Technical Efficiency and Its Determinants in Tidal Swampland Rice Farming: The Roles of Land Size and Farming Experience

2.1 Research location

The research location is in Sungai Tabuk District, Banjar Regency, South Kalimantan Province, and will be carried out from March 2025 to December 2025. Banjar Regency has an area of 4,688.50 km² with a population of more than 595,717 people. This area consists of 20 sub-districts and 290 villages. Banjar Regency is located in a wetland zone consisting of upland areas generally located in the areas of Mataraman, Simpang 4, Sambung Makmur, Pengaron, Sungai Pinang, Telaga Bauntung, Paramasan, Aranio, and Karang Intan Districts. Meanwhile, those in the Lowland zone are farmers in the areas of Sungai Tabuk, Cintapuri, West Martapura, East Martapura, Martapura, Astambul, Gambut, Beruntung Baru, Kertak Hanyar, and Aluh Aluh. The Sungai Tabuk subdistrict is a local rice-producing region in the tidal lowlands that consistently applies local wisdom in managing agricultural cultivation on its swamplands. The area of Sungai Tabuk District is 147.30 km², consisting of 21 villages. The area often experiences flooding due to tides of seawater that submerge settlements and agricultural sites. The map of the research location shown in Figure 1 is in Pematang Panjang Village, Gudang Hirang, and Sungai Pinang Lama.

2.2 Data and data collection

This study employed a multistage sampling design consisting of three stages. In the first stage, the Sungai Tabuk subdistrict was purposively selected as a local center for tidal rice cultivation. In the second stage, village quotas were determined from among the 21 villages in the subdistrict, with three representative areas—Pematang Panjang, Gudang Hirang, and Sungai Pinang Lama—being purposively selected. The third stage involved selecting respondents using simple random sampling in each of these villages to ensure that all active farmers had an equal chance of being selected. The total population of active farmers in these three villages was 850 people. Using the Slovin formula with a margin of error of 5%, a minimum sample size of 272 respondents was determined. Subsequently, this sample of 272 farmers was distributed proportionally based on the farmer population of each village. The result was 136 farmers from Pematang Panjang Village, 76 farmers from Sungai Pinang Lama Village, and 60 farmers from Gudang Hirang Village. This transparent sampling process is important to avoid errors while meeting the data requirements for Maximum Likelihood Estimation (MLE).

The research used primary and secondary data. The data collection method used structured and in-depth surveys and interviews was assisted by questionnaires according to the variables in the research objectives. Primary data were taken directly from respondents related to the technical efficiency of local rice farming and the determining factors. Secondary data were also used to provide additional and complementary information for the research, derived from books, websites, related agencies, previous research, and other literature related to the research.

The questionnaire was designed in a structured manner to meet the data analysis requirements for technical efficiency using Stochastic Frontier Analysis (SFA) and to account for the characteristics of local rice farming on tidal swampland. The questionnaire covers farmers’ socioeconomic characteristics (age, education level, farming experience, number of family members, gender), as well as information on the amount of inputs used and outputs obtained during a single growing season (land area, production, and the amount of urea, NPK fertilizer, and herbicides used, as well as labor). The final section includes questions regarding the characteristics of farming practices and cultivation techniques, as well as local knowledge related to rice farming on tidal swampland, which are described in a qualitative, descriptive manner to support the SFA analysis. These practices are used only as descriptive-qualitative supporting explanations in the Discussion section. This is done to explain the on-the-ground conditions that influence the achievement of technical efficiency in farming.

The input and output variables in the estimation of the production frontier function are calculated using the real cumulative consumption and crop yields over one production cycle in 2025. The variables for seeds (kg), NPK fertilizer (kg), urea fertilizer (kg), herbicides (liters), labor (person-days), and land area (hectares) are measured according to the amounts used in farming operations. Local rice production is measured based on the amount of paddy produced by farmers in a single growing season (kg). Data on farmer characteristics—including age, farming experience, and formal education—are measured in years, while household size is calculated based on the number of dependents. Gender is treated as a dummy variable, with a value of 1 for males and 0 for females.

Field data collection began in March 2025 and continued through December 2025, spanning 10 months. This timeline was aligned with the local rice planting season in Banjar Regency. This period included the survey, face-to-face interviews with respondents, verification of information, and the collection of supporting data from government agencies.

Data quality and accuracy were ensured through data quality control procedures. The first stage involved a pilot test with 30 farmers outside the tidal flat area. This was done to evaluate the clarity of the questionnaire’s questions and eliminate phrasing that was ambiguous or confusing to respondents. The second stage involved training enumerators to ensure a shared understanding of the research objectives and the intent behind each question in the questionnaire. Thus, in this study, face-to-face interviews with respondents were conducted by the research team with the assistance of enumerators. Enumerators are field staff who possess strong communication skills and understand research data collection techniques in the field. The next stage involved rechecking the data—specifically, synchronizing and standardizing the units of each variable—before analyzing it using SPSS and Frontier 4.1.

2.3 Data analysis

Quantitative approaches used in the research. Frontier 4.1 software was used to analyze technical efficiency using the SFA approach, while aspects of the implementation of cultivation practices were described in a descriptive manner. Before the model was estimated using SFA, all independent variables were tested for multicollinearity by analyzing the Variance Inflation Factor (VIF) values using SPSS software. In accordance with the criteria established by studies [18, 19], the model is considered free of multicollinearity if the VIF value for each independent variable is less than 10. The purpose of detecting multicollinearity is to ensure there are no strong linear relationships among the independent variables that could distort the standard errors and bias the production elasticity coefficients. Once the VIF data were obtained and the model was confirmed to meet the criteria, the model was immediately estimated using SFA. Tests for other classical assumptions, such as heteroscedasticity and autocorrelation, were not performed because SFA already addresses these issues. SFA divides the residuals into two components: random error (v_i) and technical inefficiency effects (u_i). Both components are estimated using the MLE method.

The Cobb-Douglas production function was used because its coefficients directly reflect the elasticity of production and are appropriate for the study’s sample size (N = 272). The fit of the SFA model specification was evaluated by determining the generalized likelihood ratio (LR) and gamma values. The LR test was conducted to assess the validity of the presence of a technical inefficiency effect, and the gamma parameter was analyzed to measure the proportion of error variance purely attributable to farmers’ managerial inefficiency. All physical input variables were transformed into natural logarithms (ln). This was done so that the model would follow the Cobb-Douglas production function specification, allowing the obtained parameter values (β) to be interpreted as production elasticities. The technical inefficiency variable was not transformed into a logarithmic form but was retained in its original linear form. All farmers used these inputs during a single growing season. Therefore, no farmer had zero input usage.

Analysis of the stochastic frontier production function to estimate the parameters of the Cobb-Douglas production function using MLE:

Y = β₀X₁β₁X₂β₂X₃β₃X₄β₄e^εi

Converted to linear form of natural logarithms:

Ln Y = β₀ + β₁ ln X₁ + β₂ ln X₂ + β₃ ln X₃ + β₄ ln X₄+ β₅ ln X₅ + β₆ ln X₆ + ε_i 

where,

Y = Local Rice Production (kilograms)

X₁ = Seeds (kilograms)

X₂ = NPK Fertilizer (kilograms)

X₃ = Urea Fertilizer (kilograms)

X₄ = Land area (hectares)

X₅ = Labour (man-day equivalent)

X₆ = Herbicide (liters)

β₀ = Intercept

β₁,...,β₆ = Regression Coefficient

ε_i = error term

Technical efficiency analysis uses the formula:

$T E_{\mathrm{i}}=\frac{Y_{\mathrm{i}}}{\hat{Y}_{\mathrm{i}}}=\frac{\exp \left(x_{\mathrm{i}} \beta+v_{\mathrm{i}}-u_{\mathrm{i}}\right)}{\exp \left(x_{\mathrm{i}} \beta+v_{\mathrm{i}}\right)}=\exp \left(-\mathrm{u}_{\mathrm{i}}\right)$

where,

Y_i = Actual output

Ŷᵢ = Potential output

Furthermore, the effects of technical inefficiency are analyzed with the equation:

u_i = ⸹₀+ ⸹₁Z₁+ ⸹₂Z₂ + ⸹₃Z₃ + ⸹₄Z₄ + ⸹₅D₁ + ε_i

where,

u_i = The effect of technical efficiency

⸹₀ = Constant

⸹₁,…,⸹₅ = Coefficients of the estimated parameters

Z₁ = Age of the farmer (years)

Z₂ = Farming Experience (years)

Z₃ = Farmer Education Level (years)

Z₄ = Number of family members (people)

D₁ = Gender (1 = Male; 0 = Female)

The data in Table 1 show that local rice farming on tidal lands is largely managed by experienced farmers (54%) with 19–35 years of experience. This indicates that farming experience is an important human capital factor in coping with the ecological uncertainties of tidal fluctuations. On the other hand, the data in Table 2 reveal the physical limitations of farming operations: the average farm size is only 0.94 hectares, with a harvest yield of 3,125.56 kg per season. This output remains below the national rice productivity rate due to farming constraints such as soil acidity and tidal water fluctuations.

Based on the results of the SFA analysis, it was found that the variables of land area, NPK fertilizer, and herbicides have a significant influence on production. Land area is the primary determining factor because it has the highest positive elasticity value. This means that a 1% increase in land area will increase local rice production by up to 0.9586% (assuming all other conditions remain constant). Adequate land area provides farmers with the flexibility to manage adaptive micro-irrigation systems. However, land expansion alone does not guarantee higher technical efficiency, as production performance also depends on farmers’ ability to allocate inputs effectively. Adequate land area provides farmers with the opportunity to implement adaptive traditional land management practices, such as the surjan system through the integration of planned mounds and irrigation channels. A number of previous studies have also demonstrated that controlling a larger land area has the potential to increase efficiency, as farmers have greater leeway to allocate resources and organize their land for optimal use [14, 15, 17].

The NPK fertilizer variable has a significant positive effect on local rice production. A 1% increase in NPK fertilization can boost production by up to 0.0156%. This finding underscores the importance of a balanced supply of nutrients for rice growth, particularly in tidal areas. Plants require NPK to strengthen their roots and ensure proper grain filling in local rice varieties. These findings are consistent with previous research showing that proper NPK application can increase rice productivity [20, 21]. Therefore, improvements in fertilizer management in the field should not only focus on increasing the dosage but also on optimizing the accuracy of the dosage and the timing of application to suit ecological conditions.

The use of herbicides has a significant negative impact on local rice production. A 1% increase in herbicide use reduces production by up to 0.0357%. However, we must be cautious in interpreting these results due to the potential for endogeneity bias arising from reverse causality. In reality, farmers whose fields are heavily infested with weeds tend to use more herbicides to save their crops. Thus, this negative coefficient may reflect high weed pressure rather than a direct damaging effect of the chemicals on the plants. Nevertheless, these data illustrate that weed control on vulnerable land remains a challenge because optimizing inputs does not yield a commensurate increase in output. The fact that the urea fertilizer variable had no effect indicates that marginal adjustments to the current application rate do not alter the production boundary line. Although the literature on tidal swampland often states that nitrogen uptake efficiency is low due to the structural acidity of the soil and the potential for nutrient leaching during high tide, the lack of direct soil and agronomic data in this study prevents definitive confirmation of these leaching mechanisms. Empirically, these non-significant results primarily indicate allocation inefficiencies at the farm level, confirming that simply expanding urea use within current traditional farming practices does not guarantee increased production. Other inputs, such as seeds, urea fertilizer, and labor, did not have a significant effect on production. The lack of a significant effect for the seed variable is likely due to farmers’ practice of continuously using seeds from the previous harvest, resulting in a decline in genetic quality and yield potential. Meanwhile, the lack of a significant effect for the urea fertilizer variable indicates that marginal adjustments to the current application rate do not alter the production boundary. Although the literature on tidal swampland often states that nitrogen uptake efficiency is low due to the soil’s structural acidity and the potential for nutrient leaching during high tide, this study did not directly measure soil conditions; therefore, these reasons cannot be empirically confirmed. These results confirm that the key to increasing production lies in improving management practices, not merely in increasing inputs.

Based on the SFA model, the average technical efficiency of farmers is 0.82. This means that farmers have achieved 82% of their maximum production potential with currently available technology. The remaining 18% gap reflects disparities in farmers’ management capabilities in maximizing resources. The minimum TE value is 0.23, and the maximum is 0.98. A gamma value of 99.19% confirms that fluctuations in production are caused by farmers’ technical inefficiencies, not by chance or natural disturbances (random noise). Therefore, the best strategy for improving efficiency on tidal swampland is to enhance farm management capabilities and strengthen adaptive agricultural technologies. The research findings are in accordance with the grand theory of technical efficiency [6] developed using the SFA approach [7].

The average technical efficiency value of 0.82 is consistent with previous studies. The 18% production gap indicates that farmers have not yet fully realized their production potential due to differences in management capacity, limited land area, and experience-based human capital that shapes their ability to adapt to the tidal environment. The average technical efficiency was 0.76, with one distinguishing factor influencing farmers’ managerial capabilities [1]. Farmers’ managerial skills play a crucial role in determining variations in the technical efficiency of rice farming, particularly in effectively combining available inputs [22]. Rice cultivation on tidal swamp lands in Central Kalimantan has an average technical efficiency of 0.755, where the expansion of rice-planting areas and the use of NPK fertilizer significantly increased rice yields [13]. A comparison with previous studies from several regions shows that although farmers face challenges related to high ecological vulnerability, their local ecological knowledge can help address these issues, enabling them to manage water and inputs to minimize the risk of crop yield losses.

The distribution of technical efficiency reveals a gap in capabilities among farmers. Farmers with high technical efficiency (TE ≥ 0.80) manage an average of 1.01 hectares of land and have 24 years of farming experience. In contrast, farmers with low technical efficiency (TE < 0.50) manage only 0.79 hectares of land and have 11 years of farming experience. This disparity demonstrates that farming experience is a key factor in improving technical efficiency. In line with human capital theory, years of farming experience build understanding and refine practical skills, while also sharpening farmers’ ability to make precise decisions regarding the management of production factors.

On tidal swamplands, years of farming experience have equipped farmers with the ability to determine planting schedules, manage water systems, and select cultivation techniques suited to ecological conditions. Consistent with previous research, rice productivity on tidal lands is influenced by the suitability of the technology to ecological conditions and the ability to manage the farm [23]. Conversely, limited farming experience reduces farmers’ ability to cope with the threat of ecological uncertainty, resulting in less effective input management. This mechanism helps explain how farming experience contributes to reducing technical inefficiencies in farming. However, since this study does not measure ecological adaptation quantitatively, this interpretation is presented only as a logical supporting explanation for the high technical efficiency scores of experienced farmers, rather than as a quantitative variable directly calculated in the SFA model. The results of the study are also in line with the research [13], which states that the longer the farmer's experience, the higher the achievement of technical efficiency. To achieve maximum production potential on tidal swampland, a targeted integration of experience-based farming skills and appropriate micro-water management is required [24]. In contrast to previous research findings, which suggest that experience does not always improve efficiency if it is not accompanied by the ability to adopt new technologies.

Mathematically, the sum of all production coefficients yields a value of 0.943. Although a figure below one typically indicates a trend of diminishing returns as inputs are increased (decreasing returns to scale), this descriptive indicator must be interpreted with caution. This is because some inputs, such as seeds, urea fertilizer, and labor, do not have a significant effect; therefore, this figure cannot be used as a benchmark for determining the scale of operations. Empirically, the scale of local rice production on this tidal swampland is largely determined by the dominant and significant elasticity of land area (0.9586). Thus, the figure of 0.943 is not definitive evidence of a decline in the scale of operations, but rather an indication that production growth is still constrained by limited land area.

The determinants of technical inefficiency—the farmer’s age and the number of family members—significantly reduce technical inefficiency. In contrast, educational level and gender have no effect on efficiency. The influence of age suggests that accumulated experience and maturity in decision-making may enhance farmers’ ability to manage production risks. Family size greatly aids in the availability of self-reliant labor, especially during the planting and harvesting seasons. Conversely, a high level of formal education does not guarantee better efficiency, as local rice cultivation on tidal lands requires practical knowledge and local experience related to the ecological conditions of the land.

Overall, the study’s findings suggest that increasing local rice production requires an integrated approach that combines efficient input management with farmers’ adaptive capacities—such as water management and timing of planting—which help farmers cope with the challenges posed by the uncertainty of tidal ecological conditions. However, in this study, these capabilities were not measured as quantitative variables. Therefore, future research needs to develop measurable indicators for this ecological adaptation—such as water management, the use of adaptive local varieties, and the activity of farmers’ groups—so that their contribution to technical efficiency can be calculated more precisely. From a policy perspective, increasing rice production and ensuring the sustainability of rice farming on tidal lands cannot rely solely on strategies involving the intensification of chemical inputs. Government policy must focus on strengthening farmers’ managerial capabilities, enhancing knowledge transfer among farmers, and developing technologies suited to local ecological conditions. Integrating scientific innovations with locally developed knowledge can support a more sustainable and resilient rice farming system in tidal environments.

The authors gratefully acknowledge the support provided by the Islamic University of Kalimantan Muhammad Arsyad Al Banjari Banjarmasin for this research.

The grammar in this scientific article was edited with the assistance of ChatGPT (OpenAI). However, this article has been critically reviewed to ensure its accuracy.