Assessing Groundwater Use Efficiency and Productivity across Punjab Agriculture: District and Farm Size Perspectives

Abstract

While previous studies have focused on the technical aspects of groundwater extraction and optimal cropping patterns, there is a notable lack of research on the socio-economic factors affecting groundwater access and agricultural productivity, especially at a micro-level in Punjab, India. This study, therefore, investigates the water use efficiency (WUE) and economic water productivity (EWP) of paddy and wheat crops across diverse farm sizes and districts in Punjab, offering critical insights into agricultural water management. The study is based on primary data from Punjab, India, with Sangrur, Jalandhar, Pathankot, and Bathinda districts selected for groundwater profile analysis via multistage random sampling of 246 farmers. Notably, Punjab exhibits high EWP for wheat, surpassing the national average. However, disparities exist across districts, emphasizing the importance of localized interventions. Larger farms consistently demonstrate higher WUE and EWP, underlining the significance of scale in optimizing water inputs. Factors such as farm size, crop variety, and regional variations significantly influence WUE and EWP. Tailored approaches for marginal farmers and districts with lower efficiency are crucial for promoting sustainable agricultural practices. The findings underscore the need for targeted policy interventions to enhance water use efficiency and productivity in Punjab’s agriculture sector.

1. Introduction

Water scarcity, a critical global environmental challenge in the 21st century, is deeply intertwined with agriculture, which utilizes 80–90 percent of the world’s freshwater for crop production [1]. Anticipating a 35 percent increase in the world’s population to 8100 million by 2030 intensifies global food and water needs [2]. With 40 percent of the world’s land under arid conditions, efficient rainwater use and optimized crop water productivity have become crucial [3]. Food demand is projected to rise by 2030, especially in developing countries [4]. However, limitations in irrigated area growth necessitate focusing on enhancing water productivity, particularly in dry farming systems [2]. Recognizing impending water-related challenges, it is acknowledged that advances in water management and integrated policymaking are imperative globally, prioritizing efforts to improve water productivity in agriculture [5,6,7]. Scarcer water resources prompt other sectors to turn to agriculture, which makes it essential to focus on improving water efficiency and productivity to optimize water budgets, potentially reallocating resources from lower-value to higher-value uses and environmental conservation [5,8,9]. Closing the agricultural productivity gap worldwide, including in regions like the Indus-Gangetic basin, presents a significant opportunity. This can be achieved through improved water management and innovative policy and production techniques [10]. In countries like India, where water scarcity and flooding can occur simultaneously in different areas, the uneven distribution of water resources is exacerbated by climate change. Virtual water (the amount of water, either soil moisture or renewable and nonrenewable, that is consumed in the production of agricultural goods that are then traded in the international market [11,12]) trade, estimated at approximately 1000 km³/year globally, offers a potential solution to water scarcity, though its utilization remains limited [13]. However, the problem of water scarcity can be addressed when VW is exported from water-abundant regions to water-scarce regions through agricultural commodity exports. Notably, Punjab experiences significant water loss (−4.589 TL/yr) through virtual water flows, prompting the need for institutional reforms to ensure sustainable water use [14]. Among Indian states, Punjab, an agriculture-based economy, pumps the highest volume of groundwater. The Indus Basin aquifer, the world’s second-most overexploited aquifer [15], lies beneath Punjab. NASA’s GRACE observations show this aquifer, spanning Punjab, Haryana, Rajasthan, and Delhi, depletes at 109 km³/year [16]. Punjab’s declining groundwater levels threaten environmental sustainability by diminishing wetlands, reducing river base flows, degrading water quality, and drying forest ecology. Approximately 75 percent of the 76.87 thousand hectares of gross cropped area in Punjab rely on tubewells for irrigation. The dominant rice–wheat cropping pattern accounts for 95 percent of the food crop area and 86 percent of total irrigated land. However, this reliance on tubewells exacerbates water scarcity, leading to food stress. Government policies, including higher minimum support prices (MSP) and power subsidies for electric tubewells, contribute to a monoculture where farmers predominantly grow wheat in winter and paddy in summer. As a result, farmers continue growing water-intensive crops like paddy and sugarcane due to policies of free energy and assured purchase prices, leading to significant groundwater over-exploitation [17]. Water-scarce regions like Punjab should lead in making efforts to enhance agricultural productivity and water use efficiency (WUE) through integrated land and water management policies.

This study embarks on a comprehensive exploration of the factors influencing WUE and EWP within agricultural systems in Punjab. It is hypothesized that WUE exhibits varying patterns across farm sizes and within different districts. This suggests that the efficiency with which water is utilized in agricultural practices may demonstrate distinct trends depending on the scale of farming operations and the geographical location, indicating potential disparities in resource management strategies and environmental conditions. By delving into a nuanced analysis of various determinants, it aims to unravel the intricate web of influences on water use decisions in agriculture. The multifaceted nature of factors necessitates an in-depth investigation to discern their individual and collective impact on WUE and overall agricultural productivity. Through empirical evidence in Punjab and scholarly insights, this paper contributes valuable knowledge that can inform groundwater management practices and guide future strategies for optimizing agricultural productivity in the face of escalating global water challenges.

Although several studies have explored groundwater exploitation and its impact on agricultural productivity, notable gaps remain in the existing literature. Previous research has predominantly focused on finding out optimal cropping patterns and the technical aspects of groundwater extraction and efficiency [18,19,20,21,22,23,24,25,26,27,28,29]. However, there is a lack of comprehensive studies analyzing the socio-economic factors influencing farmers’ access to groundwater resources and the resulting effects on agricultural efficiency and productivity. Furthermore, there is a specific gap in the context of Punjab, India, where micro-level studies comparing efficiency and productivity at the district level are scarce. Understanding which groups of farmers—large or marginal—are more water-efficient and productive in Punjab remains an open question that this study aims to address. The primary objectives of our study are to investigate the WUE and EWP of paddy and wheat crops in Punjab across different farm sizes and districts, examine the factors influencing WUE and EWP within agricultural systems, including farm size, crop variety, and regional variations, and to provide empirical evidence and scholarly insights to guide future strategies for optimizing agricultural productivity and addressing water challenges in Punjab and beyond. This research undertook a detailed investigation of farmer households through primary survey in districts of Jalandhar, Sangrur, Bathinda, and Pathankot conducted during July–August 2022 for the agricultural year 2021–2022 (July 2021 to June 2022). The Indian cropping season is classified into two main seasons—(i) Kharif and (ii) Rabi, based on the monsoon. The Kharif farming season is from July–October during the south-west monsoon and the Rabi farming season is from October–March (winter). The subsequent section delineates the methodology employed to acquire the insights and data essential for a comprehensive analysis. Several studies have used single or partial factor productivity to measure groundwater efficiency and productivity instead of total factor productivity [5,30]. A partial/single productivity measure relates output to a single input. Total factor productivity (TFP) relates an index of output to a composite index of all inputs [31]. This study also employs partial factor productivity (SFP).

2. Materials and Methods

2.1. Concepts of WUE and EWP

2.1.1. Historical Background and Definitions

The WUE concept was initially formulated a century ago to illustrate the correlation between plant productivity and water consumption. It was coined to quantify the biomass generated per unit of water utilized by a plant. Over the ensuing decades, numerous original research papers have explored and expanded upon this concept [32,33]. WUE is a term interpreted differently across disciplines. Originally, crop physiologists defined WUE as the amount of carbon assimilated and crop yield per unit of transpiration [34,35,36], while irrigation scientists viewed it as the ratio of total irrigation water transpired to water diverted from the source [37,38]. Crop scientists, on the other hand, defined it as the ratio of total biomass or grain yield to water supplied [34,39]. In this study, based on the core idea of obtaining more crop per drop of water, physical WUE is adopted as the ratio of grain yield achieved to the amount of water made available to the crop (irrigation water), expressed as grain yield per unit of irrigation water [40,41].

The term ‘water productivity’ (WP) also carries diverse meanings. It varies among different water users and may even differ within groups of water users. It can be defined in relation to various water-using production sectors (e.g., crop production, fishery, forestry, domestic and industrial use) [42] as the amount of output produced per unit of water involved in the production or the value added to water in a given circumstance. To a hydrologist, it means the ratio of the volume of water used productively (transpired and, in some cases, evaporated) from the area under study to the volume of water potentially available for that purpose [43,44]. In fisheries, it might mean the ratio of fish produced to the volume of water used [43]. Economists may define it as the monetary value of output divided by that of the necessary water input, while irrigation engineers could interpret it as the ratio of the amount or the value of crops produced in a farm (or catchment) to the supplied water [44]. Even within crop science, WP can take different forms, such as leaf WP, whole plant WP, and yield WP [43,44]. Essentially, WP represents the output of a given activity to the water input. In cases of inter-cropping or multiple cultures, the monetary return per unit of water input might be the most relevant measure.

2.1.2. Definitions Adopted for This Study

The definitions of the terms adopted in the study are expressed as follows:

$$\mathbf{W}\mathbf{U}\mathbf{E} (\mathbf{k}\mathbf{g}/\mathbf{m})={\displaystyle \frac{\mathbf{c}\mathbf{r}\mathbf{o}\mathbf{p} \mathbf{y}\mathbf{i}\mathbf{e}\mathbf{l}\mathbf{d} (\mathbf{k}\mathbf{g}/\mathbf{h}\mathbf{a})}{\mathbf{i}\mathbf{r}\mathbf{r}\mathbf{i}\mathbf{g}\mathbf{a}\mathbf{t}\mathbf{i}\mathbf{o}\mathbf{n} \mathbf{i}\mathbf{n}\mathbf{f}\mathbf{l}\mathbf{o}\mathbf{w} ({\mathbf{m}}^{\mathbf{3}}/\mathbf{h}\mathbf{a})}}$$

(1)

Irrigation inflow excludes rainfall (uncertain) and losses due to plant evapotranspiration or seepage for this study.

$$\mathbf{I}\mathbf{r}\mathbf{r}\mathbf{i}\mathbf{g}\mathbf{a}\mathbf{t}\mathbf{i}\mathbf{o}\mathbf{n} \mathbf{i}\mathbf{n}\mathbf{f}\mathbf{l}\mathbf{o}\mathbf{w} ({\mathbf{m}}^{\mathbf{3}}/\mathbf{h}\mathbf{a})={\displaystyle \frac{\mathbf{W}\mathbf{a}\mathbf{t}\mathbf{e}\mathbf{r} \mathbf{r}\mathbf{e}\mathbf{l}\mathbf{e}\mathbf{a}\mathbf{s}\mathbf{e}\mathbf{d} \left(\frac{{\mathbf{m}}^{\mathbf{3}}}{\mathbf{h}\mathbf{o}\mathbf{u}\mathbf{r}}\right)\times \mathbf{D}\mathbf{u}\mathbf{r}\mathbf{a}\mathbf{t}\mathbf{i}\mathbf{o}\mathbf{n} \mathbf{o}\mathbf{f} \mathbf{i}\mathbf{r}\mathbf{r}\mathbf{i}\mathbf{g}\mathbf{a}\mathbf{t}\mathbf{i}\mathbf{o}\mathbf{n} (\mathbf{h}\mathbf{o}\mathbf{u}\mathbf{r}\mathbf{s})}{\mathbf{A}\mathbf{r}\mathbf{e}\mathbf{a} \mathbf{i}\mathbf{r}\mathbf{r}\mathbf{i}\mathbf{g}\mathbf{a}\mathbf{t}\mathbf{e}\mathbf{d} \left(\mathbf{h}\mathbf{a}\right)}}$$

(2)

In Equation (2), water released denotes the volume of water released or supplied per hour during irrigation. Duration of irrigation indicates the length of time over which irrigation occurs, measured in hours. The area irrigated represents the total land area receiving irrigation, measured in hectares. For calculating irrigation water inflow, the number of hours the tubewell operated on the farm was adjusted for the cropped area under each crop, and the approximate water discharge from the tubewell, considering the diameter of the outlet, was estimated for each crop.

WP in economic terms calculated in the study can be expressed as:

$$\mathbf{E}\mathbf{W}\mathbf{P} (\mathbf{I}\mathbf{N}\mathbf{R}/{\mathbf{m}}^{\mathbf{3}})={\displaystyle \frac{\mathbf{c}\mathbf{r}\mathbf{o}\mathbf{p} \mathbf{y}\mathbf{i}\mathbf{e}\mathbf{l}\mathbf{d} \left(\frac{\mathbf{k}\mathbf{g}}{\mathbf{h}\mathbf{a}}\right)\times \mathbf{F}\mathbf{a}\mathbf{r}\mathbf{m} \mathbf{H}\mathbf{a}\mathbf{r}\mathbf{v}\mathbf{e}\mathbf{s}\mathbf{t} \mathbf{P}\mathbf{r}\mathbf{i}\mathbf{c}\mathbf{e} (\mathbf{I}\mathbf{N}\mathbf{R}/\mathbf{k}\mathbf{g})}{\mathbf{i}\mathbf{r}\mathbf{r}\mathbf{i}\mathbf{g}\mathbf{a}\mathbf{t}\mathbf{i}\mathbf{o}\mathbf{n} \mathbf{i}\mathbf{n}\mathbf{f}\mathbf{l}\mathbf{o}\mathbf{w} ({\mathbf{m}}^{\mathbf{3}}/\mathbf{h}\mathbf{a})}}$$

(3)

In Equation (3), farm harvest prices are the prices farmers receive for their produce at the location of the farm [45]. The farm harvest prices of paddy for the year 2021–2022 vary from INR 19.40 per kg in Jalandhar to INR 19.60 per kg in Bathinda, Sangrur and Pathankot for paddy and for wheat, the prices vary from INR 19.75 per kg in Sangrur and Bathinda to INR 20.15 per kg in Jalandhar and Pathankot.

2.1.3. Factors Affecting WUE and EWP

Several studies explore factors impacting water use efficiency (WUE) in agriculture. Cereal crops consistently show higher WUE compared to oilseeds [46,47,48]. WUE tends to be greater in water-stressed regions than in less-stressed areas [49,50,51]. Drip irrigation systems generally yield higher WUE than furrow systems [52]. Clay soils typically exhibit higher WUE than sandy or loamy soils [53,54]. Warm–dry climates are associated with better WUE for crops like wheat, maize, and potatoes compared to warm–wet climates [55,56]. Fertilized crops generally demonstrate higher WUE than non-fertilized ones [55,56,57]. Soil water conservation during early crop growth can enhance WUE, revealing drought’s nuanced impact on efficiency [58]. Conservation tillage methods tend to result in higher WUE than conventional tillage [59,60], and mulched fields typically have higher WUE compared to non-mulched ones [59]. Stress-tolerant crop varieties also exhibit better WUE than non-tolerant varieties [61,62]. Higher seeding rates are associated with greater WUE compared to lower rates [62,63]. While consensus on these factors’ impacts varies, global research offers an opportunity for comprehensive analysis to clarify the primary determinants of crop WUE [1].

2.1.4. Integration of WUE and EWP

Water use efficiency (WUE) and economic water productivity (EWP) are critical metrics in promoting agricultural water conservation. WUE denotes the proportion of water input absorbed by crops, while EWP measures the efficiency of water use in relation to crop benefits, quantifying water’s conversion into food [46,64]. However, inconsistencies in their definitions across disciplines lead to misunderstandings and the misuse of the terms [48,65]. WUE is sometimes incorrectly equated with biomass production per unit of water consumed, a definition more appropriately attributed to EWP [48]. It is crucial to distinguish between WUE and EWP as separate concepts; while closely related, a high WUE does not necessarily guarantee a high EWP due to the plant-specific nature of EWP [46,48]. Two crops may have the same WUE but a different EWP, as influenced by market conditions. For instance, high-value cash crops typically exhibit higher EWP than traditional cereals, highlighting the potential for increasing EWP through crop diversification towards higher-value crops.

2.1.5. Multi-Disciplinary Approach in WUE and EWP

A challenge arises from the multi-disciplinary nature of the subject, involving diverse fields such as hydrology, hydrogeology, civil and irrigation engineering, agronomy, crop physiology, and economics. Each discipline interprets the concepts of productivity and efficiency differently, emphasizing distinct measures of water. Even within these disciplines, there is variability in the terminology used or new terms coined, depending on the specific focus and approach of the study [5]. In civil engineering, conveyance efficiency is defined as the ratio of water received at the farm to water withdrawn from the water source. Moving to irrigation engineering, water-application efficiency is described as the ratio of water stored in the root zone (consumed) to water delivered to the farm, while classical irrigation efficiency is defined as the ratio of water consumed to water applied [66]. In agronomy and crop physiology, WUE is calculated as the ratio of plant biomass or yield to consumptive water use. Lastly, water productivity, also within agronomy and crop physiology, refers to the ratio of physical production or the “economic value” of production to water use [5].

2.2. Study Area

This quantitative study applies a field survey research design to examine groundwater use efficiency and productivity and their underlying determinants in Punjab, India. It focuses on the state of Punjab, known for its significant groundwater exploitation and inefficient tubewell irrigation practices. Approximately 75 percent of the 76.87 thousand hectares of gross cropped area in Punjab is irrigated through tubewells, with the rice–wheat cropping pattern accounting for 95 percent of the cultivated area and 86 percent of the total irrigated area [67]. As a result, Punjab is set to face significant food stress due to rising water scarcity [68,69,70] and the government policies of higher minimum support prices (MSP) (the prices at which the government of India agrees to purchase agricultural products from farmers), and power subsidies to farmers using electric tubewells are further complementing a monoculture in Punjab, with farmers mainly growing wheat in the winter and rice in the summer [71,72,73].

Punjab has 23 districts [74] and this survey was conducted in four districts of Punjab—Sangrur and Jalandhar, which represent groundwater-exploited districts, and Pathankot and Bathinda, which represent relatively groundwater-safe districts, as this would help in comparing and analyzing the agricultural practices across districts. They also represent the north–south extent of Punjab while covering three-fourths of the administrative divisions. The sample area is shown in the map in Figure 1.

The target population consisted of farmers of these districts who irrigated their crops using tubewells. In the first stage, Sangrur, Jalandhar, Pathankot, and Bathinda districts were purposely selected. In the second stage, three blocks from each block were randomly selected and in the third stage, from each block, 3–4 villages were randomly selected. Finally, from each village, 8–10 farmer households were targeted randomly to ensure the representativeness of the samples (

Table 1. Sample characteristics of farmers surveyed in Punjab, India.

District	Blocks	Villages	Farmers Surveyed	Average Farm Size (Ha)	Average Paddy Crop Yield (Qn/Ha)	Average Age of Farmers
Jalandhar	3	6	58	8.5	75.27	58
Sangrur	3	5	57	3.2	75.79	51
Bathinda	3	9	76	5.2	73.81	49
Pathankot	3	6	55	3.6	63.65	55
Total	12	26	246	5.3	72.13	53

). Observations and first-hand information were collected during the primary survey on farming practices and irrigation methods, and discussions were conducted with about 246 farmers to gain insights and feedback regarding their perceptions, experiences, and challenges related to groundwater management and regulations. The sample ensures a diverse representation of agricultural practices. The multistage random sampling technique employed allows for comprehensive coverage of farm sizes and practices within the selected districts. This methodological approach strengthens the representativeness of the sample. Practical considerations such as time, resources, and accessibility are also factored in while determining the sample size. The chosen sample size is a balance between statistical requirements and the logistical feasibility of conducting extensive face-to-face interviews within the study’s timeframe.

The sample characteristics of the surveyed farmers are displayed in Table 1.

The survey followed a structured questionnaire designed to capture comprehensive data on agricultural practices, water use, crop yields, and socioeconomic factors. Data collected from the surveys were carefully recorded and entered into a database. Double-checking of survey responses was performed to minimize errors. Discrepancies, if any, were resolved by cross-checking the original survey forms. The dataset was thoroughly cleaned to address any inconsistencies, missing values, or outliers. The analysis was conducted using the statistical software Stata (Stata/MP 18.0 for windows). Basic descriptive statistics were calculated to summarize the key characteristics of the sample, including the mean, median, and standard deviation. Multiple regression models were used to identify the factors influencing water use efficiency (WUE) and economic water productivity (EWP) for paddy and wheat crops. A comparative analysis was performed to examine variations in WUE and EWP across different districts and farm sizes.

The water discharge is determined based on the flow of the tubewell, which is related to the diameter of the outlet pipe. Farmers have reported the water discharge from their tubewells. All farmers possess their own tubewells, and conveyance losses are negligible due to their proximity to farms, with only evapotranspiration as a loss factor. The estimation of water is based on the irrigation frequencies (number of waterings) to wheat and paddy crops. For instance, the irrigation frequencies for paddy may increase/decrease depending on the rainfall and climate. The number of hours taken per acre to irrigate a farm is considered. The data are based on the farmers’ recollections from the previous year, but every effort has been made to minimize errors. There have been previous studies [75,76] that have calculated data using farmers’ recollections of previous years, but this study still acknowledges the limitations of recall-based data collection. The number of waterings in the entire crop cycle is multiplied by the hours farmers use to irrigate one acre of farmland. It is then multiplied by the water discharge based on the pipe outlet diameter to estimate the quantity of water applied to each crop. It is a non-experimental study approach. The study also tries to mitigate the potential for error due to reliance on farmers’ memory when reporting water use, which could affect the water use efficiency (WUE) calculations for wheat and paddy, through the cross-verification of reported data with local agricultural extension services and averages from multiple respondents to reduce individual memory bias. For simplicity, there is an assumption of consistent water application. The actual irrigation practices may vary due to several factors, including rainfall patterns and crop growth stages. Additionally, the study acknowledges that direct measurements would provide more precise data but are often impractical in large-scale field studies due to resource constraints. Despite this, the reported data offer valuable insights into farmers’ irrigation practices and serve as a basis for further investigation with more rigorous methods.

2.3. Data and Variables

This study expresses efficiency as the ratio of output to input. WUE in agriculture refers to the production achieved per unit of water used, measured in units like kg grain/ha per mm, kg/m³, or INR/m³ [60,77]. Similarly, on a broader scale, WP represents the ratio of benefits from agriculture to the amount of water consumed for those benefits. This includes physical water productivity (more crop per drop) and economic water productivity, which measures the value derived per unit of water used [44,75].

Earlier research suggests that various factors, including socio-economic conditions, environmental impacts, technological advancements, and institutional structures, can significantly impact decisions related to water usage. Factors such as the overall cultivated area and the presence of water-saving machinery, including electric and diesel motors, also play a role in influencing WUE [50,78,79]. Water consumption is intertwined with cultural and religious beliefs, making it a significant aspect of societal values. In studies, demographic details, including religion, were assessed alongside household characteristics and water usage patterns. It highlights the importance of considering religious perspectives when understanding water consumption behaviors within communities [80,81].

The selection of variables in a study is crucial for understanding the complex dynamics of water efficiency and productivity in agriculture, particularly in a region like Punjab. The dependent variable, WUE, is central to the investigation as it provides insights into the effectiveness of water utilization in agricultural practices, reflecting the sustainability and productivity of irrigation methods.

The independent variables chosen for the study encompass a range of socio-economic, demographic, and agricultural factors, each contributing to the nuanced understanding of the WUE in the region. The “Age of Farmer” is a demographic variable that could influence decision-making and adaptive strategies. “Family Members” and “Farm Size” are crucial demographic and economic variables, respectively, impacting the scale and dynamics of agricultural operations. The “District” variable acknowledges potential regional variations and distinct agricultural practices within Punjab, recognizing that water use patterns can differ significantly across geographical areas. “Highest Educational Qualification” serves as a proxy for the level of knowledge and awareness among farmers, influencing their adoption of water-efficient technologies and practices. The variable “Religion” sheds light on the socio-cultural context influencing resource allocation and farming practices [5,82]. The choice of “Tubewell (electric/diesel)” and “Depth of Tubewell” reflects the importance of irrigation infrastructure, with electric and diesel tubewells representing distinct technologies that may impact water accessibility and usage efficiency. “Internet usage by farmers” is a technological variable, indicating the adoption of modern communication and information tools, which could influence farmers’ awareness and decision-making regarding water management practices.

2.4. Regression Models

Determinants of WUE and EWP

The regression model is presented below, shedding light on the intricate interplay of the various variables within the studied context that affect WUE and EWP:

ln(WUE) = β₀ + β1.ln(Age) + β₂.ln(Family) + β₃.(Depth) + β_4.(LLL) + β₅.(Seed) + β₆.ln(Fertilizer) + β₇.ln(Labour) + β₈.(Farm.size) + β₉.(Religion) + β₁₀.ln(Tubewell.Type) + β₁₁.ln(Internet) + β₁₂.(District) + β₁₃.ln(Cultivars) + β₁₄.ln(Education) + 𝜖

(4)

where:

ln represents the natural logarithm.

β₀, β₁, β₂, …, β₁₄ are the coefficients.

ϵ is the error term.

ln(EWP) = β₀ + β_1.ln(Age) + β₂.ln(Family) + β₃.(Depth) + β₄.(LLL) + β₅.(Seed) + β₆.ln(Fertilizer) + β₇.ln(Labour) + β₈.(Farm.price) + β₉.(Farm.size) + β₁₀.(Religion) + β₁₁.ln(Tubewell.Type) + β₁₂.ln(Internet) + β₁₃.(District) + β₁₄.ln(Cultivars) + β₁₅.ln(Education) + 𝜖

(5)

where:

ln represents the natural logarithm.

β₀, β₁, β₂, …, β₁₅ are the coefficients.

ϵ is the error term.

Table 2. Variables used in the study.

Variable Name	Type	Definition	Notes
WUE/EWP	Continuous	Water use efficiency/productivity	-
Age	Continuous	Age of farmer	-
Family	Continuous	Number of household members of farmer	-
Size	Categorical	Farm size (marginal/small/semi-medium/medium/large)	Small farm as base
District	Categorical	Region/district of study	Pathankot as base
Farm price	Continous	Farm harvest prices/farm gate prices	-
Education	Categorical	Highest educational qualification	Masters as base
Religion	Categorical	Religious preference followed (Hindu/Sikh)	Hindu as base
LLL	Continuous	Number of times of land laser leveled in the last 6 years	-
Fertilizer	Continuous	Kilograms of fertilizer used per hectare	-
Tubewell	Categorical	Type of tubewell (diesel/electric)	Diesel as base
Internet	Categorical	Does the farmer access internet on phone (yes/no)	No as base
Depth	Continous	Depth of tubewell to water table (meters)	-
Labour	Continuous	Number of man-days worked on the farm per hectare	-
Cultivars	Categorical	Variety of paddy seed (PUSA44, PR126, etc.)	PR130 as base
Seed	Continuous	Kilograms of seed used per hectare	-

Note: PUSA44 is a long-duration paddy variety requiring more water, while PR126 and P130 are short-duration varieties.

depicts the variables undertaken for the study and their description.

The following section presents the results, offering valuable insights into the complex nexus between these variables and the outcomes of WUE in Punjab’s agricultural sector.

3. Results

In the upcoming section, the study delves into the detailed results and findings for the WUE and EWP of paddy and wheat crops in Punjab, India. We aim to uncover key insights into the factors influencing WUE and EWP by applying statistical techniques such as double-log multiple linear regression. Specifically, we will examine the role of farm size, crop variety, and regional disparities in shaping water management practices and productivity outcomes. Furthermore, we will explore the implications of our findings for policy formulation and sustainable farming practices in Punjab’s agriculture sector.

Table 3. Descriptive statistics of variables.

Variable	Mean	Std. Dev.	Min	Max
Age	49	13	24	85
Family members	5	2	2	11
WUE paddy	0.54	0.05	0.42	0.72
WUE wheat	0.76	0.13	0.5	0.98
EWP paddy	10.56	0.95	8.07	14.13
EWP wheat	15.20	2.51	9.88	19.28
Paddy fertilizer (kg)	271.38	31.84	200	350
Wheat fertilizer (kg)	291.54	52.29	200	370
Labour man-days paddy per hectare	104	18	80	130
Labour man-days wheat per hectare	102	10	90	120
Seed rate paddy (kg/hectare)	113	4.46	100	120
Seed rate wheat (kg/hectare)	106	8.15	100	120
Gross revenue (INR/hectare)	209,380 ($2500)	15,805 ($189)	182,188 ($2176)	245,563 ($2933)
Depth of tubewell (meters)	88	38.7	15	140
Farm prices paddy (INR/quintal)	19.3 ($0.23)	0.35	18.5 ($0.22)	19.9 ($0.24)
Farm prices wheat (INR/quintal)	19.95 ($0.24)	0.34	19.3 ($0.23)	20.5 ($0.24)

Source: Authors’ calculations based on field survey in selected districts.

depicts the descriptive statistics of variables.

The average age of the sampled farmers is 49 years, with a variation ranging from 24 to 85 years. On average, the sampled individuals have 5 family members, with the number ranging from 2 to 11. The average WUE for paddy is 0.540 kg/m³, with a standard deviation of 0.05, indicating relatively consistent efficiency across the sample farms. The average WUE for wheat is 0.76 kg/m³, reflecting a higher efficiency compared to paddy. The variability is noticeable, as indicated by the standard deviation (0.13). The average EWP for paddy is 10.56 INR/m³ or $0.13/m³ showcasing the economic output derived from water use. The standard deviation indicates some variation in EWP (0.95). The average EWP for wheat is 15.20 INR/m³ or $0.18/m³, suggesting a higher economic return than that for paddy. The standard deviation indicates notable variability in wheat productivity (2.51). On average, 271.38 kg of fertilizer per hectare is used for paddy cultivation, with a standard deviation of 31.8370, suggesting some variation in fertilizer application. The average fertilizer usage for wheat is 291.54 kg per hectare, with a standard deviation of 52.287, indicating variability in fertilizer application for wheat. On average, 104 labor days are employed for paddy cultivation per hectare, while the average labor requirement for wheat cultivation is 102 days per hectare. The average depth of tubewell is found to be 88 m with minimum depth at 15 m and maximum at 140 m with a standard deviation of 39 m. The average farm prices of paddy and wheat are found to be INR 19.3 ($0.23) and INR 19.95 ($0.24), respectively, with low standard deviation.

Table 4. Distribution of sampled farmers by farm size category.

District	Marginal	Small	Semi-Medium	Medium	Large	Total
Bathinda	13	25	15	16	7	76
Jalandhar	4	8	13	22	11	58
Pathankot	14	23	11	3	4	55
Sangrur	5	16	23	12	1	57
Total	36	72	62	53	23	246

Source: authors’ calculations based on field survey in selected districts.

provides demographic data for all four districts by farm size.

Small (less than 1 hectare of land) and semi-medium (2–4 hectares) farm sizes dominate the sample size in the study, followed by medium (4–10 hectares), marginal (1–2 hectares), and large (more than 10 hectares) farms. Regional analysis shows the dominance of small farms in samples from Bathinda and Pathankot, medium farms in Jalandhar, and semi-medium farms in Sangrur.

Table 5. Distribution of sampled farmers by education level.

District	Illiterate	5th–8th	10th	10 + 2	B.A	M.A	Total
Bathinda	11	17	15	12	8	13	76
Jalandhar	7	13	11	9	6	12	58
Pathankot	5	12	7	8	7	16	55
Sangrur	9	12	11	10	4	11	57
Total	32	54	44	39	25	52	246

Source: authors’ calculations based on field survey in selected districts.

shows the demographic sample of farmers by their level of education.

Bathinda and Pathankot have the highest number of graduate and post-graduate farmers surveyed. On the other hand, Bathinda and Sangrur have the highest number of illiterate farmers. Bathinda and Jalandhar have the highest number of farmers who have completed the schooling system. Overall, post-graduates and farmers who have achieved the highest education up to the eighth standard form the highest number of farmers surveyed.

Table 6. Distribution of sampled farmers by type of tubewell used.

District	Diesel Tubewell	Electric Tubewell	Total
Bathinda	0	76	76
Jalandhar	0	58	58
Pathankot	24	55	55
Sangrur	0	57	57
Total	24	246	246

Source: authors’ calculations based on field survey in selected districts.

provides information on the types of tubewells in four districts, differentiated by whether they are powered by diesel or electricity. Out of 246 farmers surveyed, only 24 are using diesel-powered tubewells, and all are in Pathankot district. All other 222 farmers surveyed are irrigating using electric tubewells.

The selected variables create a comprehensive framework to explore the intricate interplay between demographic, socio-economic, and technological factors in determining WUE in Punjab’s agriculture. This study applies multiple regression models to estimate the factors affecting both the WUE and EWP of paddy and wheat crops.

3.1. Farm Size, Regional Variations, and Water Use Efficiency in Paddy and Wheat

Table 7. Water use efficiency in paddy and wheat by farm size in Punjab.

	Paddy			Wheat
Farm Size	Yield (kg/ha)	Water Applied (m3/ha)	WUE (kg/m3)	Yield (kg/ha)	Water Applied (m3/ha)	WUE (kg/m3)
Marginal	4936	9700	0.512	4622	7022	0.66
Small	5091	9637	0.529	4896	6805	0.721
Semi-medium	5030	9350	0.540	5194	6731	0.773
Medium	5152	9242	0.559	5643	6675	0.845
Large	5339	9230	0.580	5597	6624	0.844
All	5089	9451	0.540	5155	6774	0.763

Source: authors’ calculations based on field survey in selected districts.

presents detailed information on different farm sizes, their corresponding crop yields, water applied for irrigation, and WUE in paddy and wheat cultivation. The farm sizes are categorized as Marginal, Small, Semi-medium, Medium, and Large. For each category, the table includes the yield of paddy and wheat in kilograms per hectare, the volume of water applied for irrigation in cubic meters per hectare, and the calculated WUE in kilograms of yield per cubic meter of water. This dataset aims to shed light on WUE across various farm sizes, offering valuable insights into agricultural practices and resource management in paddy cultivation.

In the case of paddy, the table highlights variations in yield, water application, and WUE across different farm sizes. Marginal farms show relatively low WUE (0.512 kg/m³) of paddy, indicating inefficient water utilization for yield, suggesting a potential need for optimizing water application practices. This indicates that 1000 L of irrigation water are used to produce 0.51 kg of paddy crop. Larger farms exhibit the highest WUE (0.580 kg/m³; 1000 L of irrigation water is applied to produce 0.58 kg of paddy crop). The overall WUE for all farm sizes is 0.540 kg/m³ (1000 L of irrigation water is applied to produce 0.54 kg of paddy crop), providing an average measure of efficiency across the dataset.

In the case of wheat, the average yield and resultant WUE increase from marginal to medium farms and decrease slightly for large farms. Large farms apply relatively less water compared to small farms. Marginal farms have the lowest WUE (0.66 kg/m³), while large farms have more efficiency (0.844 kg/m³). Overall, the average WUE in wheat for all sampled farm sizes is 0.763 kg/m³.

The results show a positive relationship between farm size and WUE. Relatively smaller farms are found to be less water-efficient in paddy cultivation than the larger ones. Water application also declines, and crop yield increases for larger farms compared to smaller ones. Whether small farms perform better than large farms is a largely debated issue. Several studies [83,84,85,86] indicate a positive correlation between farm size and output per household, per capita, and per hectare, with larger farms consistently yielding greater returns and efficiency. This trend is attributed to economies of scale, where larger operations benefit from enhanced financial opportunities, management practices, and marketing facilities, while small farms may demonstrate institutional constraints [84]. Larger farms typically benefit from economies of scale, which allow them to invest more in agricultural inputs and technologies [83]. These investments include higher fertilizer application rates, better pest and weed management practices, and the use of quality seeds. Larger farms generally have better financial opportunities, enabling them to adopt improved farming techniques and invest in better management practices [85]. These practices include the use of drought-resistant crop varieties, which contribute to higher WUE. Such inputs contribute to higher yields per hectare, thereby increasing land productivity and directly influencing water productivity. Larger farms are also more likely to have access to advanced irrigation systems that optimize water use, further enhancing WUE. Smaller farms often face significant institutional constraints, such as limited access to credit, inputs, and markets [86]. Additionally, water and soil conservation practices may vary with farm scale [83,84,85,86].

Table 8. Regional pattern in water use efficiency of paddy and wheat in Punjab (kg/m³).

Farm Size	Paddy					Wheat
	Jalandhar	Sangrur	Pathankot	Bathinda	All	Jalandhar	Sangrur	Pathankot	Bathinda	All
Marginal	0.481 ± 0.02	0.492 ± 0.02	0.507 ± 0.05	0.533 ± 0.04	0.503 ± 0.05	0.722 ± 0.04	0.80 ± 0.02	0.571 ± 0.03	0.683 ± 0.13	0.694 ± 0.11
Small	0.496 ± 0.04	0.508 ± 0.02	0.527 ± 0.03	0.556 ± 0.04	0.522 ± 0.04	0.76 ± 0.02	0.849 ± 0.04	0.566 ± 0.03	0.768 ± 0.11	0.736 ± 0.13
Semi-medium	0.503 ± 0.02	0.527 ± 0.03	0.553 ± 0.02	0.583 ± 0.05	0.542 ± 0.04	0.772 ± 0.04	0.857 ± 0.05	0.599 ± 0.03	0.772 ± 0.09	0.75 ± 0.11
Medium	0.523 ± 0.02	0.552 ± 0.02	0.594 ± 0.02	0.603 ± 0.03	0.568 ± 0.04	0.873 ± 0.06	0.904 ± 0.04	0.660 ± 0	0.80 ± 0.07	0.809 ± 0.08
Large	0.532 ± 0.03	0.568 ± 0	0.604 ± 0.04	0.643 ± 0.04	0.587 ± 0.06	0.899 ± 0.06	0.976 ± 0	0.684 ± 0.05	0.831 ± 0.10	0.848 ± 0.11
All	0.513 ± 0.03	0.525 ± 0.03	0.536 ± 0.05	0.575 ± 0.05	0.537 ± 0.05	0.805 ± 0.08	0.877 ± 0.05	0.588 ± 0.05	0.767 ± 0.11	0.759 ± 0.13

Source: Authors’ calculations based on the field survey in selected districts. Note: ± represents the standard deviation of the WUE of paddy and wheat crop.

presents the regional pattern in WUE of paddy and wheat in different farm sizes across sampled districts in Punjab, measured in kilograms per cubic meter (kg/m³).

Bathinda has the highest WUE in paddy among all farm sizes, indicating relatively efficient water use (0.575 ± 0.05 kg/m³). Pathankot, Sangrur, and Jalandhar followed with decreasing WUE values. The overall average WUE for paddy across all districts is 0.537 ± 0.05 kg/m³. Similarly, Jalandhar district has the lowest WUE for paddy among all districts. The WUE for paddy is lowest for marginal farms (0.503 ± 0.05 kg/m³) and highest for large farms (0.587 ± 0.06 kg/m³).

The table also shows variations in the WUE of wheat across different farm sizes and districts. Overall, large farms exhibit the highest WUE of wheat (0.848 ± 0.11 kg/m³), while marginal farms have the lowest (0.694 ± 0.11 kg/m³). Pathankot has the smallest WUE in the case of wheat (0.588 ± 0.05 kg/m³) out of all districts, while Sangrur has the highest (0.877 ± 0.05 kg/m³).

3.2. Farm Size, Regional Variations, and Economic Water Productivity in Paddy and Wheat

Table 9. Regional pattern in economic water productivity of paddy and wheat in Punjab (INR/m³).

Farm Size	Paddy					Wheat
	Jalandhar	Sangrur	Pathankot	Bathinda	All	Jalandhar	Sangrur	Pathankot	Bathinda	All
Marginal	9.33 ± 0.40	9.64 ± 0.48	9.94 ± 1.05	10.45 ± 0.8	9.84 ± 0.91	14.55 ± 0.73	15.8 ± 0.33	11.51 ± 0.62	13.49 ± 2.61	13.84 ± 2.21
Small	9.62 ± 0.8	9.96 ± 0.42	10.33 ± 0.55	10.90 ± 0.78	10.20 ± 0.78	15.31 ± 0.35	16.77 ± 0.76	11.40 ± 0.59	15.17 ± 2.26	14.66 ± 2.54
Semi-medium	9.76 ± 0.35	10.33 ± 0.51	10.84 ± 0.46	11.43 ± 0.93	10.59 ± 0.85	15.56 ± 0.79	16.93 ± 0.97	12.07 ± 0.54	15.25 ± 1.77	14.95 ± 2.03
Medium	10.15 ± 0.42	10.82 ± 0.42	11.64 ± 0.37	11.82 ± 0.61	11.11 ± 0.86	17.59 ± 1.2	17.85 ± 0.76	13.30 ± 0.11	15.8 ± 1.52	16.14 ± 1.69
Large	10.32 ± 0.57	11.13 ± 0	11.84 ± 0.88	12.60 ± 0.81	11.47 ± 1.23	18.11 ± 1.13	19.28 ± 0	13.78 ± 0.95	16.41 ± 1.98	16.9 ± 2.15
All	9.95 ± 0.58	10.29 ± 0.59	10.51 ± 0.9	11.27 ± 0.99	10.50 ± 0.95	16.22 ± 1.56	17.32 ± 1.02	11.85 ± 0.92	15.15 ± 2.19	15.13 ± 2.51

Source: Authors’ calculations based on field survey in selected districts. Note: ± represents the standard deviation of the EWP of paddy and wheat crop.

illustrates the regional pattern in the EWP of paddy and wheat across different farm sizes in Punjab. The values are expressed in Indian Rupees per cubic meter (INR/m³ or $/m³), reflecting the economic productivity of water in paddy cultivation.

Large farms have the highest paddy water productivity (11.47 INR/m³ or $0.14/m³), while marginal farms have the lowest (9.84 INR/m³ or $0.12/m³). Bathinda district displays the highest productive use of water for paddy (11.27 INR/m³ or $0.13/m³) compared to other districts. Jalandhar and Sangrur are less water-productive for paddy compared to Bathinda and Pathankot, Jalandhar being the least productive (9.95 INR/m³ or $0.12/m³).

Marginal farms have the least wheat water productivity (13.84 INR/m³ or $0.17/m³), while large farms have the highest (16.90 INR/m³ or $0.20/m³). Sangrur district displays the highest productive use of water for wheat (17.32 INR/m³ or $0.21/m³) compared to other districts. Pathankot and Bathinda are less water-productive for wheat compared to Sangrur, Pathankot being the least productive (11.85 INR/m³ or $0.14/m³).

The study reveals significant variations in WUE and EWP across different farm sizes and districts in Punjab. Bathinda stands out with the highest WUE for paddy, demonstrating relatively efficient water use, while Jalandhar displays the lowest WUE for paddy among all districts. Similarly, small farmers exhibit the lowest WUE for wheat production, while large farmers have the highest. The economic productivity of water, expressed in Indian Rupees per cubic meter (INR/m³), also varies, with marginal farms showcasing the lowest economic water productivity for both paddy and wheat. Bathinda district emerges as having the highest EWP for paddy, while Sangrur takes the lead for wheat. The findings underscore the necessity to delve deeper into the determinants behind WUE and EWP, emphasizing the diverse influences of farm size, district location, and socio-economic indicators. This nuanced understanding is crucial for formulating targeted strategies to enhance agricultural water management and resource allocation, ensuring sustainable and economically viable practices.

3.3. Determinants of Water Use Efficiency (WUE)

Table 10. Regression results of variables on WUE of paddy and wheat crops.

Variables	Paddy			Wheat
Log WUE	Coeff.	SE	CI	Coeff.	SE	CI
Log age	−0.0303 **	0.015	−0.0551, −0.0056	−0.0407	0.0317	−0.0929, 0.0115
Log family	0.0119	0.0073	0, 0.0239	0.0010	0.0154	−0.0244, 0.0263
Log depth tubewell	−0.0513 *	0.0099	−0.0676, −0.0349	0.1195 *	0.0212	0.0846, 0.1543
Log lll frequency	0.0028	0.0152	−0.0222, 0.0278	−0.0272	0.0319	−0.0797, 0.0252
Log seed	0.1928 **	0.0852	0.0527, 0.3330	0.3635 *	0.0950	0.2073, 0.5198
Log fertilizer	−0.0255	0.0533	−0.1131, 0.0622	0.2924 *	0.0945	0.1370, 0.4479
Log labour	0.0143	0.0358	−0.0447, 0.0732	0.0442	0.0909	−0.1053, 0.1937
Marginal farm	−0.0130 *	0.0040	−0.0197, −0.0064	−0.022 *	0.0086	−0.036, −0.0079
Large farm	0.0253 *	0.0065	0.0146, 0.0361	0.0522 *	0.0138	0.0296, 0.0748
Medium farm	0.015 *	0.0051	0.0066, 0.0234	0.0387 *	0.0104	0.0216, 0.0558
Semi-medium farm	0.0042	0.0045	−0.0032, 0.0117	0.0126	0.0093	−0.0027, 0.0278
Sikh religion	0.0053	0.0039	−0.0011, 0.0116	−0.0062	0.0081	−0.0196, 0.0072
Electric tubewell	0.0206 *	0.0056	0.0113, 0.0299	0.0073	0.0119	−0.0123, 0.0270
Internet usage	0.006 **	0.0031	0.0009, 0.011	−0.0114 **	0.0064	−0.0220, −0.0008
Bathinda district	0.0383 *	0.0059	0.0286, 0.0480	0.0338 *	0.0125	0.0134, 0.0543
Jalandhar district	−0.0075	0.0052	−0.0161, 0.001	−0.023 *	0.0085	−0.0370, −0.0089
PUSA44 variety	−0.0398 *	0.0040	−0.0463, −0.0333	-	-
PR126 variety	−0.0083	0.0051	−0.0167, 0	-	-
Education (12th)	0.0041	0.0042	−0.0028, 0.0111	0.0068	0.0090	−0.0078, 0.0216
Education (10th)	−0.0003	0.0042	−0.0072, 0.0066	0.0136	0.0089	−0.0009, 0.0282
Education (5th)	−0.0058	0.0057	−0.0152, 0.0036	0.0035	0.0121	−0.0164, 0.0234
Education (6th)	0.0154 **	0.0076	0.0029, 0.0280	−0.0222	0.0158	−0.0482, 0.0038
Education (7th)	0.029 **	0.0145	0.0052, 0.0528	−0.0072	0.0307	−0.0577, 0.0434
Education (8th)	−0.0045	0.0046	−0.0120, 0.0030	0.0029	0.0097	−0.0131, 0.0189
Education (BA)	0.0032	0.0049	−0.0049, 0.0113	0.0098	0.0103	−0.0071, 0.0267
Education (Illit)	0.0070	0.0047	−0.0007, 0.0147	0.0037	0.0099	−0.0126, 0.02
Cons	−0.5176 *	0.2135	−0.8689, −0.1664	−1.8279 *	0.3173	−2.3498, −1.306
Observations	246			246
F-value	28.80			27.64
Adjusted R2	0.75			0.72

Source: Authors’ calculations based on field survey in selected districts. Natural log (ln) is used. Note: * and ** refer to significance at 1 and 5 percent level.

presents the results of a double log multiple linear regression analysis, which provides insights into the relationship between the dependent variable, the WUE of paddy and wheat crop, and independent variables.

The analysis indicates that the model is statistically significant. In the case of paddy, the adjusted R-squared value of 0.75 suggests that approximately 75 percent of the variability in the dependent variable is explained by the model. The age of farmer, depth of tubewell, seed rate of paddy, size of farmland, type of tubewell used, access to the internet, and district-related regional variations (Bathinda district) are found to be statistically significant in explaining the changes in the WUE of paddy crop. A one percent increase in the farmer’s age decreases the WUE of paddy crop by 0.03 percent. Similarly, a one percent increase in the tubewell depth decreases paddy’s WUE by 0.05 percent. On the other hand, a one percent increase in the seed rate of paddy increases the WUE of the paddy crop by 0.19 percent. Different categories of farm size show varied impacts on the WUE of paddy. Larger farms (large and medium) tend to have higher WUE. Compared to small farms, large and medium farms show 2.5 and 1.5 percent higher WUE in paddy, respectively. Similarly, the marginal farm size shows 1.3 percent less WUE than small farms. Farmers’ access to the internet is also statistically significant and has a positive impact of 0.5 percent on the WUE of paddy compared to non-users. The PUSA44 variety of paddy (44% of sampled farmers) significantly explains the negative variation in the WUE of the paddy crop of 4 percent compared to more recent and less water-consuming varieties such as PR130.

Table 10 shows that the type of tubewell (electric) is statistically significant in explaining the positive impact on the WUE of paddy crop of two percent when compared to the diesel tubewell. Bathinda district is statistically significant in explaining the positive regional variation in the WUE of paddy crop of around 4 percent compared to Sangrur district. All independent variables are free from multicollinearity after checking for VIF. District dummies (Sangrur) were removed from the analysis for the same reason.

In the case of wheat crop, the model explains approximately 72 percent of the variability in the dependent variable. Variations in the WUE of wheat crops are explained by the depth of the tubewell, seed rate of wheat, fertilizer usage, internet access, size of the farm, and regional variations. A percent increase in the depth of the tubewell, seed rate of wheat, and fertilizer usage increase the WUE of wheat by 0.12, 0.36, and 0.29 percent, respectively. Marginal farm size negatively explains the variations in WUE of around 2 percent compared to small farms. On the other hand, medium and large farm sizes explain the positive variations in the WUE of wheat crops of around 5 and 4 percent, respectively, compared to small farms. Bathinda and Jalandhar districts are statistically significant in explaining the positive (3%) and negative variations (2%) in the WUE of the wheat crop, respectively, compared to Sangrur. Access to internet usage by farmers negatively impacts the WUE of wheat crop by 1 percent compared to those who do not use the internet.

The intricate interplay of variables such as crop type, soil characteristics, climate, and irrigation practices underscores the complexity of achieving optimal WUE. As we transition to the estimation of EWP, the focus will shift to assessing the efficiency with which water resources are converted into monetary benefits, recognizing that WUE alone does not capture the full spectrum of agricultural productivity.

3.4. Determinants of Economic Water Productivity (EWP)

Understanding the intricate dynamics of economic water productivity is paramount for sustainable resource management and food security. This study employs multiple log-linear regression to investigate the determinants of economic water productivity (EWP) in the cultivation of two staple crops, paddy and wheat. EWP, representing the benefit obtained per unit of water input, stands as a crucial metric in optimizing agricultural practices for water conservation and enhanced productivity. The investigation focuses on several key independent variables, including the age of farmers, family size, farm size, well depth, district-specific factors, educational qualifications, religious affiliations, economic status, internet accessibility, tubewell types, man-days worked, and kg of fertilizer used. By concurrently examining the factors influencing EWP for both paddy and wheat, this regression analysis aims to provide a comprehensive understanding of the intricate relationships that govern EWP in diverse agricultural settings. The outcomes of this research carry significant implications for developing targeted strategies to enhance EWP in crop production, contributing to the broader goal of sustainable and resilient agricultural systems. The model can be expressed as below.

Table 11. Regression results of variables on EWP of paddy and wheat crop.

Variables	Paddy			Wheat
Log EWP	Coeff.	SE	CI	Coeff.	SE	CI
Log age	−0.0344 **	0.0151	−0.0592, −0.0095	−0.0396	0.0314	−0.0913, 0.0121
Log family	0.0128 ***	0.0072	0.0009, 0.0247	0.0047	0.0154	−0.0206, 0.0299
Log depth tubewell	−0.0514 *	0.0099	−0.0677, −0.0351	0.1181 *	0.0212	0.0831, 0.1530
Log lll frequency	0.0045	0.0151	−0.0205, 0.0295	−0.0311	0.0317	−0.0834, 0.0212
Log seed	0.1922 **	0.0848	0.0522, 0.3322	0.3676 *	0.0941	0.2121, 0.5231
Log fertilizer	−0.0227	0.0530	−0.1102, 0.0649	0.2510 *	0.0943	0.0953, 0.4068
Log labour	0.0113	0.0357	−0.0477, 0.0702	0.0236	0.0906	−0.126, 0.1732
Log farm prices	0.1319 **	0.0722	0.0127, 0.2512	0.2645 ***	0.1563	0.0064, 0.5226
Marginal farm	−0.0137 *	0.0040	−0.0204, −0.0071	−0.0221 *	0.0085	−0.0361,−0.0082
Large farm	0.0244 *	0.0065	0.0136, 0.0351	0.0548 *	0.0137	0.0322, 0.0775
Medium farm	0.0148 *	0.0051	0.0064, 0.0232	0.0395 *	0.0103	0.0224, 0.0565
Semi-medium farm	0.0041	0.0045	−0.0033, 0.0115	0.0128	0.0092	−0.0023, 0.028
Sikh religion	0.0052	0.0038	−0.0011, 0.0115	−0.0060	0.0081	−0.0193, 0.0073
Electric tubewell	0.0203 *	0.0056	0.0111, 0.0295	0.0077	0.0118	−0.0118, 0.0273
Internet usage	0.0064 **	0.0031	0.0014, 0.0114	−0.0105 ***	0.0064	−0.0210, 0
Bathinda district	0.0387 *	0.0059	0.0291, 0.0483	0.0263 **	0.0124	0.0058, 0.0469
Jalandhar district	−0.0113 **	0.0052	−0.0198, −0.0027	−0.0146 ***	0.0084	−0.0286,−0.0007
PUSA44 variety	−0.0311 *	0.0042	−0.0390, −0.0232	-	-
PR126 variety	−0.0083	0.0051	−0.0167, 0.0001	-	-
Education (12th)	0.0051	0.0042	−0.0029, 0.0130	0.0087	0.0089	−0.0061, 0.0235
Education (10th)	−0.0003	0.0041	−0.0071, 0.0065	0.0143	0.0088	−0.0002, 0.0288
Education (5th)	−0.0049	0.0057	−0.0143, 0.0045	0.0045	0.0120	−0.0153, 0.0243
Education (6th)	0.0173 **	0.0077	0.0047, 0.03	−0.0178	0.0158	−0.044, 0.0083
Education (7th)	0.0330 **	0.0145	0.0089, 0.057	−0.0081	0.0304	−0.0584, 0.0422
Education (8th)	−0.0037	0.0046	−0.0112, 0.0039	0.0037	0.0096	−0.0122, 0.0196
Education (BA)	0.0044	0.0049	−0.0038, 0.0125	0.0112	0.0102	−0.0057, 0.0281
Education (Illit)	0.0070	0.0046	−0.0007, 0.0147	0.0043	0.0098	−0.0119, 0.0205
Cons	0.3811	0.2993	−0.1133, 0.8755	−1.1869 **	0.5348	−2.0703,−0.3035
Observations	246			246
F-value	29.38			25.79
Adjusted R2	0.75			0.72

Source: Authors’ calculations based on field survey in selected districts. Natural log (ln) is used. Note: *, **, and *** refer to 1, 5, and 10 percent level of significance.

provides insights into the relationship between the dependent variable, WP of paddy, wheat crop, and independent variables.

In the case of paddy, the model has a statistically significant F-statistic of 29.38, indicating that the overall model is a good fit for the data. The coefficient of determination (adjusted R-squared) is 0.75, suggesting that approximately 75 percent of the variability in the economic water productivity of paddy crops is explained by the independent variables in the model. The age of the farmer, household size, depth of tubewell, seed rate of paddy, use of electric tubewell, farm output prices and usage of the internet by farmers are significant explanatory variables. A percent increase in the farmer’s age decreases the paddy crop’s EWP by 0.03 percent. A percent increase in household size contributes positively to the EWP of paddy by 0.01 percent. Similarly, a one percent increase in tubewell depth decreases paddy EWP by 0.05 percent. On the other hand, a one percent increase in the seed rate of paddy increases the EWP of paddy crop by 0.19 percent.

In the case of paddy crop, the adjusted R-square value of 0.75 suggests that the independent variables explain approximately 75 percent of the variability in the EWP of paddy crop. The results further show that a one percent increase in paddy farm output prices contributes positively to paddy EWP by 0.13 percent. Farm size (medium and large) is statistically significant in explaining the positive change in the EWP of paddy crops of around 1.5 and 2.5 percent, respectively, compared to small farms, while marginal farm size explains negative variations in the EWP of paddy of about 1.3 percent. The type of tubewell (electric) is statistically significant in explaining the positive impact of the use of diesel tubewell on the EWP of paddy crops of around 2 percent compared to diesel tubewell. Farmers’ access to the internet is also statistically significant and has a positive impact (0.5%) on the EWP of paddy compared to non-users. Regional/district-related variations explain the regional impact on the EWP of paddy crops, where Jalandhar exhibits negative returns (1%) for farmers. In comparison, Bathinda exhibits positive returns (4%). The PUSA44 variety of paddy exhibits a significant negative impact on the EWP of paddy crop of 4 percent compared to newer varieties like PR130. The highest attained education, as primary school, affects the EWP of the paddy crop positively (around 1.5 to 3%) compared to those who have post-graduate degrees. The variance inflation factor (VIF) has been checked for all variables, and the model has been adjusted after removing Sangrur district.

In the case of wheat crop, the adjusted R-square value of 0.72 suggests that the independent variables explain approximately 72 percent of the variability in the EWP of wheat crops. Variations in the EWP of wheat crops are explained by the farm output prices (positive), depth of tubewell (positive), seed rate of wheat (positive), fertilizer usage (positive), internet access (negative), size of the farm (varied), and regional variations (varied). A one percent increase in the depth of the tubewell, seed rate of wheat, and fertilizer usage increase the EWP of wheat by 0.11, 0.36, and 0.27 percent, respectively. A one percent increase in farm output prices contributes positively to paddy EWP by 0.26 percent. Larger farms have a particularly strong positive impact on the EWP of wheat of 5 percent compared to small farms, while medium farms have a strong positive impact on the EWP of wheat of 4 percent compared to small farms. In contrast, marginal farms exhibit negative returns of about 2 percent compared to small farms. Jalandhar and Bathinda districts significantly affect EWP, with Bathinda having a positive impact (3%) and Jalandhar having a negative impact (1.5%) compared to Pathankot. The variance inflation factor (VIF) has been checked for all variables, and the model has been adjusted after removing Sangrur district.

4. Discussion

Despite high rice yields and substantial production, Punjab, India, faces food stress due to increasing water scarcity. The Central Ground Water Board report (2022) shows severe groundwater depletion, with 164% groundwater development, indicating ‘Physical’ scarcity where demand exceeds supply [87]. Government policies favoring higher minimum support prices (MSP) and power subsidies have led to a monoculture dominated by wheat and rice. Consequently, the crop diversification index has declined, reflecting the reduced cultivation of alternative crops like pulses, oilseeds, fruits, and vegetables from 2005 to 2015 [69,71,88]. This lack of diversification exacerbates water resource overuse and threatens the sustainability and resilience of Punjab’s agriculture.

The study delves into the WUE and EWP of paddy and wheat crops across diverse farm sizes and districts in Punjab, shedding light on critical factors influencing agricultural water management. The findings offer valuable insights and have significant implications for policy and sustainable farming practices.

A related study conducted on agricultural water use efficiency in India found the average water productivity for wheat to be 0.95 kg/m³, exhibiting a range of 0.69 to 1.65 kg/m³ [61]. The global average for wheat water productivity stands at 0.90 kg/m³, while Punjab emerges as one of India’s highest economic water-productive states for wheat, registering at 1.88 kg/m³ [61]. In the context of the present study, which assesses physical water use efficiency (physical WUE) in selected districts, the estimated range for these districts falls between 0.566 ± 0.03 kg/m³ and 0.976 kg/m³, contingent upon the size of the farm. The cited study [61] estimates a 0.57 kg/m³ physical productivity and a 10.85 INR/m³ or $0.13/m³ economic productivity of water in the paddy crops of Punjab. This is almost in the range of what the present study here estimates at 10.50 INR/m³ or $0.13/m³ for all the districts sampled, ranging from 9.84 INR/m³ or $0.12/m³ to 11.47 INR/m³ or $0.14/m³. Other national studies reported the WUE of paddy as 2.28 kg ha⁻¹ mm⁻¹ [89], 0.55 kg grain m⁻³ [90], and 1.64 q acre⁻¹ inch [91]. Alternatively, international studies report a WUE of paddy under similar flooding methods of 0.54 to 0.66 kg grain m⁻³ in the Philippines [92] and 0.45 to 0.73 kg m⁻³ in Japan [93]. In the case of wheat, similar national studies report WUE as 1.09 kg m⁻³ [94]. Internationally, it is reported to range from 1.53 to 2.92 kg m⁻³ in China [95]. Our findings support the consensus that larger farms tend to exhibit higher WUE and economic water productivity (EWP) [83,84,85,86]. This is consistent with other research emphasizing the benefits of economies of scale in agricultural efficiency and productivity [83,84]. A key difference between our study and others is the inclusion of variables such as rainfall and evapotranspiration [89,91,95]. While other studies may incorporate these factors to provide a broader climatic context at the district level, our study focuses on irrigation inflow to offer a more precise measurement of water use efficiency at the farm level. Differences in agricultural policies, such as subsidies and the adoption of technology and crop varieties, can also influence WUE and EWP. Our study’s context-specific focus on Punjab’s policies and practices provides a unique perspective for similar regions.

Bathinda stands out with the highest WUE in paddy, emphasizing efficient water utilization. The influence of farm size on WUE is evident, with larger farms consistently exhibiting higher WUE, while marginal farms face challenges in optimizing water use. In wheat cultivation, large farms again showcase higher WUE, emphasizing their efficient use of water resources. Factors such as the farmer’s age, farm size, tubewell type, and district-related variations played significant roles in impacting paddy WUE, and internet access also positively impacted it. Previous studies also report that variance in farm costs is also linked to tubewell type, whether electric or diesel-operated, with electricity being more economical than diesel fuel, resulting in farmers using electric tubewells incurring lower irrigation expenses than those using diesel tubewells. These disparities in input costs can influence farmers’ allocation of resources [96]. Furthermore, larger landlords enjoy the most efficient and cost-effective combination when irrigating with electric tubewells in Punjab [97]. The PUSA44 variety showed negative returns compared to newer and less water-consuming varieties of paddy like PR126 and PR130.

There are a few omitted variables in this study such as temperature, rainfall, government expenditure, and soil type, as the data are cross-sectional and these variables are inconsistent with the farm-level data used here. This data deficiency poses a challenge, as even if we use dummy variables, there is not much variation, leading to similar values. These are macro-level variables and a macro-level study that uses district-level data as the basic units of analysis can better capture the influence of these variables. Hence, in this study, these variables are not expected to influence the outcomes directly. Nonetheless, they do impact water use efficiency and productivity. Since this is not a large macroeconomic study and the districts are relatively homogeneous and lie in a single state, this study will not be able to capture the full impact of these variables on efficiency and productivity. Rising temperatures and CO₂ levels significantly affect water use efficiency (WUE). Elevated CO₂ levels typically improve WUE by reducing stomatal conductance, which decreases water loss per unit of carbon gained. However, high temperatures can counteract these benefits by inducing heat stress, thereby lowering photosynthesis efficiency and increasing water loss through transpiration [32,98]. Precipitation patterns also play a vital role. Drought conditions usually reduce WUE due to lower soil moisture, leading to increased plant stress and decreased productivity. On the other hand, excessive rainfall can cause waterlogging and reduce soil oxygen availability, negatively impacting plant growth and WUE [99]. The timing and intensity of droughts, influenced by rainfall patterns, affect WUE across various ecosystems, with droughts at different growing season stages having distinct impacts [100]. Subsidies on investment costs are more effective in promoting water-saving transformations [101]. Government policies, including subsidies, can enhance irrigation water use efficiency [102]. Soil texture is crucial in how plants respond to water stress, with loamy soils generally offering better water retention and higher WUE compared to sandy soils [103,104].

The EWP and WUE of wheat vary across farm sizes and districts, with Sangrur emerging as a notable wheat performer. Large farms consistently demonstrate higher economic water productivity, signaling the importance of scale in optimizing returns from water inputs [83,84,85,86]. Factors like the depth of tubewell, seed rate, fertilizer usage, farm size, and regional variations were influential. Both medium and large farms contributed positively to WUE, while marginal farms exhibited negative returns. Internet access had a negative impact on wheat WUE. There have been studies that may not necessarily report a direct positive relation between internet usage and farm productivity. Utilizing the internet for activities unrelated to farming may hinder the enhancement of Total Efficiency (TE) [105]. The research indicates a more pronounced correlation between internet access and productivity among younger households. The impact appears less significant in the production of crops where extensive government intervention in farming practices and pricing is present [106]. Notably, internet usage exhibits a “U-shaped” effect on efficiency in agricultural green production [107].

Household size and farm harvest prices are found to be the two differentiating factors that positively affect EWP. Furthermore, as per the definitions adopted in the study, WUE remains unaffected by the demand and supply conditions of the crops, whereas these conditions significantly influence EWP since EWP also depends on the value of the farm harvest price per hectare realized by the farmers. This is further established by the positive correlation found in the study between EWP and farm harvest price. The same has been reported by earlier studies as well [108,109,110]. Similarly, studies have also reported the correlation between household size and farm productivity [111,112].

The study identifies the need for tailored approaches for marginal farmers and districts with lower efficiency and productivity that are essential for achieving sustainable and economically viable agricultural practices in Punjab. It is essential to note that the groundwater-stressed districts (Jalandhar and Sangrur) in the state reveal higher WUE and EWP for wheat compared to paddy, while safe groundwater districts (Bathinda and Pathankot) reveal higher WUE and EWP for paddy compared to wheat. This comparative advantage is important for policymaking decisions.

5. Conclusions

The study provides insights into the WUE and EWP of paddy and wheat crops across various farm sizes and districts in Punjab. Notably, considerable variations in WUE and EWP were observed, with Bathinda standing out for its high WUE in paddy cultivation and Sangrur for wheat production. Large farms consistently exhibited higher efficiency and productivity, emphasizing the importance of scale in optimizing water use. Regional disparities highlighted the need for localized interventions to enhance WUE and EWP. Factors such as farm size, technological adoption, and regional variations significantly influenced WUE and EWP. Larger farms generally demonstrated higher efficiency, while fertilizer usage and labor variables were found to be inefficient. Furthermore, the study identified the negative impact of certain crop varieties (PUSA44) and varied internet access impact on WUE, particularly in the context of extensive government intervention in farming practices. Overall, the findings underscore the importance of targeted policy interventions to improve WUE and EWP in Punjab. Tailored approaches for marginal farmers and districts with lower efficiency are crucial for promoting sustainable and economically viable agricultural practices.

Policymakers should encourage water-saving technologies like drip irrigation, sprinkler systems, and soil moisture sensors. Providing subsidies or low-interest loans for these technologies can incentivize farmers to adopt them. Since small and marginal farmers might find it difficult to afford and manage these technologies, credit facilities should be directed at them to avoid unequal access to these resources, that benefits larger farms more than smaller ones. Promoting crop varieties with higher WUE, phasing out inefficient varieties like PUSA44, and aligning agricultural subsidies and minimum support prices with water conservation goals can improve water efficiency and productivity. Policies should incentivize sustainable farming methods, such as crop rotation, organic farming, and using green manure, and develop localized strategies for different districts. Farmers may resist change due to the risks involved, especially if the new practices are initially less profitable and the rapid implementation of new crop varieties might disrupt local ecosystems. Hence, it is essential to start these policies in small pockets as pilot projects and their success may encourage farmers in other regions. Additionally, implementing and managing subsidy programs and training initiatives on a large scale can be challenging. Ensuring that the benefits reach the intended marginal farmers without bureaucratic delays or corruption is critical and cooperation between departments and levels is essential. For regions with high groundwater exploitation like Jalandhar and Sangrur, stricter groundwater regulations and incentives for alternative practices are needed. Establishing robust monitoring systems to assess water use and productivity will help track the effectiveness of interventions and make data-driven decisions. Developing this requires significant investment in technology and human resources and ensuring accurate and timely data collection and analysis is a substantial challenge which would require cooperation from farmers, governments and other local stakeholders.

Research studies (like in the case of China, Iran, California) suggest that to support marginal farmers, implementing subsidies and low-interest loan programs for water-efficient technologies, developing training programs focused on efficient water management practices, and strengthening agricultural extension services are crucial [113,114,115]. Extension officers should work closely with farmers to implement best practices and facilitate access to high-yield, drought-resistant, and water-efficient seed varieties. Government programs and NGOs can help distribute these seeds at subsidized rates [116]. Promoting community-based water management programs, including forming water user associations, can help manage and monitor water usage collectively [117]. Improving market access for marginal farmers and linking them with value chains to sell their produce at fair prices can enhance their economic stability and enable investment in sustainable practices [118]. Developing crop insurance schemes and soil health improvement initiatives can protect farmers from losses due to water scarcity or crop failure, encouraging them to adopt innovative farming practices [119].

6. Limitations and Scope for Further Research

Despite efforts to ensure rigor, this research bears several limitations. The sample size may not fully represent the diverse population of farmers in Punjab, thus restricting the generalizability of the findings. Due to time constraints, the comprehensive analysis and comparison of WUE and EWP across different crop types (high and low water value crops) were not feasible. As such, the study primarily focuses on staple crops, and the broader comparison between high and low water value crops remains unexplored for further research. Establishing causality between variables and WUE and EWP can be complex and regression models in the study can suggest associations but may not definitively establish causation due to the presence of unobserved variables that might influence the outcomes. Finally, the cross-sectional analysis was conducted at the farm or micro level and instrumental variables such as rainfall, temperature, soil type, and government expenditure are macro-level variables. There is a data deficiency for these variables at the micro level. Given the granularity and focus of micro-level data, it is believed that the variation induced by these district-level variables is minimal and does not significantly impact the results.