Over the past several decades, the global need for boosting food production has brought incremental modernization of agriculture in developing countries toward agricultural intensification with increased use of external inputs, such as improved seeds, synthetic pesticides and fertilizers, mechanization, and irrigation facilities [1
] as well as traditional land expansion [4
], resulting in doubled or tripled land productivity and reduced food security concerns [6
However, the changes in cropping patterns and production practices in favor of intensive monocropping systems gave rise to outbreaks of pests, weed, and plant diseases, which were treated with increased use of external chemicals. The intensive use of pesticides poses a health threat to both farmers and consumers, as well as disruption to ecosystems [4
]. Moreover, repeated cultivation of the same crop in the same parcels results in the extraction of particular nutrients from soil, leading to intensive use of synthetic fertilizer to compensate for the nutrient loss [13
]. Therefore, intensive monocropping practices create negative externalities when these agrochemicals eventually make their way into groundwater or become airborne pollutants, affecting the surrounding environments [17
]. Moreover, use of mechanical soil tillage disrupts the natural composition of soil and causes soil erosion, compaction, nutrient runoff, groundwater pollution and eutrophication, and biodiversity loss, which is the underlying cause of ecosystem degradation [20
]. Besides, these changes contribute to the aggravation of climate change as degraded vegetation reduces its carbon storage capacity and releases greenhouse gases into the atmosphere [24
]. Deforestation and forest degradation due to farmland expansion also have adverse effects on recycling rainwater, regulation of wind speeds, and rising temperature [25
]. Crops and livestock yields are directly affected by harsh climatic events [28
While the resource base sustaining agriculture has deteriorated, a growing number of innovations have emerged for farming systems that are more sustainable environmentally, socially, and economically [31
]. For instance, conservation agriculture aims to minimize soil disturbances, maintain soil cover, and promote cropping diversity according to local conditions and limitations [34
]. Agroecological systems consist of environmentally friendly methods of farming that help improve or maintain soil fertility and protect against degradation of resources through biological and ecological processes while optimizing production [31
]. Other improved farming practices include organic agriculture, minimum use of synthetic fertilizer, crop rotation, multiple cropping, crop–livestock integration, and agroforestry, among other things [15
As is the case with many agrarian economies, Thailand’s agricultural sector has been in transition from subsistence farming to industrial agriculture [44
]. With the increasing integration of inputs and outputs into the market mechanism, farmers’ livelihoods tended to hinge on external inputs, such as machinery, fertilizer, improved seeds, pesticides, irrigation, and hired labor [45
]. The downsides of unsustainable practices are particularly pronounced in the northern highlands of the country. Much of the northern region lies in the mountainous landscape, where intensive monocropping agriculture occupies a large part of the forest areas, particularly maize (Zea mays.
L.) production for fodder markets [11
], which typically entails mechanical soil tillage [47
]. As these areas are located in the upstream of major rivers, the degradation of soil and water resources has adversely affected the livelihoods of the downstream population as well, adding to the externalities [48
To boost the resilience and adaptive capacity to environmental changes in the context of the Sustainable Development Goals (SDGs), a set of alternative farming systems encompassing organic farming, natural farming, integrated farming, New Theory farming, and agroforestry systems have been introduced to rural farmers in Thailand in various manners [49
]. One of the alternative land management systems that is being promoted by the government for the northern dry highlands is the Hill Pond Rice system (HPRS) [52
]. The HPRS follows the philosophy of sufficiency economy (SEP) [53
], where the farm field is divided into four principal sections, namely, (i) a hill area for forest plantation and conservation, (ii) a pond for rainwater collection and aquaculture, (iii) an organic paddy field for household consumption and supplementary aquaculture, and (iv) a field for cash crops [54
]. This farming system is based on precisely calculated land allocation and resource use, according to the altitude of the terrain, which maximizes rainwater harvesting for irrigating crops all year round [54
]. While the HPRS is intended for conservation purposes, the cash crop section caters to income generation for the households. In other words, the HPRS integrates water resource management, organic rice production for consumption, cash crop cultivation, aquaculture, and agroforestry, which is expected to help enhance farmers’ resilience and catalyze reforestation [50
Although concerned authorities and agencies have provided extension services and training programs in coordination with farmers in selected areas to date, quite a number of trained farmers have remained reluctant to convert their current cropping systems into the HPRS. While the extension efforts need to be focused on dissemination of the alternative farming practices, success would depend on the understanding of farm-level factors associated with the uptake of the HPRS.
Literature studied factors associated with the adoption of improved farming practices, such as organic farming [55
], integrated pest management [58
], and sustainable land management [51
]. In the highlands of developing countries, various studies analyzed factors influencing the adoption of soil and water conservation practices [59
]. Empirical evidence suggests that salient factors for the adoption of improved practices include knowledge transfer through social networks, such as extension agents and peer farmers [60
], economic implications (e.g., expected income increment and cost reduction) [68
], and farmers’ attitude toward human and environmental health [58
To the best of our knowledge, there has been limited quantitative research on the drivers of adoption of the HPRS thus far. The objective of this study was to assess the fac-tors associated with land conversion into the HPRS by analyzing both the likelihood of adoption and the extent of adoption using primary data collected from 253 farming households in four provinces of northern Thailand and the double-hurdle regression analysis. A better understanding of the farm-level determinants of conversion of current farming systems into the HPRS would contribute to improving the policy design for dissemination of alternative farming practices. Following this introductory section, the second section describes the methodology of this study, including the details of study sites, sampling framework, data collection, and statistical analyses; the third and fourth sections present the empirical results and discussions of the findings. Finally, the fifth section delivers conclusion and recommendations.
3.1. Socioeconomic Characteristics of the Farmers
and Table 6
descriptively summarize the binary variables and numerical variables, respectively, representing the characteristics of land converters and non-converters, along with the p
-values obtained from the inferential tests. The tables display the 19 explanatory variables included in the regression models and three other variables, namely household laborers, total household income, and household expenditure. Those three latter variables were excluded from the regression models because of the multicollinearity with one or more of the other independent variables based on the variance inflation factor.
The male ratio was higher among the converters than among the non-converters (p
= 0.040). The converters were 4.8 years younger than the non-converters on average (p
= 0.000), though the eldest farmer (81 years) was a converter. The average household size (number of members per household) was not significantly different (p
= 0.534) between the converters (4.3) and non-converters (4.4), while it was considerably higher than the national average (3.1) [129
]. The average number of household laborers was not significantly different (p
= 0.840) between the converters (2.6 heads) and the non-converters (2.5). Average frequency of participation in workforce exchange was higher for the converters (3.0) than for the non-converters (1.8).
The average landholding was not significantly different (p
= 0.617) between the converters (5.1 ha per household) and the non-converters (4.9), while it was larger than the national average of 4.0 ha [130
]. However, there was a significant difference (p
= 0.000) in proportion of land certificate holders between the converters (20%) and the non-converters (31%). The average household income was not significantly different either (p
= 0.208; THB 191,000 vs. 169,000/year), while it was much lower than the national average for farmers (THB 312,000/year) [129
]. The average proportion of off-farm income was significantly higher (p
= 0.002) for the converters (0.42) than for the non-converters (0.32). As per the average annual household expenditure, the converters spent less (THB 84,000) than the non-converters (THB 114,000) (p
= 0.006), which was much lower than the national average of THB 248,000 [129
]. Another indicator of wealth is asset holding. The number of household asset types was 7.7 types on average, with no significant difference between the two groups (p
= 0.875). The converters had greater precautionary savings on average (THB 43,000) than the non-converters (THB 13,000) (p
3.2. Extent of HPRS Adoption
presents the proportion of households implementing each of the 25 specific HPRS practices (recall Table 1
) per group. The most popularly implemented practices were zero burning (P9), planting edible plant species (e.g., vegetables and herbs) (P19), low height tree species (P18), underground species (e.g., root and tuber crops) (P20), medium height tree species (e.g., fruit trees) (P17), and tall tree species (P16). Seed sharing (P24), animal manure (P13), rice cultivation for consumption (P21), composting (P11), and bio-fermented fertilizer (P14) were also common in the study areas with at least a 50% adoption rate. Green manure (P12), bio-extracted pesticide (P15), and zero-synthetic chemicals were implemented at the 40% adoption rate.
On the other hand, the following practices were implemented only by the converters: Land use planning and design (P1), excavating water reservoirs (P2), excavating crooked canals (P3), building check dams (P4), excavating swamps (P5), use of soil from excavation to build an earth hill for planting multi-story tree species (P6), building water containers with local materials (e.g., bamboo, a mix of earth and cement) (P7), workforce sharing (P23), knowledge sharing for improved farming practices (P25), and planting food plant species on ‘the golden dike’ (P22). Of all the HPRS practices, P7 was the least implemented in the study areas.
illustrates the distribution extent of land converted to the HPRS as a result of the HPRS training. No farmer converted more than 80% of their farmland, and the majority of them converted less than a half of their farmland. The average proportion of land converted to the HPRS was 0.28 among the converters and 0.14 for the whole sample.
illustrates the distribution of the numbers of HPRS practices implemented out of the designated 25. The land converters implemented 15–25 practices, whilst most of the non-converters implemented in the range of 1–14. Thirteen percent of the converters implemented all 25 HPRS practices, 54% implemented 21–24 practices, and the rest (32%) implemented 15–20. Among the non-converters, 73% implemented 1–7 practices, while the remaining 27% implemented 8–23. The average number of implemented practices was 21.6 among the converters, 7.1 among the non-converters, and 14.5 for the whole sample.
3.3. Determinants of HPRS Adoption
presents the first-hurdle probit estimates of factors associated with whether the farmers were converters or non-converters. Out of the 19 independent variables, 9 showed statistically significant effects at least at the 10% significance level, of which 6 variables (i.e., farming knowledge, understanding of benefits, additional training, water sources, workforce sharing, access to advisory) had positive effects, whereas the three others (i.e., complexity of the HPRS, land tenure security, and negative shocks) had negative effects on the likelihood of land conversion to the HPRS.
Specifically, having one additional knowledge item raised the probability of land conversion by 7% on average, holding all the other variables constant. When perceived levels of complexity of the HPRS increased by 0.1 unit, the probability of land conversion decreased by 2%. When understanding of benefits from the HPRS increased by 0.1 unit, the probability of land conversion increased by 10%. Receiving one additional training session per year increased the probability of land conversion by 4%. Having one additional water source increased the probability of land conversion by 22%. Increasing the frequency of participation in workforce sharing by one more time a year led to increasing the probability of land conversion by 5%. Having land tenure security led to reducing the probability of land conversion by 14% compared with having no land tenure security. Having access to advisory resulted in raising the probability of land conversion by 8% compared with having no such access. Recent experience with negative shocks decreased the probability of land conversion by 18%.
The 11 other variables did not significantly influence farmers’ likelihood of converting farmland into the HPRS, namely gender, age, education level, household size, active member ratio, landholding, household asset holding, leadership potential, off-farm income ratio, and household savings.
presents the estimates of the factors associated with the extent of land conversion into the HPRS based on the second hurdle of the DHM and the tobit model. As per the DHM results, five variables showed statistically significant effects, of which four (i.e., farming knowledge, water sources, workforce sharing, and off-farm income ratio) were positive, whereas complexity of the HPRS had negative effects. As per the tobit results, nine variables showed significant effects, of which seven (i.e., farming knowledge, understanding of benefits, water sources, workforce sharing, off-farm income, and household savings) were positive, whilst three variables (i.e., complexity of the HPRS, land tenure security, and negative shocks) were negative.
Taking the DHM results, the quantitative interpretation of each significant variable is as follows. Having one additional knowledge item increased the proportion of converted land by five percentage points on average, holding the covariates constant. Increasing the perceived levels of complexity of the HPRS by 0.1 unit decreased the proportion of land conversion by 2 percentage points. Having one additional water source increased the proportion of land conversion by six percentage points. When the frequency of participation in workforce sharing increased by one more time a year, the proportion of land conversion increased by five percentage points. Increasing the proportion of off-farm income by 0.1 increased the proportion of land conversion by 2.5 percentage points.
presents the estimates of the factors associated with the number of HPRS practices implemented. Nine variables showed significant effects, of which seven (i.e., education level, farming knowledge, understanding of benefits, additional training, water sources, workforce sharing, off-farm income ratio) were positive, whereas two (i.e., complexity of the HPRS, land tenure security) were negative.
Specifically, one additional year of education increased the number of adopted HPRS practices by 0.3 on average, keeping the covariates unchanged. Having one additional knowledge item increased the number of adopted practices by 1.2 on average. When perceived levels of complexity of the HPRS increased by 0.1 unit, the number of adopted practices decreased by 0.4. For an increase in levels of understanding of benefits by 0.1 unit, the same number increased by 1.6. Receiving one additional training session per year increased the number by 0.6. Having one additional water source increased the number by 3.0. Increasing the frequency of participation in workforce sharing by one more time a year increased the number by 1.3. Having land tenure security led to reducing the number by 4.1 compared with having no land tenure security. When off-farm income ratio increased by 0.1, the number increased by 0.3.
The proportion of farmland converted to the HPRS was 0.28 on average and 0.80 at the maximum. Some farmers set aside the majority of their land for cash crops outside the HPRS. As for the number of adopted HPRS practices, the land converters adopted 21.6 practices on average, while the non-converters adopted 7.1. As shown in Figure 2
, land conversion was highly linked to the first principle of the HPRS, i.e., water harvesting and management. Most of the non-converters did not adopt the practices under Principle 1 (P1–P7), except excavating a water pond (P2). Due to the control over residue burning [131
], most of the farmers in the northern region adopted zero-burning practices (P9). Of the five principles of the HPRS, tree plantation (Principle 3) and rice cultivation were common in both groups. In particular, food staple species were widely adopted in home gardens. Moreover, there have been several promotions of tree planting development programs to restore northern forests in Thailand [132
Farming knowledge contributed to boosting the overall adoption of the HPRS. The result is consistent with Mishra et al. [98
] and Mutyasira et al. [62
] who argued that practical knowledge of improved practices, such as integrated farming, tree plantation, home gardening, organic fertilizer, and water regulation in paddy fields contributed to the adoption of improved farming practices. This is also consonant with Kariyasa and Dewi [135
] who explained the adoption of improved technologies imparted by the Integrated Crop Management Farmer Field School (ICM-FFS) in Indonesia. Conversely, farmers who had no knowledge and experience in improved techniques were less likely to adopt improved practices brought by extensionists [98
]. Education contributed to the adoption, which is consistent with literature showing that educated farmers had a better capacity to access information, learn, and understand the benefits of improved farming practices [95
]. As expected, perceived complexity of the HPRS was a deterrent to its adoption. The result was consistent with Rodthong et al. [75
], Waseem et al. [77
], and Zeweld et al. [43
] who found that perceived ease of implementation was a significant predictor of adoption of the range of improved agricultural practices.
Understanding of benefits of the HPRS also contributed to the adoption of and land conversion to the HPRS. It was noted during the discussions with farmers that they had stopped burning maize stover after acknowledging that it caused smog during the dry season, affecting people’s heath. Literature also shows that farmers with environmental concerns were more likely to adopt improved farming practices (e.g., crop diversification, organic farming, and soil conservation) [92
]. A premium price attracted farmers to opt in for organic farming [107
] while the growing orientation toward food safety among consumers justifies value addition through organic production [140
]. Pornpratansombat et al. [107
], Timprasert et al. [58
], and Farrar et al. [106
] found that the perception of positive effects on human health and soil conditions led to a greater likelihood of adopting organic farming and IPM for production of vegetables and perennial crops.
Receiving additional training increased the likelihood of land conversion and resulted in implementing more practices in line with the HPRS. The topics of training included the SEP-inspired subjects and improved farming practices (e.g., organic farming, integrated farming, New Theory farming, agroforestry, soil management, and water conservation). The result suggests that frequency of attendance to training enhances knowledge of and familiarity with HPRS practices. Several studies mentioned frequency of formal training as a predictor of improved farming practices adoption [101
It was found that the more diverse the water sources, the more the HPRS was adopted. This is consistent with Saiful Islam et al. [76
] who found evidence that the availability of irrigation systems influenced the adoption of an integrated rice–fish system. Mango et al. [143
] also found that access to irrigation equipment and water sources had positive effects on the adoption of climate-smart practices in Southern Africa.
As expected, participation in workforce sharing was a positive determinant of the overall adoption of the HPRS. Some of the HPRS practices require intensive labor, for example, excavating free-form ponds, digging small contour canals on higher ground, building an embankment, and building check dams. Literature also highlights the labor intensity of some of the improved farming practices, such as organic farming and crop diversification [116
]. Additionally, relevant information can be circulated when farmers gather for collective action. Our result is consistent with several studies showing that collective field activities positively influenced the implementation of improved farming practices [62
]. In Thailand, traditional agricultural labor exchange is a free-of-charge arrangement that provides immediate returns to households, which is referred to as “Aw Mue Aw Haeng”. This is a sort of collective action, which has been practiced for centuries. For instance, two male workers from Farm A go help Farm B harvest rice for one day; later, two male workers from Farm B help Farm A harvest rice or other crops for one day [145
Farmers with land tenure security were less likely to adopt the HPRS. In general, literature argues that land security is a positive factor for the adoption of improved farming practices, such as crop diversification and agroforestry [40
], which is particularly the case when the sustainable practice requires substantial investment into land resources [146
]. On the other hand, land tenure, depending on its type, may restrict land use patterns, especially a major conversion of farmland structure [147
]. Moreover, Wannasai and Shrestha [148
] found in Thailand that land insecurity could be an incentive to adopt perennial crops, such as fruit trees in a hope to acquire basic land use rights and entitlement to subsequent legal registration, which may explain our findings.
Those with access to advisory were more likely to convert farmland into the HPRS. Some farmers mentioned that advisors provided access to a digging machine and other relevant equipment. Yigezu et al. [78
] found in Syria that access to farm equipment significantly facilitated the adoption of zero tillage practices. Some other studies echoed that project advisors, extensionists, and trainers were key facilitators of agricultural technology adoption [60
Off-farm income had positive effects on the extent of land conversion and the number of HPRS practices implemented. Regular off-farm income, such as salary and remittance, would mitigate the problem of a long wait till reaping benefits in investment systems like the HPRS. Kassie [122
] stated that non-farm income was positively related to agroforestry adoption in Ethiopia and that households adopting non-farm activities had less time to take care of seasonal field crops, having shifted land into agroforestry systems. Moreover, off-farm income positively influenced the adoption of improved farming practices in the highlands of Ethiopia [62
]. However, Rodthong et al. [75
] showed negative influences of off-farm income on the adoption of prescribed sustainable practices for oil palm production.
The tobit estimation showed positive effects of household savings on the extent of land conversion to the HPRS, indicating that lack of savings can be a deterrent. In this relation, lead farmers emphasized that one of the main obstacles to HPRS adoption was the initial investment cost needed to convert farmland. Some farmers asked how long they would have to wait for the investment to reap benefits, while most of the farmers were in need of immediate income. Rodthong et al. [75
] found that accumulation of debt discouraged farmers from adopting the set of designated sustainable practices for oil palm production. The District Chief of Mae Cheam, Chiang Mai stated during the meeting that some farmers borrowed money from banks for purchasing farm inputs and later found it difficult to repay, ending up curtailing their expenditure on consumption.
The experience with negative shocks was a deterrent to land conversion to the HPRS, which was in line with literature. Climate hazards were the most agreed impediments that discouraged farmers from adopting new practices. Literature suggests that farmers hesitate to invest in expensive technologies in the presence of climate [68
]. For instance, farmers in low rainfall conditions used traditional varieties instead of improved seeds [68
]. Furthermore, Gebremariam and Tesfaye [125
] found in rural Ethiopia that production damage and health shocks exerted negative effects on the adoption of costly innovations (e.g., improved seeds and synthetic fertilizer). The non-converters in our study stated that productive land for maize would be wasted in harsh production environments.
Comparing the results from the DHM and the tobit on the extent of land conversion, four variables (i.e., understanding of benefits, land tenure security, negative shocks, and household savings) were significant in the tobit, but not in the DHM. This implies that these four factors were more relevant to the decision to convert farmland than to increasing the proportion of converted land, which was generally agreed on by the first-hurdle results. The triangulation also suggests that all the significant factors identified by the DHM were identified by the tobit model as well. Therefore, the DHM results seem robust not only due to the selection correction through the process with the inverse Mills ratio but also due to the consistency with the alternative model of tobit.
The set of 25 designated practices under the HPRS are intended for conservation of water and soil resources as well as for household consumption of organic rice. Yet not all the practices should be given equal weights, and there should be some prioritization and distinction. First, excavation of a water reservoir is one of the core practices, to which construction of crooked canals, check dams, and small swamps along the canals are complementary practices to support water transportation to agricultural plots. Earth hill building can be optional since soil from the excavated reservoirs is left over in cropland. Second, soil surface cover, such as mulching and ground cover plantation (e.g., vetiver grass, leguminous crops), is another core practice [150
], to which application of nutrients into soil is complementary whilst zero burning and zero agrochemicals are optional. Third, organic cultivation of rice and other staple crops is another core practice, which supports food and nutrition security for households. Fruit and long-life trees can be optional for increased food security. Fourth, crop rotation and diversity are another set of core practices, which contributes to soil nutrient management and suppression of pests [151
]. Other practices, such as seed conservation, labor sharing, and knowledge sharing are complementary to the core practices. On the other hand, construction of water management systems and land preparation involving tillage may disturb soil properties and affect the ecosystems in the long run [150
The Hill Pond Rice System (HPRS) has been promoted as an alternative to the intensive and unsustainable maize monocropping with mechanical soil disturbance, especially in the upland areas that have strategic head waters of Thailand. The implementation of the HPRS is expected to restore forest, soil, and water resources that have been degraded due to the intensive applications of tillage and synthetic inputs. Yet, there has been limited quantitative research on the adoption of the HPRS to date. The present article estimated the factors associated with the conversion of farmland into the HPRS, as well as the adoption of the practices designated under the HPRS by collecting primary data from 253 farm households in four provinces (Nan, Chiang Mai, Tak, and Lampang) in northern Thailand and conducting statistical analyses based on the double-hurdle and tobit techniques. While noting a few relatively minor differences, the results were largely consistent across the statistical models employed, where the positive determinants were farming knowledge, understanding of benefits of the HPRS, access to water sources, access to advisory, workforce sharing, off-farm income ratio, additional training, and household savings, whilst the negative determinants were perceived complexity of the HPRS, experiences with negative shocks, and land tenure security. These analytical results underscore the key roles of individual perception of the technologies in question as well as the availability of and access to different types of resources in the adoption of the HPRS.
The empirical findings lead to several policy implications toward the uptake of the HPRS land conversion and the associated techniques. First, there is a need to address the perception of HPRS practices by properly emphasizing the practicality, affordability, and relevance, as a number of the designated practices are indeed familiar to farmers through various previous interventions, such as composting, mulching, soil amendment, traditional water management, homestead livestock rearing, crop diversification, food preservation, and folk handicraft knowledge. Second, the labor constraints in implementation of some of the HPRS practices should be mitigated by promoting on-farm collective action centered around workforce sharing among peer farmers during the peak workload periods. That would be compatible with the small-scale farming systems and can help optimize investment options. Third, meetings between HPRS advisors and farmers should be reinforced for technical exchange and access to relevant equipment.
Lastly, this article has several limitations to be noted. First, the research was conducted in four provinces in northern Thailand. Hence, the findings may not be representative of the whole country or the Southeast Asian region. Second, not all the 25 designated HPRS practices are guaranteed to result in favorable or intended changes in the production environments and natural resource base. As this article focused on the adoption side of the HPRS, ex-post assessment of the environmental, social, and economic impacts of the HPRS was beyond the scope, which would be left to another article. Third, the HPRS is a relatively new system in Thailand. Thus, there is not enough information yet to assess the long-term perspectives of adoption behavior among farmers, such as on the design and mode of the HPRS training needed, and the potential non-adoption of the practices or reverse conversion of farmland.