Water Use Efficiency and Productivity of Irrigated Rice Cultivation in Nigeria: An Application of the Stochastic Frontier Approach

Abstract

In the face of declining water resources and low agricultural water productivity, it is vital to increase agricultural production efficiency and efficiency of water usage. The efficacy of irrigating rice fields in Nigeria is evaluated here using a stochastic frontier analysis. This is a parametric frontier that is assumed to have half-normal distribution, allowing the model to be separated from normal errors in a composite error model. Samples of 382 surveys were used in the study; out of the total number, 361 surveys were retrieved and used for the analysis. The analytical tools used in the study are: gross margin, water productivity analysis, and stochastic frontier approach. The results indicate an average yield of 4.69 and 4.94 tons/hectare, and net farm income of $415 and $364 for the farmers using canal irrigation and farmers using tube wells, respectively. The results also showed physical and economic water productivity of 0.51 kg/m³ and $0.11/m³ for canal water users, and 0.568 kg/m³ and $0.10/m³ for tube well users. The canal water users had a mean irrigation water use efficiency of 0.76, compared to 0.70 for the tube well users. The study recommends that effective extension services and their coverage be enhanced to provide adequate training to rice farmers on water use efficiency and the transfer of innovations and farming technologies to farmers.

1. Introduction

The current global water challenge is of concern in ensuring food security and sustainability. Agriculture is one of the primary consumers of water withdrawals globally, accounting for over 70% [1]. Due to competing water needs by different sectors of the economy, it has become a scarce resource in many countries [2]. Countries with abundant water have now entered the category of water stress [3]. To feed a growing global population, particularly in developing nations, would require more irrigated land and more efficient use of existing resources [4]. Africa has an irrigation potential of 204 million hectares, however, only 13 million hectares are currently equipped for irrigation [5]. Compared to 37% in Asia, and 14% in Latin America, only 6% of Africa’s cultivable land is irrigated [6], with more than two-thirds of this land located in Egypt, Madagascar, Morocco, South Africa, and Sudan. Rapid population growth and poor agricultural productivity have renewed calls for more irrigation development and resource use efficiency in Africa [7,8].

Low input use efficiency and small landholdings between 0.5 to 2.5 ha have characterized the farming system [9]. As a result, water and labor productivity are low [10]. Nigeria is Africa’s second-largest rice producer [11], and biggest in West Africa [12]. However, rising demand has outpaced local production, making the country the second-biggest rice importer after China [13]. Although the government has been implementing several development programs to increase rice productivity and production, less work has been devoted to research and training to improve farming households’ efficiency [14]. The average paddy production in rainfed, and irrigated agriculture is 2.0 and 3.5 MT/ha, respectively [15]. This low productivity is mainly due to low-tech agricultural equipment and low technical efficiency [16]. According to [17,18], Nigeria’s future food self-sufficiency can only be fulfilled by expanding irrigated land and improving resource use efficiency.

Rice has become Nigeria’s most important agricultural crop [19]. This is evidenced by the country’s annual consumption, which has increased from less than 1.0 million metric tons (MT) in the 1960s to more than 6.9 million MT in 2019 [20]. The country’s yearly rice intake per capita is 46 kilograms [21]. Nonetheless, during the previous two decades, consumption has expanded at a rate close to four times that of the global consumption growth [22]. Rice accounted for 10% and 6.6% of overall consumer spending, respectively [23]. In 2017, the total annual consumption was 6.4 million tons, accounting for over 20% of total consumption in Africa [22]. Although irrigated rice ecosystems account for 54% of worldwide cultivated rice acreage, and 75% of global rice output, they account for just 17% of farmed land in Nigeria [24].

Irrigation Water Usage Efficiency (IWUE) is a concept introduced by [25]. Since then, several studies have been undertaken to determine the efficacy of irrigation water utilized in various regions of the world [26,27,28,29,30,31,32]. The notion of efficiency analysis is a single-factor, input-oriented calculation of Technical Efficiency (TE), a fundamental aspect of production economics [26]. Due to increasing water shortages and the need to produce more food, water use efficiency and water productivity have captured the attention of the world’s foremost experts [33,34,35]. Apart from increasing planted land, raising the paddy yield may be done by substantially using inputs and advanced technology, or enhancing farmers’ resource use efficiency. While increasing rice yield through improved technology and input use is a positive strategy, it requires more significant input expenditure, which can be a hurdle for smallholder farmers [36,37]. Nigeria’s rice output is expected to reach 5 million tons in 2021, a 58% increase over 2013, but significantly below the local demand of 7 million tons. The average yield per hectare is only 2 tons, less than half of the global average, and the majority of the rise in output is attributed to increased planted area, rather than productivity gains (Figure 1). It is better to increase rice output by improving technological efficiency, especially in Sub-Saharan Africa, where funding is low and acquiring more input is difficult. Following on from the prior study, this study aims to evaluate rice farmers’ IWUE and water productivity in the KRIP.

Rice cultivation in Nigeria is characterized by low productivity; Nigeria’s average rice yield (2.0 t ha⁻¹), and average water productivity (WP) of 0. 35 kg m³, are not comparable to the world average’s paddy output of 4.3 t ha⁻¹ and WP of 0.58–1.23 kg m³ [38]. Various studies have studied the efficiency of paddy production in Nigeria [9,16,39]. However, there has been no study on IWUE in rice production. In seeking to address the research gap above, the study intends to compute irrigation water efficiency and water productivity in rice farming between farmers using canal water and those using tube wells, and also to identify the determinants of IWUE. Using a stochastic frontier approach, researchers used 384 paddy farms to evaluate irrigation water efficiency. The results of this research could be used by policymakers and agricultural extension workers in Nigeria to provide recommendations for increasing the efficacy of irrigation water used in rice cultivation.

What follows is a breakdown of the article’s remaining content. This is followed by a discussion of the conceptual framework and an overview of how the concept of water production and irrigation water efficiency was developed. The information and key points of the study areas are shown in the third section. Last, but not least, certain policy ramifications are discussed and summarized.

2. Conceptual Framework

IWUE refers to a rise in the efficiency with which finite irrigation water resources are used [40]. For irrigated agriculture, the IWUE can be calculated as the ratio of yield to administered water. Farmers can increase IWUE by enhancing crop management. Improving the resource’s beneficial use is the most effective method for improving water use efficiency [41,42]. Socioeconomic, institutional, and improved agronomic practices are the key determinants of improving the beneficial use of irrigation water (Figure 1). It is critical to use water resources responsibly and to improve IWUE. Physical water productivity (PWP, kg m³), increasing income based on irrigation water usage, and economical water productivity (EWP, US$ m³) are therefore critical to the long-term viability of irrigated agriculture and water use [43]. Increasing PWP in irrigated agriculture minimizes the demand for extra water, a risky solution to rising water scarcity [42,44].

Moreover, IWUE, water productivity, yield, and net farm incomes were evaluated by the use of different indicators, including socioeconomic, institutional, and farmer management practices (Figure 2).

The interaction between socio-economic indicators, institutional, and crop management practices affect IWUE at the farm level. For example, in rice production, improved cultivars can achieve a higher yield per acre than local cultivars, but the latter also have more labor demands for other cultural practices [45]. In comparison to manual weeding, the usage of herbicides can result in higher yields and water productivity, because of the decreased labor needs. Water adequacy, dependability, equity, and efficiency affect water productivity [46]. Maikasuwa and Ala [47], also found that education effectively influences small-scale farmers’ decisions on efficient water use. Contrary to this, [29], concluded that the educational level of respondents does not improve IWUE. These examples show how farm performance is affected by the interconnections between management techniques, socio-economic factors, and institutional issues.

3. Methodology

3.1. Study Area

There are more people living in Nigeria than in any other country on the African continent, making it the seventh most populated country in the world [48]. It was projected that there were 203 million people living in the nation as of the year, 2019, with 51.4% of them residing in rural regions, and with a population density of 221 persons per square kilometer [49]. The number of people living in poverty and uncertainty about their food supply is particularly high in Nigeria [50,51,52]. The Federal Government of Nigeria is the owner of the Hadejia-Jama’are River Basin Development Authority (HJRBDA), which is responsible for the administration of the Kano River Irrigation Project (KRIP). Kano State is confined by latitudes 10°3′ N to 12°4′ N of the equator, and longitudes 7°4′ E to 9°3′ E of the prime meridian [53]. The state has four (4) distinct seasons: Rani (warm and dry), Damina (wet and warm), Kaka (cold and dry), and Bazara (hot and dry) are Kano’s four diverse seasons. The average annual rainfall is 884 mm, but it differs greatly by region, from 600 mm in the north to 1100 mm in the south. August is the wettest month, with the most rainstorms and sediment deposition, while the average annual temperature varies from 26 °C to 32 °C, with a high diurnal temperature of 13.1 °C, and a relative humidity ranging from 17% to 90% [53].

As indicated in Figure 3, the project utilizes Tiga Dam to provide irrigation water to three local government areas in Kano state: Kura, Bunkure, and Garun Malam. The study area is located between 11°38′59″ and 11°37′53″ north latitude, and 8°25′28″ and 8°27′26″ east longitude [21].

The weather of the area is marked by fluctuations in both temperature and humidity that are extensive in scope and occur at a rather quick pace [54]. Humidity, at times, could rise to 100% in an area considered naturally dry. The daily lowest and most extreme temperatures are 14 °C and 35 °C, respectively [21].

Sampling Procedure and Sample Size

Farmers who use canals’ water and those that use tube wells to water their farms are the respondents who took part in the research. A multi-stage sampling procedure (MSP) was employed for the sample selection. At the beginning, all three local government areas (LGAs) of the canal irrigation command area (Kura, Bunkure, and Garun Malam) were chosen for the field study. The second phase of the sample selection was done at the irrigation community level. Twenty-four (24) communities were chosen, 12 from the irrigation command area, and 12 from the non-command site (where farmers use tube wells). The twenty-four communities were chosen purposively, due to high rice production in the communities.

Phase three of the fieldwork involved the actual survey, whereby detailed data was collected. Due to uneven population, this study conducted a proportionate random sampling of 217 canal water users and 165 tube well users. Out of the total number of 382, only 361 surveys were retrieved and used for the analysis.

Yamane’s formula was used for sample size calculation:

$$n=\frac{N}{1+N{\left(e\right)}^2}$$

(1)

where, N = total population; n = representative sample size; and e = error margin (0.05).

A semi-structured questionnaire and physical water measurement were used to collect input and output data, and quantity of water applied to each rice field selected in the study area. Trained enumerators and a supervisor were used for the data administration and water measurement.

3.2. Analytical Tools

Various statistical and econometric methods were used to examine and synthesize the acquired data (stochastic frontier model, water productivity, and net farm income). Data analysis and summarization were performed using descriptive statistics. To further investigate the efficiency and factors that influence rice farmers’ irrigation water use, an econometric model was developed in Frontier 4.11. To gather information, we employed an interview guide. In the course of the interview, we were able to gather data on a number of inputs and outputs. The inputs are estimated as labor use, including hired (permanent and casual) and family labor in person-days; pesticide and herbicide in liter hectare⁻¹; fertilizer used in kg hectare⁻¹; and surface and groundwater use for beneficiaries and non-beneficiaries in m³ acre⁻¹. We measured the output produced in kg acre⁻¹, and calculated irrigation water volume, employing an approximate estimate model, as employed by [55].

This technique involved using a 22-L plastic bucket for collecting the water that flowed through the siphons. The time required to fill the bucket was timed using a stopwatch. To ensure uniformity, the bucket was set in a tiny dugout, with a surface area roughly matching that of the irrigated patches. Daily hours of water application were multiplied by the flow rate. By multiplying the daily measured quantities by the frequency of water application during the water application period, the total amounts of water applied by the farmers could be determined. The average amount of water used during an irrigation season per acre was estimated by taking the ratio of total water use in cubic meters (m³) by the total farm area in acres.

This formula obtained the total water application for the cropping season:

$${\mathrm{X}}_{\mathrm{w}}={\mathrm{G}}_{\mathrm{r}}·{\mathrm{D}}_{\mathrm{h}}·{\mathrm{F}}_{\mathrm{w}}$$

(2)

where,

• ${\mathrm{X}}_{\mathrm{w}}$ = Approximate amount of water applied for the cropping season in ${\mathrm{M}}^3$
• ${\mathrm{G}}_{\mathrm{r}}$ = Discharge rate liter per second
• ${\mathrm{D}}_{\mathrm{h}}$ = Farmer daily hours of water application
• ${\mathrm{F}}_{\mathrm{w}}$ = Frequency of water application

The actual amount of water applied per hectare was:

$$\frac{{\mathrm{G}}_{\mathrm{r}}·{\mathrm{D}}_{\mathrm{h}}·{\mathrm{F}}_{\mathrm{w}}}{\mathrm{size} \mathrm{o}\mathrm{f} \mathrm{t}\mathrm{h}\mathrm{e} \mathrm{f}\mathrm{a}\mathrm{r}\mathrm{m} \mathrm{i}\mathrm{n} \mathrm{h}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{a}\mathrm{r}}$$

The discharge rate was estimated by applying Michael’s [56] formula:

$${\mathrm{G}}_{\mathrm{r}}=\raisebox{1ex}{$\mathrm{V}$}\!\left/ \!\raisebox{-1ex}{$\mathrm{T}$}\right.$$

(3)

where,

• ${\mathrm{G}}_{\mathrm{r}}$ = Discharge rate litres per second
• $\mathrm{V}$ = volume of the container in liters
• $\mathrm{T}$ = time taken in seconds to fill the container

The Stochastic Frontier Analysis (SFA) models take into account the possibility of output disruption by arbitrary shocks that are beyond the control of the producers. As a result, in these models, it is possible to separate the impact of technical efficiency fluctuation from the impact of random shocks on the output. Aigner et al. [57], introduced this model. An input-focused single-factor technical efficiency metric is irrigation water use efficiency [32]. The idea of input-specific technical efficiency or IWUE, in the case of this study, is illustrated in Figure 4.

For example, three farms, C, D and E, use two inputs (X₁ irrigation water, and X₂ seed to produce Y₀ output). Farm C is said to be inefficient, because it is away from the frontier. Now, let the inefficient farm, Y₀, produce an output, using X_2A amount of seed, and K1 quantity of water. The radiated contraction K1 and K2 produces a projected point, C₀, on the frontier, which is technically efficient. The technical efficiency of farm C is given by the ratio TE_c = OC₀/OC, and the irrigation water use efficiency is obtained by the given the ratio K2/K1. The IWUE measure determines both the lowest possible water use K2, and the determined potential saving in water use K1–K2, without reduction in the prevailing output level ${\mathrm{Y}}_0$.

Estimation of TE and IWUE

$$y_i=\exp (x_i\beta +v_i-{\mu }_i=\exp \left(x_i\beta +v_i\right)\mathrm{e}\mathrm{x}\mathrm{p}(-{\mu }_i)$$

or,

$$In{y}_i=x_i{\beta }_i+v_i-{\mu }_i$$

(4)

where,

• $y_i$ = Vector representing produced quantities by the production unit
• $x_i=$ Input used
• ${\beta }_i=$ Vector of co-efficient

The $v_i$ and ${\mu }_i$ are vectors that represent distinct error components.

The MLS technique offers a quick check for the detection of technical inefficiencies in data. Similarly, if ${\mu }_i$ = 0, then ${\epsilon }_i=0$. Therefore, the error term is symmetric, and there is no proof of technological inefficiency in the data. If ${\mu }_i>0,$ however, there is an indication of technical inefficiencies in the data, as the distribution of ${\epsilon }_i=v_i-{\mu }_i$ becomes negatively symmetric. The efficiency metric described does not provide an assessment of the efficient use of individual inputs, such as irrigation water, in this case. As shown in Figure 4 conceptually measuring IWUE necessitates an estimate of the unknown quantity K2. It is easy to see that $\mathrm{K}2=\mathrm{K}1\ast \mathrm{I}\mathrm{W}\mathrm{U}\mathrm{E}$ by applying the formula IWUE = K2/K1. The outcome can be written as follows, by replacing K2 in Equation (4):

$$ln{y}_i=f\left(X_i,K_i^E,\beta \right)+v_i-{\mu }_i \mathrm{where}, K_i^E=\mathrm{K}2$$

(5)

Equating Equation (4) with Equation (5) and employing the estimated parameter $\beta$ yields the IWUE.

$${\mathrm{W}}_{\mathrm{i}}=\mathrm{f}\left({\mathrm{X}}_{\mathrm{i}},{\mathrm{Y}}_{\mathrm{i}}\mathrm{\beta }\right) \exp ({\mathrm{\epsilon }}_{\mathrm{i}}\equiv {\mathrm{V}}_{\mathrm{i}}-{\mathsf{\mu }}_{\mathrm{i}})$$

(6)

where,

• ${\mathrm{W}}_{\mathrm{i}}$ = quantity of irrigation water used (i = … … … N)
• ${\mathrm{y}}_{\mathrm{i}}$ = is the amount of output produced by the farms
• ${\mathrm{X}}_{\mathrm{i}}$ = is the other production input used
• $\mathrm{\beta }$ = is the unknown coefficient to be an estimate
• ${\mathrm{\epsilon }}_{\mathrm{i}}$ = is the composite error term,
• ${\mathrm{V}}_{\mathrm{i}}~\mathrm{N}\left(0,{\mathsf{\delta }}^2\right) \mathrm{and} {\mathsf{\mu }}_{\mathrm{i}}\ge 0 \mathrm{a} \mathrm{non} - \mathrm{n}\mathrm{e}\mathrm{g}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e} \mathrm{e}\mathrm{r}\mathrm{r}\mathrm{o}\mathrm{r} \mathrm{t}\mathrm{e}\mathrm{r}\mathrm{m} \mathrm{d}\mathrm{i}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{b}\mathrm{u}\mathrm{t}\mathrm{e}\mathrm{d} \mathrm{a}\mathrm{s} {\mathrm{N}}^{+}(0,{\mathrm{\delta }}^2)$

3.2.1. The Inefficiency Model

The inefficiency of IWUE was modeled in terms of the factors that are presumed to affect the farmers’ efficiency of irrigation water use. The determinants of IWU efficiency, (μ_i), are defined by:

μ = δ₀ + δ₁Z₁+ δ₂Z₂ + δ₃Z₃ + δ₄Z₄ + δ₅Z₅ + δ₆Z₆ + δ₇Z_{7 +} δ₈Z₈

(7)

where, μ is a non-negative random term representing inefficiency; the IWU δ₀ is a constant; δ₁–δ₉ are parameters to be estimated; and Z₁–Z₈ are farm-specific and farmer-specific variables, as stated below:

Z₁ = Farmer’s age (years)

Z₂ = Education of the respondents (years)

Z₃ = Experience in farming (years)

Z₄ = Family size (number)

Z₅ = Marital status (1 for married, 0 for single)

Z₆ = Gender (Male = 1, Female = 0)

Z₇ = Contact to extension agents (1 = contact, 0 = no contact)

Z₈ = Access to credit

Z₉ = Annual income (Naira)

Along with the variance parameters δ2 and γ, δ0 and δ1–δ9 are undetermined variables that must be calculated. Here, we derive a relationship between the random error variance δv2, the inefficiency effect variance δu2, and the model variance as a whole:

$${\delta }^2=\delta v^2+\delta u^2$$

and,

$$\gamma =\delta u^2/{\delta }^2$$

Diagnostic statistics, such as the Gamma (γ) and the (${\delta }^2$) coefficients, reveal whether or not the stochastic frontier function is useful, and whether or not the assumptions about the error term’s distribution are reasonable. Gamma (γ) demonstrates that the distributional shape proposed for the composite error term is correct, and that the model is well-fitting. It indicates that the primary sources of random mistakes are not to be found in the production and cost functions.

3.2.2. Physical Water Productivity

Increasing water productivity has been argued by many researchers [58,59,60,61] as pre condition for reducing global food insecurity and poverty among farming household. Water productivity measures how much food can be harvested for a given quantity of irrigation water [62]. Physical or technical water productivity measures the yield of a crop in relation to the amount of water used for irrigation. Technical water productivity (kg/m³) is calculated by dividing yield (kg/ha) by consumptive water use (m³/ha), as stated by [63].

$$\mathrm{PWP}={fx}^{-1}$$

(8)

where, f = output produced in kg, and $x$ = quantity of water applied in m³.

The ratio of the gross value of the product to the volume of irrigation water used is known as economic water productivity [58,64]. It also relates to agricultural water use’s economic benefits and expenses [61]. It can be stated in several ways, including net profit per water unit [65]. Ronald and Marlow [64] define economic water use efficiency as the value of the product generated per unit of water consumed.

Mathematically,

$$\mathrm{EWP}={Wx}^{-1}$$

(9)

where, EWP = economic water productivity, W = value of the product produced in kg, and $x$ = unit of water consumed in m³, while a return to investments show the returns per unit of cost of irrigation water (unitless).

3.3. Validity and Reliability of the Instrument

The study used a survey method, and trained enumerators were used to administer the questionnaire. The enumerators were rigorously monitored during the survey to ensure that respondents’ responses were accurate. Before data administration, the questionnaire was pre-tested, and adjustments were made in line with the study objectives. Moreover, a reliability analysis was conducted. A Cronbach’s alpha of 0.67 was obtained, which is appropriate for sociological research, according to [66], who believe that Cronbach’s alpha over 0.6 is appropriate.

4. Results

4.1. Cost and Return Analysis of Rice Production in KRIP

Table 1. Cost and return analysis per acre of rice between canal water users and tube well users.

Variables	Canal Water Users		Tube Well Farmers
	* Cost ($/acre)	% to TC	Cost ($)	% to TC
Seed	9.08	2%	9.63	2%
NPK	44.02	9%	28.64	5%
Urea	51.11	11%	73.19	13%
Herbicide	10.53	2%	11.43	2%
Pesticide	9.99	2%	4.35	1%
Cost of fuel (acre)	0.00	0%	66.32	12%
Water charges	5.26	1%	0.00	0%
Cost of bags (acre)	8.30	2%	8.72	2%
Labor cost
Land clearing	9.53	2%	9.92	2%
Ploughing/Ridging	24.17	5%	19.14	3%
Seedbed	18.79	4%	16.87	3%
Planting	27.20	6%	14.24	3%
1st and 2nd weeding	54.34	11%	47.12	9%
1st and 2nd fertilization	8.03	2%	5.95	1%
Agro-chemical spraying	4.26	1%	5.23	1%
Harvesting/threshing	48.21	10%	60.26	11%
Winnowing/bagging	10.69	2%	11.04	2%
Irrigation labor cost	39.64	8%	41.28	8%
Transportation cost	16.17	3%	16.58	3%
TVC	399.32	82%	449.91	82%
Depreciation	8.17	2%	38.69	7%
Rent on land	78.95	16%	59.21	11%
Total cost	486.44	100	547.81	100
Revenue
Average yield (kg/acre)	1970.12		2084.11
Average price/kg of paddy ($)	0.46		0.44
Total farm income ($/acre)	902.14		912.64
Gross margin (TR-TVC)	502.68		462.73
Net farm income (GM-FC)	415.56		364.83
BCR (TFI/TC)	1.85		1.67

* Exchange rate ($1USD = N380).

presents the estimated analysis of the cost and return of rice production in the study area. The results show that the average per-acre cost of production was $484.44 and $547.61 for canal water users and tube well users, respectively. For the canal water users, the input used constitutes 29% of the production cost, labor constitutes 53%, while fixed cost constitutes the remaining 18% of the production cost. Similarly, for the tube well users, the input used constitutes 37% of the total cost, labor constitutes 45%, and fixed cost constitutes the remaining 18%. The gross margin (GM) and net farm income (NFI) obtained by the canal water users and tube wells users per acre were $502.68, $415.45, and $462.73, $364.83, respectively. These indicated that rice production was profitable in the study area. For the benefit-cost ratio, the proportion of unity indicates breakeven, less than unity indicates loss, and greater than unity indicates the profitability of the enterprises. The benefit-cost ratio is used in different analyses to compare profitability between enterprises or groups [67]. For this study, the result indicated ratios of 1.85 and 1.67 for the beneficiaries and non-beneficiaries, respectively, consistent with the conclusion of [39,68] in their analysis of rice profitability in Ebonyi state, Nigeria.

4.2. Estimation of Physical and Economic Water Productivity

Kijne et al. [69] define water productivity (WP) as an indicator of the efficiency with which agricultural systems transform water into food. Therefore, agricultural water productivity measures the output of a system in relation to the amount of water it uses. In terms of time and place, water productivity can be defined and assessed for the entire system, or for individual components [70].

Improving agricultural WP is a severe challenge of food production and safeguarding sustainable livelihoods [69]. The results of water productivity are presented in

Table 2. Economic and physical water productivity.

Indicator	Unit	Canal Water Users	Tube Well Users
Yield	(Kg acre−1)	1970.19	2084.11
Net farm income	(Dollar acre−1)	415.56	364.83
Water applied	(M3 acre−1)	3830.61	3554.92
Water cost	(Dollar acre−1)	5.26	66.32
PWP	(Kg−1M3	0.514	0.586
EWP	(Dollar m−3)	0.11	0.10
Return per m3 water cost		79	5.50

; the results show that the average water productivity of canal water users was 0.514 kg/m³, while that of tube well users was 0.542 kg/m³. This finding was lower than the average range, 0.60–1.60 kg/m³, found in other studies [71], but closely similar (0.56 kg/m³) obtained by [32]. The analysis further shows the average economic water productivity of 0.11 dollar/M³ and 0.10 dollar/M³, respectively, for canal water users and tube well farmers.

The economic water productivity was further confirmed by the return per water cost, which was 79 and 5.68, respectively, for the canal irrigation farmers and tube well farmers, meaning that, for every dollar spent on the water, canal water users had a return of 79 dollars, compared to tube well farmers, who had a return of 5.5 per dollar invested. This result was not surprising, because tube well water users spent 12% of the total cost of their production to purchase fuel for powering generators, which was just 0.08% for the canal users as water charges per production cycle.

4.3. Scattered Plot between Canal Users and Tube Well Users

The scattered plot in Figure 5 indicated that the maximum water productivity of the canal water users was 1.17 kg/m³, while the minimum was 0.72 kg/m³, with a mean water productivity of 0.51 kg/m³. The yield parameter indicated the lowest yield of 787 kg, and the maximum of 3150 kg. For the tube well water users, the statistics showed a minimum, or lowest, water productivity of 0.49 kg/m³, and a maximum of 0.96 kg/m³, with a mean productivity of 0.586 kg/m^3. Among sample farmers, the results of the analysis found a minimum yield of 760 kg/acre, and a maximum of 3225 kg/acre. The scattered plot also shows an equilibrium point of 0.84 kg/m³. At this point, both the canal water users and tube well water users could realize up to 2600 kg of paddy per acre.

4.4. Score Distribution of Irrigation Water Use Efficiency (IWUE)

The degree of IWUE is the fraction of the lowest possible to investigational IWU, restrictive on detected levels of the target output and input bundles. As a result, addressing irrigation water consumption efficiency at the crop level could help guide irrigation decisions on farms.

Table 3. Distribution of irrigation water use efficiency in rice production between canal water users and tube well farmers.

Range	Canal Water Users		Tube Well Water Users
	Frequency	Percentage	Frequency	Percentage
>30	0	0	7	4.6
0.30–0.39	3	2.0	3	2.0
0.40–0.49	2	1.3	6	3.9
0.50–0.59	7	4.6	15	9.9
0.60–0.69	16	10.5	25	16.4
0.70–0.80	46	30.3	46	30.3
0.80–0.89	71	46.7	48	31.6
0.90–0.99	7	4.6	2	1.3
Mean	0.76		0.70
Minimum	0.30		0.07
Maximum	0.91		0.90
Standard Deviation	0.141		0.17

shows a gap that needs improvement in irrigation water use efficiency between canal water users and tube well water users. The average IWUE score for canal water users and tube well users was 76% and 70%, respectively. In other words, the findings showed that the former could save 24%, while the latter could save up to 30% of the current irrigation water usage, without reducing the present output levels.

The results (Table 3) indicated that 18.4% of the canal water users were operating within efficiency score intervals between 0.30 and 0.69. Moreover, 77% of them were operating at an efficiency score between 0.70 and 0.89. The remaining 4.6% were within an efficiency score of 0.90–0.99. Similarly, for the tube well users, 4.6% were operating within the efficiency score of <30%, 15.8% were within the efficiency score of <60, 78.3% were within the score of between 0.60–0.89, while only 1.3% was above 90%.

4.5. Determinants of Irrigation Water Use Efficiency (Canal Water Users)

Table 4. Determinants of irrigation water use efficiency between canal water users and tube well users.

	Canal Water Users			Tube Well Water Users
Variables	Β	Standard Error	t-Ratio	Β	Std. Error	t-Ratio
Constant	−0.4681	1.1088	−0.4222	−1.1184	1.7303	−0.6464
Output produced(kg)	−6.0261	0.0718	−83.887 ***	0.5212	0.1799	0.2897
Seed (kg)	1.7716	0.1387	12.7714 ***	−0.0141	0.0786	−0.1799
NPK (kg)	−0.0323	0.0986	−0.3279	1.6157	0.1905	8.4813 ***
Urea (kg)	−0.0299	0.1072	−0.2793	−0.9081	0.1089	−8.3377 ***
Herbicide (liter)	1.5243	0.0978	15.5865 ***	−0.0573	0.0977	−0.5867
Pesticide (liter)	−3.4143	0.1208	−2.8255 ***	−0.2936	0.1800	−1.6315
Labor (man-days)	0.0148	0.0981	0.1508	0.4178	0.4156	1.0051
Depreciation	1.8443	0.2278	8.0978 ***	0.7844	0.2012	3.8986 ***
Farm size (acre)	−0.0079	0.0696	−0.1133	−0.6042	0.2674	−2.2597 **
Inefficiency Model
Constant	−0.2520	2.4897	−1.0123	4.7390	4.2080	1.1262
Age (years)	−0.0782	0.0347	−2.3009 **	−1.7966	0.1606	−11.185 ***
Education	0.0030	0.0260	0.1144	−0.1503	0.1800	−0.8350
Farming experience	0.1398	0.0695	2.0118 **	−0.1186	0.0276	−4.3028 ***
Family size	0.1056	0.0518	−2.0408 **	−0.1718	0.1532	−1.1216
Marital status	−0.3941	0.9001	−0.4379	2.7372	0.2583	10.5964 ***
Gender of the Household head	0.9062	1.1592	0.7817	0.7474	0.1053	7.1011 ***
Extension visit	−2.1686	1.0835	−2.002 **	−7.2871	0.7904	−9.2200 ***
Access to credit	0.6615	0.4042	1.6364	−0.7618	0.1065	−7.1508 ***
Annual income (N)	0.1747	0.1850	0.9446	−2.5793	0.3119	−8.2702 ***
Sigma square	1.1443	0.3792	3.0180 ***	2.2176	0.1817	12.2040 ***
Gamma	0.8886	0.0429	20.729 ***	0.8734	0.0973	8.9780 ***
Likelihood function	−193.57			−191.05
LR test	331.01			31.5

*** Significant at 1, ** significant at 5%.

presents the results of the estimated determinants of IWUE for rice production in the study area. The estimated coefficient (1.77) for the quantity of seed used was positive and statistically significant. This implies holding other variables constant; there could be a 1.77 increase in the IWUE for every 1% increase in the seed used. The estimated coefficient (1.52) for the herbicide used was positive and significant, suggesting that every 1% increase in the use of herbicide could lead to a 1.52% increase in IWUE. The fertilizers’ coefficient (NPK, Urea) was negative, but statistically insignificant. The coefficient of output obtained (−6.0261) was negative and statistically significant. At the 1% probability level, the positive value of the depreciation coefficient for fixed assets was statistically significant.

Part B of Table 4, which presented the determinants of the inefficiency in the irrigation water use, indicates that there was a statistically significant relationship between farming experience, family size, and extension visit for various combinations of risk factors. There was a statistically negligible negative correlation between farmer age and water efficiency (−0.078). It follows that, as one ages, their ability to conserve water dwindles. The findings provide credence to the claim that aging farmers are less productive. This may occur because farmers lose efficiency with age, and because younger farmers are more likely to embrace and adopt innovative farming practices. Consequently, this finding is consistent with [72,73]. At the 5% confidence level, the coefficient of farming experience was 0.139, suggesting that more experienced farmers were more productive. This is because farmers gain wisdom from their past failures as they gain expertise. The coefficient of extension service is a dummy variable; 1 for access to extension service, and 0 otherwise. Therefore, the negative coefficient of (−2.16) means that farmers with no access to extension services lose efficiency of 2.16%. Alternatively, a 1% increase in extension contact could increase water use efficiency by 2.16%. Education was not statistically significant; this is in line with [74], who found that education does not improve irrigation water use efficiency.

The variance of inefficiency term of the composite error term (σ²) indicates a good fit, and the correctness of the specified distributional assumption of the composite error term is found significant. Gamma, $(\gamma ={\sigma }^{2}u/({\sigma }^{2}u+{\sigma }^{2}v\left)\right)$ which shows the ratio of the deviation from frontier caused by inefficiency is found to be 89 and 87%, respectively, and is significant for both canal water users and tube well farmers.

The need to improve IWUE stemmed from rising demands for water from households and businesses [75,76]. As presented in Table 4, the quantities of NPK and Urea fertilizers, fuel use, and capital asset depreciation were all statistically significant variables influencing IWUE. The coefficient (1.6) of fertilizer (NPK) was positive and significant (5% level) while controlling for other factors. Multiple studies concluded that higher soil nitrogen levels improved water use efficiency by offsetting the inevitable production decline caused by evaporation and transpiration [77,78,79]. Depreciation of fixed assets also had a positive coefficient that was significant at the 5% level of statistical testing. Thus, it may be concluded that farmers who have access to capital employ that money more effectively. The inefficiency model found that, among tube well water users, neither level of education nor household size significantly affected the efficiency with which irrigation water was used. However, there were statistically significant relationships between age, gender of the household head, access to credit, extension services, and annual income.

At the 5% level of significance, the coefficients for age, farming experience, extension contact, access to credit, and annual income were all negative. It was highlighted how each of these factors had a unique coefficient that could be used to lessen water waste. Positive and statistically significant (at the 5% level) coefficients were found for both the gender of the household head and marital status dummy variables. While having a woman as the head of the household can boost efficiency by 0.747%, having a man as the head of the household can reduce IWUE by 2.73%, and the coefficient of extension services was negative and statistically significant, indicating that inefficiency in water use decreased with increasing extension contact. This agrees with the findings of [30], who found that farmers who had greater contact with extension service personnel also had greater productivity.

5. Discussion of Major Findings

The results in (Table 1) indicate that labor accounts for the largest proportion of production costs. This analysis validated the claim made by [80,81] that agriculture in developing nations is labor-intensive. Rice production is profitable in the area of study, as indicated by the net farm income; however, canal users obtained greater profits than tube well water users. This finding is consistent with that of [82], who reported that canal water users have a higher return, due to the inexpensive price of irrigation water. Table 2 demonstrates that users of tube well water have higher water productivity than their counterparts.

The results had two implications. In comparison, the tube well users were more curious to utilize water more productively because of the cost incurred to access the resource. Secondly, the canal water users got more return per m³ of water applied, because they paid less to access the resources. Moreover, the efficiency score indicated that canal water users’ mean efficiency score was slightly higher than that of tube well users. From the results in Table 3, we can observe that the irrigation water use can be reduced by 24% for canal water users and 30% for tube well users, without reduction in the current yield, or the yield can be increased while maintaining the current irrigation water used. Rice farmers in the study area used a flood irrigation system. Flood irrigation, on the other hand, is one of the most important issues restricting irrigation water use efficiency (Kang et al., 2017). Low irrigation water use efficiency, according to [40,83] is linked to poor timing and an uneven distribution of water applications. However, improved agronomic practices, and technical, managerial, and institutional advancements can have a positive impact on water use efficiencies [84,85].

Similarly, [41,86,87] suggested irrigation scheduling based on the crop response at each growing stage as vital for increasing IWUE. Others, such as [88], prioritized deficit irrigation. Table 4 shows the determinant of IWUE and the inefficiency model estimated by Frontier version 4.1. Results of the determinants of irrigation water use efficiency for canal water users indicated that seed used, herbicide, and depreciation on capital asset were positive and statistically significant at a 5% probability level. This indicates that each of these variables could increase irrigation water use efficiency by its corresponding coefficient. This is consistent with the findings of Obianefo et al. [89], who found that seed, agrochemical, and capital assets were significant to the production frontier. The coefficient of the output obtained was negative and significant at a 1% probability level. These results show that water efficiency decreases with an increase in the output produced; in other words, a 1% increase in the output produced reduces water use efficiency by almost 6%. This is not surprising, due to the flood irrigation system practiced in the study area. Some studies [90,91] found an inverse relationship between flood irrigation and irrigation efficiency. For tube well users, depreciation on capital assets and NPK fertilizer applied were found to be positive and significant at a 1% probability level. Many studies [92,93,94] opined that the soil nitrogen level was correlated positively to water use efficiency by enhancing the increase in yield against evapotranspiration. Urea fertilizer applied and farm sizes were negative and statistically significant, indicating over-utilization of these resources.

In the water use inefficiency model, nine influencing factors of IWUE were considered in the present study. However, four variables were found to be significant: the age of the head of the household, farming experience, family size, and extension visit. The coefficient of age, family size, and extension contact were negative, suggesting that each of these variables could reduce inefficiency by its corresponding coefficient. Ng’ombe [95]; Obianefo et al. [96] revealed that gender, farming experience, and interaction with extension agents were among the socioeconomic variables that influenced technical inefficiency. The positive sign of years of experience was probably because, as years of experience increase, age also increases, and farmers are more likely to become conservatives, unwilling to accept modern farming practices. This result is consistent with that of [97], who reported that technical efficiency decreases with increased farming experience. Equally, the finding was contrary to that of [98] who found that water efficiency increases with increasing farming experience. Education had no statistical significance, and was consistent with the findings of [29], who concluded that education does not enhance the efficiency of water use.

6. Conclusions and Policy Recommendations

Water resource management remains a critical component of achieving one of the 2030 sustainable development goals that is “ending extreme poverty, hunger, and ensuring environmental sustainability”. Water scarcity has become a growing social and economic concern for policymakers and other sectors competing for water resources. The study was conducted to assess rice farmers’ water use efficiency and productivity in the KRIP, Nigeria. The study used primary data using questionnaires and physical water measurements. Stochastic frontier, net farm income, and water productivity measures were used for the analysis.

The result show that the average per-acre cost of production was $484.44 for canal water users, and $547.61 for tube well users, respectively. The net farm incomes (NFI) obtained by the canal water users and tube well users per acre were $415.45, and $364.83, respectively. The average water productivity of canal water users was 0.514 kg/m³, while that of tube well users was 0.542 kg/m³. The analysis further shows an average economic water productivity of 0.11 dollar/m³, and 0.10 dollar/m³ for canal water users and tube well water users, respectively. The mean efficiency scores were 76% and 70% for the canal water users and tube well water users, respectively. Farming experience, family size, and extension visits were the socioeconomic drivers of water use efficiency that were statistically significant. Rice growers (canal water users) have relatively good water use efficiency in the project area. From the analysis, it could be observed that there were no significant differences between most of the indicators. However, on the basis profit realized per acre, the results showed that canal water users realized a profit 14% higher than their counterpart. Similarly, there is ample potential to increase water use efficiency for both categories of farmers. Rice farmers in the study area should focus on policies to boost operations through better cultural techniques. Farmers should be educated and mobilized on the necessity of efficient water usage amid global water stress. Water prices need to be revisited to ensure that the project’s operation and maintenance costs are covered. Desirable water prices may lead to good project maintenance, while also teaching farmers that water is no longer a free resource, but rather an economic input such as any other. Similarly, different technological innovations, such as deficit irrigation, should be practiced to manage irrigation water use, while maintaining, or even improving, the current output level.

7. Limitations

To start with, due to time and financial limitations, the survey data was only gathered in one project area. As a result, just 382 farmers make up the limited dataset used in this study. Second, there are issues with sample choice and representation. The respondents for the study were chosen using a multi-stage sample selection process; however, because the method cannot be totally random at all steps, sample selection bias may exist. The choice of 57% canal water users and 43% tube well users may not mirror the accurate proportion in the population. Thirdly, there might be some errors in the way the quantity of the irrigation water use was estimated, since the management of the canal irrigation does not measure the exact water diverted to the field canal, and the farmers using tube wells do not possess meters to observe the actual groundwater use levels.