Impact of aquaculture feed technology on fish income and poverty in Kenya

ABSTRACT

The impacts of improved agricultural technologies on smallholder households in Africa are well documented in the literature. However, the literature on the welfare impacts of aquaculture technologies, especially in the context of smallholder households, is very scanty. This paper applies the propensity score matching technique to household survey data to examine the impact of improved feed technology on fish income and poverty in Kenya. After controlling for observable household characteristics, the results indicate that improved feed technology increases aquaculture income and reduces poverty among fish farming households. Specifically, the income effect of the technology is 23–37%, with resultant poverty reduction effect of 19–23% points. Evidence from the study indicates that the likelihood of adopting improved feed in Kenya will surge with improved extension service delivery, access to government subsidized feed, and easy market access for purchasing improved feed and sale of mature fish.

KEYWORDS:

• Aquaculture
• fish farm households
• fish income
• improved fish feed
• Kenya
• poverty
• propensity score matching

Acknowledgments

We extend our appreciation to Dr Isabelle Baltenweck of the International Livestock Research Institute (ILRI), Nairobi Kenya for supervising the data collection. We acknowledge the support from the students who helped administer the questionnaires to the households and the fisheries officers in various district offices for making it possible to reach the households. The authors would also like to thank two anonymous reviewers for their great and insightful comments.

Notes

The aggregate Kenyan economy experienced a rapid decline in GDP growth from 7.1% in 2007 to 1.7% in 2009 following post 2007 election violence, coupled with the global economic meltdown (Government of Kenya 2009). Consequently, in 2009, the government of Kenya rolled out an ESP, focusing on key sectors of the economy, with the sole goal of revamping the economy and boosting economic recovery. Aquaculture was one of the sectors considered under this program. A key objective of the program was to increase economic opportunities in rural areas for employment creation and income generation.

Welfare, income and wellbeing are used interchangeably in this paper. Even though we are well aware of monetary and non-monetary measures of welfare/wellbeing, due to data limitation, we measure household welfare using household income.

Note that if the data used for the analysis is panel, then the difference-in-difference approach is warranted.

For convenience, we subsequently remove the subscript on Z_i.

Some households practice polyculture (cultivating both tilapia and catfish in the same pond) while others grow only one type. Polyculture operation is meant to diversify fish income as well as control tilapia population in fish ponds.

Feed sourced from the government usually comes as part of the aquaculture ESP. In addition to the subsidy program, a number of feed production and retail outlets are springing up in the country, albeit somewhat distanced from the fish farms. Though the subsidy program provides free feed to fish farming households, the quantity acquired by these households is not sufficient to complete a production cycle. Thus, farmers supplement the subsidized feed with either purchased improved feed or locally manufactured types. The average retail price of improved feed in the 2013 production year is about KShs 80 per kg. A substantial amount of feed is imported from Uganda (fish feeds from Uganda are branded Uga chick/feed), though there is currently no national data on the proportion of fish feed imported by source country.

The FGT poverty figures were estimated in Stata 13 using the apoverty command.

The use of PPP ensures that we do not overstate the incidence of poverty in the study regions. An alternative would be to use the nominal exchange rate for computing the poverty line. Using the nominal exchange rate will possibly overstate the incidence of poverty, and might not be consistent with household holdings, and other observable characteristics. Moreover, in order to extrapolate the results, while comparing it to other countries, it is vital to use the PPP instead of the nominal exchange rate.

The PSM analysis was conducted in Stata 13 using the psmatch2, psgraph, and pstest commands. Standard errors were bootstrapped using the bootstrap command with 100 replications.

In all model types, Pseudo R² before matching was 0.17; likelihood ratio (LR) before matching was 45.4 with corresponding p-value of 0.00; and mean standardized bias before matching, 19.40.

Note that “Treated off-support” means a household in the adoption group with no suitable comparison in non-adoption group. “Treated on-support” is an adopting household with a suitable comparison in the non-adoption category. “Untreated” is the non-adoption group.

Unmatched is the difference before PSM analysis while matched is after the matching exercise.

Additional information

Funding

This project was funded by the Feed the Future Innovation Lab for Collaborative Research on Aquaculture & Fisheries through the United States Agency for International Development (USAID), Grant No. EPP-A-00-06-00012-00 and contributions from participating institutions. Supplementary funding was also provided by the Norman E. Borlaug Leadership Enhancement in Agriculture Program (Borlaug LEAP) for data collection.