Socioeconomic Vulnerability in sub-Saharan Africa: Lessons Learned - - PDF document

1

Socioeconomic Vulnerability in sub-Saharan Africa: Lessons Learned from the Millennium Development Framework and Implications for the Sustainable Development Goals

Henry V. Doctor1 and Theresa W. Ndavi2

1Henry V. Doctor, World Health Organization, Regional Office for the Eastern Mediterranean, Cairo, Egypt; E-

mail: doctorh@who.int

2Theresa W. Ndavi, USAID, Health Policy Plus, Morning Side Office Park, Nairobi, Kenya;

E-mail: Theresa.Ndavi@thepalladiumgroup.com

Paper prepared for presentation at the XXVIII International Population Conference, Cape Town, South Africa, 29 October – 4 November 2017. Poster session: Tuesday 31 October 2017: 12:00 PM – 1:30 PM, Exhibition Hall 2

2

Socioeconomic Vulnerability in sub-Saharan Africa: Lessons Learned from the Millennium Development Framework and Implications for the Sustainable Development Goals

Abstract Global mobilization behind the MDGs produced the most successful anti-poverty movement in

• history. By 2015, the MDGs helped keep 1 billion people out of extreme poverty and led to improved

lives of many people. Nevertheless, inequalities persist, progress has been uneven, and women continue to be disadvantaged. We use data from Demographic and Health Surveys (29 countries) in sub-Saharan Africa (SSA) conducted between 1990 and 2015 to examine the relative ranking of women (𝑜=690,128) across the wealth index scale by identifying the characteristics of women which influence their likelihood of belonging to “poor” or “rich” households. Ordered probit regression results show that being young, married, urban resident, having some formal schooling, and living in Eastern Africa was associated with higher likelihood of falling within the higher categories of the wealth status index. Being head of the household or living in Middle and Southern Africa was associated with lower categories of wealth status; and women reported low wealth status ranking between the surveys. We interpret these results in line with Sustainable Development Goal 1 which focuses on ending poverty in all its forms everywhere by 2030 and the need to identify and respond appropriately to socio protection challenges in SSA. Introduction The just concluded Millennium Development Goals (MDGs) became synonymous with poverty eradication as it was the key focus in achieving and realising them. The MDGs in themselves were developed and defined to track a country’s performance in its quest towards eradicating poverty primarily through meeting the set targets within the eight defined goals. All other seven goals were linked to Goal 1: eradicate extreme hunger and poverty; and failure to meet them meant no progress towards overall poverty eradication. Extreme poverty in sub-Saharan Africa decreased by 14.2 percentage points, from 56.9% in 1990 to 42.8% in 2012. However, despite the reduction in poverty rates, the increase in population has contributed towards the increase in the absolute number of the poor from 280 million in 1990 to 389 million in 2012 (Beegle et al. 2016). Furthermore, the MDGs and their implementation were focused to a large extent on least developed

• countries. Majority of the least developed countries set highly ambitious targets at the national level

and furthermore failed to meet them. Due to the current prevailing political and social dynamics in the world, poverty eradication has currently become a global task that should be undertaken by all (United Nations, 2015). Following the momentum for the MDGs, the 17 new Sustainable Development Goals (SDGs) aim at providing an opportunity to address the existing gaps based on pertinent lessons from the MDGs as well as bearing in mind the various other emerging socio-economic issues. The growing need to end poverty still prevails in the world (World Bank and IMF 2016). Over the last few decades, natural and man-made disasters, rural-urban migration, rising unemployment, global warming and several other unplanned and undesired situations, have led to increased proportions of people living below the poverty line. This has been observed among those who have always been considered as poor as well as those that have acquired Middle Income Country (MIC) status (Sumner 2010). Hence, policies and strategies will vary significantly across countries when seeking to achieve the all-encompassing SDGs

3 aimed at eradicating poverty. Moreover, this needs to be accompanied by the relevant systems, structures and tools that must be put in place for use by countries to reduce and eventually eradicate

• poverty. Information and data must be availed to ensure appropriate target-setting with defined

milestones by 2030. Sub-Saharan Africa continues to face challenges in realising a poverty-free continent (World Bank and IMF 2017). These challenges include political instability, drought, famine and floods (which are all linked to climate changes), lack of effective social protection measures, global warming, unchecked population growth with unmatched social response, unemployment and underemployment, increasing social pressures and instability, rapid urbanisation, changing social and cultural values, gender inequalities and other social harmful practices such as gender based violence, female-genital mutilation and early child marriage. The gains made since the 1990s in improving the well-being of people in sub-Saharan Africa have been curtailed by new epidemics/pandemics such as HIV/AIDS, drug resistant Tuberculosis and Ebola coupled with weak health system infrastructure (Gostin et al. 2014). Moreover, the merging new diseases including noncommunicable diseases such as diabetes and hypertension and malignancies are on the rise which the systems can hardly respond to (WHO 2014). With respect to the education sector, adult literacy rates have generally been low in sub-Saharan Africa than the rest of the world. For example, UNESCO (2013) reported that in 2011 adult literacy rate was 59% in sub-Saharan Africa than the global average of 84%; which means that the available skill sets produced cannot contribute to a productive economic society and is also closely linked to early/teen pregnancies, new cases of HIV among young girls, poor child rearing practices and drug and alcohol abuse. The overdependence of select sources of national income for instance in the mining sector where oil reserves have dwindled and many others have resulted in not much diversity in leveraging natural resources for development (Downie and Cooke 2011). Poor governance is a feature that is not new to sub-Saharan Africa and its implications ranging from corruption, failure in political democracy and wide inequities and inequalities in wealth distribution usually with only about 90% of resources with/circulating among 1% of the population (United Nations 2004). The lack of non-responsive polices and strategies for those who are most vulnerable is another contributing factor to the challenges; and the lack of quality, reliable, real time data that would be used to inform policies, programmes and service delivery seriously handicaps the process. Despite the socio-economic achievements made in sub-Saharan Africa, women remain at a disadvantage: about 75% of working-age men participate in the labour force, compared to only half of working-age women. Globally, women earn 24% less than men; and in 85% of the 92 countries with data on unemployment rates by level of education for the years 2012–2013, women with advanced education had higher rates of unemployment than men with similar levels of education (United Nations 2015). Against this background, we examine the relative ranking of women interviewed in 29 nationally representative Demographic and Household Surveys (DHS) between 1990 and 2015 across the wealth index scale. Specifically, the objective of the study is to identify characteristics of women (aged 15-49 years) which influence their likelihood of belonging to “poor” or “rich” households. Adopting this strategy has implications on efforts to identify the profile of women and their households from which efforts to improve their health and socio-economic well-being can be targeted at.

4 Data and Methodology We used publicly de-identified data from the DHS Program (http://www.dhsprogram.com/). DHS

• ffer various socio-economic indicators including wealth quintiles calculated using information on

households’ assets and other possessions. Individual-record data files were available from the DHS Program for 29 countries (Table 1) with surveys conducted between 1990 and 2015. We used surveys from countries that had the earliest and the most recent surveys (n=58) with wealth status scale provided by wealth quintiles. The wealth quintile from the DHS distributes women into almost equal proportions in each of the wealth quintile group (20%), that is, poorest, poor, middle, richer, and richest wealth quintile. The “earliest surveys” were defined as those conducted from 1990 onwards but before 2010; and “most recent” surveys (or “latest” surveys) were those conducted from 2010 to 2014, close to the MDG deadline of 2015. We combined data from the earliest (n=244,590 women) and latest surveys (n=445,538) into a pooled data set covering 690,128 women. We then used ordered probit regression of wealth index on age, marital status, sex of household head, residence (urban/rural), schooling, sub-Saharan African region and round of survey (earliest/latest). We used Stata (version 14, StataCorp LP, College Station, TX, USA) for analyses. [Table 1, about here] Results The mean age of women in the sample was constant at 24 years during the earliest and latest survey

• rounds. At least 80% of women were married or living with a partner across the survey rounds, and

the proportion of households headed by women increased by four percentage points from 25% during earliest surveys to 29% during latest surveys. The proportion of women living in urban areas also increased by four percentage points from 27% to 31% between the two survey rounds; and women with at least some primary school increased between the survey rounds from 40% to 49%. Across all surveys, there were more women interviewed in Western Africa followed by Eastern Africa, Middle Africa and Southern Africa. [Table 2, about here] Ordered probit regression results showed that being aged 20-29 years, married, urban resident, having some formal schooling, and living in Eastern Africa was associated with higher likelihood of falling within the higher categories of the wealth status index. Being a female head of the household or living in Middle and Southern Africa was associated with lower categories of wealth status; and women reported low wealth status ranking between the surveys (Table 3). [Table 3, about here] Figure 1 displays the adjusted predictions for wealth status ranking at the means of the independent variables; and is based on post estimation of the ordinary probit model. The results show that if all the

• ther variables are set to their means or average values, the predicted probability of, for e.g., a married
• r cohabiting respondent (compared to unmarried) is the highest of all the variables considered. This

is followed by the predicted probability of a respondent interviewed in the latest survey round (0.64) when compared with a respondent interviewed during the earliest survey round. We see in Figure 1 that the lowest predicted probability is 0.03 for respondents living in Southern Africa compared with respondents living in Western Africa.

5 [Figure 1, about here] Table 4 shows the predicted probabilities for the wealth status categories evaluated at the sample means of the data. For a particular explanatory variable, the marginal effects of the five wealth categories must sum to zero by definition.1 Since our variables are categorical, the marginal effect reported in Table 4 is the change in the predicted probability based on whether a respondent falls into that category or not. Thus, for these marginal effects all the remaining variables assume their respective average values. As such these marginal effects are showing the change in the predicted probability for each wealth index class for an average respondent, according to the variable being considered. [Table 4, about here] The top panel of Table 4 shows the predicted probabilities for the five wealth categories evaluated at the sample means of the data. These probabilities indicate a strong likelihood that the average respondent is more likely to fall into the middle wealth status group. The bottom panel of Table 4 shows the marginal effects for all explanatory variables. Starting with the age of women, results show that being aged 20-24 years reduces the probability of being in the poorest wealth group or the poor or the middle wealth group. The same impact applies to the other ages groups (25-29 and 30 and over). However, being in age groups 20-24, 25-29, and 30 years and above increases the probability of being in the richer or richest category. Worth noting is the fact that these changes are relative to a woman under 20 years. The marginal effects for a woman aged 25-29 years are also stronger than for a woman aged 20-24 years. All other things being equal this suggests increased variability in wealth status rankings by age group. Women who were married or cohabiting had negative marginal effects for the first two wealth status groups (i.e. poorest and poor), but a positive marginal effect on all other wealth status categories. Being female head of the household was associated with positive marginal effects for the poorest and poor wealth status groups but a negative effect on from the middle to the richest wealth group. Residence in urban area reduces the probability of being in the poorest or poor or middle wealth

• groups. However, residing in urban areas increases the probability of being in the richer or richest

wealth status groups. Marginal effects for schooling indicates that women with at least primary schooling were more likely to be in richer or richest wealth status group relative to those who did not have any formal schooling. Residence in Middle Africa increases the probability of being in the first two wealth status groups relative to residence in Western Africa; whereas residence in Eastern Africa increases the probability of being in the richer and richest wealth status groups. The marginal effects for women in Southern Africa indicate a high likelihood of being in the first three wealth status groups relative to women residing in Western Africa. Overall, the marginal effects of women interviewed in the latest survey round shows that they were more likely to fall in the first three wealth

• groups. By implication, these results show that women interviewed during the latest survey rounds

were more likely to be poor relative to women interviewed during the earliest survey rounds. Discussion The aim of this paper was to identify characteristics of women (aged 15-49 years) which influence their likelihood of belonging to “poor” or “rich” households in order to identify the profile of women and their households from which efforts to improve their health and socio-economic well-being can

1Since the probabilities for the wealth status categories must sum to one, the change in probabilities for the wealth status

categories must sum to zero.

6 be targeted at. Results in our study show that individual factors such as age and completing at least primary education have a positive influence on the wealth status ranking of women in the countries and surveys included in the analysis. That poverty has increased between the surveys, as evidenced by the low wealth status ranking of women is a cause for concern and need concerted efforts by 2030 to ensure that sub-Saharan Africa achieves SDG 1 which calls for an end to poverty and all its manifestations and ensure social protection for the poor and vulnerable. Results also showed that living in the urban areas reduces one’s probability of being poor. However, more than half of the women in sub-Saharan Africa live in the rural areas and this has posed a challenge and continues to pose a challenge on the reduction of poverty. The rapid growth of cities in sub-Saharan Africa has consequences on the magnitude of the urban-rural divide (Brückne 2012). By implication, women living in the rural areas will fall far behind their counterparts in the urban areas with respect to key services such as access to health and social services, energy and food security. Contemporary higher education in sub-Saharan Africa continues to be highly gendered and this tend to influence the persistence of inequality across various sectors such as political, economic and social sectors (Klasen, 2003). That half of women in this study had at least primary education has consequences on strategies to enhance their socioeconomic status. Moreover, female dropout rates also remain high mainly due to lack of conducive learning environments and strong support systems. Having a large number of uneducated women in society leads to an even larger population that is vulnerable (Beegle et al. 2016). A large vulnerable population mitigates efforts in reducing poverty levels and contributes towards the existing challenges in many rural and disadvantaged communities. Majority of women in sub-Saharan Africa engage in agricultural activities particularly in rural areas. A minority of the women are engaged in paid work; and there is a large share of unpaid workers among women. This overall translates to low economic empowerment which is worsened by prevalence of gender-based violence. Women who experience gender-based violence and discrimination in sub-Saharan Africa have limited access to support systems leading to poor reproductive health outcomes. It is worth noting that there are high new HIV infections among young girls aged 15-24 years in Southern Africa while in West Africa they are low (UNAIDS, 2016). The implication for this especially in Southern Africa places these young girls under vulnerable population and they are more likely unable to access HIV services (UNAIDS, 2016). This decreases their chances

• f economic empowerment.

Generally, female representation in African parliaments tend to be low; and there are more heads of ministries, parastatals, corporations and companies who are men than women. By September 2017, Africa had only two female heads of state and government, that is, Liberia and Mauritius. Despite the existence of affirmative action policies and programmes, the lack of women as leaders in key institutions leads to poor implementation of pro-women empowerment policies. MDG 3 was the only goal that focused on gender. None of the other 7 MDGs looked at implications of gender inequities and inequalities with targets limited to outcome levels for each MDG. This was noted with concern and because of this, the SDGs have gender mainstreamed in all the 17 goals. This has ensured that many of the targets and outcomes allow for flexibility. MDGs focused more on poverty; however, SDGs cover all and responds to the needs of women in all African countries including those living in countries affected with humanitarian crises. Despite the fact that several countries in sub-Saharan Africa are in conflict and face different needs that require a multitude of interventions, these can be addressed within the SDGs. This includes interventions towards protecting the main casualties of war and conflict, that is, women and children. Africa’s interest were not reflected within the MDGs and because of this many of the countries were not able to achieve the MDGs (African Union, UNECA, AfDB and UNDP, 2016). As a lesson learnt, Africa ensured that her needs were taken care of during the development of the SDGs particularly through the Common African Position on the post-2015 Development Agenda.

7 Conclusion With mounting momentum to ensure that no one is left behind by 2030, the results from this study demonstrates existing challenges and areas that need focus to ensure that women have a platform where they can demonstrate their potential and increase the probability of acquiring life enhancing

• resources. The extent to which this can be realised by 2030 is contingent on sound policies and

commitment of national governments and the international community to uplift the status of women. For proper implementation of policies that enhance the socioeconomic status of women, there is a need to take advantage of existing data and promote collection and analysis of disaggregated data. Acknowledgments: We would like to thank the MEASURE DHS Program and the National Statistical Offices of the 29 sub-Saharan African countries for making the data publicly available with financial support from USAID. Disclaimer: The authors alone are responsible for the views expressed in this article and they do not necessarily represent the views, decisions or policies of the institutions with which they are affiliated.

8 References African Union, United Nations Economic Commission for Africa [UNECA], African Development Bank and United Nations Development Programme [UNDP]. 2016. MDGs to Agenda 2063/SDGs Transition Report 2016: Towards an integrated and coherent approach to sustainable development in

• Africa. Addis Ababa, Ethiopia.

Beegle K, Christiaensen L, Dabalen A, Gadis I. 2016. Poverty in a Rising Africa. Washington, DC: World Bank. doi:10.1596/978-1-4648-0723-7 Brückne M. 2012. Economic growth, size of the agriculture sector, and urbanization in Africa. Journal of Urban Economics 71(1): 26-36. Downie R, Cooke JG. 2011. Assessing risks to stability in Sub-Saharan Africa. Washington, DC: Center for Strategic and International Studies. Gostin LO, Lucey D, Phelan A. 2014. The Ebola epidemic: A global health emergency. JAMA 312 (11): 1095-1096. Klasen, S. 2003. Low schooling for girls, slower growth for all? Cross‐country evidence on the effect

• f gender inequality in education on economic development. The World Bank Economic Review 17(2)

315–316. Sumner A. 2010. Poverty and the New Bottom Billion: What if Three-quarters of the World’s Poor Live in Middle-income Countries? IDS Working Papers Volume 2010 (349): 1-43. doi: 10.1111/j.2040-0209.2010.00349_2.x United Nations. 2015. The Millennium Development Goals Report 2015. New York: United Nations. United Nations. 2004. United Nations Convention Against Corruption. New York: United Nations. UNAIDS 2016. Global Aids Update 2016. New York: United Nations.

• UNESCO. 2013. Adult and youth literacy: National, regional, and global trends, 1985-2015.

Montreal: UNESCO Institute for Statistics. World Bank and International Monetary Fund [IMF]. 2016. Development goals in an era of demographic change. Global monitoring report 2015/2016. Washington DC: The World Bank and IMF. World Health Organization. 2014. Global status report on non-communicable diseases 2014. Geneva: World Health Organization.

9

Table 1: Countries and Demographic and Health Surveys included in the analysis for 29 Sub-Saharan African countries Country Earliest Survey Latest Survey Observation timea Western Africa (n=12) Benin 1996 2011-2012 16 Burkina Faso 1993 2010 17 Cote d’Ivoire 1994 2011-12 18 Ghana 1993 2014 21 Guinea 1999 2012 13 Liberia 2007 2013 6 Mali 1995-96 2012-13 17 Niger 1998 2012 14 Nigeria 1990 2013 23 Senegal 1997 2014 17 Sierra Leone 2008 2013 5 Togo 1998 2013-14 16 Middle Africa (n=4) Cameroon 1991 2011 20 Congo (Brazzaville) 2005 2011-12 7 Congo Democratic Republic 2007 2013-14 7 Gabon 2000 2012 12 Eastern Africa (n=11) Comoros 1996 2012 16 Ethiopia 2000 2011 11 Kenya 1993 2014 21 Madagascar 1997 2008-09 12 Malawi 1992 2010 18 Mozambique 1997 2011 14 Rwanda 1992 2010 18 Tanzania 1996 2010 14 Uganda 1995 2011 16 Zambia 1996 2013-14 18 Zimbabwe 1994 2010-2011 17 Southern Africa (n=2) Lesotho 2004 2014 10 Namibia 1992 2013 21

aObservation time calculated based on the upper bound of the year. For example, the 2010-2011 year

uses 2011 as the end point. Latest surveys defined as those from 2010 with the exception of Madagascar (2008-09). Source: DHS StatCompiler (www.statcompiler.com); Last accessed 9 December 2016.

10

Table 2: Descriptive results of the study population from Demographic and Health Surveys in 29 sub-Saharan Africa conducted between 1990 and 2015 Characteristics Earliest surveys Latest surveys All surveys Number (Percent) Number (Percent) Number (Percent) Wealth index Poorest 62,318 (19.1) 73,304 (20.0) 112,618 (19.7) Poor 44,267 (21.5) 74,586 (20.4) 118,852 (20.8) Middle 42,270 (20.5) 75,757 (20.7) 118,051 (20.6) Richer 41,447 (20.1) 75,392 (20.6) 116,850 (20.4) Richest 38,504 (18.7) 67,080 (18.3) 105,583 (18.5) Mean age (SD)* 24.3 (6.5) 24.5 (6.5) 24.5 (6.5) Married/Living together No 42,376 (18.3) 71,254 (19.5) 113,989 (19.0) Yes 189,566 (81.7) 294,904 (80.5) 485,955 (81.0) Sex of household head Male 174,631 (75.3) 259,532 (70.9) 434,162 (72.6) Female 57,314 (24.7) 106,625 (29.1) 163,940 (27.4) Residence Rural 170,157 (73.4) 251,696 (68.7) 421,843 (70.5) Urban 61,791 (26.6) 114,461 (31.3) 176,262 (29.5) Formal schooling None 139,765 (60.3) 185,923 (50.8) 325,649 (54.5) At least primary 92,171 (39.7) 180,211 (49.2) 272,421 (45.6) Region Western Africa 99,947 (44.8) 200,691 (54.8) 300,620 (51.0) Middle Africa 20,221 (9.1) 16,257 (4.4) 36,466 (6.2) Eastern Africa 91,676 (41.1) 135,881 (37.1) 227,512 (38.6) Southern Africa 11,125 (5.0) 13,328 (3.6) 24,448 (4.2) Number of women 232,661 (100.0) 367,289 (100.0) 599,950 (100.0)

Note: *SD – standard deviation

11

Table 3: Ordered probit regression analysis of wealth status ranking on selected individual characteristics, Demographic and Health Surveys conducted between 1990 and 2015 Characteristics Coefficient z-statistic p-value Age group <20 (r)

• 20-24

0.03 7.24 0.000 25-29 0.04 7.59 0.000 30+ 0.01 1.64 0.100 Married/Living together No (r)

• Yes

0.06 10.78 0.000 Sex of household head Male (r)

• Female
• 0.12
• 26.17

0.000 Residence Rural (r)

• Urban

1.26 291.03 0.000 Formal schooling None (r)

• At least primary

0.43 108.65 0.000 Region Western Africa (r)

• Middle Africa
• 0.23
• 30.73

0.000 Eastern Africa 0.09 22.46 0.000 Southern Africa

• 0.02
• 1.46

0.000 Round of survey period Earliest (r)

• Latest
• 0.10
• 28.42

0.000 Prob > F 0.000 Number of observations 557,271 Note: r – reference category.

Table 4: Predicted probabilities and marginal effects from the estimated ordered probit model of wealth status ranking, Demographic and Health Surveys conducted between 1990 and 2015 Poorest Poor Middle Richer Richest Predicted probabilities 0.1530 0.2180 0.2492 0.2423 0.1376 Marginal effects Respondent is aged 20-24 years

• 0.0078
• 0.0047
• 0.0001

0.0053 0.0074 Respondent is aged 25-29 years

• 0.0087
• 0.0053
• 0.0002

0.0059 0.0083 Respondent is aged 30 years and above

• 0.0019
• 0.0011
• 0.0000

0.0013 0.0018 Respondent is married/cohabiting

• 0.0140
• 0.0081

0.0001 0.0094 0.0125 Household head is female 0.0298 0.0169

• 0.0005
• 0.0201
• 0.0261

Lives in urban area

• 0.2301
• 0.1762
• 0.0656

0.1233 0.3486 Attended at least primary schooling

• 0.0988
• 0.0600
• 0.0029

0.0662 0.0955 Lives in Middle Africa 0.0591 0.0290

• 0.0047
• 0.0387
• 0.0447

Lives in Eastern Africa

• 0.0212
• 0.0129
• 0.0005

0.0144 0.0201 Lives in Southern Africa 0.0036 0.0021 0.0000

• 0.0024
• 0.0033

Latest survey round 0.0244 0.0150 0.0007

• 0.0166
• 0.0234

12

.2 .4 .6 .8 Adjusted predictions

Lives in Southern Africa Lives in Middle Africa Respondent is aged 30 years and above Respondent is aged 25-29 years Household head is female Respondent is aged 20-24 years Lives in urban area Lives in Eastern Africa Attended at least primary schooling Latest survey round Respondent is married/cohabiting

ranking at the means of the independent variables Figure 1: Adjusted predictions for wealth status