40 0.52 0.51 0.42 0.53 0.49 0.46 0.45 Mean family income (current peace) 1 471 989 2 427 1 804 1 203 2 970 2 541 1 755 4 405 3 736 2 818 5 867 Index, current price 100 100 100 123 123 123 173 178 182 254 285 242 Index, constant peice 100 100 100 109.0 110.8 111.3 122.5 128.6 133.0 127.7 147.6 125.3 Mean urban family income/ mean rural family income 2.45 2.47 2.51 2.08 Mean per capita income (current pesos) 258.5 31k.1 432.5 607.8 Index, current price 100 100 100 120.3 167.3 293.6 Index, constant price 100 100 100 107.0 118.7 125.4 . not available Mission calculations. Source of basic data: Bureau of the Census and Statistics Fanily income and expenditures surveys. Note that the definition of rural and urban has been changed several times over the period under discussion. (See FIES, 1971 p. xiii.) The change seems to have been particularly important between 1961 and 1965, judging by the fact that a higher share of families were classified asp ur in thlatter year (7)%) than in the former (66%). The definitional changes between 1965 and 1971 worked in the same direction but less strongly. Source: ILO, Sharing in Development, 1974, p.10, with slight adjustment to the 1971 figures to take account of the distribution calculated by the BCS and presented in FIES. 1971, p.xxi. The data for 1956/57, 1961 and 1965 agree with the breakdown of total income from FIES, 1965, page xxvii. The figures for income per capita ;ere calculated by the author. Deflation is by the consumer price index for the Philippines. (See Central Bank of the Philippines, Statistical Bulletin.) For March 1, 1956/February 27, 1957, the Manila index was used since the national one was not available. The increase was calculated as only 1.50 percent over that period.) Indices for the otherperiods were: 1961: 112.5; 1965t 141.0; May 1970/April 1971: 198.84. In deflating rural and urban incomes, we used the consumer price indices of "outside Manila" and "Manila" respectively. Comparisons among income and expenditure surveys taken at different points of time are fraught with a variety of dangers. Evidence of differential underreporting, as in the present case, obviously raises doubts and calls for as many cross-checks as possible. One important variable in this regard is consumption expenditures. It is also of much interest in its own right independently of its value as a check on income statistics; it is often considered to be a better proxy for perma- nent income than is current income.l/ Data on consumption from these same FIES run counter to some of the more pessimistic implications of the income data. Table 3 presents trends in the consumption distribution by families. That the top 5% has lost is again indicated.2/ The recorded share of the next 1% also fell slightly from 29.5 to 27.8%. The figures suggest an increase in the con- sumption share of the bottom 20%, from 5.0 in 1956/57 to 5.9% in 1970/71. Such improvement as is registered over 1956/57 to 1970/71 occurred entirely over 1956/57-1961, however, with the total breakdown (rural and urban together) changing very little since then, although there was some worsening of the rural distribution and some improvement of the urban dis- tribution. This fact raises the possibility that the improvement recorded for the period as a whole may not be a real one since, as indicated earlier, the 1956/57 survey seems to have achieved substantially less complete report- ing of consumption than the succeeding ones, assuming the national accounts figures are reasonably accurate.3/ Further there is a strong suggestion that food expenditures were partIcularly underreported, which underreporting if it were equally or more severe for the lower income (or consumption) groups would lead to an overestimate of consumption inequality. With improved reporting, this bias would disappear. The 1956-1961 trends are 1/ And lifetime consumption is probably a better welfare indicator than lifetime income. 2/ It may well be that top incomes and bottom ones are the least accurately reported, in the former case because of fear of disclosure and in the latter more because of complexity, importance of home consumption, etc. Any conclusion with respect to changes in the income share of, say, the top 5% would have to be verified by other types of information before warranting high confidence. The data on expenditure are presumably substantially more reliable, although desire to conceal income probably implies in some cases a desire to conceal certain types of expenditure too. 3/ Table 1 shows a considerable increase in the ratio of the FIES consumption estimates to that of the national accounts in both 1956/7-1961 and 1961- 1965, so for the observed 1956/7-1961 trend to be particularly biased the relative underreporting by expenditure classes would have had to change more than over 1961-1965. Table 3 /a Consumptionp Distribution, 1956 - 1970/71 Families Grouped by 1956/57 1961 1965 1970/71 Family Consumption Total Rural Urban Total Rural Urban Total Rural Urban Total Rural Urban Lowest 20% 5.04 5.98 7.52 5.08 5.65 6.79 5.24 5.92 6.92 6.2 2nd 20% 9.03 10.32 12.65 9.78 10.25 12.23 9.67 10.18 12.08 10.52 3rd 20% 13.02 14.68 16.93 14.56 14.57 16.87 14.41 14.76 13.66 15.68 4th 20% 20.03 21.03 22.97 22.36 21.10 23.20 21.37 21.98 22.45 24.59 Top 20% 52.88 47.98 39.94 48.22 48.43 40.91 49.31 47.16 43.00 45.23 Top 10% 35.42 31.66 24.30 31.09 32.49 24.29 32.73 30.81 26.71 26.79 Top 5% 23.38 21.00 15.29 18.73 21.43 15.06 18.80 19.37 16.74 16.50 Index of Quintile Inequality Gini Coefficient Mean family Con- sumption: 1285 866 2116 1793 1332 2688 2877 2142 4621 4479 3474 6810 Index: Current Prices 100 100 100 139.5 150.3 127.0 223.9 247.3 218.4 348.6 401.2 321.8 Index: Constant Prices 100 100 100 123.7 135.4 114.9 158.5 178.5 159.6 175.2 207.8 166.6 Mean Urban Family Consumption Mean Rural Family Consumption 2.44 2.02 2.16 1.52 * Note the non-constancy of the definitions of rural and urban areas over 1956-1971; see footnote to Table 7. /a Including taxes paid and gifts/contributions to others. Sources and methodology - Figures are based on data on expenditure by expenditure class in the four FIES surveys. Making these estimates involved guessing the "intra-expenditure class distribution of families by expenditure levels" leading to some degree of error in estImates. (The surveys do not indicate whether such interpolations were undertaken in calculations of the income distribution.) Fstimating the expenditures of the top expenditure class is particularly difficult; for most of the middle expenditure classes, the mid-point was taken as the mean expenditure for the class. For the higher ones, a lower average was assumed, to make the total expenditures consistent with the reported total. The need for this sort of interpolation implies tha that the table is less precise than the income distribution table. Any important biases are likely to be common to all of the years, however. consistent with this interpretation, though of course others are also possible.1/ A measure of the non-coincidence of the income and consumption distribution is provided by the shares of total consumption by income categories (Table 4). Families by particularly low (high) incomes have a much higher (lower) share of consumption than they do of income. The lowest quintile of families by income had 3.7% of income but 5.9% of con- sumption. In Manila, the top 10% of families by income had 35% of the income but only 24% of consumption. (See Table A-2.) In the rural areas the consumption share of low (high) income people has risen (fallen) moderately; (recall that the income shares have moved in the opposite direction). In urban areas the same trend is more marked; overall the increase of the bottom quintile is considerable -- 5.1 to 8.6% -- although a good share of it was registered during 1956-1961. The decline of the top 5% of all families was dramatic, from 24.9 to 15.7%, and more or less continous. In summary, consideration of the trends in the family distribution of consumption, either by income classes or by consumption classes, throws serious doubts on any pessimistic conclusions which might be drawn from the recorded trends in family distribution of income; if anything, some improvie- ment seems to be suggested by those data.2/ But given the probable degree of 1/ The national accounts (together with our population figures of Table A-4) indicate an increase of 9.9% in consumption per capita between 1956 and 1961, and a higher increase in income per capita (14.1%). The FIES shows an increase in per capita income of 5% and in per capita consumption of 21.4%. The latter figure is implausible. Further, the dramatic increase between these two years in the recorded share of consumption going to food (see Table A-1) is totally inconsistent with a rapid increase in total consumption per capita. If in fact the food share in 1956 were 54% instead of the reported 46.1% and all other items were reported with the same accuracy in the two yearsf this would imply a relative under- reporting of total consumption vis-a-vis 1961 of about 17%, in which case the implied change in consumption per capita would be roughly con- sistent with that implicit in the national accounts. By 1965 and 1970/71 the FIBS data on total consumption were.relatively close to that of the national accounts. Income was still far below the national accounts estimates (Table 1). In agriculture, it is relevant to note the 1970/71 FIES recorded an income generated of 72.6% of (our national accounts based estimate of) wage compensationplus entrepreneurial income in agri- culture; in non-agriculture the corresponding ratio was 69.8; when the FIES consumption figure is related to the same denominator, the ratio is about 90%. (In non-agriculture, however, it is only 82%. This is partly because the savings rate may well be higher in non-agriculture.) 2/ Many data and conceptual problems would have to be resolved befure any strong presumption in either direction can be created. Distribution may well have been worsening, but the existing evidence does not indicate it. It is conceivable that the reported data tend increasingly to underesti- mate the high incomes relative to the low ones especially when real appre- ciation of assets is taken into account (not included in national accounts income figures nor in most household surveys including those taken in the Philippines). Table 4 Consumption 'Shares by Income Class, 1956, 1961, 1965, 1970/71 Families Grouped 1956 1961 1965 1970/71 by Family Income Total Rural Urban Total Rural Urban Total Rural Urban Total Rural Urban Bottom 20% 5.95 8.03 5.18 7.63 10.71 6.81 8.98 11.22 7.88 8.56 10.57 8.62 1 00 2nd 20% 9.49 12.44 8.84 11.75 14.31 10.76 12.18 14.64 11.14 12.29 13.86 12.27 1 3rd 20% 13.29 16.42 13.30 15.26 17.00 14.93 15.08 18.00 14.74 15.84 17.28 15.94 4th 20% 20.07 22.07 19.98 20.43 22.32 20.08 20.63 22.00 18.41 21.10 21.55 22.66 Top 20% 51.18 41.01 52.68 44.90 35.66 47.42 43.13 34.14 47.83 42.19 36.71 40.49 Top 10% 36.07 25.46 36.19 35.27 21.95 30.92 29.19 20.09 30.39 26.54 22.49 23.76 /a Top 5% 24.91 15.37 18.27 20.30 13.58 19,26 19.50 11.64 419 15.69 13.38 14.62 //a Rough estimate since whole of top 5% fell in the open-ended top income category. Source: The FIES Studies for the four years in question. underreporting of consumption in the early years and its apparent decline over time, recorded trends in consumption distribution may not be better reflections of trends in, say, permanent income than are the recorded income distribution trends. At least the aggregate level of income under- reporting seems not to have changed much. The recorded consumption distri- bution trends can be substantially affected by whether consumption reporting improved more for families near the bottom or near the top of the distribution. 1/ Demographic Aspects of Family Income and Consumption Distribution Trends- A possible interpretation of the different movements of family in- come and consumption distribution trends is that low income families (e.g. young or old ones) are increasingly able to consume well above their income levels when same are very low. This development may be associated with in- creasing capacity of young families to borrow and of older (e.g. retired) ones to live on past earnings or on gifts, and in such a case would probably be a favorable omen. Its applicability in the Philippines is consistent with the increasing real average income trends revealed by the national accounts and the FIES, though as discussed later, trends in consumption patterns and in certain wage rates are not obviously consistent with the calculated income trends. Other major problems with the income distribution data as they emerge from the FIES relate to (a) the lumping together of persons and families of all different ages, (b) the fact that families vary in size and composition, and (c) the inability to take account of differences in the price vectors applicable to different families, Before asking how changes in the demographic and price components may have affected trends in income distribution, it is helpful to review those components briefly. Age of Household Head and Income Kuzu.tets has drawn attention to the importance of life cycle factors in the observed family income distribution. Average household income varies substantially by age of household head>rising by over 5O% between the categories 25-3 and 55-64. In the U.S., Israel and Taiwan, the relative income between highest and lowest 10 year category within the range 25-64 was about 30%, 12% and 25% respectively. The greater variation in the Philippines may be associated with a more strongly positive association between family size and age of head than characterizes the other countries, especially the U.S. and Israel, but this cross is not presented in FIES, 1971. 1/ A useful discussion of how demographic variables affect the observed family distribution of income, including a discussion of the 1971 FIES, is Simon Kuznets, "Demographic Components in Size-Distributions,of In- come," Yale Economic Growth Center Discussion Paper No. 226, April 1975. - 10 - Though most of the family income variance associated with age of household head relates to the age groups lying in the range 25-64, a considerable share of poor families are at the two age extremes.1/ Table 5 shows that in 1970/71 whereas families with household head under-25 years (over 65) made up only 5.0 (7.8) percent of all families, they composed 8.2 (12.8) percent of families with recorded income of less than a thousand pesos per year. These two categories accounted for 25.1% of all families with less than 500 pesos per year and 19.1%of thosewith 500-999 pesos while making up only 12.8% of all families.2/- Family Size and Family Income The positive correlation between family income and family size is very important. When family income is the measure employed, the larger families appear higher in the distribution than they would were per capita income of the family the classifying variable. Whereas families of ten or more persons had median income four times that of one person families in 1971 (2,911 to 1.,095) the per capita income of the latter was about three times higher (1,095 to 360). The pattern for rural areas is comparable.3/ (See Table 6.) To get some feel for the effect of this relation on the estimated equality of income distribution, and beginning with the informa- tion on family income by family size, one can estimate a distribution of income by persons by assuming that each family had an equal intra-family distribution of income among members.h/ As Table 7 shows, this distribution 1/ The share of families with head<25 or>65 is around 20% in the U.S. and Israel but below 13% in the Philippines (Kuznetq, oE.Cit.3 p. 30). 2/ Kuznets (op.cit., p. 30) notes that the relative family income of fami- lies with household head 65 and over is not low in the Philippines (0.95) as it was in the other countries in his sample (U.S., Israel and Taiwan). Accordingly he concluded that the two extrne-age classes (<25,7-65) "contribute little to the disparity associated with the age of head variable" (p. 30). Their exclusion had little effect either on the Gini coefficient or on Kuznetst "Total Disparities Measure" (p. 39). As Table 5 shows, the group with household head 65 years and up does include a quite disproportionate share of low income households; this is consistent with Kuznet's observation that their relative income is not low, since the income variance for this set of families is far above that of all families taken together. (See FIES 1971, p. 128). It would be of great interest to know whether much of this high income variance is asso- ciated with high variance of family size. 3/ Part of the lower income per capita in larger families is irrelevant since the child/adult ratio is higher. But even after allowance for this, a negative association does exist. (See below.) 4/ Note that the concept here is not the same as the commonly used distribu- tion of income among members of the labor force (usually referred to as the personal distribution of income). - 11 - Table 5 /a Poverty and Age of Household Head All Families _Household Head Household Head less than 25 65 or over Share of all families 5.03 7.80 Share of families with recorded incoaes"_ <1000 pesos 8.16 12.80 Share of families with recorded incomesL2 <2000 pesos 7.03 9.37 Rural Families Share of all families 5.50 7.80 Share of families with recorded incomes Z1000 pesos 8.51 12.07 Share of families with recorded incomes £ 2000 pesos 7.37 8.94 /a For more detailed statistics on this question in rural areas see Table A-3. /b These families constitute 17.3% of all families. Ic These families constitute 41.2% of all families. Source: FIES 1971 pp. 12. .. . .. . . Table 6 a/ Family Income and Income Per Capita, by Family Size, 1971 Family Size 1 _ 3 4 5 6 7 8 9 :?1110 Rural and Urban: Median Family IncoMe 1,095 1,588 1,912 2,264 2,378 2,526 2,751 2,797 3,280 3,911 Median Family Income/ Family Members a/ 1,095 794 637 566 476 421 393 350 364 350 Rural: Median Family Income 835 1,296 1,548 1,822 1,873 1,913 2,228 2,341 2,572 3,331 Median Family [ncome/ Family Members a/ 835 648 516 455 375 319 318 293 286 300 a/ Whereas income per capita was the variable we sought to use here, it could only be estimated indirectly and approximately, so the variables median family income and "median family income/family members" are presented. For all families, mean income is 3736 and median income 2454. Source: FIES, 1971, pp. 3-6. Table 7 A Comparison of Distribution of Income by Families and by Persons, 1970/71 Family Income Per Capita Income Lowest Quintile 3.7 3.9 2nd Quintile 8.2 8.5 3rd Quintile 13.2 13.2 hth Quintile 21.0 21.1 Top Quintile 53.9 53.4 Top 10% 36.9 36.7 Top 5% 24.3 24.5 Source: The distribution of family income is the same as presented in Table 2. That of per capita income is calculated from FIES 1971, by using the classification of famly income by family size to con- vert the figures on family income to per capita income by family incame-family size cells and regrouping according to per capita income. For families of a given size and income category, some positive correlation was assumed between size and family income. is very similar to that of family income; the positive relationship between family income and number of persons apparently just offsets the negative relationship between per person income and family size.3/ For families with 7 or more persons, family income tends to be two or more times higher than that of families with 1 or 2 persons.4/ Further details on the relationship between family income and family size in rural areas, along with same evidence on how family coposition varies with income, is shown in Table 8. The median and mean family size are 60% or more higher in the second highest income category than in the bottom one, and the number of working members is about twice as high; the number of adult equivalents, however, is only two thirds higher. 5 1/ Note furtherc that the methodology employed presumably underestimates the skewness of per capita income due to the assumption of perfect equality within the family. On the other hand, as mentioned earlier, treating adults and children in the same way tends to overstate real inequality. 2/ Kuznet (op.cit., p. 99) notes that "the TDMs and Gini coefficients remain large for the Philippines, even when we shift from income disparities among households (classified by income per household) to disparities among persons (similarly classified). If we had shifted the whole distri- bution to a per person basis (as was done for the other three countries), total disparity in the distribution of income among persons, by per person income per household, would still have been quite high. For the country as a whole, the TEK would have to be significantly above 61.8, and the Gini coefficient significantly above 0.430--wider disparity than in any of the three countries covered above. The rather distinctively wide in- equality in the conventional size-distribution of incomes among households in the Philippines would probably remain after conversion to a per person basis." 3/ There is no simple general rule for predicting the difference between the two distributions compared here. If per capita income is equal or almost so, grouping people into families of various sizes clearly increases the measured skewness. But when the negative relation between per capita income and family size becomes strong enough it dominates and the effect to reduce measured skewness. 4/ See TIM, 1965, and FIES, 1971, pp. 3-5. / Comparative data on family size and other demographic characteristics are presented for the various years of the FIES, in Table A-4. Table A-5 presents information on median and (estimated) mean income and consump- tion by size of family in 1970/71. Income per person and per adult equi- valent is a-sharply declining function of family size in both rural and urban areas. Table 4 A PARTIAL PROFILE OF FAMILY CHARACTERISTICS BY FAMILY INCOME CLASS: RURAL, 1970/71 Approximate Family members Share of families Share of families Share of share of other than Number Working During Annual family with female with male head and families families husband, wife Survey Week income Family Members head with wife, With wife with husband Children and children (excludes Unpaid Family Income per Income per Adult Income per Consumption Consumption Adult b present present under 18 under 1 Helpers) Working Member Equivalent person per adult per person Equivalents Mean Me dian Present Working Memo Nedian mean Median (1) (2) (3) (4) (5) (6) (7) (B) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (500 4.45 4.2 .174 .733 .117 .907 .826 2.4 2.1 .317 1.04 1.0 328 105 77 494 361 3.25 500-999 4.95 4.7 .124 .840 .132 .954 .876 2.7 2.4 .410 1.19 1.1 635 210 153 552 401 3.60 1000-1499 5.26 5.1 .118 ,883 .150 1.00 .882 2.9 2.6 .478 1.25 1.1 998 328 237 632 457 3.81 1500-1999 5.57 5.4 091 .880 .150 .971 .909 3.1 2.9 .590 1.30 1.1 1,336 432 312 702 506 4.02 2000-2499 5,63 5.3 .064 .906 .151 .970 .936 3.2 3.0 .524 1.34 1.2 1,669 555 397 801 $73 4.03 2500-2999 5.98 5.9 .075 .920 .201 .995 .925 3.5 3.2 .560 1.45 1.2 1,888 663 458 871 602 4.13 3000-3999 6.15 6.1 .066 .909 .170 .975 .934 3.5 3.2 .741 1.52 1.3 2,263 782 559 892 638 4.40 4000-4999 6.39 6.4 .072 .906 .218 .978 .928 3.5 3.2 .984 1.67 1.4 2,662 958 696 1,023 743 4.64 5000-5999 6,66 6.9 .082 .905 .283 .987 .918 3.6 3.3 1.155 1.84 1.6 2,973 1,126 821 1,130 824 4.86 6000-7999 6.76 6.6 .088 '912 .383 1.00 .912 3.5 3.3 1.348 2.01 1.9 3,414 1,370 1,015 1,299 962 5.01 8000-999 6.60 6,6 .107 .857 .372 .964 .893 3.4 3.1 1.343 1.92 1.8 4,647 1.820 1,351 1,653 1,226 4.90 10,000-14,999 6.97 7.1 .081 .892 .424 .973 .919 3.8 3.5 1,276 2.17 2.0 5.419 2,320 . 1.686 1,794 1,304 5.07 15,000-19,999 7.20 7.1 .074 .842 .313 .916 .926 3.6 3.2 1.758 1.82 1.7 9,348 3,138 2,359 2,370 1,782 5.40 20,000 6.41 6.4 .124 .898 .441 1.00 .876 3.6 3.2 .934 1.94 1.8 14,778 6,212 4,463 2,761 1,984 4.61 Total 5.66 5.5 .092 .878 .162 .970 .908 3.1 2.8 1.41 1.2 1,998 . 686 498 845 614 4.11 a) Assuming a child under 18 - .5 adult, b) Excluding unpaid family workers. atp EstLimsted/.098 for the country as a whote in the National Demographic Survey of 1973 (See Population Institute, University of the Philippines, Weightal Marginals on Records 1 through 8. National Temoraphic Survey of 1973, 1ec.2/p.4) d) Estimated at .866 for the country as a whole in ibid, Estimated at .964 for the country as a whole in ibid, Sources sod meuhodology. Based on PIES, 1971. Column 2 com from page 5 and column I is based on figures presen ted there, Column 3 e based on the figures of p.127. Columns 4 and 5 are calculated on the basis of information on page 126. Columens 6 and 7 are based on the assumption that whet. the wife is the head of the hos.hold, no husband is present. If this assumption is Incorrect, the figures of column (7) are Undretim ted and those of Column (6) are overestimated. Sot judging by the presentation In the National Demographic Survey, that presumption is correct, Coloss 9 comes from page 139 and Column 8 is based on data shoWn there. Columna10 is Column I mins Colussn (7) minu Clum (). Column 12 ces from page 125 and Column 11 is calculated on the basis of the ame dsta. Columns 13 and 14 u.e Incom data from page 2 and the estimate of adult equivalent shown in Colusn 18, which Is based n the assumption that children under 18 cons.ume oehalf a much a adults. Columns 15, 16 and 17 are also based on or directly from data Don pages 1 or 2. - 16 - Conceptually, the best single indicator of whether distribution is improving or worsening over time would probably be the trend in the distribution of lifetime consumption by persons, appropriately discounted. The above rather confusing pieces of information do not permit any conclu- sions with respect to trends in this variable. Table 3, in the absence of other evidence, suggests constancy in the distribution of consumption among families over 1961-71; the recorded improvement between 1956 and 1961 must be viewed with some caution due to the apparently particularly high underreporting of consumption in the former year. Meanwhile, the variance of family size as between high income and low income families appears to have decreased over time, thus possibly creating a false impression of improving distribution over tim or cancelling out what would otherwise have appeared as a worsening.-!/ As between 1956 and 1971, the relative family size of high income families compared to low income families decreased; in 1956-57 the top quintile of families by income had 1.55 times as many members as the bottom quintile, while in 1970/71 the ratio was only 1.38. 2/ This suggests that income (consumption) distribution may have been moving less favorably than Table 2 (Table 3). The disproportionate representation of families with either young ( 25 or Old ( 65) family heads among the poor seems to have increased at least between 1965 and 1970/71, the only two years for which this breakdown is available. Whereas in 1965 they represented only about 14% of the lowest quintile, in 1970/71 this share was about 20% (Table 5). This factor may have tended to hide some degree of improvement in the distribution of income and consumption. Unfortunately, with only two points of time observed, this conclusion is quite tentative. Frice Differences The 'share of thepopulation found in urban areas has increased substantially over the pe iod in question. I/ Prices of many key items in I/ As Table 7 indicated, family distribution and per capita distribution were quite similar in 1970/71. If the distribution of per capita income had been the same in 1965, and the relationship between per capita income and size of family the same, then family income distribution would have been more skewed, as was in fact the case. 2/ As indicated in Table A-h, mean size of family is virtually constant at 5.8, although the median size is lower in 1956 than in each of the follow- ing surveys. 3/ It is interesting than between 1956 and 1961, when the recorded family consumption distribution improved, the family size/income relationship did not alter. When it did begin to alter, not much improvement in the consumption distribution was observed. , This whole argument is tenuous for several reasons, so cannot be stressed too much. First, there is no general theoretical presumption that the existence of a range of family sizes would imply a family income or consumption distribution either more equal or less equal than for persons. The negative association between family size and income per capita must exceed a certain positive level for the former to be true, since family size variance per se creates a tendency to inequality. Our evidence indicated that the net (Continued...) 17 the family budget (especially food) are substantially higher in the urban areas, perhaps as much as 25 or 30% for Manila; 5/ but since many items, including public goods and services are cheaper in the urbai areas, a detailed study would be required before conclusions could be reached as to relative piece.$/ Tentatively, the FIXS data would seem to suggest that distribution did not change mnch over the 15 year period, and that any movement was more likely in a positive than a negative direction. Both the income and consump- tion distributions show a declining share for the top groups. 'While the (Continued frm previous page) effect in 1970/71 was about zero. We then presumed, in the above argument, that the relationship between per capita income and family size was the same for previous years. This assumption could be invalid, in which case the effect alleged in the text would not exist. / Although this does not show up in the FIES figures, according to which that share actually fell from 33.5% offa-milies surveyed in the March 1957 sample, to 30.1% of those surveyed in the 1971 sample. This was due to changing definitions of "rural" and "urban" over the years; with a fixed definition (and depending on just what it was) the share would increase by perhaps 4.5% over this period. (See K.C. Zachariah, "Migration in the Philippines with Particular Reference to Less Deve- lped Regions," mimeo, 1975, Table 4.) Changes in definition are described in the 1971 FIES (p. xiii). 5/ A recent study has suggested that the cost of a subsistence concumption bundle is about 0O % as high in rural areas as in Manila. (See Develop- ment Academy of the Philippines, Measuring the quality of Life: Philippine Social Indicators, Development Academy of the Philippines, 1975, p.11. 6/ There is the possibility that the rural price vector is preferred by rural people given their preferences, and the urban one by urban, people. - 18 - bottom quintileys share of recorded income fell, its share of consumption rose. However, biases resulting from the use of aggregate family income and consumption data seem more likely to have helped generate an artificial improvement, than the opposite, so overall judgement must be quite tentative. A best guess would be that as of 1970/71, the long-run consumption stream of the best off five percent would involve per adult equivalent con- sumption about eight times that of the bottom 20% 1/ The trend in this variable cannot be assumed necessarily to follow that of single year con- sumption distribution since changing age structure, family size, debt capa- city and other factors could alter the relationship between the two. More intensive analysis will be required to judge its movement. B. Wage Rates and Other Micro-Type Evidence on Distribution Trends In an economy like that of the Philippines, when wage rates corresponding to low income occupations are rising substantially less rapidly than income per member of the labor force or income per capita, it seems a reasonable first presumption that distribution is worsening. Account must, however, be taken of the fact that when the system is under- going substantial structural change with increase in the share of labor force found in higher skill sectors, the typical wage rate will rise less j/ We assume that the current consumption distribution is the more relevant base from which to judge trends in lifetime consumption. Table 3 indicates that for consumption expenditure in year 1970/71, the ratio was 13.1 (9.7 in rural areas and 10.6 in urban areas). For income the ratio was 26.3 (20.5 in rural areas and 19.7 in urban areas). But the top five percent of income earners only consumed 7.3 times as much per family as the bottom quintile (5.1 rural and 6.8 in urban areas). The 13.1 ratio cited above is probably the more relevant one for judging the relationship between long-run consump- tion per adult equivalent of the well off and the poor. As between the top 5% and the bottom 20% of families by 1970/71 income, the number of adult equivalents was 46% higher for the latter. If anything, the ratio would probably be higher for the consumption distribution, say 1.5. Thus 1970/71 consumption per adult equivalent would be 8. times higher for the top 5 percent of families ranked by 1970/71 consumption tends both to have a higher variance across families than longer period consumption and to have somewhat different memberships from year to year, the comparable ratio (to the previous one) would be lower when longer run consumption is compared, probably falling in the range 7.5-8-.. All the above assae implictly that conaumption underreporting was proportionally the same for low and high conwuming families. In 1970/71 consumption reported in the FIES study was only about 13-14% below the figure arrived at in the national accounts. The latter may of course be underestimated as well. High underreporting might occur at either end of the welfare scale; assuming 10% more consumption underreporting for poor (rich) families leaves us with the conclusion that the ratio in question fell in the range 6._-8 2.*. Pending further evidence, about 8.0 is a plausible best estimate. (Continuedv..) - 19 - rapidly than will the average wage for all occupations .1/ The other major factor to be borne in mind is the distribution of income between labor and capital. The 1960s in the Philippines was a period of dramatically unsatis- factory wage movements, according to most of the recorded statistics. The evidence on declining real wages for farm workers, industrial workers, and some other categories (Tables 9 and 10) is matched by a decreasing labor share in industries reporting to th6 Annual Sarvey of Manufacturers, both the larger scale firms of 20 workers and up and the small ones.(Table 11). For the former, the estimated property share rose from 69.7 in 1956 to 78.8 in 1971; the rise occurred over 1956-1963, a period characterized by decreasing real wages (there were further real wage decreases in subse- quent periods). For small manfacturers of 5 to 19 workers the property share rose from 48.2 to 67.8% in these years, again the increase ended by 1964.2/ The wage index of skilled workers in industrial establishments in Manila and suburbs fell from 123 to 106 ower 1956-63 and that for unskilled workers fell from 113.4 to 102.8 It then remained fairly constant over the rest of the 1960s for latter group but fell again sharply to 91.3 for the skilled workers in 1971; both fell further in the rapid inflation of the succeeding years. (Continued from previous page) Table 3 indicates that for consuaption expenditure in year 1970/71, the ratio was 13.1 (9.7 in rural areas and 10.6 in urban areas). For income the ratio was 26.3 (20.5 in rural areas and 19.7 in urban areas). But the top five percent of income earners only consumed 7.3 times as much per family as the bottom quintile (5.1 rural and 6.8 in urban areas). / It must also be remembered that if workers per family or workers per dependent are rising (falling), these wage rates seed not rise as fast (need to rise faster) as income per capita for the relative income of persons dependent on said occupational categories to remain constant. 2/ Lecnardo Sta. Romana III, "j Stuly of Trends in Entrepreneurial and Property income in the Philippines (Tables)n (Tables, III-b and III--). Particularly characterized by increases in the property share were wood, wooden furniture, printing, rubber (here the sudden jump was from 1956 to 1957), and machinery. Possibly these samples were not equally complete in the various years, although this is not hinted at by the total value added series. Possibly the process involved a decline in imputed income, although it does not appear plausible that this was a major factor since even in 1956 a considerable majority of the employees reported were engaged in the large manufacturing firms, aq suggested by the fact that they accounted for 84% of the wage bill of a reporting firms. - 20 - Table 9 Agricultural Real Daily Wage (Prices of 1965) Years IL/a Author Hicke-McNicol1Lo T) (T) (3) 1950 3.32 1951 3.16 1952 3.69 1953 4.19 1954 4.42 1955 4.45 1956 3.64 3.85 1957 3.84 3.63 3.87 1958 3.80 3.59 3.84 1959 . 3.85 3.54 3.77 1960 3.68 3.36 3.69 1961 3.49 3.26 3.63 1962 3.41 3.25 3.55 1963 3.43 2.86 3.61 1964 3.03 2.78 1965 2.93 1966 2.98 1967 3.09 1968 3.04 1969 2.75 1970 2.4h 2.56 1971 2.25 2.44 1972 2.17 2.58 /a Estimated by the ILO team in preparation of Sharing in Development: A Programme of Enployment, Iaity and Growth for the Philippines, ILO, Geneva, 1973, using data fran BAECON Wage Surveys, or made available directly by the Bureau of Agricultural Econamics (BAECON). Deflation was by the consumer price index for areas outside Manila. /b Estimated by the author on the basis of data presented in Bureau of Agricultural Economics, Department.of Agriculture and Natural Resources, Farm Wages 1956/57- 1964/65; 1966/67-1967/68 and Urban Wages 158/59-196T/ Bureau of Printing, 1973, Manila. /c Hicks and McNicoll, op.cit. Table 10 Real Wage Rates (1965 Prices)"" Index of Real Wages Salaried Employees and Wage Earners in Selected Non-ARricultural Industries 1949-1972 in Industrial Estab- in the Philipines by Snt Di v i n 1 - lishments'in Manila and Suburbs All Workers 1ining & Quarrying Manufacturing Electricity, etc. Conerce Trans. and Com Skilled Unskilled Salaried Wage Salaried Wage Salaried Wage Salaried Wage Salaried Wage Salaried Wage Year Laborers Laborers Employees Earners Employees Earners Employees E Employees Earners Employees Earners Kn9joyees Earners 1949 10.51 6.76 1950 10 19 5.74 1951 8.82 5 72 1952 9.58 6.54 87.5 89.8 73.9 70.1 96.2 89.8 111.6 91.4 88.2 87.3 79.2 107.9 1953 10.14 6 97 93.8 95.2 81.6 78.9 100.0 98.6 122.0 92.8 89.4 94.5 87.1 109.5 1954 10.35 6.99 99.0 101.6 89.5 83.1 98.4 105.0 125.1 103.7 97.9 100.1 94.1 115.0 1955 10.45 7.27 103.5 105.6 97.6 88.7 102.6 108.7 126.4 104.7 101.1 105.4 97.4 117.6 1956 10.21 7.19 102.0 104.2 88.3 91.1 105.0 102.3 123.1 95.0 104.7 112.0 99.0 121.2 1957 10,00 6.99 102.4 106.0 88.0 92.4 103.3 105.6 122.8 104.9 104.0 110.0 103.2 116.8 1958 10.02 6.80 104.4 103.5 93.8 93.8 102.6 107.6 120.5 101.0 104.6 101.6 110.0 114.3 1959 10.27 6.92 110.2 108.9 103.0 101.4 105.7 106.8 127.5 104.4 108.3 111.1 114.4 121.2 1960 9.85 6.64 111.9 110.7 112.5 103.7 103.5 104.5 123.8 105.9 110.2 122.2 111.2 116.3 1961 9.67 6.70 112.6 109.3 114.6 107.3 103.9 106.4 123.6 101.4 111.0 118.1 - 109.9 114.1 1962 9.25 6.52 109.8 104.9 111.9 103.3 102.5 105.5 119.9 96.9 107.9 111.2 106.6 108.9 1963 9.02 6.52 107.4 100.3 106.1 102.2 103.4 104.3 119.3 93.3 105.9 100.8 104.8 103.8 1964 8.48 6.07 101.4 96.8 98.6 101.0 100.1 99.7 111.8 91.5 101.0 97.6 98.9 96.8 1965 8.51 6.34 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 1966 8.47 6.45 99.7 104.7 93.4 102.8 99.7 102.6 116.1 107.4 98.7 102.8 - 96.8 107.4 1967 8.35 6.36 95.4 104.2 82.5 110.9 99.7 101.6 119.4 106.9 93.1 101.7 95.0 100.0 1968 8.81 5.92 97.9 102.4 85.1 111.0 L100.6 100.9 126.2 103.4 93.6 97.1 95.9 100.8 1969 9.10 7.10 100.5 104.6 89.2 115.3 104.3 102.9 125.8 109.5 96.2 95.0 98.2 100.3 1970 8.45 6.88 91.9 98.3 77.8 103.2 99.4 102.0 112.1 102.4 93.3 91.4 89.0 92.5 1971 7.77 6.42 86.3 92.7 74.5 101.6 95.4 101.5 98.8 92.2 88.3 87.0 83.6 83.8 1972 83.8 92.8 70.8 99.6 92.5 103.1 96.8 91.1 87.3 87.0 81.7 87.9 T 2he source cited in (b) does not indicate whether fringe benefits are included or not, but the presumption is that they are not (as with the other series). / Includes only money wages, overtime pay and pay for sick leave, holidays and vacationaj exoledes =uch privileges am food ration, lodging, recreation, etc. (See Statistical Bulletin, December 1973, p. 383.) I Source: Elizabeth Ong, "An Analysis of Non-Agricultural Real Wages in the Philippines, " miseo, 1973. Deflation is by the Consumer Price Index in anila. Original sources (listed in Central Bank of the Philippines, Statistical Bulletin where the wage series appear regularly) are described as "private firms for the Manila industrial establishment series and n787 cooperating establishments for the other series (see Statistical Bulletin, December 1973, p. 383). Table EVIDENCE ON THE WAGE SHARE IN MANUFACTURIN G All Reporting Large Small Establishments Establishments Establishments 1(2) (3 1956 32.? 30.3 51.8 30.9 29.6 44.0 27.9 26.6 41.3 I 27.5 26.2 41.6 1960 26.7 25.9 36.9 24.9 24.1 41.9 22.7 22.0 38.0 24.0 23.5 32.7 1965 26.6 26.2 33.7 27.0 26.5 34.5 a 24.0 23.2 36.5 23.6 23.0 35.7 1970 20.4 19.7 34.3 1971 21.7 21.2 32.2 a) Establishments with 20 workers or more. b) Average total emnployment of establishment with _-19 workers. Source, Methodologys Cols. (1) - (3) are from Romana, op. cit., Table II-A., III-B and III-C. It comes originally from BCS, .Annual Survey of Manufacturers. A small labor cost category, *extra benefits" was excluded.- FI - 23 - Note that our agricultural real wage Feries tends to parallel those for unskilled urban workers; there was a substantial increase up to some time in the mid-1950s, after which the general trend has been down. But the decline is more marked in the case of the agricultural real wage and no levelling off occurred in the late sixties as appears for the Manila unskilled manufacturing laborers.or the wage earners in most of the sectors included in Table 10. (Commerce is an exception; there the decline for wage workers was of the same magnitude as that indicated by our preferred agricultural wage series - column 2 of Table 9.) Meanwhile the wage series indicate that salaried workers, who also gained during the 1950s, have suffered general and in some cases very substantial declines since then. The decline in most of the real wage series raises the question of whether they can be consistent with the evidence from the family income and expenditure survey data to the effect that incomes of wage earning families as well as of all families were rising during the 1960s. The consistency or lack thereof between national accounts and FIES data which generally indicate increasing income and consumption levels per family and per capita (Table 12), and the declining wage series calls for more detailed probing; only for the last inter-FIES period, 1965 to 1970/71 does the disaggregation of FIES data by occupation of household head permit some rough comparisons with the wage series. It is worthnoting first that changing oogqpgational structure des account for pa:rt of the apparent discrepauy: for all families the increase in real income per family.according to FIES, was 4.2%. (See Table 13). If the family income trends for each( category over 1965-1970/71 distinguished in Table 13 were applied to their shares of all families in 1963, the calculated income increase is -1.0%. In this sense structural change was associated with an about 5% increase in income per -'amily. Meanwhile (see Table 14) there is a broad agreement between FIES annual family income trends a@RqindePendent_wage series by occupations indicating that: 1/ (a) white collar workers and professionals suffered substantial real income decrease; the wage series of Table 10 suggests a fall of 10.9%; the FIES statistics a fall of 1h.4% (when the three subgroups distinguished are appropriately weighted),-See Table 1. 1/ The two types of data wemld not be expected to show identical trends since: (a) FIES data are annual, the wage series monthly (or in agriculture, daily) earnings; (b) FIES data are family income, wage series refer to individual workers; (c) wage series exclude fringe benefits; FIES earnings reflect all in- come (in principle at least); and (d) wage series are based on'a sample of 787 corporations (as of 1973); these would tend to be of above average size; FIES reflects all earners. Table 12 Evidence on Aggregate Income Consumption Trends, 1965 to 1970/71 National Accounts and FIES 1970/71 'urrent Prices % Incrase, 1965 Prices 1965 1965-1970/71 A. Aggregate Trends 1 Personal Income Per Capita National Accounts: Estimate A 592 921 652 10.2 Estimate B 656 889 630 -4.0 FIES 433 647 459 6.0 2. Consumption Per Capita National Accounts 556 814 577 3.8 FIES 490 776 50 12.3 3. Income Per Family, FIES 2541 3736 2648 4.2 b. Consumption Per Family, FIES 2877 4479 317h 10.3 Sources and Methodogy: Estimates A and B are based on income data and consump- tion, tax and savings data, respectively. Population figures used are estimates of T. Ruprecht,made for the ILO mission. In estimating per capita values from the FIES data, we assumed average family sizes of 5.87 in 1965 and 5.77 in 1970/71, based on the data of Table A-4. Table 13 Family Income and Consumption Trenda. 196g-1 /70{1, by Occupation of Household Head (F(ES Statibticl) 1965 1970/71 Income per Percent of Total Consumption Thousands of Income per % Consumption % Thousands of % Family family Income per Family Yamilies .Faily O per Faxily Chare Fanilies o Professional, Technical ad 8,507 8.4 8,528 128 < 6,082 ( -28.5 < 6,931 <-18.7 228+ 78.1+ Related ..((8,581) ((9,780) Excluding 1rofessionals and 10,081 5.9 9,648 76 7,281 -27.8 7,788 19.3 132 73.7 Technicians (10,273) (10,989) Administrative, Executive and 11,682 6.4 10,675 71 7,101 -39.2 7,898 -26.o 125 76.0 anagerial (11,141 Proprietors, Wholesale, amd 4,885 5.2 4,819 138 4,145 -15.2 4,561 - 5.4 246 78.2 Retail (5,848) (6,435) Clerical Workers 4,694 6.7 5,202 185 - 5,038 A 7.3 t, 6,255 _ 20.2 201+ 8.6+ -(7,109) ,,(8,826) Sales Workers 3,443 5.3 3,844 199 3,444 0 4,233 10.1 223 12.0 (4,859) (5,972) Farmers, farm laborers, 1,651 36.2 2,030 2855 1,736 5.1 2,211 8.9 316 10.7 et. al. (2,449) (3,120) Workers in Mines 3,277 0.35 3,077 14 3,066 -6.4 3,377 9.8 15 7.1 (4,326) (4,765) a Transport Workers 2,561 4.21 3,166 214 2,755 7.6 3,252 2.7 314 46.7 (3,888) (4,589) Craftamen, Factory Operatras 2,530 11.8 2,888 608 2,562 1.3 3,226 11.7 753 23.8 et l.. (3,615) (4,552) Other Manual Workers 2,007 1.6 2,601 104 2,045 1.9 2,680 3.1 1514 8.1 (2,886) (3,782) Service Workers 2,868 5.1 3,399 225 3,123 8.9 3,833 12.8 293 30.2 (4,407) (5, 408) Unaployment without work 3,085 9.0 3,128 379 3,314 10.8 3,369 7.7 582 53.6 experiance,or not in-labor (4,719) (4,754) force All Groups 2,541 100.0 2,877 5126 2,648 4.2 3,174 10.3 6347 23.8 (3,736) (4,479) Al Included mail carriers and messengers 1970/71; it is not clear whether they were included in 1965 or not. Engineers, dask officers, ato., were included under the heading ' Workers in Transport anid Comunications Occupations in 1970/71 but in 1965 the were probably under professionals, etc. and clerical. Since the way they were allocated in 1965 is unknown we have left them unallocated in 1970/71,, and have indicated the way in which mle numbels in the two cited categories are biased. - 26 - Table 14 Evidence on Income Trends by Groups, 1965 to 1970/71: FIES Family Income Data and Selected Wage Series from Other Sources 1965 1970/71 Current 1965 Percent incre8 lyU7T7Vr-t1 Prices Prices Income Consumption Occupational Groups 1. Agricultural Labor (J)1' Daily wage (Table 9, col. 2) -12.2 (i) Annual Lncome, Families whose Household head is agricultural laborer 1283 2248 1593 24.2 11.3 2. Manufacturing Labor (i) Manufacturing Wage Earners, Yonthly Earnings a 1.7b (ii) Anual Income, Families whose obsehold head is craftsman, factory operative, etc. 2530 3615 2562 1.3 11.7 3. Transport Workers a (i) Wage Earners, Monthly Earnings -12.0 (ij) Annual Income, Families whose Household head is transport worker 2561 4271 3027 7.6 2.7 4. Commerce a (J) Wage Earners, Monthly Earnings -10.8 (ii) Annual Income, Families whose Household head is sales worker 3443 4859 3444 0 10.1 5. White Collar (i) Salaries Employees, Monthly Earningsa -10.9 (ii) Annual Income, Families whose Household head is: a) professional, technical etc. 8507 6082 6931 -28.5 -18.7 b) administrative, executive and management 11682 10020 7101 -39.2 -26.0 c) clerical 4694 7109 5038 7.3 20,2 All three - all families 7257 8372 5933 -18.2 All three - weighted 6769 8176 5794 -14.4 by selective numbers of wage workers in the 1971 BCS Labor * Force Surveyb a) Data from Table 10. b) The ratios of wage earners/total employed was 937, 47.5 and 98.1 for groups (a), (b), and (c) respectively. Absolute numbers in May 1971 were 547, 77, and 455 thousands. Here the percentages are applied to the 1965 and 1970/71 totals of the respective FIES. - 27 - (b) blue collar workers in manufacturing registered neither gains nor losea.1/ Large discrepancies exist in the cases of transport workers, agricultural workers and sales workers, with the wage series indicating a decline in each case and the FIES data constancy or an increase. Part of the discrepancies probably relate to changes in the inclusions of these categories between the 1965 and 1970/71 FIES. Many other factors could also be involved.2/, / Taking the various statistics at their face value, the improvement of income distribution registered between 1965 and 1970/71 is a result of relatively greater wage increases (smaller decreases) for persons and fami- lies at lower wage levels; this is suggested both by the wage series and the 1/ The wage series of Table 10 do not include fringe benefits. A test to see whether.this biases the series downward over time suggested that it does not. In 1959 the ratio of finge benefits to basic payroll was 3.03 and in 1969 it was 4.54. (See Bureau of the Census and Statistics, Annual Survey of Manfactures. 1959 and 1969, respectively.) 2/ In respect of the agricultural workers, FIES presents data on families whose main source of income is agricultural wages as well as on families whose household head is an agricultural laborer. In 1970/71 the latter category included 267 thousand families while the former was much larger, 680 thousand. Taking each figure literally, the indication is that there are many families whose household head is not classified as an agricultural laborer but for which that activity is the main source of income; many families have a small plot -- if there is any ambignity as to how to describe himself the individual might be expected to prefer the designa- tion "farmer" to "agricultural laborer" -- but may work also as a laborer, along with.other family members. The tendency to underestimate income in kind and to think of income as "cash income" may even contribute to an overestimate of the number of families whose main source of income is agricultural wages. In 1970/71 the 287 thousand families whQse head was an agricultural laborer had average recorded income of 2248, while the 680 thousand for whom agricultural wages was the main income source averaged 2600 pesos. Differences in family size, age structure, etc., between the two groups are not known. 3/ Note that the BAECON data give daily wages. The BCS household surveys indicate no large changes in hours worked per week for agricultural laborers; unfortunately no data seem to be available to permit a judgment as to whether weeks worked per year per family might have increased or not. - 28 - FIES data. Further, there has been an increase in the relative income of farmers vis-a-vis proprietors in other sectors; families whose household heads were agricultural operator.- had an average increase of 3.2% while proprietors in wholesale and retail commerce, for example, registered a decline of 15.2%. While one might well question the validity of such sharp declines in average income as are indicated in Table 14 for families whose household heads are professionals or administrative executive-wamagers 1/ (and conclude that underang of high incomes was particularly severe in tKe 1970/71 FIES), it must be noted that various independent pieces of evidence point in the same direction. Further, the very rapid advance of middle and higher educa- tion in the Philippines, coupled with the relatively slow growth of output characterizing this period would be expected to bring about.a decrease in the relative in.come of the better educated. C. Conclusions Distribution of income and consumption by families appears not to have altered much over the period 1956-1970/71 in the Philippines. FIBS studies suggest if anything some improvement, while trends in certain wage series have been negative and tend to suggest the opposite.2/ A partial reconciliation has been attempted above for the period 1965-1970/71, with the conclusion that the declining wage rates do not constitute strong evidence that distribution did not improve during that period (or that the incomes of poor families did not rise). Structural change has been rather rapid, with an increasing share afthe labor force in the higher paying occupations. Only with better knowledge of the relative underreporting of different income and consumption groups and of the strengths and weakness of the cited wage series can a really satisfactory picture be painted. 1/ Note that, with these groups expanding very rapidly (presumably faster than before), an increasing share of persons in them were probably young and low on the income ladders characterizing their occupational cate- gories; some of the average income decline is presumably due to this factor, and is therefore somewhat illusory. But this would not likely explain the full declines. 2/ Especially the agricultural wage series. Since many industrial workers have above average income the decline in their wages would not be suggestive of increasing inequality. - 29 - Table A-1 Family Consumption Pattern: Comparison Over Time Percent of Total (Figures in Parenthesis are the share of total expenditures made on cash purchases of the item) Consumption Going to: 1956 1961 1965 1970/71 Food: Total 46.1 (19.7) 53.8 (39.8) 53.7 (43.3) 53.7 (44.2) Urban 40.1 (16.4) 48.4 (42.5) 46.1 (43.1) 47.1 (44.5) Rural 53.6 (24.5) 59.5 (36.9) 60.6 (43.5) 59.3 (43.9) Manila 36.2 42.3 (40.8) 40.4 (38,8) 41.5 (40.3) Cereals: Total 20.2 (13.0) 21.1 (15.3) 19.6 (14.6) Urban 15.3 (12.5) 14.6 (13.2) 13.9 (12.9) Rural 25.3 (13.6) 27.1 (17.3) 24.5 (16.0) Manila 10.8 (10.2) 10.0 ( 9.6) 11.1 (10.7) Implicit Food Expenditures for Total Population 2,344.0 M 4,270.2 X 7,917.7 M 15,270.8 M Source: The FIES for the four years in question. - 30 - Table A-2 Consuption Shares by Income Categories, Manila, Share of Share of Income Consumption First Quintile 5.54 8.73 2nd Quintile 9.04 12.21 3rd Quintile 13.67 16.61 4th Quintile 21.08 23.15 5th Quintile 50.67 39.30 Top 10% 35.04 23.71 Top 5% 24.39 14.16 Source: I 1971, p. 1. , -- _ , ��,, . ; - эi - таЪ1а � -1 D1вtr1Ъ¢цт of Tar113n+ Ът lувевl ЯО�иеУлlд Нвад апгl Fri1T Iмювв, 1971, Suca1 . Income CLee _ 25 25-Э4 Э5-44 45-54 55-64 б5+ (Peeos рег 7еаг) Himber то аг MAbkг аг вг (tЬоивапде){of а11 fam111ee (thoueandв) (ot а11 1шn12iee at (tlиuaa�a} (о[ ац 1ап111ва at (thоива¢дв) (of а11 t�lцвв at (thouвande) (ot а11 ta111sn at (thousanda) (nf ац faa111aa at �,� У�� а at tltiп iтсаив 1ew1) thlв lдсоте ]вvвl) thla lтсоте 1вvв1) th1a lксоsе 1вче1) thlв lтсова 1в�е1) thle lаоовв 1ave1) 5сю 2э.42 е.о2 ь9.аг 23.91 б7.з1 zз.о5 4].I,o v,.]8 4].65 ц,.гб 4т.т5 16.э5 zse 5оо-999 61.о0 В.71 ]85.47 2Ь.50 171.В9 2b.56 1го.5г 17.22 9о.В6 12.98 71.9Т 1о.28 10оо-1499 $1.оо 7.бб 195.29 29.32 162.27 г4.з6 1]5.о0 1Т.27 95.91 14.40 46.Т1 7,о1 � цоо-]99э з2.и 5.z4 176.74 ге.74 ]7з.n9 28.w ]15.9z ]е.в5 . ВО.7т 1з.�з зб.бе 5.9Ь 6]5 2ооо-2499 22.94 5.13 1Эо.92 29.г9 121.40 27.]6 8Э.7г 1В.7з 58.о5 12.99 зо.1о 6.73 1+47 2500-2999 14.15 4.0Э 79.64 22.69 1оз•37 29.45 87.40 24.9о 42.28 12.о$ 2Э.53 6.7о Э51 зооо-з999 2э.42 4.е, 109.]0 г1.60 ]59.87 31.66 ]15.9z 2z.95 63.1о ]г.5е зз.22 - 6.58 ,о5 i#оо0-Ь999 7.Э2 2.54 5о.19 17.Lз Вб.54 ЭО.оS 72.68 г5.24 48.59 16.87 22,84 7,9з 288 5�-5999 3,So 2.з2 31.ЬL 18.Вз 45.бе 2Т.19 51.52 30,67 26.5о 15-77 9.Оо 5.36 16� 60.1D-7999 1.22 0•64 Э0.55 15.91 5о.48 26.29 61.64 32.1о 39.12 го.38 9.00 4.69 1i'г 8о00-9949 ].95 2.14 ]5.z7 ]ь.78 z6.44 z9.o5 2ь.б8 29.зz ]5.те 17.з4 э.в] 4.19 10,ооа-11�,499 о.49 0.66 8.73 11.80 26.44 35.7Э 17.48 23.62 14.51 19.61 5,В8 7.95 74 15,�-19,999 0.98 5.16 Э.27 1Т.21 2.40 12.63 4.Ь0 24.21 5.68 29.89 2.77 11,.58 19 г0,�+ - - L.ЭЬ 1Ь.77 4.В] 1В.5о b.44 24.77 8.2о 31.5L г.77 10.65 26 - Tota1 243.99 5•50 1,090.99 24.61 1,201.99 27.11 927.92 20.77 6Э1.00 14.2Э Э46,Оз 7.80 4�� i j а) Fros FIS3� 1971: Р. 729 � з -31- Table 1-4 Number of Families, Median Size of Family, kverage and Median Number of Children Under IS Years of Age Living with the Family, Median Number of Family Members Employed by Incoc:e Class 1957. 1961, 1965 and 1971 Number of Children Median Under 18 Years of Number of Po?ulation (Population Number of Age Livin g With ' Family Implicit (Ruprecht FIBS based Fami ties of Median Six. Mean Size Family Members Participation Income Category EstLatep) Ertiatates) (thoupands) Families of Family of F-ily Average Median Employed Race (1) (2) (3) (4) (5) (6) (7) (a) (9) (10) 1 9 5 7 Philippines 25,228 22,524 3,959 100.0 4.9 5.7 a.&. n.a. 2.1 36.8 Under P250 164 4.1 2.7 3.7 1.6 43.2 ilua to P499 727 18,4 3.9 4.7 1.8 38.3 1'500 to P749 721 18.2 4.5 5.3 2.1 39.6 P750 to P999 561 14.2 3.1 5.8 2.3 34.5 P1,000 to F1,499 654 16.5 5.3 6.o 2.2 36.7 P1,500 to P1,999 414 10.5 5.7 6.3 2.1 33.3 P2,000 to P2,999 338 8,5 6.0 6.8 11 .1 2.8 41,2 P3,000 to 34,999 226 5.7 6.3 6,8 2.2 32.4 P 5,000 and over 154 3.9 6.7 7.1 2.3 32.4 1 9 6 1 Philippines 27,903 25,665 4,426 100.0 6.1 5.8 n.a. n.a. u.a. Under P500 751 17.0 4.7 4.4 1 500 to P999 1,296 29.3 5.7" 5.4 P1,000 to 1-1,499 788 17.8 6.4 6.0 P1,500 to F1,999 531 12.0 6.5 6.2 V2,000 to P2,499 298 6.7 7.3 6.s '02,500 to 112,999 181 4.1 7.6 7.2 P3,000 to 103,999 219 5.0 7.2 6.8 0 P4,000 to 1-4,999 105 2.4 7.1 6.9 5 000 to V5,999 79 1.8 73 7.2 6:000 to P7,999 84 1.9 7.8 7.4 98,000 to 09,999 30 0.7 7.2 6.7 VlO.ODO and over 63 1.4 7.8 7.3 1 9 6 5 Phillpoinet 31,417 30,116 5,126 10010 6.2 5.8 3.2 3 5 2.3 39.7 Under 9500 593 11.6 4,8 4.7 2.6 2:7 2.0 42.6 10500 to P999 908 17.7 5.8 5,0 2.9 3 3 2.3 46.0 V1,000 to 01,499 856 16.7 5.7 5.3 3.1 3 3 40.0 P1,500 to V1,999 2.2 693 13.5 6.2 3.8 3.3 3 6 2.2 37.9 02,000 to F2,499 508 9.9 6.6 6.2 3.5 3:8 12.2 35.5 02 , 500 to K , 999 387 7 6 6 7 6.3 3 6 4.o 2.2 34.9 P31000 to P 31999 458 8:9 7:0 6.5 3:4 3 a 2.5 38.5 P4,000 to P4,999 235 4.6 7.3 7.o 3.5 4:0 2.6 37.1 P5,000 to P5,999 143 2.8 7.9 7.3 3.8 3.9 2,6 35.6 P6,000 to P7,999 129 2.5 6.9 6.7 3.2 3.3 2.7 40.3 F8,000 to P9,999 79 1.5 7.5 7.2 3.2 3 7 3.0 46.7 P10,000 and over 136 2.6 7.7 7.4 3.1 3:4 2.9 39.2 1 9 7 1 Philippines 37,789 36,602 6,347 100.0 5.5 5.8 3.1 2.8 1.2a 20.9 a Under P500 329 5.2 4.1 4.4 2.3 2.0 1.0a 22.7a P500 to j,999 767 12.1 4.7 $.a 2.6 2.3 1.1a 22.0a Pl,ooo to P1,499 773 12.2 3.0 5.2 2.9 2.5 1.1a 21.1a P1,500 to F1,999 748 11.8 5.4 5.6 3.1 2.9 1.1a 19.6a F2,000 to 1-2,499 611 9.6 5.2 5.6 3.1 2.8 1.2a 21.4a P2,500 to P2,999 517 8.1 5.6 5.9 3.4 3.0 1,2a 20.3a P3,000 to 93,999 794 12.5 3.8 6.0 3.3 3.0 .,a 20.Oa 1-4,000 to F4,999 475 7.5 6.1 6.3 3.3 3.0 1.4a 22.2a P5,000 to V5,999 316 5.0 6.8 6.8 3.5 3.2 1.6a 23-53 P6,000 to P7,999 03 6.4 6,3 6.6 3.2 3.0 1.7 a 25-B& V8,000 to V9,999 226 3.6 6.4 6.6 3.1 2.9 1.8a 27.3& P10.000 to T14,999 234 3.7 6.5 6.8 3.1 2.9 1.9 a 27.911 P15,000 to 919.999 71 Ij 6.7 6.8 3.3 2.9 1.9a 27.9a 020.000 and over 81 1.3 6,6 6,7 2.9 2.5 .Be 26.9a a) As of the t970/71 PIES survey, the deLinition of "family members employed" van changed from the prior concept of "number of members employed during the survey week", where employment might have been an a full-time or part-time ban is as working employer, own account worker, wage and salary worker or unpaid fam ly worker-- ee ?I S, 1965, page 67) to " embera employed for pay or profit during 'he survey week", not including unpaid family workers in family farm or enterprise. (FIES, 1971, p, L24) Sokirce: Family Income and Expenditures, BCS Survey of Households Bulletin. ... . ...... .. . . ... .... . .. .. .... - 33 - Table A-5 Family Size, Income, and Consumption: Rural and Urban Areas Compared 1970/71 Size of Family 1 2 3 4 5 6 7 8 9 >10 Rural Income: Median 835 1,296 1,548 1,822 1,873 1,913 2,228 2,341 2,572 3,331 Mean a/ 1,204 1,869 2,232 2,627 2,701 2,759 4,890 3,376 3,709 4,803 Consumption: Median 1,043 1,659 2,058 2,391 2,629 2,857 3,128 3,376 3,623 4,541. Mean a/ 1,307 2,079 2,574 2,996 3,294 3,580 3,919 4,230 4,540 5,690 Income Per Person: Median 835 648 516 456 375 319 318 293 286 308 Consumption Per Person: Median 1,043 829 686 598 526 476 447 422 403 =420 Median Consumption Median Income 1.249 1.280 1.329 1.312 1.404 1.493 1.404 1.442 1.409 1.363 Urban Income: Median 2,548 2,792 3,217 3,607 3,929 3,948 4,689 4,634 5,242 5,662 Mean a/ 3,763 4,124 4,752 5,328 5,803 5,831 6,926 6,844 7,742 8,363 Consumption: Median 3,306 3,622 3,769 4,624 5,408 5,212 6,031 5,828 6,915 7,240 Mean a! 4,298 4,704 4,700 6,011 7,030 6,776 7,840 7,576 8,990 9,412 Income Per Person: Median: 2,548 1,396 1,072 902 786 658 670 579 582 524 Consumption Per Person: Median '3,306 1,811 1,256 1,156 1,082 867 862 728 . 768 67Q C/Y 1.297 1,297 1.172 1.281 .1376 1.320 1.286 1.258 1.319 1.279 a/ Direct data are not available for mean income and consumption by family size, so the mean is roughly estimated on the assumption that the miean/median ratios for all families hold for each family size. These ratios are: for rural areas, income 2818/1954 = 1.442; for consumption 3474/2772 - 1.253; for urban areas, income 5867/3972 = 1. 477;Consumption 6810/5250 - 1.300. Probably income (especially) on consumption distribution within a given family size group is less skewed for larger sizes, so our estimates of the means may be too low for small family size and too high for large family size. b/ Assumed to average 10.8 members.