6. Technical Change in Wetland Rice Agriculture

Paul Heytens

Indonesia's wetland rice farmers use a wide spectrum of production tech-nologies. A variety of fertilizers and pesticides are available at subsidized prices; four different land preparation technologies are practiced through-out much of Indonesia; new seed varieties appear almost yearly; and several institutional arrangements for planting and harvesting coexist even in the same village. Farmers' selection of inputs and levels of use differ across the rice-producing regions, and these decisions affect pro-ductivity and the demand for labor.

The purposes of this chapter are threefold. The first goal is to investi-gate the dynamics of technical change since the inception of the Green Revolution and to describe the changes that have occurred in input use in wetland rice agriculture. A second aim is to examine how technical change has affected employment and farm incomes from rice production on Java. Answers to these questions indicate the extent to which the expansion of rice production since the late ig6os has promoted rural employment and income growth directly, important criteria in assessing alternative future rice strategies. A final task is to discuss the factors likely to affect the adoption of labor-displacing technology in the future.

Current Input Use in Indonesian Rice Production

The most striking observation from the farmer surveys is the ubiquity of the Green Revolution package of inputs. High-yielding seed varieties have been disseminated rapidly since the first improved seeds from IRRI were introduced to Indonesia in the late ig6os. Within the wetland sys-tems of Java, adoption of HYVs is virtually complete. At present, the adoption rate is over go percent in the wetlands and about 83 percent for all rice land. This widespread replacement of traditional seed varieties with short-duration HYVs has transformed the nature of wetland rice agriculture in Indonesia from one of low yields, nonuse of purchased inputs, and single annual rice crops to one of high yields, high levels of purchased inputs, and multiple rice crops.

Another vivid impression from the field surveys is the regional variation in wage rates, labor-hiring arrangements, and degree of mechanization, both within Java and between on- and off-Java areas. The patterns of input use and institutional arrangements in wetland rice cultivation are spread like a mosaic across the Indonesian landscape, determined by relative wages, technical constraints, and the level of rice profits. This diversity of input use is discussed below in connection with each individual task in rice cultivation.

Land Preparation

In Indonesia, four individual technologies-hand hoes, draft animals, two-wheel tractors, and four-wheel tractors-are available for land prepa-ration. Because of topographic and plot-size constraints, mechanical prob-lems, and differing factor endowments of farmers, many technologies coexist in the same village. Very small plots tend to be hand-hoed. Farm-ers who own their own draft animals are likely to use them, on their rice fields. Farmers who do not own animals generally use whichever tech-nology is most cost-effective. The presence of large rocks or steep slopes sometimes prevents the use of tractors and draft animals. Finally, in many areas of Java, hand hoes are still used alongside animal traction or tractors, particularly to repair bunds and to turn corners difficult to reach with animals or machines.

On a regional scale, choice of technique appears to depend primarily on the relative prices of inputs. Of the three survey areas in Java, only land preparation in Majalengka, West Java, can be classified as partially mecha-nized. Hourly wages for hoers and animal power in Majalengka are higher than in either Klaten, Central Java, or Kediri, East Java. Consequently, the prevailing rate per hectare for plowing and harrowing by a two-wheel tractor (including some hoeing) in 198'7 is competitive with other tech-niques in Majalengka, but low wages make tractor rental a less profitable venture in the other areas of Java. Hourly real wages for animal traction and hoeing in Kediri and Klaten would have to rise to levels prevailing in Majalengka before tractors would become an attractive alternative. The findings in the field mirror the situation in Java on a larger scale. Only West Java is tractorized to any significant degree.

In both of the off-Java fieldsites, the mechanization of land preparation is more advanced. Wages in Again, West Sumatra, and Pinrang-Sidrap, South Sulawesi, are much higher than in any of the Java sites. Wages are higher because of shortages of manual laborers and draft animals at peak demand times such as during land preparation. In both areas, population density per hectare of sawah is much lower than in any of the Java fieldsites.

Generally, the level of tractorization in Indonesia is very low. Since the mid-1970s, the total number of tractors in Java and the outer islands has increased severalfold, but the actual percentage of land cultivated by machine is still very small. For example, the number of two-wheel trac-tors in Java increased from 3,996 in 1981 to 9,496 in 1986 (BPS 198-7). But if one hand tractor cultivated as much as thirty-five hectares per year, total tractorized area would not have exceeded 7 percent of total culti-vated wetland area in 1986. The corresponding figures in West Sumatra and South Sulawesi for 1986 are 3.5 and 1o percent if mini- and small, four-wheel tractors are included at a capacity of seventy hectares per year. (On Java, only two-wheel tractors are used on rice land, but off Java the smaller four-wheel tractors are also used.)

Many explanations underlie the low level of tractorization in Indonesia. There are no agronomic reasons a priori (Binswanger 19-78) and no evi-dence empirically in Indonesia (Lingard and Bagyo 1983; Maamum et al. 1983) that tractor preparation confers any yield advantage over other land-preparation techniques. On Java, the generally low level of wages and technical factors, especially the small size of plots and the imprac-ticality of using tractors in hilly areas, are the main constraints on mecha-nization of land preparation. The constraints on greater tractor use for the off Java areas are probably more varied, although topographic limitations and the greater difficulty in obtaining and servicing tractors in the outer islands are important.

Government policy recently has not promoted tractorization. Although many owners purchased tractors at government-subsidized nominal in-terest rates of 12 percent (real rates of 2 to 4 percent) per annum since the mid-1970s, these concessionary rates are no longer available. Nominal borrowing rates have been over 20 percent per year for commercial loans during much of the 1g8os. Moreover, tractors are not cheap in Indonesia. High costs of assembly (because of tariffs on imported parts and high-priced domestic parts) and large distribution costs (because of supply monopolies) contribute to high prices. Domestic prices of two- and four-wheel tractors were estimated to have been roughly 25 percent above world prices in 1988.


The transplanting of seedlings is practiced universally in the wetland rice systems of Indonesia. Seedlings are transplanted in rows by women on the main plot three to six weeks after the seeds have been planted in a nursery. This practice predated the dissemination of high-yielding variety seeds and other inputs. In this sense there has been no recent technical change in planting techniques. Minor technical innovations, such as using a small board to lay out grid marks for plant spacing and placing seedlings in straight rows, have made the task easier and faster, but the fundamen-tal task remains the same. More interesting planting issues are the changing institutional and labor-hiring arrangements that farmers employ to obtain labor, discussed in Chapter 5.


Weed growth is a problem in all wetland rice systems. The severity varies depending on the degree of water control. Weed problems tend to be greater when sawah alternately becomes wet and dry, a common phenomenon in rainfed and lower-productivity sawah, where water con-trol is uneven. But weeds are present on all sawah. There are two basic approaches to dealing with weed problems. Farmers can either try to prevent weed growth or remove weeds after they appear. Rice farmers in Indonesia overwhelmingly take the second approach.

With the advent of straight-row transplanting of rice seedlings, it has become possible to use hand tools for weed removal during the early stages of rice plant growth. Farmers have used various homemade hand tools, including a rotary weeder that is pushed along the top of the soil. Although no evidence was available from the survey, other researchers have reported that the introduction of this practice halved labor use in the early weeding (Sinaga 1978; Collier et al. ig8za). In the later stages of the cropping season, the increase in size and breadth of the rice plant pre-vents the use of all implements to remove weeds. Therefore, weeding later in the season must be done by hand.

In Pinrang-Sidrap, farmers spray herbicides to reduce weed growth on higher-productivity, irrigated plots. Use of herbicides dramatically re-duces or completely eliminates weed growth, thus allowing for large re-ductions in labor requirements. No herbicides were used in any of the Java fieldsites or in Agam, West Sumatra. In Java, the reasons probably stem from the low costs of labor and the perception that chemicals are ineffective in reducing weed growth. Also, herbicide use has not been promoted in government intensification programs in Java as it has in South Sulawesi. In Agam the reasons are less clear but probably stem from a lack of knowledge and availability; the higher labor costs there could make herbicide use an attractive weed-control option.

Fertilizer Use

Fertilizer subsidies have been one of the cornerstones of Indonesia's rice development program, and use per hectare is high in comparison with other rice-producing countries in Southeast Asia. The application of fertilizer has risen dramatically since the late ig6os. Nutrient sources have become more diversified in recent years, as indicated in Tables 6.1 and 6.2. Urea makes up a large but declining portion of total use; triple superphosphate (TSP) accounts for a good portion of the remainder. Fer-tilizer use is much higher on Java than in the outer islands, averaging 276

kilograms of urea, 99 kilograms of TSP, and 3 kilograms of other fertilizer per hectare on Java's wetlands and io6 kilograms of urea, 5o kilograms of TSP, and 6 kilograms of other fertilizers per hectare in wetland areas off Java (BPS, 1986). Fertilizer use in the Java survey areas was somewhat higher and more diversified than these overall averages (see Appendix 4.1, Chapter 4). Starting in 1987, farmers with higher-productivity, irri-gated sawah in Kediri, Klaten, and Majalengka started using potassium chloride (KCI) and ammonium sulfate (ZA) as well as increased levels of TSP as part of the INSUS Packet D (general intensification) and SUPRA

INSUS (super intensification) programs. For example, SUPRA INSUS recommendations for the 1989 wet season are that Java participants apply a standard package consisting of z5o kilograms of urea, loo kilograms of TSP, 75 to loo kilograms of KCI, and ioo kilograms of ZA (depending on the sulfur content of the local soils).

Fertilizer use at the off Java fieldsites was lower than at the Java sites but somewhat higher than the overall average for the outer islands. Sur-vey farmers off Java tended to use only urea and TSP. The more recent intensification programs recommending use of ZA and KCI had not begun in off Java areas at the time of the field surveys.

The yield advantages from applying chemical fertilizer were clear to farmers participating in the field surveys. All stated that their yields had risen in response to higher fertilizer applications. Among survey farmers, fertilizer use was greater on the higher-productivity systems with good water control; fertilizer applied in a more stable and fertile crop environ-ment was considered more likely to pay off and less risky than fertilizer applied in a variable environment. In the good-control sawah systems, farmers tended to apply less fertilizer during the wet season to reduce the risk of lodging. Lodging typically is not a problem during the dry seasons.

In virtually all of Indonesia's wetlands, fertilizer is broadcasted onto the rice paddy. Fertilizer generally is applied three times in a season-at the time of transplanting and twenty to thirty days and thirty to forty-five days afterward. In all survey areas, farmers used family labor to spread fertil-izer. An average fertilizer application per hectare can easily be finished in a day by two or three people. Hence family labor usually is sufficient, particularly on the small sawah plots that characterize wetland rice pro-duction on Java.


The most commonly described technological change in Indonesian rice agriculture is the replacement of the traditional ani-ani (hand knife) with the sickle to harvest rice. As noted in Chapter 5, this shift involves considerably more than a change in technologies. The ani-ani has been associated in the literature with Geertz's notion of involution and the traditional bawon or share harvest open to all who wish to participate. The sickle, on the other hand, has been associated with the closing of the harvest and the tebasan system in which farmers sell their standing rice crop for cash to a middleman who conducts the harvest. In the early 1970s, the shift to sickle harvesting tended to involve a substantial change in social power relations (Collier et al. 1974; Hayami and Hafid 1979). As time passed and high-yielding seed varieties spread, adoption occurred for efficiency and technical reasons. By 1983, the sickle was the major harvesting tool in wet rice cultivation and was used by 70 to 8o percent of farmers (BPS, 1983). A more recent estimate for the wet season 1986-8-7 placed sickle use at 85 percent of all rice farmers in fifteen major rice-growing provinces.

The widespread adoption of the sickle along with HYVs makes sense for technical and economic reasons. The sickle is the most convenient and efficient tool to harvest the shorter and thicker-stemmed HYVs, which tend to lodge and ripen at the same time. With an anti-anti the harvester cuts each individual panicle, a task that is much easier to perform on the taller, traditional seed varieties. The increased cropping intensity made possible by the shorter-maturing HYVs made speed more essential, again greatly favoring the use of the sickle. Finally, because of the sickle's greater speed, farmers need to hire far fewer people to harvest a hectare of rice with a sickle than with the anti-anti, making harvests much easier to arrange in times when labor supplies are tight.

In the future, the major input-substitution issue in harvesting likely will be with threshing. Manual threshing (e.g., hitting paddy stalks on the ground or stomping by foot) is still by far the most commonly used meth-od. But in the 198os the number of fuel- and pedal-driven threshers in Indonesia rapidly increased. On Java, the number of machine threshers (both manual and fuel-driven) increased from 13,670 in 1981 to 54,313 in 1986. The corresponding numbers for West Sumatra were 176 in 1981 and 1,136 in 1986 and for South Sulawesi, 91 and 15,o89, respectively.

These increases are rather dramatic, but they do not represent a very large portion of cultivated area (see below). Threshing by machine does not necessarily confer a cost advantage over manual techniques. For ex-ample, in Agam both the thresher owner and traditional harvesters re-ceived about a 1o percent share of the harvest, and the farmer had to pay extra to have the padi cut before machine threshing. Returns with ma-chines are often higher because of reduced harvesting losses (estimated to be 5 to 1o percent of total output) and a higher-quality output-there are fewer -brokens, drying is easier, and winnowing is more effective. In addition, machine threshing is much faster and less labor-intensive, a big advantage in the labor-scarce areas off Java.

The Impact of Technical Change on Farm Labor Demand and Income in Rice Production

Technical change can influence labor demand and farm income in two ways. It can do so directly by changing the labor-intensity and yield per hectare and also through intensification of crop production-raising crop-ping intensities on existing land. Change in the labor-intensity for a par-ticular task usually implies a change in the number of workers employed in the agricultural sector. Raising cropping intensities increases the frequency of employment opportunities during the year and thus increases the total employment per worker.

Technical Change and Labor Demand

As shown in Table 6.3, total cultivated wetland rice area in Indonesia increased by 35 percent between 1969 and 1987. The corresponding fig-ures for Java and off Java are 24 and 5z percent, respectively (the of Java figure includes a large increase in low-productivity, tidal swamp area in the ig8os). On Java, the rise resulted mainly from an increase in cropping intensity; physical sawah area increased very slightly. Additional crops were made possible by adoption of the much shorter-duration HYV seeds (crop duration was cut from 6 to 3.5 months on average), which could be exploited as a result of public investments in new irrigation systems and rehabilitation.

Considerable field evidence on labor use by task in wetland rice agri-culture in Java has been gathered on a regular basis by research teams since the late 196os. Collier and his colleagues, in particular, have been active in the collection of such data. The various studies are far-ranging both in regional scope and time period and conclude, generally, that labor use per hectare per crop on Java has declined since the initial spread of HYVs.

Early studies show little difference in labor-intensity for preharvest

labor between local/national varieties and HYVs around 1970. Collier and Birowo (1973) in a survey of 622 farmers from zo Javanese villages found preharvest workdays per hectare for local and HYV varieties to be about the same (z4o). These findings were corroborated by the studies of Soelistyo (1975) in East Java and Montgomery and Sisler (1974) in the Yogyakarta area. Montgomery's surveys showed that both high-yielding and local varieties used 318 workdays per hectare. Soelistyo found that there was no significant difference between high-yielding and traditional varieties in labor use per hectare for preharvest activities in irrigated areas.

The shift from an open harvest with the ani-ani to a closed harvest with the sickle (yields were about the same among seed varieties in the early 1g7os), however, resulted in a large drop in labor use per hectare with HYVs. Inferences from data in Collier's studies (1973, 1974) place the figure as high as i,ooo work hours per hectare in areas where farmers shifted from harvesting traditional varieties with an ani-ani to HYVs with a sickle.

As the decade of the 1970s passed and the yield and purchased-input gap between local and high-yielding varieties widened, labor require-ments increased for some tasks of HYV cultivation. For example, fertilizer applications became more time-consuming as usage levels increased. Similarly, harvesting and threshing took longer as yields increased. But more generally, the time required to cultivate a hectare of wetland rice decreased. In a broad-ranging study of many villages, Collier et al. (198za) estimated that labor-use requirements per hectare in wet rice cultivation including harvesting fell by 1o percent from 1969 to 19-78. In a recent resurvey of villages and a reassessment of old evidence, Collier et al. (1988) estimated that labor use per hectare remained about the same through the 1970s but declined fairly substantially after ig8o. The decline was attributed to decreased employment opportunities for women in har-vesting, threshing, and weeding and for men in land preparation resulting from adoption of the technical innovations discussed above.

A number of problems arise from comparisons of such village-level studies of labor use patterns carried out at different times. Each study deals with peculiar ecological conditions and agronomic practices, size distribution of farms, enumeration techniques, and measurement errors. The regional coverage of the studies cited above varied markedly, and the average size of farms differed from one study to another. In all studies, data, especially on labor inputs in the traditional ani-ani harvest, are open to a considerable margin of error. These factors help to explain the huge range in estimates of labor inputs for each activity in the survey samples. Although useful and individually valid, the earlier survey work is not very helpful in quantifying the impact of the Green Revolution on total labor demand in the rice sector.

A different approach is taken here. Data on labor use per hectare from the field surveys are used to estimate current intensity in wetland rice agriculture on Java in 1987. Labor use per hectare in earlier periods is estimated by assessing the apparent impacts on labor requirements per task resulting from technical change, as documented in the work dis-cussed above. Estimates of earlier labor use thus are obtained by working backward from recent survey evidence using old survey evidence as a guide. The aggregate impact on labor demand then is assessed by estimat-ing the extent of the spread of various changes and multiplying by culti-vated area. This scheme provides a rough estimate of the impact of the Green Revolution on labor use in Java-whether aggregate labor demand has increased, decreased, or remained about the same in Javanese rice agriculture.

Survey estimates of labor use per hectare by task per season for wetland rice in Java for 1987 are presented in Table 6.4. The table includes mecha-nized and nonmechanized cases and does not distinguish between hired and family labor. Total work hours per hectare through harvesting in 198-7 range from 1,46o, if all tasks are nonmechanized, to i,oio, if land prepa-ration and threshing are mechanized.

As discussed above, land preparation in wetland rice cultivation on Java remains largely unmechanized. Estimates based on calculations from BPS survey data place tractor use at about 7 percent of total cultivated area in 1987 and 3 percent in ig8o (see above). Although data are not available, tractors are assumed not to have been used in 1969. Labor hours for both

mechanized and unmechanized land preparation (Table 6.4 assumes a combination of hand hoeing and animal traction for nonmechanized land preparation) in the earlier time periods are assumed to be the same as in 1987.

Planting was not subject to major changes over the time period. Minor technical innovations (e.g., spacing boards) appear to have lowered the time required to complete the planting task by 25 hours per hectare in ig8o and 1987 relative to 1969. Weeding done early in the cropping season when rice seedlings still are small was affected by the introduction of hand-pushed weeding tools. Time savings allowed by this innovation were offset somewhat by greater weed growth resulting from higher fertil-izer use. Weeding time is assumed to have dropped by 25 hours per hectare in ig8o and 198-7 in comparison with ig6g. Because application of fertilizer and spraying of pesticides are associated with the spread of HYV seeds, labor requirements for these activities are set to zero in the ig6g estimation. Fertilizer application levels are higher in 1987 than in ig8o, and that task is assumed to be io hours per hectare less time-consuming in ig8o. According to the BPS cost of production surveys, pesticide usage levels did not change much between ig8o and 1987 so spraying time is assumed to be the same in both years.

The most important changes that have occurred in rice cultivation since ig6g are in harvesting. The shift from the ani-ani to the sickle allowed for considerable time savings in cutting paddy. These time savings have been offset very slightly by rising yields. The introduction of the sickle, ceteris paribus, has halved at least the number of labor hours required to har-vest. Harvesting time with the sickle in ig8o, then, is assumed to be half that in 1969 with the ani-ani. Because yields are higher, harvesting time per hectare is assumed to have risen 25 hours in 1987 in comparison with ig8o. In 198-7 harvesting required 375 hours per hectare. Harvest time is assumed to be 35o hours in ig8o and loo hours in ig6g. Adoption of the sickle on Java was estimated to have been carried out by at most 3 percent of wetland rice farmers in ig6g, about 75 percent in ig8o, and well over go percent in 1987. Data on adoption of the sickle apply to farmers rather than cultivated area and are considered to be rough estimates. For the estimation, sickles are assumed to have been used on o percent of culti-vated area in ig6g, 75 percent in ig8o, and ioo percent in 1987.

Another technical change in harvesting is machine threshing. It is as-sumed that two-thirds of existing threshers are pedal-driven and one-third fuel-driven (the number of each type is not distinguished in the aggregate BPS data) and that pedal threshers are used for 1o hectares per season and fuel threshers 2o hectares per season. Each type of thresher is estimated (from agricultural survey data) to account for 2 percent of har-vested area in ig8o and 8 percent of harvested area in 1987. Machine threshers were not present in Java in 1969. Labor hour estimates for machine threshing from the farm surveys are given in Table 6.4.

The results of the calculations of total labor demand are contained in Table 6.5. Annual labor demand in wetland rice cultivation on Java de-clined by 1.6 percent between 1969 and ig8o. Labor demand in the ig8os has risen as a result of an expansion in cultivated area. Still, overall labor demand in 1987 is estimated to be o.5 percent lower than it was in 1969. These results corroborate the conclusion from other work that on-farm labor demand in wetland rice production has not changed much on Java since the Green Revolution began. Increased demands, particularly from increased cropping intensities, have largely offset reduced demands for particular tasks (especially harvesting). Although stable in the aggregate, labor demands have become more evenly distributed through the year. The per capita employment obtained from rice probably has increased over the past two decades, whereas the total work force has declined.

Technical Change and Farm Income

The spread of the Green Revolution package of inputs and other related technical change (e.g., sickle harvesting) also affects the rural economy through changes in farm incomes. The full impact of changes in income will be discussed in Chapter 8. The task here is to estimate the changes in farm income from the production of rice that have occurred on Java as a result of the adoption of HYVs and infrastructural investments.

Both large- and small-scale farmers can use HYV seeds and chemical fertilizers profitably. Because of adoption of HYV seeds, increased fertil-izer use, and better water control, rice yields have almost doubled in Indonesia's wetlands. In 1969 yields on sawah in Java were 2.6 tons of unmilled rice per hectare; by 1987, yields had increased to about 5 tons of gabah per hectare. Similarly, HYV seeds and irrigation investments en-abled farmers to grow on average an additional half crop per year.

The labor coefficients reported in Table 6.4 are the starting point for the analysis. Data from the farm surveys are used to calculate on-farm prof-itability in 198-7 for a representative one-hectare Javanese wetland rice farm using nonmechanized production technologies (i.e., tractors and machine threshers are not used). Nonlabor input use, wage, yield, and

output price data from various cost of production surveys published by the Central Bureau of Statistics are used to calculate on-farm profitability in 1969 and ig8o.

This analysis indicates that annual incomes per hectare from rice cultivation in Java have almost tripled in real terms since 1969. These results are summarized in Tables 6.6 and 6.7. Table 6.6 shows the impact of higher yields only and indicates that rice profits (in constant 1987 prices) increased from Rp 271,000 per crop (per hectare) in 1969 for traditional seed varieties to RP 334,000 per crop in ig8o and RP 455,000 per crop in 1987 for high-yielding seed varieties. Table 6.7 includes the impact of higher average cropping intensities and indicates that on a per hectare basis annual real returns to wetland rice cultivation on Java have increased almost i5o percent since 1969. The rate of increase in income was greater in the ig8os than in the 1970s.

The Future of Labor Use in Rice Farming

During REPELITA V, if real wages continue to rise, mechanization of threshing and land preparation will spread more widely both on and off Java. Beyond profitability considerations, other influences could affect mechanization of rice agriculture in the future. One is the desire to reduce the drudgery involved in labor-intensive farming. Related to drudgery reduction is a desire on the part of farmers and laborers to raise the prestige of agricultural occupations. Both of these aspirations point in the direction of greater mechanization. Although the importance of these factors can only be guessed at, the 1987-88 survey found considerable sentiment in this regard, as have other researchers (Siregar 1986).

Changes that standardize cropping schedules in a particular region, thus raising the seasonal peaks in the demand for labor, also will affect the spread of future mechanization. Mechanization of land preparation has been induced in parts of West Java as a result of the synchronization of water-delivery schedules (Manning 1986). Similarly, tightness in inter-seasonal water schedules also induces mechanization; farmers must finish the previous crop quickly to be ready for the scheduled water delivery for the following one. In addition, the integrated pest management program will intensify seasonal peaks in labor demand through the simultaneous planting and harvesting of all contiguous sawah plots in an area.

The rate of mechanization will be influenced by farmers' ability to make investments in new technology. This is directly related to price and farm-ers' income. The cost of mechanical technology depends on government trade, tax, and credit policies. As discussed above, the present mix of government policies (e.g., trade taxes, supply monopolies, and credit policies) discriminates against the purchase of capital equipment. Gov-ernment policies, therefore, do not encourage the mechanization of rice agriculture in Indonesia.


As shown in Chapter z, farmers' returns are affected by government price policies for rice output and purchased inputs. Profits from rice production generally are very high (see Chapter q.). If high profits are maintained in the future, farmers, particularly those who cultivate good-control sawah, will continue to have high savings that could be used to purchase new farm implements. This ability to self-finance new capital investments will give them more flexibility to respond to rising wages.

Since the inception of the Green Revolution, there has been no change in aggregate employment in rice agriculture on Java. Real incomes in rice farming rose by two and one-half times per hectare between 1969 and 1987. Yet despite an increase in cultivated area, aggregate employment stayed about the same. Even if cultivated wetland rice area continues to increase in the future, input substitutions could cause rice production not to absorb many more farm laborers. Nevertheless, as shown in Chapter 5, rice farming done with mechanized technologies is still more labor-inten-sive than that of substitute crops.