Production risk and technical efficiency of fish farms in Ghana

Highlights

• This paper considers the stochastic frontier model, whilst incorporating the flexible risk properties to analyze production risk, technical efficiency and its determinants of fish farms in four regions of Ghana including Greater Accra, Volta, Western and Ashanti regions.

• The paper employs a two year panel data of 320 farms making a total of 640 observation through a random selection.

• With respects to model fitness, the study demonstrate that the translog production function which allows for squared and cross product terms is more suitable to derive policy conclusions in the data.

• Production risk associated with input choice is also identified to be present in the activities of fish farms in the study area. Hence, exclusion of risk component in the frontier model might result in bias estimates.

• The paper accounted for intercept change with dummies for including hired and family labour with zero cases in the model and demonstrated that there could be bias estimators of the parameters in the frontier production function without inclusion of theses dummies.

• It is also revealed that deviation from the frontier due to technical inefficiency is relatively larger compared to the deviations due to random noise effects.

• Empirical evidence from the paper indicate that the expected elasticities for all inputs including hired labour, family labour, feed, fingerlings and other cost are positive and have reasserting influence on fish farming production in the study area.

• Findings from the computed return to scale indicates that fish farms in the study area exhibit increasing returns to scale which means that on the average, farms operate in the first stage of the production process and this calls for increase in size of operation to take advantage of economies of scale.

• The paper also demonstrates that all the inputs variables except fingerlings influence production risk significantly. Hired labour, family labour and other cost are found to be a risk decreasing inputs, whilst feed and fingerlings are identified to be risk increasing inputs.

• The predicted overall mean technical efficiency indicates that on average fish farms produce 74% of the potential (stochastic) frontier output, given the present state of technology and input levels. This implies that 26% of technical potential output is not realized.

• The paper demonstrates that the mean technical efficiency scores for the farms when the production risk component is excluded produced a value of 87%. This revelation suggest that technical efficiency score is an exaggerated opinion when the production risk component is omitted from the conventional stochastic frontier model.

• The paper also reveals that the combined effects of factors involved in the technical inefficiency model are important in explaining the variation in production of fish farms in Ghana, although individual effects of some variables are not significant.

• Specifically, the paper demonstrates that larger farms take advantage of economies of scale to enhance efficiency, whilst technical efficiency also improved over the two production years.

Abstract

This paper adopts the stochastic frontier model with flexible risk to analyze production risk, technical efficiency and its determinants of fish farms in four regions of Ghana including Greater Accra, Volta, Western and Ashanti regions. The study employs a two year panel data from 320 farms making a total of 640 observation through a random selection. The findings demonstrates that the translog model is best fit for the mean output function, whilst the input variables: hired labour, family labour, fingerlings, feed and other cost are identified to positively influence fish farm output at an increasing returns to scale. The paper also finds that feed and other cost are risk increasing inputs, whilst hired labour, family labour and fingerlings are identified to be risk decreasing inputs. The estimated average technical efficiency score of 0.74 shows that efficiency is compromised when the production technology is modeled without the flexible risk property which gave a score of 0.87. Additionally, the results show that age², experience, pond area, gender, pond type and fish farming education significantly influence technical efficiency of farmers. This paper also highlights that technical efficiency of fish farms in the study area improved over time. In conclusion, the paper notes that on the average, 26% of potential output is lost due to technical inefficiency and production risk and given the present state of technology and input level, the possibility of enhancing production can be achieved by reducing technical inefficiency by 26% through adoption of practices of the best fish farm, whilst taking production risk into consideration.

Introduction

The government, research institutions, the private sector and other major stakeholders under Ghana's fishery sector development goal emphasized the development of inland fisheries and fish farming in particular with the goal of achieving sustainable yields and to compliment the annual domestic supply (mainly from traditional marine fisheries) of about 420,000 metric tons (MT), which is around 52% less of what the country demands (Frimpong and Adwani, 2015). In view of this, policy interventions by government directed banks to enhance finance for fish farming at subsidized interest rates in the 1980s (Wijkstrom and Vincke, 1991). This initiative motivated the young and old; and male and female farmers to consider fish farming either as a full-time occupation or as a part-time business. Majority of these farmers entered the industry with little or no technical knowledge and experience and as such relied on government extension agents or family members with formal fish farming education (FFFE) for technical advice. Formal fish farming education has recently been made easily accessible at government Universities and some private commercial farms. Predominantly, the earthen and the cage pond system have been practiced by fish farmers in the country with the cage ponds limited to lake and river bodies. These farms have diverse sizes depending on whether it is a subsistent or commercially operated. Race and Pen culture systems are still not properly developed in Ghana. Tilapia and catfish are the two predominant species in extensive cultivation. Tilapia (mainly Oreochromis niloticus) constitutes nearly 88% of the total fish farm production, whilst cultivation of catfish (Clarias gariepinus and Heterobranchus longifilis) takes about 10%, with the remaining 2% accounting for other species such as Heterotis niloticus (Onumah et al., 2010).

However, after about three decades of introduction, the fish farming industry has been marked by low production. In 2016, it was estimated that total annual production stood at about 52,000 MT (MT). Studies have been undertaken to assess the possible constraints that has led to the low production by a number of authors and organizations (Onumah et al., 2010; Seini et al., 2002; Wijkstrom and Vincke, 1991; and Frimpong and Adwani, 2015). Their findings include among others; undeveloped factor markets, limited funding for pond production, wrong pond siting, and lack of enough technical personnel. In addition to these constraints, Villano and Fleming (2005) note that technical inefficiency and production risk are very often associated with farming practices including fish farming in developing countries. Up to date, no comprehensive study has been undertaken to investigate technical inefficiency and production risk involved with the fish farming industry in the country. A number of factors make the production of pond fish risky over time in Ghana. These include; fish diseases, unreliable rainfall, extreme weather conditions and poor management practices. Onumah et al. (2010) consider the conventional technical efficiency estimation technique to assess the performance of fish farms in Ghana but this paper fails to account for production risk and time varying technical efficiency effect. Thus, assessing the influence of production efficiency and risk on the performance of fish farms in Ghana whilst accounting for the effect of time is worth studying.

The use of the stochastic production function proposed by Aigner et al. (1977) and Meeusen and van den Broeck (1977) can be seen as more realistic for predicting technical efficiency than the deterministic frontiers pioneered by Farrell (1957) and Aigner and Chu (1968). However, a significant aspect of production which has not been adequately accounted for in the stochastic frontier analysis is production risk. This makes the extent of technical efficiency considerably compromised in studies of farm performance in risky production environments. Risk plays a vital role both in input decision and production of outputs to achieve efficiency.

Work done by Just and Pope (1978) paved the way for understanding production under risk. Their study consider production risk measured by the variance of output, whilst focusing on a specification that allows inputs to be either risk-increasing or risk-decreasing as noted also by Kumbhakar and Tveteras (2002). Against this backdrop, this paper seeks to analyze production risk, technical efficiencies and its determinants to derive stakeholder conclusion for policy interventions.