Consuming sex: the association between modern goods, lifestyles and sexual behaviour among youth in Madagascar

A newer paradigm: exploring transactional sex motivated by “consumption”

The more conventional understanding of transactional sex and women’s heightened vulnerability to HIV is explained through women’s disproportionate economic disadvantage arising from gender discrimination. Women are described as passive victims with little choice but to resort to “survival sex,” exchanging sex for basic needs for themselves and their children [13, 23, 24]. This relationship has also been described in the context of economic globalization as macro-level economic crises have micro-level consequences that lead to increased likelihood that women have to rely on transactional sex [25, 26].

More recent work has shed light on a different set of factors leading to transactional sex- less poor young women seek the stuff of “modernity” through engaging in economically-motivated sexual relationships. Young people, and their parents, in many different contexts across sSA describe a perception of “moral decay” that has accompanied the liberalization of economies, and an increased reliance on monetized capital accumulation [3, 19, 27, 28]. Increasingly documented in ethnographic research is the notion that young people, especially women, are actively using their sexuality to obtain modern “designer” goods (i.e. cell phones, fashionable clothing) that are increasingly available but, for most, unaffordable. This perceived linkage between interests in consumption and sexual behaviour has been documented in urban and peri-urban contexts in sub-Saharan Africa; [1, 2, 12, 17, 29–32] and in Madagascar in particular [3, 18, 33]. Very few studies examining transactional sex in the rural context draw from this paradigm (but see [28, 34]). While evidence is mounting in the ethnographic record, the extent to which interest in the consumption of modern “designer” goods motivates risky sex has yet to be examined quantitatively. A few studies do however provide some understanding about the details of the transaction. In a study of women in Soweto, South Africa, Dunkle (2004) [7] quantitatively assessed what goods women who reported having had transactional sex received. Luke [29, 35, 36] examined the amount of the “transfer” offered by men in their informal exchange relationships with women and found that higher transfers are associated with riskier sexual behaviour (namely forgoing the use of a condom).

More recent reviews of the literature have suggested that there may not be such a clear dichotomy between the survival sex of the passive victim and the consumption sex of the active sexual agent, rather, these distinctions are far more nuanced [22, 37]. Kuate-Defo coined the phrase “economic vulnerability” to capture both relationships motivated by “survival” as well as conformity, fashion, social status and pride among peer group members [15]. Another rendering presented by Jennifer Cole described young women on the east coast town of Tamatave, Madagascar who actively pursued relationships with foreign men through accessing the “right” clothes and accessories, both toward social and economic advancement as well as toward the hopeful eventuality of finding a foreign spouse [3, 38] (see also [39] for similar findings among university students in Zimbabwe).

We drew from two contrasting sites in Madagascar to examine the association between interests in consuming modern goods and sexual behaviour, and the extent to which regional differences play a role. We examine these associations in the large national capital city of Antananarivo and its peri-urban and rural surroundings, and the northern port town of Antsiranana.^a While households in the compared urban areas report similar expenditure levels and close to similar average educational attainment, the employment sectors differ, as do socio-cultural profiles.

Transforming socio-economic contexts

More recent political and economic transformations in Madagascar increase the likelihood that women and men have become more “economically vulnerable,” and therefore that women may have had to rely on their sexuality to buffer uncertainty. In the last 10 years Madagascar has endured two political crises, the latter of which began in 2009 and is ongoing. In the last thirty years Madagascar has transitioned from isolationist Marxist-Socialism to globally-integrated neoliberalism [3, 40]. The relatively late and then rapid insertion into the global economy catalyzed a number of changes. Most significant and relevant perhaps are the introduction of export processing zones (EPZs), the relaxation on import tariffs, the increases in the tourist sector, and the increasing ineffectiveness of tertiary education [41–47]. These different aspects of the transforming economic reality in Madagascar affected the two study sites similarly in some cases and differently in others. Antananarivo, located in the center of the country, is arguably the most “global” city in Madagascar. The capital city of nearly 2 million inhabitants, it was most impacted by the expanding EPZ sector, as well as the influx of imported goods and specialty stores. Antananarivo now hosts numerous internet cafes, and has seen a dramatic rise in shopping centers and boutiques, restaurants and bars in the last 20 years. These service points cater not only to foreign tourists and business people, but more recently to a small population of wealthy Malagasy. Antsiranana is about one tenth the population of Antananarivo and is located on the northern coast, built at the bottom of a large, U-shaped bay and boasting a large, functioning port; the second largest in Madagascar, it is a destination for naval and cargo ships, as well as tourists en route to its nearby beaches and national parks.

Foreign-direct investment resulted in increased possibilities for low skilled workers in Madagascar. Madagascar experienced near unmatched increases in the export-processing zone (EPZ) sector as compared to the rest of SSA through the late 1990s (20% annual increase between 1997–2001) [42]. EPZs are almost exclusively located in urban or peri-urban Antananarivo and young women make up most of the workforce [41].

Alongside export-led growth strategies, loan conditionalities of the IMF and World Bank and trade agreements have led to reductions in import tariffs, and subsequently, imports continue to outpace exports [43, 44].^b While data are difficult to locate for import trends for designer clothing and boutiques in particular, technology imports implied by the rate of change in cell phone subscriptions provide some indication. Not unlike the trend in other sub-Saharan African countries, there has been an exponential increase in cell phone subscriptions over the last two decades. While 2 in 1000 individuals had a cell phone in 1994, this was up to 30 in 1000 by 2005, and had increased to a remarkable 300 in 1000 by 2009 [45].

Another significant “import” are tourists, who have been coming to Madagascar at a rate of increase of about 20,000 tourists per year since the early 1990s. (ibid) Following recovery from a crippling but short-lived political crisis in 2002, the rate of increase jumped to 30,000 or 40,000 additional tourists per year [45]. The investment in the tourist industry in Antsiranana has soared in the last 15 years: between 2004 and 2008 alone, the estimated annual number of tourists to the city and surrounding region nearly doubled from about 15,000 to almost 27,000; particularly significant for a city of only about 200,000 [46].

There are some worrying indicators with regard to social and economic mobility in Madagascar. Reliance on foreign direct investment has been accompanied by gains in employment opportunities for low-skilled jobs, as major corporations have tended to import more technically skilled workers. In addition, emphasis on privatization has reduced public sector and stable employment possibilities. Unemployment among 25 year olds in the capital city of Antananarivo in the late 1990s was almost twice that experienced by the previous generation. One researcher commented that in SSA cities such as Antananarivo, a university diploma “has become a risk factor for finding employment” [47]. Evidence from the World Bank shows that from 1993 to 2005 change in the share of total unemployment for those with tertiary education more than doubled from 4.1% to 9.3%; and was more severe for men than for women. During the same time period, the share of total unemployment for those with a secondary education declined significantly from 37% to 24% [45]. While those with a tertiary level education comprise only a small percentage of the total population (about 3%), these trends carry significance for understanding the pathways to economic and social mobility for society as a whole.

Transformations and distinctions in sexual behaviour

Regional distinctions in family formation and sexual and reproductive behaviour in Madagascar have been noted in the ethnographic record and can be deduced from more recent demographic data collection [3, 48–52]. (It is important to note here that administrative divisions in Madagascar officially changed from 6 provinces to a further decentralized 22 regions (faritra) in 2007- both the cities of Antananarivo and Antsiranana were the urban centers of provinces of the same name until 2007. We refer below to these older administrative divisions.) For example, Demographic and Health Survey data from 2004 show that among 25–29 year old women the weighted median age at first sex in the province of Antsiranana was 15.7, while it was 18.7 in the province of Antananarivo. Age at first union (either formal marriage or cohabitation) is correspondingly much younger in Antsiranana than Antananarivo (in 2004 median age at first union was 18.2 in Antsiranana, and 20.7 in Antananarivo) [53]. This distinction may be explained in part by a 2 year gap, as of 2004, in the mean level of educational attainment for 25–29 year olds in Antananarivo as compared to Antsiranana at the provincial level (as opposed to within the respective cities, where the education gap has closed) [54, 55].

There are also differences in the reporting of multiple sexual partners (2 or more sexual partners in the last 12 months) across these regions. According to Youth Sexual Behaviour survey data, among young never married women (ages 15–24) in Antananarivo, the percent reporting having had multiple partners in the last year rose from 5% in 2004 to 9% in 2008; in Antsiranana the corresponding figures moved from 7% to 23% [56–58]. Among men, the pattern is similar, but the magnitude is higher in both areas. These regional distinctions in reported sexual behaviour are not totally unexpected. Ethnographers have documented the relative openness and acceptance of premarital sexuality within certain ethnic groups in Madagascar (for example see: [3, 52, 59]). The area including and surrounding Antananarivo is primarily inhabited by those who self-identify with the Merina ethnic group. The Merina were more highly exposed to and accepting of European missionary efforts beginning in the early 1800’s than many other groups in Madagascar. Urban elite Merina in the city of Antananarivo identify as Christian and contrast their (self-described) chastity with the sexual behaviour of other ethnic groups and social classes; explaining this distinction in large part through Christian ethos, and using it as a basis for ethnic-related discrimination [60]. Many Merina sociologists would suggest that these distinctions are exaggerated, and that Merina are simply less inclined to disclose their sexual practices than other ethnic groups ([61, 62]. It is the case, however, that not only reports of risky sexual practices, but also HIV prevalence rates, are among the lowest in Madagascar within the Antananarivo province (it is also the case that infrastructure, education levels, and access to healthcare are better in this region as compared to others) [63, 64].

National surveillance studies of unmarried youth beginning in 2004 asked about engagement in transactional sex, wording the question as follows: “Have you ever received gifts or money in exchange for sex?” In many parts of Madagascar, this exchange is implicit among a normative set of prescriptions in sexual relationships. Therefore, the observation that there are differences between the provinces of Antananarivo and Antsiranana in the prevalence of reporting of engagement in transactional sex should come as no surprise. In both Antananarivo and Antsiranana the percentage of women who report having received money or material items in exchange for sex rose between 2004 and 2008; in Antananarivo from 1.2% to 5.6%; in Antsiranana, from 22% to 57% [56–58]. In parts of northern and eastern Madagascar a woman was traditionally expected to receive remuneration as a token of respect following sexual relations [52, 59]. Therefore, the meaning of this finding from these surveillance studies is ambiguous. It is difficult to know how these questions are understood by the respondents, and a reported rise in this behaviour may just refer to a rise in sexual activity among this population group- a finding worth consideration, but not the intended meaning of the question. In addition, no study has recorded sexual behaviour alongside ethnic identity (governmental-funded surveys have not been permitted to include questions regarding ethnicity since the early 1970s), place of residence and socio-economic indicators. In this study we are better able to differentiate the influence of ethnic identity from other known influences on sexual behaviour.

Research questions and study expectations

The increasing evidence of the linkages between transactional sex or multiple sexual partners and HIV has led to calls for greater investment in prevention efforts that try to address these behaviours, particularly among youth [65, 66]. Work on understanding the determinants and associations with these behaviours, from the structural to individual level, is therefore imperative for HIV and STI prevention efforts. We contributed to this area of inquiry by asking: Is there an association between young people’s consumption of “modern” goods and reporting of risky sexual behaviours? Specifically, we examined linkages with the reporting of multiple partners in the last twelve months or having ever engaged in transactional sex.

Research describing sex for consumption has often been based in urban contexts. Certainly access and exposure to the stuff of modernity are more apparent in urban landscapes. Limited research has also found independent effects of the urban environment on risky sexual behaviour [67]. This relationship has been explained by both the anonymity, and often the greater diversity (ethnic, religious, linguistic) of urban centers, which together may reduce the effectiveness of communal-based social sanctioning systems [67, 68]. We explored the linkages between consumption and sexual behaviour across type of residence expecting to find a higher prevalence of reported riskier behaviour, levels of consumption, and expressed interest in consumption in urban areas.

Finally, we expected to find strong socio-cultural associations with sexual behaviour, particularly for women. Distinctions in belief systems and practices regarding sexual permissiveness and the meaning of monetary exchange within sexual relationships differ between the cultural groups inhabiting the regions of Antananrivo and Antsiranana. Specific cultural belief systems have been described together with increasing reliance on market economies as important influences on patterns of transactional sex in various parts of sub-Saharan Africa ([69] in Lesotho; [52] in Madagascar; [70] in Kenya;[2] in South Africa).

In order to address the relationship between consumption and sexual behaviour, we first determined the meaning of modern goods and a modern lifestyle in the study sites of interest through formative qualitative research. We then measured the consumption patterns and reported sexual behaviours of youth in the urban, peri-urban and rural context and in two regions. Below we detail the methodological and analytical approach.

Sexual behaviour outcomes

We modeled ‘reporting having had multiple sexual partners in the last year’ as well as ‘reporting having exchanged sex in order to receive gifts or money in one’s lifetime’. Respondents reporting having had two or more sexual partners in the last year were considered as having had multiple partners. We focused on what distinguishes those who reported only having had one partner from those who reported having had multiple partners.

We carefully worded the question designed to assess whether or not respondents had engaged in transactional sex in their lifetime given the normative understanding of exchange within sexual relationships. We stated the motivation behind the exchange asking “Have you ever had sex with the intention of receiving gifts or money in return?”^c We expected this wording would result in a lower percentage of respondents answering in the affirmative as compared to other surveys, particularly in Antsiranana.

Explanatory variables

Explanatory variables included “modern” goods and lifestyle characteristics and perceptions of their importance; region and type of residence; and ethnic identity.

Control variables

The literature points to important socio-demographic and socio-economic determinants of sexual behaviour for young men and women that we captured in our analyses using the following variables: age, education, household asset-based wealth, last month’s expenditures, whether or not the respondent received money from their parents, and household composition.

Education has been associated with lower levels of sexual behaviour, such as delayed initiation of sexual behaviour, or lower rates of multiple partners [72]. We use highest level of education attained in single years (range 0–17).

Research is somewhat equivocal on the relationship between socio-economic status and likelihood of reporting multiple sexual partners or transactional sex. A number of studies support the expected gender dynamic wherein lower economic status is associated with risky behaviour for women [73–77]; while relatively higher socio-economic status is associated with risky sexual behaviours for men [21, 35, 78, 79]. We used three variables to assess respondent’s background socio-economic status as well as disposable income: household asset based wealth, respondent’s last month expenditures, and whether or not a respondent reported their parents as a source of their personal income (as a binary variable). For the latter, the respondent was asked “What all do you do to earn money?” and “pocket money from parents” was among the response categories. We included this variable as it represents respondent’s personal expenditure capability independent of monies they might receive from a sexual partner; it is associated with household wellbeing; and it may also indicate parental attitudes about the extent to which young women or men should be supported as opposed to contribute to the household during this life stage. Research has suggested that parents may allow or subtly encourage girls’ to seek boyfriends who will support the girl [34].

Household-based asset wealth was constructed from a multiple correspondence analysis (MCA) of 19 variables describing characteristics of the home and the household’s major assets.^e Results from MCA have been found equivalent to principle components (PCA) based asset-indices; but MCA is better suited to categorical variables [80]. The household wealth index score was divided into quintiles and then further reduced to a binary variable distinguishing the two poorest quintiles from the three non-poor quintiles.

We collected last month’s income and expenditure data from our individual youth respondents, not their households. We rely on expenditures, as these data are known to estimate income from all sources more accurately than reported income. Last month’s total expenditures is included as a categorical variable with five categories based on the sample distribution ranging from less than 18,000 Malagasy Ariary (about 9 USD) to more than 100,000 Malagasy ariary (about 50 USD). We use this variable only for modelling men’s sexual behaviour as women may be reporting money they have received from sexual partners (29.6% of young women reported their partner as a source of income as compared to only 0.7% of men).

Finally, we include respondent’s household composition, as among never married youth, living with both parents has been found to have a significant impact on sexual behaviour in some parts of sSA [81]; and in Madagascar sociologists have suggested lack of parental oversight as a determinant of risky adolescent sexual behaviour [33, 50]. Following exploratory analysis that showed no statistically significant difference in outcomes for those living with both versus either parent, we operationalized this variable in to three categories: respondent lives with either or both parents; respondent lives with spouse or partner; or respondent lives with others (family or non-family members).

Analysis plan

Data were double entered into Epi-Data and transferred into Stata (Version 10) and SPSS (Version 14) and after cleaning and re-coding data, a series of descriptive statistics were obtained to assess distributions of outcomes, controls and explanatory variables of interest. We present all of our findings stratified by sex, as the prevalence and determinants of sexual behaviour are known to differ by sex. Gendered expectations for sexual behaviour often result in a sexual permissiveness granted to men that is not extended to women [78, 82, 83]. Thus, in most settings men are more likely than women to report having had multiple or multiple and concurrent partners ([78, 84]) and given persistent gendered-power structures, women are more likely than men to report having received goods or money in exchange for sex (but see [79, 85]).

Comparing contextual factors between urban Antananarivo and Antsiranana

Tables 2 and 3 describe the male (n = 1175) and female (n = 1080) study populations, respectively, by region and place of residence. These tables indicate that the urban populations of Antananarivo and Antsiranana were comparable for both sexes in terms of level of education, and age. While the likelihood of reporting living with a spouse or partner was about the same across cities for both sexes, it was more common that respondents reported living with at least one parent in Antananarivo as compared to Antsiranana, where respondents were more likely to report living with extended family members. Although respondents were less likely to live with their parents in Antsiranana, they were as likely (for women), or more likely (for men) to receive financial support from their parents. Last month’s median expenditures differed between urban men from Antananarivo (20,000 Malagasy Ariary; ~10 USD) as compared to Antsiranana (10,000 Malagasy Ariary, ~5 USD). This finding suggests that men in Antsiranana had less spending power than in Antananarivo, consistent with the finding also shown in Table 2, that these men additionally reported having purchased fewer modern goods. That said, households in urban Antsiranana tended to be wealthier than households in urban Antananarivo, overall.

For modern goods and lifestyle (lamaody) characteristics, socio-cultural or structural factors help explain some of the differences found across urban sites. Going to a nightclub is considered more socially acceptable in Antsiranana. In both locations, nightclubs are considered less appropriate for women to attend as compared to men, hence lower percentages of women reported having frequented a nightclub. Internet use was also considered lamaody- the internet was much harder to access in Antsiranana, where there were, as of 2009, very few functioning internet cafes; as compared to multiple such sites in Antananarivo. Finally, for alcohol consumption, given a significant percentage of the respondents in Antsiranana were Muslim (31% of men and 18% of women), one would expect lower reported levels of alcohol consumption in Antsiranana as compared to Antananarivo- this was the case for women, but not for men. The reported differences by region concerning perceptions around fashion did not provide a clear pattern; while women and men in Antsiranana were both more likely to suggest they were often fashionable; women in Antananarivo were more likely to suggest that fashion was very important (12.4% versus 5.6%).

Comparing contextual factors across residence type in Antananarivo

The socio-economic and demographic distinctions by place of residence (rural, peri-urban and urban) within Antananarivo often followed expected trends, particularly among women. Education levels rose from rural to urban place of residence for both sexes; and for women the percentage who reported they lived with a spouse or partner fell from rural to urban residence. As expected, given women tend to marry younger than men, a greater percentage of women reported residing with a spouse or partner, regardless of place of residence. For both men and women it was most common to report living with extended or non-family members in urban residence areas. Monthly expenditures were highest overall among rural men at 24,000 Ariary (about 12.00 USD), reflecting the greater likelihood that these men were heads of households and the primary household income earner. Finally, the percentage of households in the bottom two quintiles of the asset-based wealth index was much higher in rural areas as compared to peri-urban or urban areas.

The same trends from rural to urban were found for some modern goods and lifestyle characteristics. For both men and women, cell phone use, internet use, and having frequented a nightclub all rose from rural to urban place of residence. This was not the case, however, for reported alcohol consumption, where it was high in the rural areas, particularly for men, where 39% reported having consumed alcohol in the last four weeks, a difference of more than 11% from the next highest residence category. While expenditures on fashionable or modern goods rose from rural to urban place of residence for both sexes, it is notable that urban men in Antananarivo outspent all other residence, sex or region categories. There was no clear type of residence based trend in perceptions around fashion. While we expected that women and men in urban areas would value fashion more, we found that in Antananarivo peri-urban women were most likely to suggest they were often fashionable (19.7%); and peri-urban men were most likely to suggest they felt following fashion was important or very important (8%).

Region, residence and sexual behaviour

Reported sexual behaviour differed by region, although for men, unlike in previous studies, the contrast across region within urban areas was almost negligible, 58% in Antananarivo as compared to 61% in Antsiranana reported ever having had sex; and 33% versus 35% reported having multiple partners. For women, however, large differences remained. Only 42% of urban women in Antananarivo as compared to 62% of women in Antsiranana reported ever having had sex; and only 5.5% of urban Antananarivo women reported multiple partners as compared to 19.4% in Antsiranana. Similarly, women’s reports of having ever had sex with the intention of accessing money or materials in return were much higher in Antsiranana as compared to Antananarivo, although the percentages in our study were much lower than in previous studies, likely due to the change in the wording of the question. Regardless, a substantial difference remained with less than one percent of women reporting transactional sex in urban Antananarivo as compared to nearly 20% in urban Antsiranana.

Descriptive statistics provided only limited support to our hypothesis that the urban setting would be the riskier setting for sexual behaviour. For women, there appears to be a trend toward increased likelihood of reporting having had multiple sexual partners in the last year moving from only 1.8% in the rural areas to nearly 4% in the peri-urban zone and finally 5.5% for urban women. For men, this pattern was observed for having ever had sex, where far more young urban men reported having started sexual activity as compared to both their rural and peri-urban counterparts (58% as compared to 43% or 46% respectively). Apart from these findings, however, there was a stronger indication that within Antananarivo the peri-urban environment may be the riskier one for women, as the percentage of women who reported ever having had sex or ever having had transactional sex was highest in this zone of residence; the latter is particularly striking as nearly all of those who reported having practiced transactional sex resided in the peri-urban zone.

Study limitations

This research effort was limited in at least three ways. First, the analysis is hampered by a relatively small sample size and low proportion of women, in particular, who reported either outcome behaviour (7.3% for multiple partners and 5.9% for transactional sex). We did not have enough power to detect difference in models including interaction terms, nor were multivariate region and sex stratified models robust. Therefore, we were not able to test the extent to which urban residence modified the relationship between consumption or lifestyle and sexual behaviour in our multivariate models, or examine in detail the extent to which associations between consumption and sex differed by region. Second, we are also limited by this being a cross-sectional study. We therefore cannot make any claims about the causal direction of these relationships, hindering the capacity to understand to what extent consumption motivates sexual behaviour. Third, and related, our independent and outcome variables are not sequenced appropriately to demonstrate causality within the context of a cross-sectional study design. We had to balance concerns with recall bias alongside our ability to more appropriately predict sexual behavioural outcomes. We used time frames that were consistent with other studies (where applicable), appropriate to the Madagascar context (going to an internet café may not be a frequent occurrence for some), and allowed for recall. Nonetheless, we stress that these models cannot make any causal claims. Our results do provide evidence of association in a new area of research examining the intersections between consumption, lifestyles and culture. Further research is warranted to capture a longitudinal understanding of the motivations for engaging in transactional sex or having multiple sexual partners.