Effective demand for climate-smart adaptation: A case of solar technologies for cocoa irrigation in Ghana

ABSTRACT Given the generally low adoption of early climate change response technologies among tree crop producers in sub-Saharan Africa, stakeholders interested in the commercialization or scaling of such technologies will require empirical evidence of their market prospects. Using a double-bounded contingent valuation approach, the study evaluated the willingness and ability of 523 Ghanaian producers to invest in solar-powered irrigation pumps (SPIPs) for cocoa irrigation. The sample was split into three segments based on farm size: resource-poor, resource-limited, and resource-rich. Our results show that effective demand increased across the resource segments, with resource-endowed farmers more likely to demand SPIPs than resource-limited or resource-poor farmers. Also, while willingness to invest (WTI) depended on resourcefulness (land), farmers’ ability to invest was directly related to their resource (income class) endowment. We found that WTI across the resource segments was positively influenced by income, education, livestock ownership, credit, and extension services and negatively affected by household size and age of cocoa trees. Among others, we propose that promotional strategies for SPIPs should incorporate well-planned initiatives for income diversification and microcredit services to improve the financial position of the resource-poor and limited segment to encourage the adoption of these technologies.


Introduction
Climate change poses an existential threat to human livelihood (Bellprat et al., 2019;Dwivedi et al., 2022;Ornes, 2018), and the potential to derail global efforts toward poverty alleviation, food security, and sustainable development (Lipper & Zilberman, 2018).Climate change impacts manifest in reduced precipitation, increased flooding, rising sea levels, extended droughts, and higher temperatures.The future precipitation and temperature are subject to significant uncertainty, projected to increase over time (Kaini et al., 2020(Kaini et al., , 2021;;Nepal, 2016;Wester et al., 2019).For instance, future annual precipitation is expected to increase overall, with drier winters and wetter monsoons, increasing precipitation during the monsoon and post-monsoon (Kaini et al., 2020).The average annual temperature is expected to increase, especially during winter.
These variations in climatic conditions are expected to have devastating effects on production activities within climate-sensitive sectors like rain-fed agriculture (Hoffmann et al., 2022;Yang et al., 2022).Kaini et al. (2021) projected that the climate change impacts on the hydrological regime of the Koshi River basin could increase cropping areas for monsoon paddy rice and winter wheat due to future high river flows in Nepal.In Egypt, the irrigation water requirement is projected to grow by 6.2 and 11.8% in 2050 and 2100, respectively, while the yield is expected to fall by 8.6 and 11.1% in each of those years (Mostafa et al., 2021).In Iran, irrigation water requirements rose by 40-80% during 1986-2015, leading to a 60-100% decline in winter wheat yields (Mirgol et al., 2020).However, Kaini et al. (2022) found that for most climate scenarios, changes in future irrigation water demand are negligible, which contrasts with the life-threatening implications of climate change observed in other parts of the world.Nonetheless, by the end of the century (2071-2100), biomass and yields had significantly decreased, showing that even irrigation would not lessen the effects of climate change on crops.Notwithstanding, demand for irrigation systems is anticipated to increase because of the predicted future climate variability and uncertainty.There is, therefore, the need to stimulate the development of irrigation and to adapt the existing irrigation systems to climate change.
Smallholder farmers in developing countries are vulnerable to climate-related shocks.Increasing interest has been in ensuring food and livelihood security through adopting climate-smart agriculture (IPCC Intergovernmental Panel on Climate Change, 2022).Adaptation strategies to climate change impacts are the most effective option for farmers to respond to adverse climate change impacts efficiently, especially if farmers' adaptation actions are tied to appropriate and effective government policies (Dang et al., 2019).Depending on the level of investment required, adaptation strategies can be classified as 'early' and 'passive'.Early response strategies involve deliberate efforts and require substantial investments by farmers, development organizations, governments, and scientists (Nowak et al., 2022).Climate-smart irrigation technologies are a good example of early response strategies.Passive responses to climate change are implicit actions that require no substantial investments at the farm level (Tripathi & Mishra, 2017).While passive strategies have increasingly been applied, early responses, which demand substantial investment, have been limited (Acheampong et al., 2018;Perez et al., 2021).The limited access has been attributed to farmers' incapacity to pay the overhead cost of such technologies (Akrofi et al., 2019;Gwambene et al., 2015;Phiri et al., 2020).
In the cocoa sector, farmers in Ghana and other West African countries are increasingly becoming vulnerable to climate change, particularly climate-induced water stress (Ameyaw et al., 2018;Bunn et al., 2019).Cocoa farmers have taken adaptive measures in response to climate change effects, such as increased wilting and mortality rates in young cocoa trees and reduced production areas and productivity due to climate-induced pests and diseases, and changing temperatures and rainfall (Asante et al., 2017;Denkyirah et al., 2017;Wongnaa & Babu, 2020).Passive responses are dominant adaptation mechanisms to reduce adverse effects (Codjoe et al., 2013;Denkyirah et al., 2017;Kosoe & Ahmed, 2022).These responses include shade management, improved cocoa varieties, crop diversification, switching from cocoa planting to short-rotation crops, and income diversification (Amfo & Ali, 2020;Asante et al., 2017;Denkyirah et al., 2017;Kosoe & Ahmed, 2022).Based on primary data collected from 20 communities in the Dormaa West and Bia East districts of Ghana, Afriyie-Kraft et al. (2020) described nine adaptation practices used by Ghanaian cocoa farmers: extension advice, loans, petty trading, reliance on other cash crops, casual labor, crop diversification, basic farm management, reliance on food crops, and cutting back on expenses.
Adaptation strategies for the cocoa sector in West Africa could include encouraging tree shading, production intensification, and a revival of deforested areas through cocoa farming (Schroth et al., 2016).A growing early response area of interest for cocoa farmers, stakeholders, and the government is the application of irrigation to counteract climate-induced water stress (Gadeberg, 2022).Hucheon et al. (1973) initiated the earliest inquiries into the effectiveness of irrigation for cocoa production, and research has continued over the years (Carr & Lockwood, 2011).In climate-induced water stress, evidence has shown that irrigated cocoa fares better than rain-fed cocoa (Bunn et al., 2019).Yet, Afrifa et al. (2009) and van Oosten et al. (2022) highlighted the lack of popularity of irrigation among cocoa producers.
In recent years, the Ghana Cocoa Board (COCOBOD) and various private sector actors have trailed solar-powered irrigation pumps (SPIPs) (Cocoa Post, 2019).Promoting solar technologies requires farmers to bear the investment cost of the SPIPs.However, Bunn et al. (2019) argued that high upfront investment in SPIPs may hamper demand.With high costs, developers of SPIPs and cocoa irrigation technologies need to understand the demand for and the market segments of their products with the highest potential for use.
Meanwhile, the market potentials of such irrigation solutions are unknown.While Hucheon et al. (1973) and Carr and Lockwood (2011) studied the agronomy of cocoa irrigation, and Gebrezgabher et al. (2021) studied solar photovoltaic technology for small-scale irrigation in Ghanaian farming, there have been no studies on SPIPs in cocoa.There is no evidence of effective demand for SPIPs among cocoa farmers, measured by willingness to invest (WTI) and ability to invest (ATI).This study thus seeks to address whether Ghana's cocoa growers are willing and able to invest in SPIPs as an early response adaptation strategy for climate change.Specifically, we ask how willing different resourceendowed farmers are to invest in SPIPs.What factors influence their willingness to invest (WTI)?And what is the level of financial ability to invest?
This article advances insights into market demands for SPIPs and cocoa irrigation technologies.These insights are crucial for technology developers to understand their prospective market and tailor their business models to different demand segments.The insights also have implications for emerging policies and interventions towards early response climate adaptation in the cocoa sector.Ghana Cocoa Board's Productivity Enhancement Programmes (PEPs), which include solarbased irrigation for cocoa production, can use the recommendations of this study to improve farmer access to the program.

Theoretical framework and the empirical model
Underpinning the WTI analysis is the random utility theory (RUT).The RUT postulates that the consumer's or household's resource allocation decision (e.g.WTI in SPIPs) is determined by the utility maximization behavior based on three rationality assumptions.First, an economic agent or consumer, in making a choice decision from J alternatives, will select the alternative that maximizes utility subject to resource constraints or endowment.The utility assigned to each alternative depends on the attributes of the alternative (Aj) and the characteristics of the decision maker (Xi).It is also assumed that the consumer considers the costs and benefits of each alternative before making the final decision.
Second, all consumers have the same ATI in the goods or services under consideration.Such a 'one-sizefits-all' framework is problematic, especially when analyzing WTI in capital-intensive goods such as a SPIP.To allow for an in-depth analysis of determinants of WTI across different resource segments, this study was built on the frameworks of Akrofi et al. (2019) and Tesfaye et al. (2021).The study's population was split into three segments based on farm size: resource-poor (0.1-6 acres), resource-limited (>6 < 16 acres), and resourcerich farmers (>15.9 acres).The nuanced nature of the WTI analysis sets our framework apart from conventional contingent valuation studies.The land was used as a proxy for wealth since all else being equal, cocoa farmers with larger farms are expected to obtain higher cocoa revenues than those operating on smaller farms.
Third, all farmers have the same irrigation water requirement per acre, requiring a SPIP with the same capacity per acre.This assumption allowed for the comparison of WTI responses across different resource segments.We then proceeded to specify the WTI responses as an indirect utility function for each respondent under the assumption that the utility associated with using SPIP for cocoa production is positive and higher than without it.
Let the indirect utility function for i th respondent be represented by The i th respondent may choose the responses 'yes, yes ', 'yes, no', or 'no, yes' (i.e 1)), εi is error term assumed to be logistically distributed for all cases.The empirical model for the WTI is given as: The estimated parameters were interpreted as marginal effects, which indicates the effects of a marginal change of the variables conditioning WTI in the SPIP on the probability of saying 'yes'.Therefore, the marginal effects were estimated as follows:  2014) reported a strong relationship between the variables influencing the adoption of motorized pumps and other water-lifting technologies in Ethiopia.Farmers' decisions to utilize water-lifting technologies were influenced by socio-economic and demographic considerations, as well as spatial diversity in biophysical characteristics (e.g.rainfall, topography, soil, and water conservation).Additionally, it is anticipated that the age of cocoa trees would have a negative impact on WTI since younger plants are at their peak and may yield greater amounts (Wongnaa et al., 2022).Still, older trees are more likely to be infested by pests and diseases and are typically unproductive (AfDB Africa Development Bank Group, 2019).Also, cocoa production is seasonal, and revenues are unreliable and unstable (Aneani et al., 2011;Staritz et al., 2022)., Thus, Farmers with diversified income sources and higher incomes are expected to have greater WTI.Others (Adejuwon, 2019;Jambo et al., 2021;Kohansal et al., 2009;Oyinbo et al., 2019;Woldemariam & Gecho, 2017;Yigezu et al., 2013) found a positive relationship between access to extension services and adoption of irrigation.Furthermore, factors that influenced the WTI in irrigation water included age (Weldesilassie et al., 2009), household size (Alhassan et al., 2013;Bane, 2005;Claessens et al., 2012;Fadeyi et al., 2022), credit access (Aneani et al., 2012;Balana & Oyeyemi, 2020;Chirambo, 2016;Nakawuka et al., 2017;Urama et al., 2019), land size (Woodhouse et al., 2017), household income (Fisher & Carr, 2015;Habtemariam et al., 2019;Makate et al., 2019) and off-farm income (Fadeyi et al., 2022).

Study areas
The primary data were collected from cocoa farmers from eleven communities in nine districts across Ashanti, Bono, Ahafo, Western South, and Western North regions (Figure 1).These districts lie within the deciduous, semi-deciduous, and rainforest agroecological belts in Ghana, evincing bimodal rainfall patterns.The soil types, drainage, vegetation cover, and rainfall regimes in these zones support cultivating tree crops, including cocoa.

Data and survey design
Semi-structured questionnaires were administered by trained enumerators under the supervision of the lead researcher.The questionnaire was designed to elicit information on production, demographic, socioeconomic, technical, and institutional factors, and other key variables of interest.Using a multi-stage sampling procedure, we sampled 523 cocoa household heads who are the decision-makers for irrigation investment.The regions were selected based on being the highest cocoa production regions.The purposive sampling method was used to select at least one district from each region (Table 2), and the leading cocoa-producing communities were randomly selected.Since irrigation is not applied in Ghana's cocoa production, the final sampling stage randomly selected at least forty cocoa producers per town.Districts with more cocoa-producing communities and high concentrations of cocoa growers received a higher quota of respondents.The determination of sample size and distribution of respondents was guided by expert opinions and field observations, as no sampling frame was available from secondary sources.
As is typical for contingent valuation studies, a hypothetical market was created for SPIPs using an average initial bid price of GHS 30,000 for a standard pump that can irrigate at least one acre of cocoa.The pre-determined market price ranges between GHS 25,000 and GHS 30,000 for a typical SPIP, including the pump, panels, controller, sensors, and pressure switch.However, expectations were that the price could increase by about 15% at the time of its introduction due to exchange rate fluctuations in the country.As a result, the first bid was generated using the adjusted average price, and the second bids were calculated based on 15% upward and downward adjustments in the initial bid.This range of values was tested in a pilot survey to prevent start-up price bias (see Knapp et al., 2018;Tesfaye et al., 2021) and was determined to be appropriate before the execution of the final survey.
Respondents were then asked to express their WTI at the opening offer, with the subsequent question depending on their initial response.If someone said 'yes' to the opening offer, they were then asked their opinion on the subsequent bid price and so on.Those who said 'no' received the next reduced offer.There were four possible responses: 'Yes, yes ', 'yes, no', 'no, yes' and 'no, no'. A 'no, no' response is considered as a zero WTI, while the rest of the responses were considered as acceptance.As suggested by Champ and Boyle (2003), follow-up questions were presented to respondents with 'no, no' responses to control for protest bids.However, individuals with protest responses (e.g.zero WTI responses) were excluded from mean WTI estimations to control for protest zero bids bias (Mezgebo et al., 2013;Tesfaye et al., 2021).

Data analysis
Data were first entered into SPSS software for cleaning.The data were subsequently transferred into STATA for rigorous statistical analysis.WTI was analysed using descriptive statistics such as means, frequencies, percentages, and tests of mean differences (analysis of variance).The binary logistic regression model was used to elicit the factors that influence WTI.The ability-toinvest method was used to analyse farmers' ATI in SPIPs.Respondents were grouped based on the number of years it would take them to recoup the market price.Based on empirical analysis, it was determined that farmers would typically contribute about 18% of their yearly gross income (both on and off-farm sources) toward the SPIPs.Farmers perceived SPIPs to be capital-intensive innovations and so required high upfront outlays to adopt them.The percentage the farmers selected reflects the kind of technology and the repayment period.Farmers thought that marketers and  promoters of SPIPs would require them to repay within a short period.This amount was the foundation of the 'ability to invest' analysis.Depending on their repayment capacity, respondents were grouped according to the required repayment periods.

Socio-economic profile of sampled cocoa growers
The socio-economic profile of respondents in relation to their WTI is presented in Table 3. Results indicate that cocoa production in the study areas was maledominated.Some 84% of the cocoa farming population were males.Males mostly head traditional cocoa farming households.A female becomes the head when the husband is deceased or incapacitated.It is, therefore, not surprising that males dominated the sample population.Furthermore, male farmers had better access to productive resources, especially land for cocoa production, than their female counterparts.The data showed that the average age of a typical Ghanaian cocoa household head in the sample area was about 52 years, slightly below the national average of 55 years.With a life expectancy between 55 and 60 years (Kodom et al., 2022), cocoa producers in the study area can be described as an aging farming population, which affects production systems and productivity.
On average, a typical cocoa farming household has eight members, indicating a large family size.This large household size is likely to have implications for household expenditure and the availability of labor for farm work.The average cocoa farmer in the sample has spent eight years in school, equivalent to primary education according to the Ghana Education system.From this observation, the farming population may be described as educated at the primary level.
Cocoa growers in the study area cultivated 12.2 acres of cocoa farms.The mean age of a typical cocoa tree on cocoa farms was 17 years.The total household annual income distribution among the respondents, using the official exchange rate as of August 2022, was 49% earned between GHS 7,000-30,000 (USD 870.26-3,729.67),25% earned below GHS 7,000 (USD 870.26), and 26% earned above GHS 30,000 (USD 3,729.67).There was an uneven income distribution among respondents, with a relatively higher concentration of farmers in the middle-income category.More than half of the respondents had diversified their income sources i.e. 53% participated in off-farm generating activities such as trading and carpentry work, 54% engaged in livestock production, and 55% cultivated other crops in addition to cocoa.
Cocoa production is seasonal, and its revenues are unstable and unreliable (Aneani et al., 2011;Staritz et al., 2022).Income diversification is a strategy to generate extra income to compensate for shortfalls in cocoa revenues.Farmers in the study area reported receiving four annual extension visits, translating into one extension contact every three months.The infrequent extension contacts could have affected the adoption of improved production techniques.Only 37% of the respondents had access to credit.This result is not new, as lack of access to favorable credit from formal credit institutions is the most critical challenge facing Ghanaian farmers, and cocoa growers are no exception (Aneani et al., 2012;Urama et al., 2019).
Responses of willingness versus unwillingness to invest (UWTI) in SPIPs varied across different resource segments, as shown in Table 4.The percentage of UWTI farmers was higher than those of WTI.However, the mean differences between the WTI and UWTI were statistically significant for almost all the study variables.This suggests that socio-economic, institutional, resource endowment, and farm-level factors could account for the variations in WTI across different resource segments.

Willingness to invest in solar-powered irrigation pumps
WTI responses to double-bounded WTI questions and WTI amount are presented in Tables 5, 6 , and 7, respectively.In general, 37.7% of the respondents indicated their WTI in the SPIPs at the initial bid price of GHS 30,000 (USD 3,729.67).The distribution of this   6).
The results from segment permutations (Table 7) reaffirmed the significant differences in WTI amounts across different resource segments.Although the number of resource-rich farmers willing to invest was significantly higher than those in the resource-poor and resource-limited segments, this amount is less than the pre-determined market price of the SPIS.This finding is consistent with previous studies (e.g.Acheampong et al., 2018;Perez et al., 2021;Tesfaye et al., 2021), highlighting the lack of smallholder access to irrigation.While Akrofi et al. (2019), Perez et al. (2021), andPhiri et al. (2020) reported smallholders' inability to afford the overhead cost of irrigation technologies as a primary factor for the limited irrigation access, Tesfaye et al. (2021) ascribed it to high acquisition and operation costs.
Furthermore, when the opening offer was raised from GHS 30,000 (USD 3,729.67) to GHS 34,500 (USD 4,289.12), a negative correlation was observed (Table 5).The proportion of respondents willing to invest decreased by 33.3%, 23.4%, and 22.0% for the resource-poor, the resource-rich, and the resource-limited segment, respectively.This result aligns with the earlier studies on contingent valuation that found a negative correlation between the opening bid price and WTI (Tesfaye et al., 2021).The WTI is expected to reduce at a higher bid price as rational consumers will likely demand less for a good or service as its price increases, depending on the product type under consideration.Our study showed a small reduction of the WTI across the three segments.This means that cocoa growers in the study areas are generally aware of the increasingly erratic rainfall due to climate change and irrigation needs.Farmers show strong interest in solarbased irrigation to secure their livelihoods with watersecure agriculture.The extraction and usage of water resources are also governed by socioeconomic factors prevailing at the local level (Kaini et al., 2020).As such, findings on these factors are presented in Tables 8, 9 , and 10.

Determinants of WTI decisions
The binary logit regression model was used to estimate the determinants of WTI decisions, and the estimated coefficients (Table 8).This, together with marginal effects, are presented in Table 9 (estimates for the pooled sample) and Table 10 (estimates for resourcepoor, resource-limited, and resource-rich segment).The overall significance of the models was fit, as shown by the Wald test estimates.Pooled sample results show that farmers within the middle-and high-income groups were more likely to invest in SPIPs than those in the lower-income group (Table 3).This also reaffirms the positive influence of resourcefulness on farmers' investment decisions.For instance, belonging to the high-    income category increases WTI by 21.6% and 13.1% for the middle-income category compared to being resource-poor.This concurs with economic theory and the literature (Mezgebo et al., 2013;Tesfaye et al., 2021), as higher and middle-income earners have a greater capacity to pay for capital-intensive goods such as SPIPs.
Limited credit access has been described in the literature as a major constraint to technology adoption (Nakawuka et al., 2017).De Fraiture and Giordano (2014) show that 80% of smallholders used their savings to acquire irrigation equipment, likely implying smallholders' low credit access.However, credit constraints might not be entirely associated with supply-side factors.Demand-side factors such as risk-averse behavior and limited information access could also affect smallholders' access to credit (Balana & Oyeyemi, 2020).Improving access to credit or alternative financing schemes could mitigate the capital constraints, enabling smallholders to benefit more from participating in market-oriented, high-value irrigated production (Balana et al., 2020).Similarly, Chirambo (2016) found that African microfinance institutions can be sustainable mechanisms for funding climate change initiatives while promoting rural development and financial inclusion.The credit variable was significant at 1% with a positive marginal effect of 17.5%, which aligns with many empirical studies (e.g.Amankwah & Egyir, 2013;Balana et al., 2020;Tesfaye et al., 2021).
Keeping livestock as an income diversification strategy increased the probability of accepting irrigation technology by 10.2%.Livestock production is expected to generate additional income for households which, in turn, help lessen their liquidity constraint.It is also possible that wealthier farmers likely take more risks to invest in new technologies (Tesfaye et al., 2021, Zannou et al., 2020;Gebregziabher et al., 2014).
Education had a positive significant impact on WTI-one additional year spent in school increased WTI by about 1.5%.It is clear from the ATI analysis that farmers who were highly educated, had a high income, or were a member of the upper class were better placed to meet the investment cost of the SPIP.More educated farmers were more likely to appreciate the benefits of modern technologies such as SPIPs due to their better understanding of climate change impacts on their farms (Mezgebo et al., 2013).Also, as agriculture is a dynamic occupation, conservation practices, and agricultural production technologies provide better knowledge and opportunities.If the household head is literate, he/she will be very prone to accept extension services and irrigation use as confirmed by, for example, Jambo et al. (2021), Woldemariam and Gecho (2017), Yigezu et al. (2013), and Kohansal et al. (2009).
Household size had a significant negative impact on WTI.An additional household member would reduce WTI by 1.0%.Larger households likely spent more of their annual income on household expenses.The financial burden reduced the number of WTI farmers.Gebregziabher et al. (2014) found no relationship between adopting motorized pumps and family size in Ethiopia.Previous studies found a positive and significant relationship between family size and WTI in waterlifting technologies (Alhassan et al., 2013;Bane, 2005;Mezgebo et al., 2013;Tesfaye et al., 2021).The age of cocoa trees has a significant (1%) negative impact on WTI in the SPIPs.When cocoa trees aged by a year, it resulted in a 0.6% reduction in WTI.Elderly trees are less productive as they have grown past their prime and are prone to pests and diseases, with downward cascading effects on cocoa yields and profit.In contrast to our finding, Wongnaa et al. (2022) found that the age of cocoa trees had a positive effect on the profit efficiency of cocoa production.Accordingly, farms with older cocoa plants are more profitable than those with younger trees.The impact of the age of the cocoa tree on profit was described as 'two sides of the same coin'.(Wongnaa et al., 2022).Older cocoa trees increase profitability since they may have more branches and fruiting nodes than younger trees.Meanwhile, as the younger trees are at their peak, they may yield bigger amounts.These findings were explained by the simplicity of managing older cocoa plants and the requirement for little input on these older tree farms.The PEPs program implemented by COCOBOD aims to improve yields and secure cocoa revenues by promoting the pruning and rehabilitation of aged, disease-infected, and unproductive cocoa trees (AfDB, 2019).This will require strong collaboration between the Ghana Cocoa Board and SPIP promoters.Any interventions to improve production or technical efficiency of cocoa production at the farm level could benefit SPIPs.
Willingness to invest decisions within the resource-poor segment was influenced by cocoa land size, the age of the cocoa tree, and education (Table 10).Farm size had a positive impact on WTI among the resource-poor segment; an additional acre of land would result in a 5.4% increment in WTI for them.The potential for land expansion would be higher among the resource-poor if these farmers were optimizers; resource-poor farmers could increase their income to invest in irrigation.Similar studies on water-lifting technologies reported counter-intuitive results (e.g.Gebregziabher et al., 2013;Tesfaye et al., 2021;Woodhouse et al., 2017).The age direction of the cocoa tree variable was consistent with that of the pooled sample but with a relatively higher marginal effect of about 0.7%.Consistent with the pooled sample, education positively impacted resource-poor farmers' WTI with a reduced 1.1% marginal effect.
As shown in Table 10, WTI among resourcelimited segments of the cocoa farming sample was influenced by education, age of the cocoa tree, income category, credit access, and extension contact.The magnitude and direction of farmer education and the age of cocoa trees were similar to the results obtained for resource-poor farmers.Resource-limited farmers in the high-income category were likely to increase WTI by 23.1%-one of the highest marginal effects among the four models.Formal credit access was also positively impacted as WTI increased by 24.1, 1% more than the marginal effect of the income variable.The significant effects of financial factors such as credit and income offer additional proof that high costs act as a barrier to investing in capital-intensive but productivityenhancing technologies like solar-based irrigation by resource-constrained farmers.Also, an additional visit from an extension agent is likely to increase WTI by 1.9%.Extension agents help to inform smallholder farmers about new technologies and their benefits.Frequent contact with extension agents is expected to improve information access and adoption probabilities (Oyinbo et al., 2019;Adejuwon, 2019).To improve climate literacy, SPIP promoters should collaborate with the Ghana Cocoa Board to strengthen extension agents' capacity to educate farmers on adopting climate-smart irrigation technologies.
Willingness to invest among resource-rich farmers was influenced mainly by education, household size, cocoa land size, income, and off-farm activity.Education had a positive impact on resource-rich farmers' WTI in irrigation.An additional year spent in school was likely to increase WTI by about 2.2%, which is higher than the marginal effect of education in the pooled sample and across the other segments.Household size had a negative impact on resource-rich farmers' WTI.Smallholder farmers' household size measures labor availability and financial commitment (Claessens et al., 2012).Our finding on household size is consistent with Fadeyi et al. (2022), who suggested that the larger the smallholder farmer's family, the lower the likelihood of adopting new technologies due to the higher financial commitment.To address the negative effect of large household sizes on the SPIP WTI, large cocoa farm families should, if possible, be prioritized in any future credit schemes accompanying the sale of SPIPs.Such an action would lessen the financial burdens of large families, thus improving their SPIP WTI.
Here, land size had a negative impact on WTI, contrary to our expectations.Resource-rich farmers may have diversified their investment portfolios into other non-farm and on-farm enterprises, as evident in their income from off-farm, livestock, and other crops.Also, resource-rich farmers with larger cocoa lands could be making enough profit that offset the adverse climate change impacts, such as low yields.
Thus, investing in cocoa intensification might not be a priority for resource-rich farmers with larger landholdings.
The income categories, middle-and high-income for the resource-rich, recorded the highest WTI marginal effects, 27.3%, and 33.7% respectively, across the study.90% of resource-rich farmers in these two categories of income were the highest, with 41% of resource-poor farmers found in the low-income category (Table 4).Finance was noted to empower smallholder farmers to purchase, operate, and maintain new technologies (Fisher & Carr, 2015;Habtemariam et al., 2019;Makate et al., 2019).Finance was identified as a factor that influenced technology adoption by 81 out of the 128 articles reviewed by Fadeyi et al. (2022).This could be the major reason for the relatively higher willingness to invest among the resource-rich farmers, compared to the other resource segments.
Participation in off-farm income activity had a significant positive impact (at 5%) on WTI among the resource-rich segment.This is because off-farm income often serves as an alternative for smallholder farmers to overcome financial constraints, which can be used to adopt new technologies (Fadeyi et al., 2022).

Farmers' ability to invest in solar-powered irrigation pumps
Table 12 shows the descriptive statistics of the percentage of total income (on-and off-farm) after deducting expenses that farmers are ATI in SPIPs each year (consumption, saving, and input and labor cost).Farmers in the resource-rich segment have more available income with an average annual income of GHS 38,730.31(USD 4,815.05)compared to that of GHS 22,248.26 (USD 2,765.96),and GHS 12,430.17 (USD 1,545.35) of resource-limited and resource-poor, respectively (Table 11).
Table 12 lists the average ATI and the proportion of respondents who can invest between years 1 and 5. Farmers were WTI on average 17.86% of their annual household income on solar water-lifting technologies.More than 88% of respondents within the resourcepoor group could not meet the investment cost of the SPIPs based on their indicated income levels within the 5 years.Only 12% could meet the overhead cost, of which the majority (6.4%) could recoup the pre-determined market price by the fifth year.According to Tesfaye et al. (2021), resource-poor farmers were constrained by a lack of financing options to use more capital-intensive, efficient water-lifting technologies (WLTs), such as fuelpowered motorized pumps.Thus, they invested in less efficient, low-cost, and labor-intensive WLTs, such as pulleys, operating below their production capacity frontier.72% of respondents within the resource-limited segment could not meet the pre-determined market price.The resource-rich segment had the highest percentage (56%) of respondents who could meet the investment cost, of which the majority would be able to repay the loan by the third and fourth year.The ATI of most of the  respondents was lower than their WTI.This corroborates the findings of Gwambene et al. (2015) in their study to assess smallholder farmers' practices and understanding of climate-smart agriculture in the Southern Highlands of Tanzania.

Conclusion
In this article, we have provided estimates of Ghanaian cocoa farmers' willingness and ability to invest (WTI and ATI), providing essential information on the effective demand for solar-powered irrigation pumps (SPIPs).The sample comprising 523 cocoa farmers from nine cocoa districts of Ghana was categorized into three resource segments to determine how WTI and ATI vary across.The findings indicate that, generally, many cocoa farmers show strong interest in SPIPs, even at a higher bid price.However, the ATI across the three segments is low.The resource-rich segment shows a statistically significant difference between the three segments.Most of the resource-rich farmers were willing to invest more.However, the amount they are willing to invest is below the minimum pre-determined market price of SPIPs.This indicates the need to support all cocoa farmers to improve their access to irrigation.Key variables of investment decision-making include income, credit access, and household size.Financial constraints are an important obstacle to WTI and ATI in solarbased irrigation technologies.Policies and interventions that promote the use of SPIPs should be accompanied by well-thought-out income diversification and soft credit schemes such as cash credit and term loans, on-credit sales, or irrigation equipment leasing arrangements.These will improve the financial capacity of resource-constrained farmers to meet the SPIP investment cost.High interest in SPIPs is good for farmer-led innovation scaling, which could trigger a win-win adaptation.

Table 1 .
A priori expectations of variables used in the binary logit regression model

Table 2 .
Distribution of sampled respondents

Table 4 .
Summary statistics of study variables of different cocoa farmer segments

Table 5 .
Summary of responses to double-bounded WTI questions Source: Authors' computation.The GHS (Ghana Cedi) is the official Ghanaian currency; The official exchange rate during the survey (August 2022): USD 1 = GHS 8.0436.

Table 6 .
Mean willingness to invest statistics for different cocoa farmer segments

Table 7 .
Test of statistical difference

Table 8 .
Results from logit model estimation for different cocoa farmer segments

Table 10 .
The marginal effect of covariates in the logit model for different cocoa farmer segments Marginal effect: dy/dx is for discrete change of dummy variable from 0 to 1.

Table 11 .
Descriptive statistics of resource endowment