Urban agriculture and farmers’ willingness to pay for treated wastewater: Insights from vegetable producers in the greater Accra metropolis of Ghana

Abstract

Urbanisation and water scarcity pose a challenge to urban vegetable production and livelihoods in major cities across the world. At the same time, increasing demand for fresh vegetables coupled with the high profitability of vegetable production enterprises has made it attractive and indispensable as it creates jobs along the entire crop value chain. This study analysed the efficiency performance of urban vegetable producers and their willingness to pay (WTP) for treated wastewater using data collected from 214 farmers. By applying the data envelopment analysis (DEA) and logistic regression models, the results show that the current production system is technically inefficient as 33% more output could have been produced using the existing resources. The mean amount that farmers are willing to pay for treated wastewater to be delivered to their farms is GH¢66.7 per month. Technical efficiency influences farmers’ WTP for treated waste water. Other drivers of WTP for treated wastewater include experience in vegetable production, type of irrigation practised, source of water used for irrigation, the volume of water applied per day, extension access, association membership, market demand for products, and access to market information. Strengthening public–private partnerships in the treatment and utilization of water resources and providing technical training to urban producers could enhance the efficient use of treated wastewater in agriculture.

Keywords:

• Efficiency
• treated wastewater
• willingness to pay
• vegetable production
• Ghana

PUBLIC INTEREST STATEMENT

Smallholder farmers form the central pillar of most agrarian economies. The availability, accessibility, and adoption of improved production technologies are anticipated to trigger productivity increases, better incomes for farmers and enhance their welfare. Financing the development, scaling-up, and uptake of improved technologies is critical to improving the welfare of smallholder farmers. Evaluating the impact of improved technologies delivered to smallholder farmers to ascertain the level of changes that has taken place as a result of the intervention has wider implications for governments, donors, programme implementers, and beneficiaries. This study assessed the impact of technology adoption on smallholder farmers’ welfare in Ghana.

1. Introduction

Wastewater reuse, especially for irrigated agriculture, is growing globally due to the increasing scarcity of refreshed water resources (Drechsel et al., 2022; Mara & Sleigh, 2010). The increasing reuse of wastewater can be attributed to the high cost of freshwater, unavailability of freshwater, and the increasing demand for freshwater both for agriculture and non-agriculture use (Mara & Sleigh, 2010; Woltersdorfa et al., 2016). In recent years, the use of treated wastewater is increasingly being promoted by governments and development practitioners due to its health implication for producers and consumers (Becerra-Castro et al., 2015; World Health Organization, 2006). Wastewater recycling is a viable engineering solution to the challenge of freshwater supply (Rice et al., 2016) and an efficient approach to managing wastewater in cities (Del-Haghi et al., 2020). Wastewater use not only affects soil productivity and fertility but also poses serious health risks to humans and the environment (Antwi-Agyei et al., 2016; Mara and Sleigh, 2010Becerra-Castro et al., 2015).

Wastewater reuse has been broadly categorized into treated and untreated. Untreated wastewater is “sewage from the household, municipal, and industrial sources,” while treated wastewater is “wastewater that has gone through cleaning processes to improve its quality” (Rice et al., 2016: pp. 4). This study adopts this definition to examine the efficiency of treated wastewater reuse in smallholder urban vegetable production and farmers’ willingness to pay (WTP) for the product. A private company (Sewage Systems Ghana Limited, SSGL) is currently exploring options to increase its investments in the treatment and delivery of wastewater to support urban agriculture production. Improvements in wastewater treatment plants are vital to reducing the costs of wastewater treatment (Hernandez-Chover et al., 2018). Treated wastewater has been applied in fish farming (Gebrezgabher et al., 2015) and vegetable production with positive outcomes (Amponsah et al., 2016; Antwi-Agyei et al., 2016; Azanu et al., 2018; Magwaza et al., 2016; Woltersdorfa et al., 2016). It is also a good source of nutrients (nitrogen, phosphorus, potassium, and magnesium) for plant growth and development (Woltersdorfa et al., 2016). Nonetheless, high levels of faecal coliform bacteria and worm eggs in wastewater have been reported which makes it unfit for irrigation (Ensink et al., 2002). Also, the huge investments required for the operation and maintenance of wastewater treatment plants are a concern and the costs and benefits associated with the activity remain unclear (Hernandez-Chover et al., 2018).

Global estimates show that over 200 million urban dwellers depend on urban agriculture which supplies 15–20% of the World's food (Armar-Klemesu, 2000). Urban agriculture significantly affects households’ income positively and improves livelihoods (Mupeta et al. 2020). In Ghana, rapid population growth coupled with urbanisation and migration is expected to cause more people to reside in urban areas. In the Greater Accra Region, only 51% of the population has direct access to portable clean water and 48% of the water is used for agricultural purposes (Water Resources Commission, 2018). Thus, freshwater demand far exceeds its supply due to economic growth and population increase (Adank et al., 2011). Weak enforcement of laws and legislation is also an issue (National Water Policy, 2007; Yeleliere et al., 2018). Urban agriculture will be critical to increasing the urban food supply and creating jobs for the teeming population. At the same time, urban agricultural production will be constrained by water shortages, land access, and the effects of climate change. Technology adoption in the water value chain coupled with internalising externalities would be relevant in addressing these constraints. Hence, the need to find alternative water sources that support production activities for improved urban food and nutrition security. The growing interest in investing in wastewater treatment facilities to support urban agriculture production is very essential and topical. Understanding the current level of production efficiency of urban vegetable producers and their WTP for treated wastewater is crucial in supporting the investment decisions of waste treatment companies, policymakers, investors, and the government. As the demand for water is increasing, alternative water resources for production need to be explored for sustainable production.

2. Literature review

2.1 Use of treated wastewater in Ghana

The possibility of recovering nutrients from treated wastewater through the use of advanced technologies for crop production exists (Iqbal et al., 2022; Magwaza et al., 2016). As the demand and price for freshwater rise, treated wastewater has become a necessary alternative to agricultural production. Furthermore, the sale of treated wastewater becomes a new revenue stream for companies engaged in the water supply value chain, hence, creating employment opportunities for people (United Nations World Water Assessment Programme, 2017). As previously observed by Obuobie et al. (2006), the use of wastewater, whether treated or untreated, provides indirect economic and social benefits for smallholder irrigated vegetable farmers, traders, and consumers. These benefits include employment for farmers and traders, easy access to fresh vegetables, high profits, and family food supplementation (Obuobie et al., 2006; Wongnaa et al., 2019). The use of treated wastewater for fish farming provides new business opportunities for entrepreneurs while improving the nutritional status of citizens. Empirical evidence shows that, in Ghana, consumers are willing to pay for fish (Catfish and Tilapia) produced using treated wastewater (Gebrezgabher et al., 2015). The treatment and reuse of wastewater in urban farming have been shown to enhance greenhouse gas mitigation and directly save groundwater (Miller-Robbie et al., 2017).

The number of wastewater treatment plants in Ghana has been increasing as various reports suggest from 44 in 2001 to 66 in 2011 (Drechsel & Keraita, 2014; Environmental Protection Agency EPA, 2001; Murray & Drechsel, 2011). Of the 66 treatment plants identified, only 4.5% were centralised urban wastewater treatment plants and 16.7% were fully functional. The majority (63.3%) of the wastewater treatment plants were either partially functional, non-functional, or under construction. Some of the functional and partially functional plants are owned and operated by private enterprises such as hotels and restaurants (Drechsel & Keraita, 2014). The fewer number of wastewater treatment plants in urban areas is a hindrance to treated wastewater reuse relative to the large number of smallholder-irrigated vegetable producers. A number of guidelines have been adopted to help minimize crops contamination through wastewater irrigation. However, most wastewater-irrigated vegetable farmers do not apply these guidelines partly because they do not have the capacity to treat and there is a lack of enforcement of regulations (Amponsah et al., 2016; Antwi-Agyei et al., 2016; Obuobie et al., 2006). Also, many farmers are unaware of the regulations guiding wastewater usage in vegetable production (Drechsel & Keraita, 2014). A knowledge gap, therefore, exists in the use of treated wastewater. Furthermore, by-laws of the Accra Metropolitan Assembly acknowledged that wastewater can be used but it must be treated to a satisfactory level (AMA, 2017 pp. 160). This could result in changing water demand and the availability of wastewater treatment plants.

2.2 Production efficiency and farmers’ willingness to pay for treated wastewater

Urban agriculture impacts significantly on households income, food supply, earnings, savings and has the potential to improve the livelihoods of people (Bolang & Osumanu, 2019; Mupeta et al., 2020). It is a key strategy for supporting viable, productive, and sustained urban habitats (Ruma & Sheikh, 2010). Knowledge and experience are necessary for understanding the perceptions of the public and acceptance of wastewater reuse (Rice et al., 2016). The challenges associated with waste management and use are linked to various factors including social, disposal, water governance restrictions, and allocation system provisioning (Adu-Boahen et al., 2014; Water SMART Solutions Ltd, 2019). These challenges do not only limit community development efforts but impact directly on economic growth. Legislations have also not kept pace with advances in waste reuse technology and what really drives the implementation of wastewater reuse remains unclear (Water SMART Solutions Ltd, 2019).

Antwi-Agyei et al. (2016) assessed the level of awareness, knowledge, and health risk associated with wastewater use among farmers and market salespersons. They found that health risk awareness was high among salespersons and consumers than farmers and that consumers do not prioritize health indicators when buying products from vendors. The purchasing behaviour of consumers was based on cost, friendship, convenience, taste, and freshness of produce. Farmers’ awareness of health risks did not influence their adoption of safer farm practices, suggesting the need for improved knowledge of proper irrigation management among urban irrigators (Pratt et al., 2019). In examining the link between risk attitudes of individuals and the choice of consumers for fruit and vegetable consumption in Italy, Giampietri et al. (2021) reported that risk aversion positively influences consumers’ choice of recommended vegetables consumed. Previously, Ruma and Sheikh (2010) examined the status of wastewater reuse in urban agriculture and its implications on households in Nigeria. The reuse of wastewater was found to be a full-time occupation for households and sole income-earning activity that cities cannot do without. Yet, no official priority has been given to this crucial activity in the urban planning process. A bibliometric analysis of wastewater use in agriculture shows that research in the field has improved significantly at the technical level but the elimination of heavy metals and social aspects remains (Lopez-Serrano et al., 2020).

Hernandez-Chover et al. (2018) analysed the efficiency of wastewater treatment facilities in view of the huge investments required for operation and maintenance. Scale economics was found to significantly influence the efficiency of wastewater treatment processes (impact directly on operational costs). In assessing the costs and benefits associated with treated wastewater use in peri-urban agriculture, Ensink et al. (2002) found the increasing use of wastewater and a clear WTP for untreated wastewater by farmers. The growing use of wastewater was attributed to the lack of money for treatment as well as the scarcity of freshwater. The production of vegetables with high demand (e.g. lettuce) has been found to be profitable in Ghana, and the use of irrigation in open spaces allows significant profits (van Veenhuizen & Danso, 2007; Wongnaa et al., 2019). Since profitability is directly linked to efficiency in production, farmer experience, access to credit, and household size positively influence profitability (Wongnaa et al., 2019). Adeoye (2020) in assessing the efficiency of vegetable production reported that farm size, seed quality, amount of fertilizer and agrochemicals are directly linked to vegetable output. The main factors that significantly influence the technical efficiency production of vegetables were extension contact, household size, and level of education. However, the mean efficiency varies from the south to the north, suggesting that geographical location impacts technical efficiency performance. Previously, Agbonlahor et al. (2007) revealed a mean technical efficiency (TE) of 0.74 for urban vegetable producers in Nigeria and the main sources of inefficiency identified were labour used, land, and seed resources. Akamin et al. (2017) revealed that farmyard manure was the most productive input though equipment and labour also matter in the TE analysis of vegetable farmers. Females and educated farmers were significantly more effective in production, and the mean TE was 0.67. As farm size increases, farmers’ TE reduces. Furthermore, Rosli et al. (2020) estimated TE among pepper producers in Malaysia and reported a mean value of 0.518, suggesting inefficiency in production. Educational level, membership to a farmer-based association, full-time pepper farming, and participation in pepper training are significant factors influencing TE.

Chopra and Das (2019) assessed the willingness of urban households to pay for the operational and maintenance costs of a local wastewater treatment plant. They found the willingness of consumers to substitute a wastewater treatment plant to cover their non-potable water uses should freshwater prices rise high. The co-provision of such public goods can become an important supplement to urban municipal finance. Zekri et al. (2016) estimated WTP for treated wastewater for 400 irrigation farmers in Oman. The ownership and management of wastewater treatment plants was by private companies who were willing to sell excess treated wastewater to farmers to recover their costs. The average amount farmers were WTP for the treated wastewater was 0.111 OR per m³ and the income of farmers and market-orientation of the farm significantly influences WTP. Del-Haghi et al. (2020) revealed that about 65% of farmers were willing to pay for treated wastewater at a price of 1,500,000 Rials per ha, similar to freshwater irrigation charged. About 18.3% of farmers expressed their WTP the highest bid of 1,800,000 Rials per ha with a greater majority (91.7%) willing to pay the lowest bid of 1,200,000 Rials per ha. Furthermore, Donkoh (2019) examined farmers’ WTP for weather forecasts in northern Ghana using the contingency valuation framework and the double hurdle model. The average WTP for weather forecast through text messages was GH¢122.15 annually. Sex, education, aim of production, adaptive capacity, and climate information services significantly and positively influence WTP, while age and access to credit negatively influence the same.

Farmers’ WTP for treated wastewater is influenced by several factors such as health risk perception, education, and living place (Del-Haghi et al., 2020). Bolang and Osumanu (2019) revealed that sex, education, age, household size, finance, and formal workload influence formal sector workers participation in urban agriculture. Ambreen et al. (2020) reported that farmers’ WTP for treated wastewater is positively influenced by age, health risk, education, income, land ownership, area of cultivation, farmers’ awareness of the benefits of wastewater use, and productivity perceptions. About 70% of the farmers were willing to pay for the treated wastewater. Similarly, Alcon et al. (2010) revealed that family size and recreational purposes influence households WTP more to preserve the ecological status of the Segura River. Education and information impact greatly on farmers’ WTP, while lower profits negatively influence it (Del-Haghi et al., 2020). Farmers with high health concerns made lower bids on WTP (Alfarra et al., 2013).

In sum, the review shows that studies that have jointly analysed the efficiency performance and farmers' WTP for treated wastewater are rare which this study sought to address. Government involvement in wastewater treatment is minimal as the private sector tends to dominate the process. Perceptions of health risk and farmers’ knowledge and application of the guidelines relating to wastewater use in vegetable production require further deepening. High operational and maintenance costs of treatment plants are the major challenge hindering the delivery of treated wastewater to farmers for production. Finally, a number of factors have been revealed that influence farmers’ decisions to use treated wastewater including the cost, accessibility, income, and farm size. Attention should be paid to these variables to help increase acceptability and efficiency in production.

3. Methodology

3.1 The study area

The study was conducted in the Greater Accra Region of Ghana where about 90% of the fresh vegetables consumed are produced within the city by smallholder farmers using mainly untreated wastewater from drains (Mara and Sleigh, 2016). There are over 1,000 smallholder vegetable farmers in the city, cultivating about 47 ha in the rainy season and 116 ha during the dry season (Obuobie et al., 2006). Watering cans are largely used to collect wastewater from drains, dugouts, and streams for irrigation. In some instances, water pumping machines are connected to the drains and the water pumped to the fields. Common vegetable crops normally produced are onion, cabbage, lettuce, Amaranthus spp., pepper, eggplant, and spring onion, which are sold to market women and food vendors locally. Vegetable production is done all year round due to the availability of untreated wastewater.

3.2 Sampling, data, and variables

Data was collected from 214 farmers operating in six (6) sites across three districts in the Greater Accra Region (Table 1). About 28% of the sample was collected from the Korlebu site (highest) with the least sample (4%) obtained from the Okponglo site. Sewage Systems Ghana Limited, a private company based in Accra, is promoting the uptake of water-smart solutions in the city of which wastewater is an integral part. Under the initiative, the company will treat wastewater and supply it to urban vegetable farmers for reuse in production. To ensure sustainability and efficiency in production, it is necessary to assess farmers’ WTP for the treated wastewater.

Table 1. Vegetable production sites and sample selection

District	Site/Area	Frequency	Percentage (%)
Ayewaso West	CSIR Compound	38	17.76
	Dzorwulu	30	14.02
	Roman Ridge	27	12.62
	Legon	23	10.75
	Okponglo	9	4.21
Ayewaso Central	Plant Pool	28	13.08
Ablekuma South	Korle-Bu	59	27.57
Total		214	100.00

Lists of vegetable farmers operating at the various sites were collected from their respective association executives, and random sampling was applied. The selection of the sites was based on production activities and the number of farmers operating in the area. Sites considered to be actively engaged in all-year-round production and marketing were selected.

There are different types of techniques used to treat sewage water that make it suitable for agricultural irrigation. According to Iqbal et al. (2022), micro-filtration techniques result in a significant reduction in Biological Oxygen Demand (BOD), Chemical Oxygen Demand (COD), total Suspended Solids (TSS), and Total Bacterial Count (TBC) of wastewater, thus making it safe and suitable for irrigation. For wastewater discharged from industrial sectors, the Microbial Fuel Cell (MFC) technique is recommended (Iqbal et al., 2022). Electrocoagulation and coagulation are utilized to remove oxidable chemicals from wastewater. However, greater efficiency is associated with the electrocoagulation process in terms of COD removal. The characteristics of wastewater have been categorized into physical parameters, chemical, and biological. The physical parameters include total dissolved solids, colour/turbidity, temperature, hardness, and electrical conductivity. Chemical parameters of importance include COD, the concentration of cations and anions (Nitrogen, Phosphorous, Chlorine, Sulphates, and Calcium), and heavy metals. The biological parameters include BOD and microbial life in wastewater (bacteria, protozoa, virus, etc.). These parameters are used in evaluating the quality of wastewater for agricultural use. The key variables included in the study and their measurements are captured in Table 2 based on various literature sources consulted.

Table 2. Variables and measurement

Variable	Symbol	Measurement	Expected Sign
Seed Cost	SEEDC	Cost of seeds (GH¢)	negative
Quantity of seed	SQ	Quantity of seed planted (kg)	positive
Quantity of fertilizer	FQ	Quantity of fertilizer applied (kg)	positive
Cost of fertilizer	FERTC	Fertilizer cost (GH¢)	negative
Quantity of pesticides	PQ	Quantity of pesticides applied (litres)	positive
Cost of pesticides	PESTC	Cost of pesticides (GH¢)	negative
Volume of water	VWATR	Volume of water used daily for irrigation per hectare (m^3).	Positive/negative
Water fees	WATERFEE	Amount spent on water bills or in lifting water per month (GH¢)	negative
Labour	LAB	Total number of people working permanently on the farm	positive
Total revenue	TR	Total farm income obtained per year (GH¢)	positive
Experience	Experience	Number of years a farmer is engaged in vegetable production	positive
Land size	LAND	Area put under vegetable cultivation for the 2019 season (Ha)	positive/negative
Type of irrigation practiced	TYPIRG	1 if watering can technology is used, 0 otherwise	positive
Water source for irrigation	WSIRG	1 if the main source of water used is tap/borehole, 0 otherwise	positive
Water availability	WAV	1 if water is available for all year round production, 0 otherwise	Positive/negative
Irrigation frequency	IRFQ	1 if crops are irrigated daily, 0 otherwise.	positive
Access to extension services	EXTENSION	Number of extension visits received by a farmer from extension agents.	positive
Demand for produce	DEMAND	1 if there is demand for the vegetables produced locally, 0 otherwise	positive
Access to market information	MINFO	1 if farmer have access to market information, 0 otherwise	positive
Farmer perception	PERCEPTION	1 if farmer perceive the quality of treated wastewater to be good, 0 otherwise	Positive/negative
Guaranteed market	GMAKT	1 if farmer have access to guaranteed markets, 0 otherwise	positive
Production efficiency	EFFICIENCY	Estimated technical efficiency score for each farm	positive
Association membership	ASSOCIATION	1 if farmer belongs to a producer/marketing/processing group, 0 otherwise	positive

Source: Authors compilation from various sources

3.3 Theoretical framework: Data envelopment analysis

The use of multiple inputs to produce multiple outputs is a characteristic feature of most irrigated vegetable farm enterprises as assumed in this study. Every vegetable farm enterprise sets weights through an optimisation process for efficiency maximisation subject to constraints. The main idea behind the Data Envelopment Analysis (DEA) approach is that a firm that uses fewer inputs relative to its peers in achieving the same output level is technically more efficient. DEA uses a linear programming technique to create an envelope around the frontier. Both input and output data are usually observed for each farm, and technical efficiency (TE) is computed using the model below:

(1) $Mi{n}_{\theta \lambda }\theta ,$(1)

Subject to $-y_i+Y\lambda \ge 0,$

$\theta x_i-X\lambda \ge 0,$

$NI{ }^{\mathrm{\prime }}\lambda =1,$

$\lambda \ge 0$

Where: $\theta$ = scaler, λ = vector of constants, NI = vector of ones, $x_i$ = input vector, and $y_i$ = output vector. $\theta$ and λ are used by the model to solve simultaneously for each farm and consideration is given to the largest contraction radial for the input vector in the technology setup. Where the contraction corresponds with $\theta$ represents the TE score for the i^th farm and the values can range from zero to one. A score of one indicates an efficient operation of a farm.

Certain constraints are usually imposed to allow for model estimation. The first constraint restricts the output produced by the i^th farm from exceeding that of the frontier. The next constraint restricts the proportional decrease in output, given that $\theta$ is minimised to the input use achieved. The final constraint is the convexity condition which gives rise to variable returns to scale (VRS) specification which assumes optimal scale operation of the farms (Fraser & Cordina, 1999) and that an increase in input amounts does not automatically result in a proportionate increase in output. This epitomizes the situation of smallholder urban vegetable producers in the Greater Accra Metropolis, making it appropriate in this study. To facilitate comparison, allocative efficiency (AE), as well as economic efficiency (EE), were computed (Coelli et al., 2002) for each farm based on Fare et al. (1994) as contained in equation 2(2) $Mi{n}_{\theta \lambda }{\theta }^k,$(2) .

(2) $Mi{n}_{\theta \lambda }{\theta }^k,$(2)

Subject to $-y_i+Y\lambda \ge 0,$

${\theta }^k{x}_i^k-X^k\lambda \ge 0,$

$x_i^{n-k}-X^{n-k}\lambda \ge 0,$

$NI{ }^{\mathrm{\prime }}\lambda =1,$

$\lambda \ge 0$

Where k = variable input. The k^th input is inclusive of$x_i^k$ and $X^k$ and exclusive of $x_i^{n-k}$ and $X^{n-k}$ input columns. The value of ${\theta }^k$ arising from the constraints captures the maximum reduction in the k input within the technology, Ceteris paribus. All other variables remained the same as in equation (1)(1) $Mi{n}_{\theta \lambda }\theta ,$(1) .

The empirical analysis was conducted following a one-step approach to estimate the efficiency level of individual farms and identify the sources of inefficiency. Using econometric software (LIMDEP version 10), the DEA model was computed and efficiency scores for each farmer were determined alongside the factors that influence efficiency performance. The factors which influence technical efficiency performance differentials among the sampled farmers are not reported in the current study since that is not the focus of discussion. However, the TE scores estimated were used as an independent variable in the WTP model to establish whether it impacts farmers purchasing decisions for the treated wastewater or not.

3.4 Contingency valuation method and logistic regression

The theoretical framework adopted for this study is underpinned by the theory of consumer behaviour where a rational consumer aims to maximise utility from any bundle of goods subject to a given construction. The Contingency Valuation (CV) approach has been implemented in diverse fields of study including health economics (Donaldson et al., 2006; Smith, 2003), agriculture and climate science (Donkoh, 2019; Idris et al., 2022; Lazaridou et al., 2019), and water resource (Ambreen et al., 2020; Chopra & Das, 2019; Del-Haghi et al., 2020). Analysis based on a single question and that of the double-bounded CV methods have been used, though the latter yield more efficient estimates (Hanemann et al., 1991). Nonetheless, the double-bounded approach is without criticism as it could introduce bias arising from indignation (Watson and Ryan, 2007). The issue of bias, therefore, constitutes the main limitation in using CV methods. Strategies that minimized these biases were employed using the single-question approach as reported in recent studies (Donkoh, 2019; Idris et al., 2022). The survey was explained to farmers in terms of the hazards posed by using untreated wastewater to produce vegetables for consumption and the solution in terms of treatment and the benefits that comes with it. Arrow et al. (1993) emphasised the usefulness of CV methods in generating relevant information that supports firm decisions. Hence, its adoption in the current study.

Previous studies have utilized the logistic regression model to uncover the factors that influence WTP (Collins-Sowah et al., 2013; Donkoh, 2019; Idris et al., 2022). The approach is suitable for examining the relationship between categorical predictors and binary categorical dependent variables (Gujari, 2004). We employed the logit model to analyse the association between farmers’ WTP for treated wastewater and some explanatory variables. The logistic model of the WTP is represented as:

(3) $log\left(\frac{B_i}{1-B_i}\right)=\alpha +\gamma Z_i$(3)

Where

$\left(\frac{B_i}{1-B_i}\right)$ represent the odds ratio in favour of WTP; $\alpha$ is the intercept; $\gamma$ is the vector of coefficient to be estimated; and $Z_i$is the vector of explanatory variables that influence farmers’ WTP for treated wastewater. The log of the odds is assumed to be a linear function of the explanatory variables

The empirical model estimated for the i^th farm is of the form:

(4) $WTP={\alpha }_0+{\alpha }_1(Experience)+{\alpha }_2(Extension)+{\alpha }_3(LAND)+{\alpha }_4(TYPIRG)+{\alpha }_5(WSIRG)+{\alpha }_6(WAV)+{\alpha }_7(IRFQ)+{\alpha }_8(VWATR)+{\alpha }_9(Demand)+{\alpha }_{10}(MINFO)+{\alpha }_{11}(Perception)+{\alpha }_{12}(GMAKT)+{\alpha }_{13}(Efficiency)+{\alpha }_{14}(Association)+{\epsilon }_t$(4)

Where:

WTP = willingness to pay for treated wastewater and ${\epsilon }_t$ = error term. The definition and measurement of all other variables remain the same as captured in Table 2.

4. Results and discussions

4.1 Descriptive statistics

Descriptive statistics used in the efficiency analysis, as well as for estimating farmers’ WTP for treated wastewater, are captured in Tables 3 and Table 4, respectively. From Table 3, farmers spent on the average GH¢51.6 on seed and GH¢268 on fertilizers per Ha for the 2019 season. The amount spent on pesticides amounts to GH¢332 per Ha and GH¢115.4 was spent as water fees or in lifting water using motorised pumping machines. About 93.8% of farmers expressed their desire and WTP for the treated wastewater and the average amount they were willing to pay was GH¢66.7 per month. While the minimum amount was GH¢5, the maximum was GH¢350 per month. The mean volume of water to be supplied by Sewage Systems Ghana Limited to meet the demand for farmers is 4.68 m³ per day. The mean total revenue obtainable by farmers is GH¢8,410, suggesting that the industry could likely be profitable. Only 38% of farmers were using pipe-borne water for irrigation with the majority dependent on streams and dugouts/wells for water.

Table 3. Descriptive statistics of the variables used (efficiency analysis)

Variable	Cases	Mean	Std. Dev.	CV (%)	Min.	Max.
SQ	204	15.65	19.41.	124.02	0.20	95.0
SEEDC	204	51.46	57.48	111.69	20.0	384.6
FQ	197	311.66	74.65	23.95	0.0	7500.0
FERTC	201	268.00	312.01	116.42	0.0	2400.0
PQ	205	7.26	16.19	223.0	0.0	200.0
PESTC	206	332.75	46.85	14.07	0.0	3300.0
WATERFEE	96	115.41	146.04	126.54	10.0	1200.0
VWATR	214	4.681	0.278	5.93	0.28	407.31
TR	214	8410.64	891.46	10.59	500.0	56000.0
LAB	195	2.10	1.96	93.33	1.0	11.0

Table 4. Descriptive statistics (determinants of WTP for treated wastewater)

Variable	Obs	Mean	Std. Dev.	CV (%)	Min.	Max.
Experience	214	14.10	10.62	75.31	1.0	46
Demand	214	0.929	0.255	16.68	0.0	1.0
IRFQ	214	0.714	0.307	42.99	0.0	1.0
WSIRG	214	0.384	0.110	28.64	0.0	1.0
WAV	213	0.680	0.467	68.67	0.0	1.0
TYPIRG	214	0.579	0.514	88.77	0.0	1.0
Land	214	0.051	0.042	82.35	0.001	0.29
Efficiency	212	0.771	0.363	47.08	0.017	1.0
MINFO	212	0.952	0.212	22.26	0.0	1.0
Association	195	0.830	0.375	45.18	0.0	1.0
Perception	214	0.864	0.343	39.69	0.0	1.0
GMAKT	212	0.726	0.446	61.43	0.0	1.0
EXT	213	0.892	0.311	34.86	0.0	1.0
WTP	212	0.938	0.240	25.58	0.0	1.0

The majority (42.9%) of the respondents were within the youth category of below 35 years. About 38.3% of the respondents did not have any form of education, 21.4% had completed primary school but the numbers diminish as they move higher to tertiary level. About 7.9% of farmers had Arabic education and could therefore keep records on their farm operations. The majority (94.8%) of respondents were mainly engaged in vegetable farming as the main livelihood activity with lettuce (47.6%) and leafy vegetables (30.2%) being the dominant types produced. The greatest concentration of the production of leafy vegetables is done at the CSIR compound while that of lettuce is at the Korle-Bu site. This is important in linking producers to markets and in reducing search time and cost to consumers. The main sources of information for farmers were extension agents (54.8%) and neighbouring farmers (25.3%). Access to extension services is not an issue as revealed by 89.2% of the respondents. The mean number of extension visits received by farmers for the 2019 production season was 10. The majority (79.6%) of farmers were open to learning new farming and marketing methods. However, males were more open (79.8%) than females (66.6%). Most farmers (57.9%) were using the watering can technology for irrigation. The average household size is five persons and on average two persons work permanently on the farm. Farmers have an average of 14 years of experience in vegetable production, and 83% were members of farmers groups. Access to market information was high among farmers (95%). About 80.3% of farmers currently use less than 10 m³ per day, while 42.8% of farmers irrigate their farms on a daily basis. Farmers’ WTP for treated wastewater is largely influenced by its availability, absence of pathogens/bacteria, and the availability of nutrients. This outcome is consistent with the findings of Idris et al. (2022) who reported high willingness among visitors to pay for environmental preservation.

The Coefficient of Variation (CV) is a useful metric for evaluating the relative dispersion of the data around the sample mean. While some variables such as seed quantity and seed cost had a higher CV, others such as volume of water applied and pesticide costs were low (Table 3). These variations would likely impact the technical efficiency performance of firms as well as the risks and returns associated with vegetable production using wastewater.

4.2 Technical efficiency estimates

The model estimation included 11 variables in total. Total revenue was used as the main output variable that serves as the dependent variable. Five input variables, namely quantity of fertilizer (FQ), the quantity of seed (SQ), the quantity of pesticides (PQ), total labour (LAB), and volume of water applied (VWATR) per Ha were included in the model as input variables. To facilitate the estimation of economic efficiency and allocative efficiency, four price variables, namely, seed cost, fertilizer cost, pesticide cost, and amount of money spent in lifting water or water fees for irrigation, were included. The estimated results (Table 5) revealed that vegetable farmers in the Accra Metropolis are inefficient in all dimensions studied (technical, economic, and allocative). The mean technical efficiency (TE) revealed is 0.77 suggesting that about 33% more output could be achieved using the present resources if the farmers were technically efficient. More, therefore, needs to be done in terms of information sharing and good agronomic practice adoption to help bring these farmers to operate on the frontier and be able to pay for and use treated wastewater sustainably for production. This result compares favourably with those reported in previous studies that analysed the efficiency performance in vegetable production. For instance, in Malaysia, Rosli et al. (2020) reported a mean TE of 0.518 for pepper farmers. Similarly, in Cameroon, Akamin et al. (2017) found a mean TE of 0.67. Furthermore, Rajendran et al. (2015) found a mean TE of 0.67 for vegetable producers in Tanzania and the performance was variable across the four regions studies due to agroclimatic variability, infrastructural facilities, and access to extension services. Tsiboe et al. (2019) reported TE estimates for okra, pepper, and tomatoes to be 0.54, 0.74, and 0.58, respectively, and diseconomies of scale were observed in the vegetable production system. In rural Kenya, Alulu et al. (2021) recently found higher mean TE values for contract farming participants producing chilli pepper and spider plants (0.66 and 0.24) relative to non-contract participants (0.12 and 0.15), respectively. Inefficiency in vegetable production is therefore widespread, but how that impacts farmers' WTP for treated wastewater remains to be established empirically.

Table 5. Efficiency estimates under VRS assumption using DEA

Variable	Mean	Std. Deviation	CV (%)	Minimum	Maximum
TE (Input oriented)	0.775	0.235	30.70	0.277	1.000
TE (Output Oriented)	0.520	0.361	69.42	0.017	1.000
Economic efficiency	0.357	0.277	77.59	0.011	1.000
Allocative efficiency	0.457	0.287	62.80	0.018	1.000

Source: Computed from survey data, 2020

Furthermore, the results show mean economic and allocative efficiencies of 0.35 and 0.45, respectively, suggesting that the cost of factors of production (inputs) significantly impacts resource allocation decisions with implications on the efficiency performance of urban vegetable farms. These statistics are not far different from what has been documented in the vegetable production efficiency literature. For instance, Adams et al. (2020) reported economic and allocative efficiency scores of 0.52 and 0.68, respectively, for small-scale irrigated vegetable-producing farms in northern Ghana.

4.3 Distribution of technical efficiency scores

About 17.7% of the farms were technically efficient as they operate on or closer to the frontier output (Figure 1). The TE scores distribution shows that about 51.8% of the farms were operating below 50% efficiency. This is a concern and confirms the widespread inefficiency in production among urban vegetable farmers (Adams et al., 2020; Akamin et al., 2017; Rosli et al., 2020; Wongnaa et al., 2019). This suggests the need to strengthen extension services delivery and training for farmers to help improve their production practices.

Figure 1. Distribution of technical efficiency scores.

4.4 Determinants of farmers’ WTP for treated wastewater

The logistic regression model was applied in determining the factors likely to influence farmers’ WTP for treated wastewater. The goodness-of-fit test (Table 6) and the specification error test (Table 7) shows that the model is well fitted.

Table 6. Logistic regression results (Dep. Var: WTP for treated wastewater)

Variable	Odds Ratio	Std. Err	t-value	p-value	[95% C. I.]
Experience	0.838	0.060	−2.46	0.014**	0.728	0.965
LAND	0.970	0.035	−0.81	0.417	0.903	1.043
TYPIRG	5.045	3.725	2.19	0.028**	0.903	21.450
WSIRG	0.450	87.978	1.72	0.086*	0.592	3.464
WAV	40.754	40.754	87.978	1.720	0.086	0.592
Association	−0.132	0.078	−1.70	0.090*	−0.285	0.020
IRFQ	3.930	3.439	1.56	0.118	0.707	21.842
VWATR	1.008	0.005	1.89	0.059*	0.999	1.017
Extension	0.136	0.061	2.22	0.027**	0.015	0.256
Perception	0.998	0.001	−1.36	0.173	0.995	1.001
Demand	0.001	0.002	−2.54	0.011**	5.42e-06	0.207
MINFO	0.011	0.019	−2.63	0.008***	0.004	0.319
GMAKT	1.360	0.539	0.78	0.437	0.625	2.958
Efficiency	0.775	0.210	1.22	0.012**	0.032	0.412
Constant	44573.8	259.2	1.84	0.066**	0.483	4.11e+09
Pseudo r-squared		0.585	Number of obs			212
Chi-square		44.846	Prob > chi2			0.000
Akaike crit. (AIC)		59.816	Bayesian crit. (BIC)			104.123

*** p < .01, ** p < .05, * p < .1

Table 7. Link test-specification error for WTP for treated wastewater

	Coef.	Std. Err	z	P>z	[95% Conf. Interval]
_hat	1.089	0.313	3.480	0.000	0.476	1.702
_hatsq	−0.024	0.019	−1.310	0.192	−0.061	0.012
_cons	−0.008	0.619	−0.010	0.989	−1.221	1.205

Note: Observations =212, Number of covariate patterns = 212, Pearson chi2 (161) = 72.70, Prob > chi2 = 1.000

The odd ratios revealed that VWATR has a value of one, suggesting that the exposure does not affect the odds of the outcome. The variables TYPIRG, IRFQ, WAV, and GMAKT had values greater than 1, indicating that the exposures are associated with higher odds of the outcome. Except these, all other variables included in the regression model had odd ratio values less than one, suggesting that the exposures are associated with lower odds of the outcome.

Farmer experience is statistically significant and positively related to WTP for treated wastewater at a 5% level. This is relevant considering the fact that experience influences input use and impacts directly on production efficiency. As documented in Ethiopia, the level of experience in irrigation significantly affects farmers' WTP for wastewater (Alebel et al., 2009). Farmer experience influences the profitability of vegetable farming and has the potential to attract farmers to pay for treated wastewater and other inputs (Wongnaa et al., 2019).

The type of irrigation (TYPIRG) practised also impact farmers’ WTP for treated wastewater. Farmers who use the watering can technology for irrigation are more likely to pay for the treated water than those using motorised irrigation machines. Similarly, the source of water (WSIRG) used has a positive and significant relationship with farmers’ WTP for the treated wastewater. This suggests that controlled water source users such as taps, boreholes, and wells are likely to pay for the treated wastewater to be delivered compared to those using water from drains. This could be attributed to the fact that the latter hardly pay for the water and may be trapped in the mentality of free services. Thus, farmers who use drain water are less likely to be willing to pay for the treated wastewater.

The volume of water required for irrigation (VWATR) positively and significantly influences farmers’ WTP for treated wastewater (Table 6). The small-sized nature of most farms makes it easy to irrigate since not much volume of water is required. With shorter dry spells in southern Ghana relative to the northern parts, the demand for large volumes of water by farmers is not an issue. Furthermore, the demand for the vegetables produced has a positive and significant relation with farmers’ WTP for treated wastewater. This is due to the fact that high demand will translate into higher revenues for farmers; hence, their ability to pay for treated wastewater (invest more into the vegetable production business). Zekri et al. (2016) showed that farms that are market-oriented impact positively on farmers’ WTP for treated wastewater. Kini et al. (2020) showed that the distance travelled by consumers to production sites of organic vegetables influences demand in Burkina Faso. The intended use of vegetables and consumer health awareness are also significant determinants of demand.

Similarly, access to market information has a positive and significant relation to farmers’ WTP. This is because such information reduces search costs and creates more demand for the produce. Thus, consumers who were initially unaware of the existence and exact place to buy such products, get to know. This, therefore, widens the scope of the market for the product. Relational factors including trust and access to market information influence demand for vegetables (Kini et al., 2020). Also, Donkoh (2019) showed that farmers’ access to climate information services positively and significantly impacts their WTP for weather information in Ghana. Access to information is therefore key in WTP analysis.

Additionally, the results show that the efficiency level of production matters for farmers’ WTP for treated wastewater. Farmers who are more technically efficient produce higher outputs. With better market conditions in place, such farmers can earn higher revenues which act as a stimulus to increase investments in farm inputs. However, labour market imperfections affect efficiency in production and technology adoption (Chavas et al., 2005). Participation in pepper training influences the technical efficiency of production (Rosli et al., 2020) with a likely impact on revenues. Farmers with higher gross income and profit are WTP more for the improvement of water supply (Azahara et al., 2012; Tziakis et al., 2009) and higher household income relates directly to the mean WTP for irrigation water (Tang et al., 2013). Annual revenues generated from farms significantly affect farmers’ WTP for wastewater in Ethiopia (Alebel et al., 2009). Household WTP for wastewater increases with an increase in revenues from crop production arising from efficiency in production. Thus, low levels of TE are indicative that most farmers are suffering from access to technology, which affects their WTP for the treated wastewater.

Furthermore, farmers’ access to extension services significantly impacts their WTP for treated wastewater. Extension contact influences the technical efficiency of vegetable production positively (Adeoye, ibid) with a likely impact on access to information (production and marketing). This could also result in the observed positive effects on farmers' WTP for treated wastewater. Donkoh (2019) reported the positive effects of extension access on farmers’ WTP for climate services in Ghana. Finally, association membership of farmers is negatively related to their WTP for treated wastewater. Farmers who belong to a production or marketing group are more unwilling to pay for the treated wastewater. This is significant and highlights the effect of peer influence on farmers’ decision-making. This could also be attributed to strong rules within the associations likely to influence the behaviours of their members. Rosli et al. (2020) observed that membership to a farmer-based organization directly influences the technical efficiency of farmers. However, the findings of the current study are contrary to this earlier study.

5. Conclusions and policy implications

Urban agriculture contributes to food and nutrition security, incomes, and the livelihoods of urban dwellers. Urban agriculture and wastewater reuse are growing yet little is known about the efficiency level of producers and how that influences farmers’ WTP for treated wastewater, especially, in the face of growing freshwater scarcity, increasing population growth, and rising food prices in urban areas. This study analysed the efficiency performance of smallholder vegetable producers and their WTP for treated wastewater in Ghana. The novelty of this study lies in the nexus between technical efficiency and farmers’ WTP for treated wastewater which has not yet been explored in the literature. Using a sample of 214 urban vegetable producers selected across six major production sites in the Greater Accra Metropolis, the study first applied the DEA framework to analyse the technical efficiency level of production of each farm. At the second stage of analysis, the efficiency scores together with other socioeconomic and demographic variables were then applied in the logistic regression framework to identify the factors that influence farmers’ WTP for treated wastewater. The findings revealed that though few farm units currently operate at the efficiency level expected, on average, inefficiency exists within the farm units considered. More specifically, the potential to increase output by 33% with the same level of resources is achievable provided all farmers were operating on the frontier. The majority of farmers (93.8%) expressed their WTP for the treated wastewater to be delivered to their farms and are willing to pay an amount of GH¢66.7 per month. This amount is far below what farmers currently spend on water fees (GH¢115.4). However, with economies of scale in production, the treatment plant (SSGL) could cover its operational costs, especially in the long run. Technical efficiency in production positively influence farmers’ WTP for treated wastewater. Other factors found relevant in driving WTP for treated wastewater are experience in vegetable production, type of irrigation practised, source of water for irrigation, the volume of water applied per day, extension access, association membership, market demand for produce, and access to market information. These findings though specific to the Greater Accra Metropolis reflect the dynamics of wastewater reuse in many cities in developing countries.

Strengthening the technical capacity of farmers through pieces of training could help improve their skills for more efficient production and better use of inputs. Strengthening collaborations and partnerships with private wastewater treatment plants could help sustain the treatment and supply of wastewater for all year round production. Initiatives and policy options that facilitate market information dissemination and create local demand for the vegetables produced are a sure way to encourage urban agricultural production and the use of treated wastewater.

Acknowledgments

The authors are grateful to participants of the International Conference on Development Economics organized by the French Association of Development Economics (AFEDEV) held in CERDI, Clermont-Ferrand, France from the 30^th of June to 1^st July 2022 for their contributions. We are grateful to Mr. Giulio Schinaia who served as a discussant for this paper and provided useful comments.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research did not receive any funding support.

Notes on contributors

Abdulai Adams

Abdulai Adams is a Lecturer at the Department of Economics, Simon Diedong University of Business and Integrated Development Studies, Ghana. He obtained his PhD in Economics from the University of Zululand and has worked as a Research Scientist, Project Manager, Grants Coordinator, and Technical Advisor with a number of international organizations. He has been involved in managing large-scale projects and grants. His research interest covers microfinance, innovation economics, food security, applied economics, and agricultural economics.

Samuel Sekyi

Samuel Sekyi is a Senior Lecturer at the Department of Economics, Simon Diedong University of Business and Integrated Development Studies, Ghana, and a Senior Hall Tutor. He holds a PhD in Economics from the University of South Africa. His research interest covers applied micro-econometrics, agricultural economics, and health economics.

Irrshad Kaseeram

Irrshad Kaseeram is Professor of Economics at the Department of Economics, University of Zululand, South Africa. He has served as the head of the department and deputy director of research and innovation. His research interest includes monetary economics, financial economics, agricultural economics, and macroeconomic modelling. He has published widely in reputable Journals.