Targeted characteristics and use of socially marketed preventive health goods: evidence from condoms in sub-Saharan Africa

ABSTRACT

Social marketing is a popular method for allocating targeted publicly funded preventive health goods in poor countries. However, low demand among targeted groups may inhibit take-up relative to non-targeted groups, ownership may not result in use, and there exists little large-scale evidence on how use of socially marketed goods varies by targeted characteristics. I assemble national household survey data from 27 sub-Saharan African countries to examine how use of one of the most common socially marketed preventive health goods (i.e. male condoms) varies by the main targeted characteristics (i.e. low income, low educational attainment, and high HIV risk).

The results suggest that the majority of condoms used are socially marketed condoms, engaging in transactional sex is associated with an increased likelihood of using a socially marketed condom brand, and low-income/low educational attainment are not associated with increased likelihoods of using socially marketed brands. The fact that distribution targets low socioeconomic status groups and relative use remains low suggests that weak demand for condoms among these groups inhibits use. Policymakers should consider mechanisms to increase demand and to further refine targeting efforts.

KEYWORDS:

• HIV/AIDS
• preventive health
• publicly provided goods
• social marketing
• sub-Saharan Africa
• targeting

JEL CLASSIFICATION:

• H40
• I15
• J10

Acknowledgments

I thank Willa Friedman, John Hoddinott, Adrienne Lucas, Isaac Mbiti, Mark Taylor, and Josh Wilde for many excellent comments. The findings, interpretations, and conclusions expressed in this paper are those of the author and do not necessarily represent the views of the aforementioned individuals or agencies. This study did not receive any funding and there are no conflicts of interest. All errors are my own.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ Social marketing of condoms in poor countries appears to have originated in India in the 1960s (Behrman 1989).

² Condoms and insecticide-treated bed nets (ITNs) are two of the main donor funded socially marketed preventive health goods in sub-Saharan Africa. In 2013, donors funded approximately 1.7 billion male condoms (UNFPA (United Nations Population Fund) 2013) and 136 million ITNs in sub-Saharan Africa (WHO (World Health Organization) 2013. World Malaria Report 2013).

³ As I discuss in Section 3, the brand data distinguish between socially marketed and non-socially marketed brands, whereas the source data do not explicitly distinguish between socially marketed and private sources.

⁴ The main measure of material standard of living in the household survey data that I use to measure condom brands, the Demographic and Health Surveys (DHS), is consumer durables ownership. Young (2012) demonstrates that this measure may be superior to typical measures of gross domestic product (GDP).

⁵ See Elbers et al. (2007), Dinkelman and Schulhofer-Wohl (2015), Ravallion and Chen (2015) and Karlan and Thuysbaert (2016).

⁶ See Alderman (2002), Stifel and Alderman (2006), Alatas et al. (2012), Nose (2014) and Karlan and Thuysbaert (2016).

⁷ See Stifel and Alderman (2006), Pradhan, Saadah, and Sparrow (2007), Alatas et al. (2012), Ravallion and Chen (2015) and Karlan and Thuysbaert (2016).

⁸ See Jacoby (1997), Teklu and Asefa (1999), Niehaus and Sukhtankar (2013), Nose (2014), and Muralidharan, Niehaus, and Sukhtankar (2016).

⁹ Behrman (1989), Brent (2009) and Terris-Prestholt and Windmeijer (2016) provide what appear to be some of the few economic analyses of social marketing programs.

¹⁰ For example, see Boone, Farley, and Samuel (1985), Meekers, Agha, and Klein (2005), Brent (2009), Chapman et al. (2011), Evans et al. (2011), Alkema et al. (2013), Darroch and Singh (2013) and Terris-Prestholt and Windmeijer (2016).

¹¹ The countries and survey rounds are as follows: Benin (2011), Burkina Faso (2010),Burundi (2011), Comoros (2011), Congo Democratic Republic (2013), Cote d’Ivoire (2011), Ethiopia (2003), Ghana (2014), Guinea (2012), Kenya (2014), Liberia (2013), Madagascar (2008), Malawi (2010), Mali (2012), Namibia (2013), Niger (2012), Nigeria (2013), Rwanda (2014), Sao Tome and Principe (2008), Senegal (2010), Sierra Leone (2013), Swaziland (2006), Tanzania (2009), Togo (2013), Uganda (2011), Zambia (2013) and Zimbabwe (2010).

¹² As noted elsewhere (Gersovitz 2005), the DHS do not routinely ask respondents about the price paid for condoms.

¹³ Unlike the DHS question for brand, the source question does not include ‘do not know’ as an option.

¹⁴ The DHS do not ask about socially marketed sources and ‘private source’ includes socially marketed sources.

¹⁵ To harmonize education measures across countries, I define primary-school completion as completing 7 or more years of schooling and secondary-school completion as completing 11 or more years of schooling.

¹⁶ Total household consumer durables owned is the sum of indicator variables for each of the following consumer durables: (improved) floor, refrigerator, television, radio, bicycle, motorcycle, and car.

¹⁷ One exception is secondary completion is associated with a roughly 0.6 percentage point reduction (p < 0.01) in the likelihood of using a government brand relative to those who have only completed primary school.

¹⁸ One explanation for this association is that total consumer durables owned is associated with an increased likelihood of using a condom of any type. Consistent with this explanation, total consumer durables owned is not associated with the likelihood that the brand variable is missing conditional on condom use (see Table 4).

¹⁹ These are auxiliary regressions because the distinction between private and NGO sources is not particularly clear, with many quasi-public (e.g. social marketing) organizations distributing condoms through private sources.