Rainfed revolution in Northeast Thailand

Southeast Asian Studies

Volumn 46, Issue 3, 2008, Pages 289-376

  Grandstaff, Terry B ^a   Grandstaff, Somluckrat ^a   Limpinuntana, Viriya ^b   Suphanchaimat, Nongluck ^b  

^a Khon Kaen University ^*  (Thailand)

^b KHON KAEN UNIVERSITY   (Thailand)

Author keywords

Agricultural development; Economic change; KDML105; Northeast Thailand; Rainfed rice; RD6; Remittances

Indexed keywords

AGRICULTURAL DEVELOPMENT; ECONOMIC DEVELOPMENT; ECONOMIC GROWTH; EMIGRATION; HOUSEHOLD INCOME; MIGRANT WORKER; REMITTANCE; RICE; URBAN AREA; VILLAGE;

ASIA; EURASIA; SOUTHEAST ASIA; THAILAND;

EID: 67650073113     PISSN: 05638682     EISSN: None     Source Type: Journal    
DOI: None     Document Type: Article

References (95)

1. In December 2007 the Ministry of Social Development and Human Security announced the results of its new Human Security Index (similar to quality of life). The press was very surprised to hear that the Northeastern provinces on average were rated the highest in the country [Bangkok Post 2007]. In a day or two the issue seemed to fade from notice.
2. List of acronyms used in this report can be found at the back, after the main text.
3. SES [2004]: 77.5% of Northeastern village households have piped water and 98.9% have private indoor bathrooms. PHC [2000: 2]: 90.2% of all households in the Northeast have safe drinking water.
4. CDD [2003]: 29% of rural village households in the Northeast reported having at least one mobile phone, compared to 34% for all such villages in the country. CDD surveys villages designated as poor, which includes most rural villages in the Northeast.
5. PHC [2000]: Villagers still cook primarily with charcoal (33%) and wood (30%).
6. PHC [2000] put the portion of rural house made of cement and brick at 14%, cement/brick and wood 30%, and otherwise permanent 49%. SES [2004] reported 20% cement and brick, 41% cement/brick and wood. Visual impression from recent widespread observation in Northeastern villages suggests a large majority of houses have been improved [see also Grunbuhel et al. 2003: 68 footnote 15].
7. Even with full development, only 16% of cultivatable land in the Northeast could be serviced by a major irrigation system, because of limitations of the type of terrain and water sources in the region [AIT 1978; Limpinuntana et al. 1982: 53-54; also see Long 1966: 357 and Pendleton 1962: 148]. By 2006, less than 1% of agricultural land in the Northeast was serviced by a large or medium sized system, the lowest of all regions in the country [calculated from OAE 2006: Tables 127 and 129; 2007: Table 128]. Somrith and Awakul [1979: 113, Table 1] estimated 90% of the paddy land in the region was "shallow or intermediate rainfed land" and only 8% of it was irrigated.
8. Rainfed rice fields in Northeast Thailand are "rainfed shallow drought- and submergence- prone" and "rainfed shallow drought-prone type II" [Mackill et al. 1996: 4-10]. In most areas both types can be found at the local level. "Drought is the major problem in rainfed areas" [Pushpavesa et al. 1986: 168]. When rainfall is scarce, as it frequently is, drought is widespread within the affected area. Flooding occurs less often, but many of the "lower paddies" that are typically part of farm holding are flood prone as well. ("The drought-prone area in the northeast is subjected to submergence from time to time" [Pushpavesa et al. 1986: 169].)
9. The principal numerical data underlying the analysis in this report are included in the tables and in the appendix, but the analysis here stops short of mathematical modeling and statistical testing.
10. Village data are now called "non-municipal" because in 1999 everything non-municipal except villages (i. e., urban and rural sanitary districts) was moved to the municipal category. However, in order to keep the same categories clearly identified for comparison over time, in this report we simply keep to the old definition of "village" (even though some are now suburbs) and group all others as "non-village" (see Table 1, notes 1 and 2). Changing definitions make it more difficult to measure urban growth (e. g., the "municipal" registered population jumped from 1.32 million in 2000 to 3.28 million in 2002, in Table 1, line 1.2). Nevertheless, real, evident, widespread substantial urban growth has certainly taken place.It is not the result of changing definitions.
11. Computed from CDD [2003] (already rural biased) data. 4,430 villages were sampled (representing 14% of all villages in the Northeast and containing 15% of population [calculated from DOPA 2007b]). All provinces were non-randomly sampled. Within the province, districts were selected and all villages within the district sampled. All provincial capital districts were selected but also intentionally offset by choosing one or two of the districts in the province that were most remote from every capital district (i. e., remote from capitals in neighboring provinces as well).
12. As one villager in Selaphum District, Roi Et Province, complained about the lack of employment opportunity, "We really need someone to come and build a factory here." Even some whole provinces may lack urban-type employment opportunities, e. g., Yasothon [see Funahashi 1996: 626-627; DIW 2008].
13. If they were away from home long enough to make more money from other sources than agriculture, they would probably have been away long enough not to have been counted as a household member by survey definition. In NSO censuses and surveys, if people stayed away for three months or more, they were usually not counted as living in the household, except in a probable minority of cases when they stayed away for three months or more but had no other "permanent residence" such as ship crews or traveling salesmen [e.g. ACIS 1993: 19], or stayed away for up to six months specifically for education or training outside the country [e. g., SES 2004: 31]. (N. B., the reader is cautioned to check the more detailed Thai language definitions in these publications in order to best understand these criteria.)
14. The year 1993 is used for the purpose of data compatibility because available data on the number of people registered in and out of the region only went back to 1993 on the government website [DOPA 2007a]. In the past, many Northeasterners stayed registered in the Northeast despite working year-round elsewhere, causing, for example, mass exodus from Bangkok prior to elections. For the year 2000, 21.40 million people were registered but only 20.83 appeared in the census [PHC 2000], less of a difference than in the past (Table 1, lines 1.1 and 2.1). There are now more incentives to change registration (e.g., to get children into school in the new location) and the process is now much easier than before. A one-stop household registration at the destination has been set up (starting in 1996) and is now available on-line at most district offices throughout the country. Nationwide, censuses have always missed a small portion of the registered population for one reason or another, but less and less over time. Censuses are conducted in the dry season when municipal populations swell, but this should not affect the findings here because the criterion is "usual place of residence" during the 12 months ending March 31.
15. Analyses of seasonal and longer-term migration in this report are based on available data and partially arbitrary definitions of household membership (see footnote 13 above), and should not be assumed to necessarily imply qualitatively different work situations. (Working away from home for four months would not be much different than for three months, while working in a permanent job would be very different from working for only four months, etc.)
16. We use the term "supported" for what the SES calls "economically inactive" households, because, as we show, many of these households are not economically inactive, even though they received more than half of their total income from remittances, pensions or property income. In later years, data are reported on property income vs. the two others combined. In 2004, households receiving more than half of their income from remittances (and pensions) accounted for 23.4% of all households in the Northeast and property income 0.5%, and in villages 24.2% and 0.5% respectively [SES 2004: Tables 11.1, 11.3].
17. 17)In 2002, 23% of the households in the Northeast had most of their income from support ("economically inactive households") (Table 1, line 3.11). The rest of the country without the Northeast had 13% [computed from NSO 2002: 7 and Table 11.1].
18. Income survey data may not be reliable enough for precise analysis, and averaging also obscures differences among survey units (e.g., "village" household data obscure differences between farm and non-farm households). Nevertheless, assuming a reasonable degree of continuity in survey methodologies, income and expenditure data can help reveal broad trends (with cross-checking, and, e.g., with special caution toward figures on home- produced, non-cash income). However, the reader is cautioned about using the data presented for purposes beyond these limited usages.
19. 19)"Economically inactive households" are sub-classified as "receiving assistance or pensions" and "receiving property income." In the Northeast Region, the vast majority are in the first category, 98% of all households, 99% in "non-municipal areas" [computed from SES 2004: Tables 11.1, 11.3]. In the rural villages there are very few pensioners. (Before the 1990s, when there were few supported households in the Northeast, there would have been a greater percentage of pensioners among them, but mostly not in villages.) For example, of the income coming from "current transfers" ("assistance payments," "pensions and annuities" and "terminal pay") in all the "non-municipal areas" (including those that may have become suburbs), 94% came from "assistance payments" (95% including in-kind income "received free") [computed from SES 2004: Table 4.3]. It can be safely assumed that the vast majority of these "assistance payments," providing a majority of household income, came from close relatives (also consistent with additional data available only for 1990-see footnote 23 below).
20. In 2002, about 88% of the "supported" households in the Northeast were in villages: 0.241 × 4.3/(0.232 x 5.1) (4.3 and 5.1 from Table 1, lines 3.2 and 3.4; 0.241 and 0.232 from SES [2002: Tables 12.3 and 12.1]).
21. Based on calculations for the year 2002, a minimum of 0.45 million and a maximum of 0.81 million supported households had members still working in agriculture. Statistics used were: percent distribution of members of supported households by 10 year age intervals by household socio-economic class; percent of all household members age 15 and above for all households in the region; supported household members as a percent of all household members in the region; distribution of supported households by number of income earners in the household [all from SES 2002: 114]; percent distribution of household members by socio-economic class and occupation (agriculture, construction, etc.) [SES 2002: 119]; and number of population in the Northeast in 2002 (from Table 1, line 3 of the current report). Supported households have a much larger percentage of "economically inactive" members. In 2002, 34.9% of the males and 43.8% of the females age 15+ (vs. 11.8% and 20.1% for farm households). In 2004, it was 37.4% of the males and 44.4% of the females (vs. 12.0% and 19.8% for farm households) [SES 2002: Table 12.1; 2004: Table 12.1].
22. Income from farming was only 397 baht/month, vs. 3,532 for farm household mainly owning land whose income was mainly from farming [SES 2004: Table 4.1]. It is possible some of these supported households have no agricultural land but do agricultural labor and/or raise livestock (mainly cattle). Because of access to fodder from public and private land, raising livestock does not depend on agricultural holdings. For example, in 2003,78% of all agricultural holding households raising livestock had less than 2 rai (0.32 ha.) of agricultural land [ACIS 2003: Table 18.2]. However, cattle raising is labor intensive, so probably not well suited to many supported households, because of their smaller size. And given their degree of support, they might be less interested in working for others as agricultural laborers than in looking after their own fields. For these reasons, supported households containing people who participate in agriculture are probably agricultural holding households.
23. For 1990, additional data are available on all the households who received assistance [SES 1990: Table 14]. This table includes any degree of assistance for all 10 types of households, not just "economically inactive households" (but a limitation is that at that time "economically inactive households" were only 10% of all households, compared to 24% in 2004 (Table 1, line 3.11)). In 1990, a total of 31.4% of all households in the Northeast received some degree of assistance, and nearly half of those (46%) were households whose income was mostly from their own farm land. (Own-farm households made up 55.1% of all households in the Northeast, and 26.3% of them received assistance payments.) In second place were the vast majority (88%) of the "economically inactive households" or 28% of all those receiving assistance. All other categories were less than 5% of all households receiving assistance. Of the total amount of money sent, "economically inactive households" received 57.6% (receiving households averaging 1,747 baht/month), farm-owning households 24.2% (averaging 451 baht/month), and the rest again with no category receiving more than 5%. Most households received assistance from outside the Northeast-about three-quarters overall, and slightly more for the own-farm households than the economically inactive ones. Most senders were sons and/or daughters, again about three-quarters overall and slightly more for the own-farm households. Second were from spouses, overall 12% but with the economically inactive households being much more characteristic of spousal support: 20%, vs. only 8% for the own-farm households. Almost all the remaining sources were other relatives. (It should be noted that payments were also sent out of Northeastern households, especially for children in the Northeast and Bangkok, etc., but from only 8.5% of households) [SES 1990: Table 15].
24. Villagers explained to Funahashi the strategy of "baeng kan pai ha ngan" -parents or close relatives taking over the work responsibility in a couple's rice fields while they go live and work elsewhere but remit cash, or taking care of the couple's children while they attend school. "As the parents grow older, the young couple (particularly the youngest daughter and her spouse) will return to the village" [Funahashi 1996: 628]. This phrase might be translated as "splitting up to go get work," but specifically for the purpose of helping each other to make a living as a family, in other words, a joint strategy.
25. From the PHC censuses, there were 5.7 persons per village household in 1980 and 4.2 in 2000 (computed from Table 1, lines 2.7 and 2.8). For agricultural holding households there were 5.7 in 1983 and 4.0 in 2003 (computed from Table 1, lines 5.1 and 5. 5). It might be thought that household sizes have fallen because of increasing preference for nuclear family housing, but there is little support for this in SES survey data: 76% of village households without any "sub-families" in 1990, 76% in 1994, 72% in 1998, 78% in 2002 and 77% in 2004. Average household sizes shrink when people have fewer children, but it is also probable it is because people leaving home and not returning have not (yet) caused any significant abandonment of houses. In most cases, at least someone is staying in the house and "minding the farm." This is also consistent with an increase in single person households: from 2.8% in 1990 to 6.7% in 2004.
26. J persons/household-see Table 1, note 3b), 55% of all households and people in the Northeast (65% and 66% of "village" households and people, respectively) (divisors in Table 1, lines 3.1-3.4). [For unknown reasons, this estimate is somewhat lower than in NESDB and World Bank [2005: 136-137] which reported that farming households in the Northeast were "over 60 percent of all Northeastern households" in 2002 compared to "42 percent. in the North," (their accompanying bar chart shows figures even higher than in the text, but the axis may be mislabeled). In their background study [Agrifood Consulting International 2005: 302], 58% grew rice compared to 31% in the North. These figures were attributed to a 2002 SES "supplement," a special additional survey requested by the World Bank, not publicly available from NSO (NSO Household Economics Statistics Group, personal communication August 2007).]
27. Village household income increased from around 2,200 baht/month in the early 1980s (2,289 baht/month in 1981; 2,165 in 1986) to 8,727 in 2004 in current value (not adjusted for inflation) [SES 1981: Table 4; 1986: Table 4.3; 2004: Table 4.3]. Using the same multipliers sourced for Table 7A below, in constant 1998 baht these would be about 4,400 and 8,000 baht/month, respectively. So real household income nearly doubled, but household size decreased from more than 5 to 4 persons (footnotes 25 and 26 above), so per capita household income more than doubled (4,400/5= 880; 8,000/4= 2,000).
28. Net agricultural cash income can only be directly compared with net non-agricultural cash income, but the latter cannot be calculated because only expenses on agricultural inputs were reported by OAE. The "inputs" (additional costs) incurred in earning non-farm income (transportation, meals, accommodation, etc.) were not separated out from other household expenses in the OAE data (summarized in Table 7C). For example, if a farmer earned wages from non-farm work, all those wage would be reported as non-agricultural income, without deducting, for example, the bus fares to and from the work site, an expense which would not have been incurred if the farmer had simply stayed home on the farm instead of going off to work.
29. These increased remittances after the mid-1990s were probably used for housing improvements and taking care of aging parents (who were often also taking care of the grandchildren) and other non-agricultural purposes, as we argue below that the biggest agricultural changes occurred earlier.
30. Remarkably, CDD [2005] found a total of 171,183 shops and stores in 30,862 villages in the Northeast, an average of 5.5 per village. But more remote villages might have fewer stores.
31. With economic growth, there has been reduced consumption of rice in traditional rice- eating societies [e.g., Gehlhar and Coyle 2001: 10; Regmi et al. 2001: 21]. In South Korea, yearly milled rice consumption decreased nearly 30% between 1980 and 2000 (from 132 to 94 kg/person) [Nam 2002: 114 note 28]. In Thailand it apparently dropped 15% from 1990 to 2002 (from 119 to 101 kg/person) [Isvilanonda 2006: 3]. Among villagers, it fell to an average 114 kg/person-128 kg/person for the poorest 25%, 91 kg/person among the wealthiest 25% of village households. Even if we assume Northeastern farm families consume about 200 kg of unmilled paddy rice per person (133 kg milled-a little over Isvilanonda's highest villager consumption level of 128 kg), that would still be a saving of about 100 kg/person over the traditional Northeastern diet, which was about 300 kg/person for consumption alone, 400 kg for all purposes [see Fukui 1991: 528 and Suzuki et al. 1999: 58], although common estimates for traditional paddy rice consumption per person in Thailand were often lower than this [e.g., see Fukui 1993: 401 note 2].
32. Reduced consumption of home-produced rice means more rice can be sold instead, or some/ paddy land might be used for other things. However, the ability to sell more rice as a result of reduced home consumption should not be thought of, in and of itself, as raising farm family income. Instead, it is the other way around. Rising incomes allow people to diversify their diets-more meat and dairy products, more noodles for lunch, etc., and less rice. But rice needed for social purposes is still not sold, and might even still be nearly 100 kgs/ person/year (400 minus 300, see footnote 31 above). This can be so important that Leach [1960: 52-53] saw it as a principal factor in maintenance and change in ethnic identity. Highland rice-growing peoples, more heavily dependent on swidden agriculture, could not change to the ethnic identities of lowlanders unless they were also able to adopt agricultural practices that would give them greater rice surpluses to meet social obligations expected under Theravada Buddhism.
33. Agricultural wage labor replacing exchange labor accompanied the transition process. While 11% from off-own-farm cash income in 1986 (Table 7A, line 1.5) seems a little high, it may reflect off-season agricultural work in other regions that was partly characteristic of that era (e.g., cutting sugarcane in Kanchanaburi, harvesting irrigated rice in Central Thailand).
34. The number of people in agricultural holding households who worked mainly on other farms is known but they were only a very small percent, although it has been increasing, especially among small holders (Agricultural Census data, not shown). The unknown number who do such work, but less than "mainly," is unknown but undoubtedly much higher.
35. Since the early 1990s, rice sales have been about half of total rice production: 56% in 1994, 45% in 1995, 39% in 1998, 44% in 2001, 50% in 2002 and 54% in 2004 (production data from Appendix Table 4; data on rice sold available in OAE [1989-2005]. This is consistent with percent of rice sold in case study data (49% in 2000 from Pholthanee et al. [2002: 106]; 48% in 1991-92 from Nakada [1996: 616]), although there is a great deal of local variation. For the whole of Yasothon Province, for example, rice sold as a percent of rice produced in the same year was over 80% in 1995 and 40% in 2001 (same OAE source). Farmers cope with variation in rainfall, for example, by waiting to see how the next crop is doing before deciding to sell the excess glutinous rice being stored for subsistence purposes [Nakada 1996]. However, some of the rice produced is seemingly still unaccounted for, which could mean higher total income from rice. If the data on rice consumption (footnote 31) are calculated out, even assuming no reduction in rice used not for direct consumption purposes, by the early 2000s there should still be about 70% left for sale, more than the 50-60% sold. If the data are substantially correct, what happened to the "missing rice"? Some of it was probably bartered for other goods, a common traditional practice still used in the village [e.g., Nakada 1996: 624-625] and often not adequately accounted for by survey methods. It is also highly likely rice is being taken away by absent relatives not defined as household members by survey definitions. There are many informal reports from people going back on a short visit who "fill the vehicle" with sacks of milled rice for the return trip. Over time, this rice may be directly consumed (and used for social purposes). As one long-time Bangkok taxi driver from the Northeast said, "I have never had to buy rice." But much of it was surely also cooked and sold in value-added form, especially with somtam and kai yang ("papaya pok pok" and grilled chicken) [cf., Funahashi 1996: 627, 634-635]. It is highly likely that the right to take rice away in this manner, for whatever use, is one of the prime features of "joint investment" as discussed in this report.It might be argued that since most rice sold was non-glutinous while most rice not sold was glutinous, that the additional in-kind income could not have matched the income from rice sales. Non-glutinous has had a higher unit baht value, but from 1990 onwards, glutinous price was above 80% of non-glutinous in most years, similar enough for the in-kind income to nearly match the average income from sales, even if all the rice sold had been non-glutinous, which was never the case.
36. We can agree with the farmers in the ACIS 2003 survey that the majority of overall (cash and in-kind) net income (not just gross income) of agricultural holding households came mainly from farming, with rice playing a central role, if we agree to a few assumptions, using the information in Tables 7A and 7B. First, assume the net rice income from sales in 2004 was equal to about one third of net cash income in Table 7B (estimated from the trend of percentages under line 1.1 in Table 7A). Then, if about half the rice was not sold (footnote 35), net agricultural income would increase by about a third due to in-kind income from rice contribution-raising net income to about 27,000 baht (in constant 1998 baht), without having to deduct any additional expenses because all agricultural expenses have already been deducted. Next assume that, while rice was the main source of in-kind income, many others (fish, poultry, vegetables, both natural and farmed, and occasionally pigs and cattle, etc.) also contributed, if only half as much as rice. That would raise the total to 30,000 baht (from 27,000). Next, note that the net agricultural cash income came to 43% of the agricultural cash income (20,285/47,363), meaning agricultural expenses came to 57%. It is very likely that all business expenses, not just agricultural expenses, rose over the years, with rising incomes. Cash expenses incurred for non-agricultural work, usually away from home (transportation, housing, prepared food, etc.) should be high as the farmer usually has to shift temporarily from a semi-subsistence rural home environment into a much more monetized urban economy. If we assume those expenses were 50% of wage earnings (lower than agriculture's 57% of gross cash sales), net non-agricultural income would be about 25,000 baht (1/2 of 50,253), which is less than the 30,000 baht estimated for agriculture, and less than the 27,000 baht estimated from including only rice as in-kind income. Two further considerations are relevant. First, the tendency for under-reporting income and over-reporting expenditure in survey data of this type is commonly known. Therefore, real net agricultural income is likely to be greater than computed in this report. Second, trends in gross agricultural income are also indicative of net agricultural income, because of assumed economically rational behavior of the farmers themselves. They would not have kept making these agricultural investments unless net agricultural income had remained positive, or kept increasing them without a promising trend.
37. Ruaysoongnern and Suphanchaimart [2001: 70-71] proposed the expansion in planted area of non-glutinous rice for sales purpose in the 1990s as the "beginning of commercial farming" of rice in the Northeast. In a case study of a village in Khon Kaen, Vityakon etal./ [2004: 458, 460] characterized the changing rice varieties and increased sale of rice during the period 1987-94 as a new substage in increasing commercialization of farming in the Northeast. Others have studied related aspects. Miyagawa [1996] studied the spread of KDML105 as a cash crop in the early 1990s (discussed below). Nakada [1996] studied dynamics of rice consumption and sales. The quantified, region-wide analysis in this subsection of the present report concerning income findings reveals the surprisingly large and growing role of rice as being the main cash crop (fieldcrops such as cassava having long been the main cash crops for most Northeasterners). Also unexpected is agricultural income keeping pace with non-agricultural, connected to the unexpected persistence of the small family farm. Agricultural growth is inherently slower than manufacturing, and much more so for rainfed agriculture on small family farms. The authors do not claim that these trends will or can continue (discussed at the end of the report). 38) In Table 8, note that the number of agricultural household planting rice only for sale was and has always been negligible (less than 1%). Author fieldnotes in the early 1980s stressed farmers' own family subsistence needs as their overwhelmingly main priority in growing rice. Fukui [1991: 528]: by the early 1980s, "rice has never become a commercial crop in this village." Rigg [1986:
38. 38] in the two villages studied in Mahasarakham: "only one fifth of the households interviewed marketed (or were intending to market) any of their [rice] harvest in the 1982/83 crop year." As further evidence of how important Northeastern farmers thought rice was for direct subsistence purposes, Thomas [1988: 88, 98, 101] found that farmers saw a simple trade-off between [non-rice] cash crops and off-farm work (they/ simply picked the one they thought paid best at the time), but rice for subsistence purpose was so important that "no recorded opportunity costs are sufficient to reduce allocations of land and labor to rice."
39. As will be seen below, recent changes in agriculture allow people to be away more of the year. Peak periods for being away are in the dry season in January through April but also after planting and before harvest during the wet season (in September and October) [Miyagawa 1995: 196; Nakada 1996: 620 Fig. 4; Limpinuntana 2008].
40. The Northern Region had the next biggest difference, with 0.46 million people displaced from agriculture in the dry season, or 13% of the 3.42 million in the wet season [LFS 2006: Rounds 1 and 3].
41. In one village in Yasothon in 1992, half the people in the village (52%) were out during the off-season [Nakada 1995: 523], but this might be higher than average now. As will be discussed below, 1992 was in a transitional period for the Northeast when agricultural labor was particularly needed at key times, which may have led to an unusually large number coming back to the village during the wet season. Also, rice farming in this village was relatively productive, perhaps making seasonal absence more likely than permanent migration [cf., Fukui 1996: 691]. But other income-earning opportunities were not within commuting distance [Fukui 1996: 690], so more people might have left, or stayed away longer than if there were more off-farm opportunities closer at hand.
42. In Fig. 2 (and Fig. 3), the high amount of total "agricultural" land compared to land actually used for various agricultural purposes should not be taken to mean agricultural land is vastly underutilized. For whatever reasons (e.g., poor soils), this category apparently includes large parcels of land not suitable for agricultural purposes.
43. For some reason (possible data revision?), Fig. 2 shows a slight increase in paddy land between 1990 and 1991, while Fig. 4A shows a slight decrease in area planted to rice in that year. This slight increase in Fig. 2 is much less than the series of increases in Fig. Ah/beginning after 1987 (but comparisons over time prior to the period of increased stabilization are difficult-see footnote 47 below). In any case, if both tables are substantially correct, we can safely say that the increasing trend in rice area planted in Fig. 4A was not due to further expansion of paddy land.
44. The percentages used to compare recent times with the early 1980s in this paragraph are conservative estimates. If production data for 1981 and 2005 are used [points which are respectively just barely below and above the linear trend line for Fig. 4B], the production increase is 94% [Limpinuntana 2008]. If the trend line itself is used, production gain between 1980 and 2006 would be about 87%. However, it is not clear exactly what this means in the Northeastern context. Before the late 1980s, rice area planted and production in the Northeast were so variable from year to year that even averages over many years can be misleading [cf., Fukui 1993: 191ff]. For this report, we averaged two different periods, 1980- 87 and 2003-06, and did a simple form of sensitivity analysis on the more variable earlier period as well, recalculating excluding 1987, a low production year (in order not to overestimate the increase), and again excluding 1983, a high production year (in order not to underestimate). Average wet season rice area planted in 1980-87 was 4.52 M ha, or 4.57 excluding 1987 or 4.48 excluding 1983, vs. 5.24 in 2003-06, a gain of 16%, 15% or 17% respectively. Similarly, average early wet season production was 6.24, 6.32 or 6.07 M tons, vs. 10.21 M tons in 2003-06, a gain of 64%, 62% or 69%. For the yields, in the early period it was 1.38, 1.38 or 1.35 tons/ha, vs. 1.94 in 2003-06, a gain of 42%, 42% or 45%. (Wet season data are used throughout this report because we are focusing on rainfed agriculture-dry season rice area planted in the Northeast has always been less than 3% of the wet season, at most; production always less than 5%, at most.)
45. This should not be taken to mean that variable rainfall is no longer damaging (especially floods). Figs. 5A-C contain regionally aggregated data, obscuring local variation, of which there is still a great deal, affecting local production from year to year (e.g., see footnote 35 above).
46. This has seemingly been true through 2006, but may not necessarily be true thereafter (discussed below). Data more recent than 2006 was available (including farmgate price data shown in Appendix Table 9) but we have not used it in this report. Very recent government data like this is usually forecasted or preliminary and subject to revision (N. B.).
47. "Rainfall distribution is generally bimodal. The rainy season commences in mid-May with the southwest monsoon coming from the Indian Ocean on the western side of the peninsula. The monsoon trough gradually moves north-It leaves a dry spell of 2-3 wk before it turns back, moving from north to south from August to the beginning of October" [Pushpavesa et al. 1986: 167]. Photoperiod-sensitive rice varieties are particularly adapted to tropical monsoon rainfed conditions: "among the photoperiod sensitive varieties, the lower the latitude of distribution, the higher the sensitivity." [Vergara and Chang 1985:18]. The principal advantage of photoperiod sensitivity is that it "ensures that the rice crop will flower near the end of the monsoon" [Pushpavesa and Jackson 1979: 139]. Among other advantages: photoperiod-sensitive rice "can be left in the seedbeds for prolonged periods. without serious damage," permitting transplanting when water is "closest to optimum. [and] land preparation and transplanting can be staggered." [Pushpavesa and Jackson 1979: 139; also see Mackill et al. 1996: 8].
48. RD6 and KDML105 have increased production by yielding better per area planted but also because they are better adapted to field conditions, allowing more paddy land to be used more of the time. Per area planted, yield of wet season glutinous rice in the Northeast in 2006 was 29% higher than in 1984-85, 39-43% higher than yields in 1986-88, and 27% and less for most of the years thereafter (data for Fig. 5C). We know that by 1989 RD6 accounted for most of the glutinous rice grown, so the data are consistent with RD6 providing a better yield, but variety-specific data are lacking and in any case it is difficult to say just how much better-yielding it was than varieties it replaced, since yields varied so greatly from year to year in the previous era. Under controlled conditions from three different tests in the Rice Department (in Thai language at DOA [2008]), RD6 had yield advantage of 17%, 27% and 29% over NSPT (Niaw Sanpahtawng), the most popular and reportedly best-yielding glutinous rice in the Northeast prior to RD6. Under test station controlled conditions, yields are almost always higher than farmers can expect in their fields, but relative differences among varieties ought to be more realistic. Since RD6 does better on problem soils, the two higher percentages seem more likely to represent the Northeast, and might be too low, depending on the soils used in the tests. But RD6 didn't replace only the better-yielding NSPT, which was planted only in the lower paddy areas with the better soils. It also replaced other poorer-yielding glutinous varieties, so the yield advantage over them would be higher still. Under controlled conditions, RD6 yields higher than KDML105, but as Fig. 6C shows, this is apparently not true in the farmer fields of the Northeast. In sum, it would seem that the 40% increase in yields of rice noted in the data for Fig. 4 can be attributed to the spread of these two varieties. In the Northeast today, RD6 does not yield appreciably better than "native" varieties still in use, but that is probably because in the few areas where those varieties were retained, they always yielded comparatively well, thus they were not displaced (e.g., in areas near the Mekong River with better soil and more reliable rainfall).
49. RD6 qualities from authors' experience, villager interviews, DOA [2008] and Miyagawa [1995: 192; 2003: 45]. Despite its phenomenal success, the authors believe not everyone thinks RD6 tastes better than traditional varieties, and having only one, uniform taste all the time can make people wish to eat a different variety at least occasionally.
50. KDML105 and RD15 together are often referred to as KDML but are officially designated as "Thai Horn Mali Rice" by the Ministry of Commerce in the Government Gazette (volume 118, special part 109d dated 2 November B.E. 2544) (A. D. 2001). This name and its trademark logo are registered with the Department of Intellectual Properties in Thailand and in the approximately 50 countries around the world to which these two varieties are exported.
51. Since the Khon Kaen Rice Testing Station was a major tester of RD6, villagers nearby were probably among the earlier adopters. In research training the authors conducted in several villages in Phra Yuen District, Khon Kaen Province in 1990, villagers claimed a total switch to RD6 from all other glutinous rice varieties had already been completed by that time.
52. In 221 villages contacted by Miyagawa where glutinous rice was grown in the early 1990s, 90% grew RD6 and 30% grew no other glutinous rice crop at that time [Miyagawa 1995
53. 191]. So 60% of villages must have been growing some other glutinous varieties somewhere in the village, but from the government statistics, it should have been on only a small percentage of the rice land.
54. The price of glutinous rice rose unusually in 1997 and 1998 (Appendix Table 9), presumably because of the economic crisis of 1997 which forced so many people to return home. Presumably more glutinous rice was planted and less was sold for that reason, in order to be able to feed the returnees.
55. In a non-random sample survey around the Northeast in the early 1990s, Miyagawa [1996: 554] found "KDML" (not all KDML105, but surely mostly) was the main non-glutinous variety in 79.9% of the villages, the only non-glutinous variety in 32.4% of them, and only 6.9% of them did not grow any KDML at all.
56. In North and Northeast Thailand, the recommended harvest date for KDML105 is November 15 and for RD6 November 21 [DOA 2008].
57. At the time these changes took place in the late 1980s and early 1990s, switching to planting only KDML105 and buying glutinous rice for consumption would have been seen as far too risky. However, this is not to say that more glutinous rice farmers could not later adopt such a strategy (but if glutinous rice becomes more in demand on the market, the price would rise and more of it would be grown for the purpose of sale). Percent of Northeastern agricultural holding households planting only non-glutinous rice for the main, wet-season crop increased from 22% in 1983 to 25% in 1988, 26% in 1993 and 1998 to 28% in 2003 [computed from ACIS Tables 3.1 (1983), 7.1 (1988), 5.2 (1993), 5.2 (1998), 6.2 (2003)]. This gradual expansion would logically have started in the southernmost areas where non-glutinous rice is the main staple, but more recently spread up into the Tung Kula area among some of the glutinous rice eaters as well. In any case, these changing percentages are apparently not a differential demographic phenomenon, i. e., if the non-glutinous rice eating population had expanded relatively faster. The village population in the four provinces where non- glutinous rice eaters are most concentrated (Nakhon Ratchasima, Buriram, Surin and Sisaket) has remained at about 33% of regional village population over the years [from PHC 1980; 1990; and 2000].
58. KDML105 also probably benefited from the later reduction of area planted to RD6, not just the earlier reduction of area planted to or reserved for former glutinous rice varieties. The downward slope of the linear trend line of a scattergram of area planted to RD6 and KDML105 for the period 1989-2005 indicates an inverse relationship (using data in Appendix Table 6).
59. Former Rockefeller Foundation/IRRI plant breeder (with the Thai Government Rice Division for many years) Ben R. Jackson, PhD (personal communication, June 7, 2005): "I do not consider either of these varieties to be HYV because if large amounts of fertilizer are used they become leafy and tall and do not produce the amount of extra yield. However, if a modest amount of fertilizer is applied they do very well in the impoverished NE soils." See also Naklang [1997: 131] and Nakamura and Matoh [1996: 586]. RD6 and KDML105 respond
60. similarly and reportedly received equal fertilizer when applied (Fig. 6C, author interviews, also Wijnhoud et al. [2003: 124]).
61. It is unclear why chemical fertilizer costs should increase more in the late 1980s but usage increase more in the early 1990s. But if all the data in both Table 10A line 2 and Table 10B line 1 are correct and represent the same number of holding households in any given year, then the cost per kilogram of chemical fertilizer increased between around 1986 to 1991 but then decreased between 1991 and 1995, so this might help explain the seeming discrepancy.
62. More people were working in jobs where it was difficult to get off work, especially on short notice, and this was probably the main reason for sending money instead. But it might be argued that the concentrated timing of the harvest period, known in advance, could actually have improved the chances of getting a brief leave from non-farm work.
63. In Table 10A it seems likely that some of the expenses for hiring a mechanical plow and its operator were reported as hired labor (in line 1) instead of as equipment rental (subsumed in line 4).
64. As a woman in Sakon Nakhon said, "My kids work overtime in the factory so I work overtime, too" [making a joke of it, using the slang term "tham o" - "o" for overtime]. In a village in Khon Kaen, even a relatively well-off headman and his wife said they did part- time wage labor on their neighbors' farms. According to villagers in Roi Et, it is considered doing a favor for a neighbor so they were paid a little more than an outsider (e.g., 160 instead of 140 baht/day). Trucks also come from other villages not yet finished harvesting, looking to hire people already finished with their own harvesting.
65. Rainfed rice fields are seasonally fallowed (or planted to other crops), unlike many irrigated fields planted to rice throughout the year. Pests are therefore usually less of a problem. Additionally, in many of the mini-watersheds, local water impoundment (fish ponds) at the bottom of the slope would be polluted by run-off from the fields above. The rice paddies themselves are also an important source of many natural foods [Somnasang et al. 1988], such as fish, frogs, crabs and snails, which could be harmed or contaminated.
66. Northeastern farm families stored rice from a good harvest to make up for the bad years in between, but this was not an expendable surplus.
67. This effect might logically have occurred earlier in villages close to major cities, like Don Daeng, which Miyagawa studied, where nearby off-farm work was already available.
68. The timing and details of a technological change (not reflected in the data in Tables 10B-C) would be worth looking into: the changes in the transmission gearing for the two-wheel tractor. It may have allowed more women to operate it and perhaps greater use for local hauling and transportation, easier and cheaper than having to remount the engine in an itaen farm truck.
69. With respect to productivity per labor at various times, data for Figs. 7A-C are based on overall agricultural labor, not rice labor (data not available at regional level), and the "seasonal" labor force data are based on a weekly snapshot, which included people who came back only for a short period. If the rice labor figures had been used, the production per person-hour might have been higher than implied in Fig. 7C, and the rise in later years might have been relatively higher compared to the initial changeover when more people came back to help transplant and harvest. With respect to the influence of the price of rice, the index trend in rice prices from late 1980s through the 1990s may have been slightly positive but with wide fluctuations, while wage rate kept rising [see Isvilanonda 2002: 218 Fig. 1]. Constant farmgate price of KDML seems to have been rising slightly over time, exhibiting a moderately increasing trend (Appendix Table 9). It did rise in 1988 and 1989 but went consistently down thereafter until 1996 when it started on another upward trend for a few years, slightly passing the 1989 level in 1997, 1998 and 2004. The rice premium (export tax) was eliminated in 1986 and international exports of KDML apparently were first recorded in 1988 [Rice Exporters Association 2008], but it is unlikely that had a major effect on farmgate prices at that time. The price of glutinous rice also followed a similar pattern in those years, with negligible exports. Rice production suffered in 1987 (e.g., Figs. 5A-C), so that may have been the cause of higher prices for a year or two thereafter. The fact that KDML105 is a higher-value rice that seems to achieve its best quality when grown in the Northeast has helped make up for the still relatively lower yields inherent under rainfed conditions in the region. But that was nothing new in the 1980s-fragrant rice had long been higher-value, and KDML105 itself was released in 1959. And RD6 was widely expanded primarily for the purpose of subsistence consumption, rather than sale (as discussed in the current report). In sum, it is very unlikely that farmgate prices could have been a major cause of a massive changeover to RD6 and KDML105 in the late 1980s.
70. Villagers in one village in Khon Kaen confirmed smaller holdings achieved higher rice production perrai.
71. Farm family income from remittances was smaller in the early 1980s (Table 7A, line 2.1), but at that time most people stayed at home on the farm most of the year, so most non-farm earnings were probably classified as off-farm income, rather than as remittances from people living elsewhere.
72. A growth rate of 12% is unusually high for net cash agricultural income, but that is because it is starting from a very low base, presumably because of the sudden start-up of rice sales. Agriculture inherently cannot sustain such a high growth rate [Mellor 1995a: 5-6].
73. Maruyama [1963: 166] argued that the "initial conditions" for many types of positive feedback processes may actually be trivial, such as whatever initially caused the particular location of a city on a homogeneous plain, and only the "deviation-amplifying mutual positive feedback networks" could explain why the city grew. More recently, economists concerned with geographical factors have made similar points [e.g., Krugman 1993b: 131]. These points are not disputed here. But the arrival of RD6 was not trivial. It became one of the key factors within the positive feedback process. Its capabilities in this particular environment were unmatched by any other glutinous photoperiod-sensitive variety available at the time [cf., Rigg 1995]. This is also not to claim that environmental constraints are always the main constraints on resource practices.
74. Krugman cites McCarty for coining this relational dichotomy (McCarty, HH. 1940. The Geographic Basis of American Life. Westport, CT: Greenwood Press). There is much literature on the subject of spatial concentration(s) of industrialization within countries [e.g., Krugman 1999], including in Thailand, but it is beyond the scope of the current report.
75. We have chosen not to expand our discussion in this report into the literature debating various aspects of "virtuous cycles" between agricultural and non-agricultural economic growth, which would require a much more far-ranging discussion. In Northeast Thailand, the relationships we discuss seem fairly straightforward, and conform closely to Mellor's views except where noted.
76. Village household expenditure on manufactured products was calculated from SES 1986- 2004 Statistical Table 2 and deliberately underestimated by excluding a wide range of expenditures indirectly involving purchase of manufactured goods (school fees, medical costs, transportation costs, etc.). The relatively minor ceremonial category (weddings, etc.) was included, but discounted by 50%.
77. While Northeastern rice production per hectare is relatively low, production per person is not as low, because of larger average holding size. For an early example of doubting the suitability and sustainability of rice cultivation on the poor, largely unirrigated soils of Northeast Thailand, see Long [1966: 356-357]. Lowland rice can be grown even on poor soils because of its "semi-aquatic nature" [Grist 1953: 11; Pendleton 1939: 43; 1943: 27]. In Northeast Thailand in earlier times, Pendleton observed that additional nutrients came from livestock droppings, run-off from adjacent forest (especially after burning the understory), from fallowed [upper] paddies where livestock graze, from litterfall from trees in paddies [also Vityakon 2001], and from run-off of nutrients from livestock manure [etc.] from the village area above the paddies [Pendleton 1943: 27, 30, 22; 1962: 72-74, bracketed comments added]. (Termite mounds might be another direct or indirect source [see Pendleton 1939: 43; 1942].) Wet season crops suffer from reduced sunlight during the monsoon but this is partly compensated for by photoperiod-sensitive crops that mature after the clouds lift [Farmer 1979: 309], and the Northeast gets a relatively great deal of sunlight [Pendleton 1962: 128], even during the peak monsoon period.
78. There is another important difference between what has happened in Northeast Thailand and what seemingly occurred in some other Asian settings such as Taiwan. Despite the increasingly high-quality major road networks, transportation time and costs to other regions still limit diversification into higher-value agricultural activities, and there is little evidence of small local enterprises growing into major industrial contributors [cf., Mellor 1995b: 321-322]. This does not seem surprising in locations so distant from a country's major region of industrial and urban growth [e. g., Krugman 1999].
79. In any case, the coupling of seasonal farm and non-farm work should probably not be seen uniquely as "a defining feature of late twentieth century capitalism" [Hart 1996: 269]. The world's first industrial revolution in England exhibited a similar pattern lasting for many years, apparently until grain prices fell due to imports [Sokoloff and Dollar 1997].
80. Most could sell rice, but even those very small farms who could not sell benefited if they did not have to buy rice, or had to buy less rice (farm size discussed above under "Differences by Size of Land Holdings").
81. The effect of seasonal absence contributing to more rice available for sale has not been addressed in this report [cf., Nakada 1995]. While labor is needed during rice growing, much less labor is needed at other times and people who leave the household for extended periods leave more rice behind for others to eat in their absence. Historically, this may have saved lives [see Fukui 1993: 145-146]. It may be that at some times in some places (e.g., with fewer nearby non-agricultural employment opportunities), seasonal migration has been so large as to be a major cause of rice surplus [see Nakada 1995; also Funahashi 1996; Fukui 1996]. But overall, the effect of increased production as discussed in this report is likely to be a larger cause. For example, in 2003 even if half the roughly 11 million farm people were away 10 out of 12 months, at 200 kg/person/year, that would be a savings of 0.9 million tons, less than 10% of the 10.2 million tons produced in that year (data for Fig. 4B in Appendix Table 4), much less than the roughly 60% production gain discussed in this report. It should also be noted that any increase in rice sales because of seasonal absence would not be reflected in overall household income. That is because, in government data, absent people are counted as either household members or not for the survey purpose. If they were not members, their contributions were via remittances as discussed in this report. If they were members, their income and expenses were already included. Whatever additional income the household earned from their absence came from their earnings. If they were not able to take rice with them, they probably had to pay more for their food than the farmgate value of home-produced food they left behind.
82. For description of the particular type of rainfed rice adaptation in Northeast Thailand around the 1970s and early 1980s, see Ng [1970: 32-33]; Demaine and Dixon [1972: 52ff]; Dixon [1978: 7]; Limpinuntana et al. [1982: 67, 103ff]; Rigg [1985]; Grandstaff and Grandstaff [1987: 56- 58]; Grandstaff [1988]; Thomas [1988]; Fukui [1991: 527-528; 1993: 181, 291, 308-309, 389 note 19, 401 notes 13-14]; Rambo [1991]; and Vityakon et al. [2004]. Zimmerman [1999: 150] writing in 1931 may have been the first to mention varieties with different maturations matched to "level of land and the supply of water" in the Northeast.
83. See also Takaya and Tomosugi [1972], who describe an area in Buriram where paddies on lower, treeless valley bottoms were "established before 1930" whereas paddies with trees in them on valley slopes were "recently opened" when they surveyed the area in 1972. This is consistent with Janlekha [1968: 30, 41-42] in an area in Nakhon Ratchasima in the 1960s where land was not yet scarce and forest was still being cleared for paddies [cf., Falkus 1991: 64]. Zimmerman [1937: 387] asserted that "practically all agriculture [in "Northeastern Siam" c. 1930] is irrigated." [also cited in Vityakon et al. 2004]. Certainly much of the Northeast was very sparsely settled until very recently but there is still doubt about the specific cultivation practices in various places and periods. Boontawee [1980], in an autobiographical novel set in Yasothon around the time of his childhood in the 1930s (he was born in 1928), described rice shortages forcing people to undertake an oxcart trip to the Chi River in order to catch fish to exchange for rice, and implied this kind of trip was common. So irrigation if used was inadequate, and Miyagawa [1996: 570] says that in Don Daeng in Khon Kaen "old villagers" remembered that in their youth people had to take fish to other villages to exchange for rice because of flood damage. Pendleton was more qualified than Zimmerman to assess land use and probably more widely traveled in the region in the 1930s. He warned readers that Zimmerman and Andrews "failed adequately to interpret the observed land-use conditions and agricultural circumstances" [Pendleton 1943: 20 note 20; see also 1939: 43-44]. He also noticed the later expansion into uplands in Nakhon Ratchasima [1962: 149], but said rice agriculture throughout the whole region was primarily unirrigated [1962: 139, 148]. He observed that "not all the paddy land is planted every year; a considerable part is left fallow." [Pendleton 1943: 27]. He noted the fallowed [upper] paddies which helped fertilize other paddies (footnote 75 above), and noted that trees in paddies were also widespread at that time [ibid.: 22, also 27]. In any case, it is logical that rice agriculture would still have become more elaborated and less productive per labor under greater population pressure by the 1970s.
84. In the old days, farmers feared that the penalty for guessing wrong in doing something else instead of planting rice would be starvation. Clearing land on the frontier was a way to keep an existing adaptation going in a new place, but many people, even now, would prefer to stay in their original village, where they already have valuable place-related investment and social relationships, if that were a viable alternative. And even today, choosing to abandon family rice-growing entirely is a very serious decision. There are "sunk costs and associated irreversibilities" [Coxhead and Plangpraphan 1998: 11-12] or filters that inhibit restarting later, for example, loss of technical knowledge, deterioration of infrastructure ("landesque capital") especially paddy floors and dykes which took many years to build, and as we have heard many times, "I've been away too long-I can no longer endure that level of physical labor" (or Subhadhira et al. [2004: 57]: "used to a Bangkok lifestyle"). Another example: Once the space under the houses (tai thun) was enclosed, raising water buffaloes would be much more difficult [Simaraks et al. 2003: 328; Funahashi 1996: 631]. Some villagers also mentioned people not wanting animal droppings on now-paved village roads and not wanting smells (and insects) from keeping big livestock so nearby (author fieldnotes).
85. Marginal return (marginal product) is the additional output (e.g., kgs of rice) that comes from adding a unit of a particular input (labor, land, fertilizer, etc.) while all other inputs remain unchanged. When marginal return to labor declines, people ought to become more interested to look for new technology, new land, or alternative forms of employment. If nothing else changes, at the theoretical point where marginal return to labor becomes zero, the local situation becomes hopeless-no amount of additional labor can get any additional product.
86. In matching varieties to micro-niches, the "rules" were straightforward. In Don Daeng in Khon Kaen, long, medium and short duration varieties were for lower, middle and upper paddies, respectively, but this was influenced by the yearly weather and land types within the holding [Fukui 1993: 179-181]. When these are taken into account, the actual planting patterns were highly logical. This is supported by data from the same era elsewhere in the Northeast as well. In unpublished analysis for a presentation at the NERAD Farm Systems Workshop held at NEROAC January 25-28, 1982, the first author of the current report and Mr. Nusit Chindarsi and colleagues analyzed data gathered by Mr. Nusit in 1981 in Nong Paen Village, Na Muang Subdistrict, Selaphum District, Roi Et Province. Of the 547 paddy plots in this village 25% appeared to be "illogical" in terms of the micro-niche rules. However, a simple set of other rules involving landholding size (small, medium, large) and composition of paddy plots within the holding (lower, middle, upper), could account for farmer choices of types of varieties planted (glutinous, non-glutinous; short, medium or long duration) on 513 (94%) of the paddy plots within the village. Key to the analysis was household priority on sufficient glutinous rice to meet subsistence needs.
87. Apparently even shifting cultivation systems can be elaborated in this manner, planting and harvesting different species at different places and times, increasing weeding, etc. [Phengchanh 2007: 96, passim], or combined with small wet-rice paddies (e.g., Karen in North Thailand) and other innovations that can support increasing populations [Brookfield 1984: 24, 36-37; cf., Geertz 1963: 25ff]. It seems possible that tropical and subtropical agroecosystems in general, with their more plentiful rainfall and sunlight, allow a greater degree of agroecosystem elaboration than in temperate zones, like the natural ecosystems in these zones. On varying terrain and with some seasonality, the number of potential elaborations would logically be further increased. (Grist [1953: 8] noted that traditional rice was absent from equatorial zones because of insufficient seasonality.) In Northeast Thailand, even the rice paddies alone are a "multiculture of flora and fauna" [Heckman 1979: 204] and the main source of natural foods for most villagers [Somnasang et al. 1988].
88. Brookfield [1984: 18, Fig. 1] also neatly summarized Geertz's insights in the same diagram, with wet rice producing a greater marginal output before tailing down at a higher labor input, compared to swidden rice (but both consistent with Ricardian diminishing returns).
89. This continuing need for variety improvement of rainfed rice must certainly include rainfed glutinous rice. It is consumed as the main staple by about half the population of Thailand (people in or from the North and the Northeast) and is increasingly being consumed by others as well as diets diversify throughout the world. It is also being increasingly used as an ingredient for many manufactured products by the world food industry.
90. According to data supplied by email from OAE (October 10, 2007), rainfed rice production in Thailand in 2006 was 12.92 million tons irrigated and 16.67 million tons non-irrigated. But the breakdown by season shows 1.41 million tons non-irrigated rice (on 0.35 million hectares) in the dry season, which could only occur if the rice was in fact irrigated (the yield of 4.0 tons/ha also suggests Green Revolution photoperiod-insensitive high-yielding varieties).This is because "irrigated" rice in this data means only rice grown on land within a recognized irrigation system command area. If that production were re-assigned, irrigated would be 14.33 and non-irrigated 15.26 million tons. However, if that same amount of apparently irrigated dry-season land were also considered to be irrigated in the wet season as well, then irrigated would be 15.74 and non-irrigated 13.85 million tons. In any case, there would still be an enormous amount of mainly rainfed rice production in Thailand-certainly not marginal in any sense.
91. Another possible "time bomb" forcing sale of agricultural land might be household debt. In our judgment, and based on available data, this issue requires separate treatment and could not be assessed in the present report.
92. "There comes a stage in agriculture's development process where land has to be consolidated and farm size has to increase in order to take advantage of economies of scale" [Kay 2002: 1084].
93. Worldwide issues in the increasing privatization of grain seed production and patenting of plant genetic material are beyond the scope of the present report, but they may affect the welfare of billions of people, positively, negatively or both (e. g., see http://grain.org/brief- ings/?id= 35).
94. In addition to locally initiated improvements, there have been many recent reports concerning the expanding number of foreign men (mostly Europeans and Japanese) marrying women in Isan villages, building houses and setting up internet communications, etc. It is difficult to anticipate what longer-term effects this could have, but it is another indication that Northeastern villages have become attractive residential settings.
95. Perhaps the Northeast was never meant to sustain a large population [Pendleton 1939: 43- 44], but that could change after enough investment (see "second nature" [Krugman 1993b]). In the United States, arid Nevada has had major population increases. It will presumably depend on many things in addition to the natural environment-Thailand's economic conditions (including wage rates), investment capabilities and goals, existing infrastructure and investments already made, range of technologies available, etc.