|CERES No. 058 (FAO Ceres, 1977, 50 p.)|
by Ingrid Palmer
Aggregate data for 1974 place Indonesia's per caput GNP at $150 compared with neighbouring Malaysia's $600, the Philippines' $310 and India's $130. But it is difficult to believe that Malaysia's average standard of living is four times that of Indonesia's. Moreover, those who have observed living standards in both India and Indonesia believe that poverty is clearly more acute in the former. Nevertheless, nobody has disputed the presence of widespread and severe poverty in the densely populated island of Java ever since the studies undertaken at the turn of the century were discussed in a shocked Dutch Parliament.
Food consumption is a common indicator of the level of poverty in rural Java. The Survey Sosial Ekonomic Nasional (SUSENAS), relating food sufficiency to income sufficiency, took 240 kg of rice equivalent per person per year as the "poverty line." Penny and Singarimbun point out that this amount is regarded by the peasants as cukup (enough), and state that this measure is traditionally seen as 120 kg of rice plus a surplus of another 120 kg for sale. For a family of five, then, cukup means an annual income equivalent of 1 200 kg of rice. Using the SUSENAS income poverty line of 240 kg of rice plus maize per person per year, the data on per caput food production shown in Table I provide an idea of average trends in poverty in Java over a forty-year period.
The 1969-70 SUSENAS, which provided data on production and consumption of food staples by province (Table 2), illustrated some interesting points. The better-off island of Sumatra is included for comparison. First, the geographical variation is very great. Second, in better-off west Java and Sumatra, consumption is greater than production whereas poorer central and east Java production is greater than consumption. There would appear to be an export of food from the poorest areas. Third, there is a marked seasonal variation in consumption
But what does cutup mean in terms of farm size? Let us suppose that a household of five persons requires 1 200 kg of rice for "sufficiency."
If the land yielded as much as 3.5 tons of (milled) rice per hectare a year, only 0.34 ha per household is needed (assuming this is net of input costs). Since land quality is so variable in Java, this one figure is not very helpful.
The literature of the pre-green revolution era includes suggestions of between 0.5 and 0.7 ha as sufficient. But one writer has claimed that 0.25 ha of first-class sawah under high-yielding rice varieties will support a family of five.
The data in Table 2 are only averages. To appreciate the possible range of individual consumption levels, it is necessary to know something of the social relations of production, which are dominated by differentiated access to land.
According to population data, with a farming land total of 10.1 million ha, land averaged 0.16 ha per rural resident in 1975. Thus, an egalitarian distribution of land would lead to operating household plots averaging well under I ha. Clearly, if farming has remained profitable, it must be because there has been resistance to fragmentation beyond a certain level. This is borne out by data A 1903 poverty survey described 45 percent of farms as having less than 0.5 ha, whereas the 1963 Agricultural Census specified only a slightly higher figure: 52 percent. Moreover, over the same period the proportion of farms greater than I ha merely changed from 21 to 22 percent. Yet, over that sixty-year period, Java's population rose from 29 to 66 million. Some more land has been brought under cultivation and the decline of the sugar plantations has also placed more land under staple foods. But the last significant extensions of cultivated land were made prior to 1920. On the evidence, then, between 1903 and 1961, something between 24 and 28 million rural residents (80 percent of a 30 to 35 million population increase) may have been added to the category of landless households.
In 1970, of 9.4 million farming households, the bottom 20 percent averaged less than 0.1 ha, while the bottom 60 percent averaged 0.2 ha. Thus, in 1970, an estimated 1.9 million households operated farms too small to be included in the Census, and can be regarded as virtually landless. Unfortunately, the figure for the absolutely landless in 1970 is not available, but according to the 1963 Agricultural Census they amounted to 21 percent of all rural households. Combining this Census and the 1970 Survey, we can say that in 1970 at least 41 percent of rural households had no land or less than 0.1 ha (and that at least 81 percent had no land or less than 0.2 ha). According to the 1971 Population Census, then, about 31.2 million rural Javanese belong to virtually landless households.
Since 1967, there have been rice intensification programmes, mainly concentrated on Java, which have contributed to a substantial increase in output. There has also been a general economic rehabilitation in the country after the decline of the mid-1960s. Agregate data would suggest that the standard of living of the Javanese has risen and that economic development is at last affecting the poor. But these changes have been accompanied by changes in the set of social relations governing production, and a distributional analysis is required to understand which groups of people have really benefited.
The notable data contribution to distributional analysis is in the form of household and per caput weekly expenditure and consumption data obtained from the several SUSENAS surveys, which were undertaken periodically from 1963/64 onward. Here the surveys for 1963/64, 1964/65 and 1969/70 are used. The annual rate of inflation was high during the first two surveys, 122 and 339 percent respectively, compared with only 15 percent for 1969/70. With such high and varying rates of inflation, any evaluation of trends from income, expenditure or consumption data has to be circumspect as lagged effects of inflation are likely. There are also problems of price, as well as income, effects on the composition of essential goods purchased. Seasonal influences are bound to be present too.
The results of the three SUSENAS surveys are shown in Table 3. The effect of inflation is clearly seen. The first and third periods recorded show a more egalitarian distribution of per caput expenditure. However, whereas 1969/70 shows less inequality at the top end of the distribution, there is greater relative poverty at the low expenditure end than in 1963/64. Since 1969/70 marked a period of substantial price stabilization but was too soon to reflect structural changes apart from those induced by the first few years of the rice intensification programmes, the only conclusion that can be drawn is that the price stabilization of the later 1960s redressed much of inflation's regressive effects on income distribution.
The percentile expenditures were measured from a cumulative frequency curve using the original survey data of frequencies and class intervals. These data are ratios. For example, the figure for P9O is the ratio of the income of the top 11th percentile (individual's expenditure) to the median income. Thus Pi = (100 x Yi)/ Y50
The percentile distribution of per caput weekly consumption of rice (Table 4) demonstrates vividly the impact of the rapid acceleration of inflation in 1964/65, although it must be said that rice imports were reduced to negligible amounts during 1965. The decline of rice consumption of the first and second decile groups was dramatic, although due to interpolation difficulties with only five class intervals for 1963/64, that period's consumption figures might be too high. Actual per caput rice consumption in 1969/70 had not achieved the 1963/64 level for any decile group; and as a proportion of 1963/64 consumption the quantities of the lowest expenditure classes declined most of all.
1. The survey data were given in the form of average consumption of the food item for each per caput expenditure class. The frequencies used pertained to each expenditure class. Cumulative frequency curves were drawn on these class averages. Whatever the disadvantages of this method, they should not affect the relative positioning of the three curves and therefore the relative readings from them, except perhaps for the first and last decile readings, which are placed in parentheses.
2. For 1969/60, the frequencies from which percentages were calculated were of numbers of persons in each per caput expenditure class. For 1963/64 and 1964/65 the frequencies from which percentages were calculated were of numbers of households in each per caput expenditure class. If household size remains constant over all expenditure classes, there would be no difference.
A major cause of this deterioration was undoubtedly changes in
the system of government rice distribution. In 1963/64, the poorest could rely
on this cheaper rice for almost 10 percent of their rice consumption, but in
1969/70 for only 2 percent. Moreover, whereas in 1963/64 the per
quantity of government distribution rice was 8.3 times greater in the highest than in the lowest expenditure class, in 1969/70 it was 34.8 times greater, indicating a shift of government rice subsidies toward the better-off.
Deeply affected by inflation
The period 1963/64 was part of the declining period of the late President Soekarno's administration. How would the better earlier years have compared with 1969/70? Unfortunately, we do not have comparable expenditure data for the period 1958 to 1961. However, P.R. Deuster compared real income changes of different categories of rural residents between 1959 and 1968 and showed how inflation affected categories of rural residents differently. He developed separate cost-of-living indices based on the question, "For a man on such an income how much must his (money) income have risen to be on the same standard of living?" Applying these indices, he came to the estimates for quartile real income shown in Table 5.
Each set of quartiles was obtained for each category of rural resident separately. Thus, each row in Table 5 refers to a separate "population." From the listing of the socio-occupational categories, it can be assumed that average income rises as one moves down the table. Thus, it can be concluded that for the first four (farming) groups, the lower income categories were more deeply affected by inflation, as their ratios to 1959 real income were lower. Moreover, by moving horizontally across the table from the right to the left, it is seen that within each socio-occupational category, the originally worse off (the first quartile) suffered greater proportional falls in real income than their better-off peers (the third quartile). Thus the effects of inflation can be read in two ways: by moving vertically and horizontally across the table.
Inroads into employment
The varying influence of inflation by size of farmer is marked. Small farmers at each of the three points in their own distribution were worse off, while medium and large farmers were better off in the later period. The farm labouring category underwent a process of stratification with median and (upper) third quartile labourers doing a little better in 1968, but with the poorer farm labourers being much worse off than in 1959. The recent changes in rice cultivation are believed to have brought about a further stratification within the landless labouring class. But from Deuster's data, it would appear that this had been occurring prior to the technological and commercial innovations of the green revolution. This supports a now widespread contention that increasing concentration of land and changes in tenurial arrangements had been occurring long before the green revolution, which was barely under way by 1968.
The promotion of higher yields has been profitable for those who had adequate access to credit and whose family's labour mobility was not restricted by the burden of debt labour.
Because exchange of land title is not easy in Indonesia, the land sales that have occurred do not fully reflect the trend toward concentrating control over land in fewer hands. The Agro-Economic Survey provides some evidence to show that there has been a shift from the landless renting-in land to the larger landowners adding to their farms by renting-in more land.
But if the better-off farmers can spearhead an agricultural revolution that increases rice output and labour inputs per hectare, does not this raise both supply and effective demand for a staple food?
Rice cultivation in Java has been highly labour intensive for many decades and the increase in preharvest jobs using the latest high-yielding rice varieties has not been nearly as great in Java as in other parts of southeast Asia. The Agro-Economic Survey has provided data to show that on good irrigated land, preharvest labour requirements increased (approximately) from 150 to 200 man-days per hectare, but much less on poorer land.
On the other hand, new harvest and postharvest practices have made inroads into employment, the former alone leading to falls of 104 man-days per hectare where it has been applied. Instead of allowing hundreds of men, women and children to harvest a field and receive between one ninth to one seventh of the crop, a much smaller group of men is hired for the job now. Although the latter earn more as individuals, the total return to labourers seeking income has declined. It needs also to be borne in mind that while the new preharvest practices apply only on good rice land which can take the new varieties, the new harvest practices can be applied under all conditions.
No longer tenable
The new village rice mills have displaced at least half of women's hand-pounding of rice with only a small fraction of these lost jobs being replaced by jobs in the mills-for men.
It is possible, therefore, to discern a new differentiation among landowners and among the landless, which can largely be seen as an acceleration, as a result of recent rice intensification programmes, of earlier trends in rural class formation and access to land.
From two separate sources (SUSENAS and Deuster's thesis) came evidence that inflation worsened the unequal distribution of income. This in equality declined somewhat as inflation abated toward the end of the 1960s, but the official data, such as they are, suggest that standards of living among the poor had not regained their 1963/64 level by 1969/70, and must therefore have clearly been lower than in the late 1950s. In addition, the nature and direction of the process of concentration of land control and of labour market stratification, which have been greatly accelerated by the rice intensification programmes in Java since 1966, and which are encouraging a new set of social relations of production, point toward further inequality of income. At the same time the old idea (always questioned) that on Java the intricate patron-client relationships of agricultural involution provided everybody with a niche in "shared poverty" now appears to be no longer tenable under the impact of the private returns to aggressive technocommercial innovations.