Report No: AUS0002718 . Cameroon Juvenocracy: Youth-shaped IGA A Tale of Two Countries: Labor Market Profiles of Youth in Urban and Rural Cameroon . September, 2021 . SPL . Document of the World Bank . . © 2017 The World Bank 1818 H Street NW, Washington DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved This work is a product of the staff of The World Bank. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Attribution—Please cite the work as follows: “World Bank. 2022. A Tale of Two Countries: Labor Market Profiles of Youth in Urban and Rural Cameroon. © World Bank.” All queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. A Tale of Two Countries: Labor Market Profiles of Youth in Urban and Rural Cameroon Ioana Botea and Mitja Del Bono1 Abstract Low productivity – rather than absolute employment levels – is the main labor market challenge in Cameroon, where vulnerable employment in the form of subsistence farming or self-employment in the informal sector remains the norm. Low-skill, low-pay agricultural work is prevalent in rural areas, while more productive employment in urban areas is increasingly accompanied by high unemployment and inactivity among educated youth holding out for public sector jobs. Labor market vulnerability, either detachment or weak-attachment, is thus particularly acute among youth (ages 15 to 35), who are often uninterested in agriculture and unable to access better opportunities in urban areas. Using an advanced statistical technique, we identify distinct profiles of youth experiencing labor market vulnerability. The largest group in urban areas (35 percent) consists of men with some education who work full-time in the informal sector. In rural areas, almost half (45 percent) of youth continue to work the land at a subsistence level. A clear pattern of gender inequality emerges: all detachment groups are majority women, with high inactivity rates among married women in rural areas and highly-educated but discouraged women in urban areas. Women are also overrepresented among the most vulnerable employed groups, especially in rural areas where they often work as unpaid family support. Tailored employment support interventions are, therefore, needed to promote inclusive productivity and effectively utilize the country’s human capital. 1World Bank, Washington, DC, USA. Corresponding authors: Ioana Botea (ibotea@worldbank.org) and Mitja Del Bono (mdelbono@worldbank.org) 1 Table of Contents Abstract................................................................................................................................................................. 1 1 Introduction.................................................................................................................................................. 3 2 Country Context: The Cameroonian Labor Market ..................................................................................... 4 2.1 Labor Market Overview ............................................................................................................... 4 2.2 Youth-Specific Vulnerabilities ..................................................................................................... 6 3 Youth Profiles: Methodology .......................................................................................................................10 3.1 Reference Population ..................................................................................................................10 3.2 Labor Market Vulnerability Indicators.....................................................................................11 3.3 Methodology ................................................................................................................................12 4 Youth Profiles: Results ...............................................................................................................................13 4.1 Urban Results ...............................................................................................................................13 4.2 Rural Results.................................................................................................................................18 4.3 Priority Classes .............................................................................................................................23 5 Conclusion and Recommendations...............................................................................................................25 6 Bibliography ...............................................................................................................................................27 Annex 1. Latent Class Analysis Methodology ....................................................................................................29 Annex 2. Additional Results .............................................................................................................................32 Pair-Wise Correlation Coefficients ........................................................................................................................34 2 1 Introduction As typical in the region, low productivity – rather than absolute employment levels – is the main labor market challenge in Cameroon. High labor force participation (74 percent) and low unemployment (4 percent) rates mask two important issues. First, underemployment is prevalent among those who are employed: 58 percent earn less than the national minimum wage and 28 percent work less than 40 hours a week. While government has made reducing underemployment a main objective of its Growth and Employment Strategy,2 available survey data suggests a persistent upwards trend instead. Second, economic activity and unemployment rates vary substantially across socio-demographic characteristics. Youth, women, and those with low education are especially disadvantaged, with stark differences between rural and urban areas. Labor market vulnerability is particularly acute among Cameroonian youth (ages 15-35), who are not only more likely to be in precarious employment but often remain economically inactive. Young people in Cameroon are twice as likely as non-youth to have exited the labor force (19 percent vs. 10 percent), with even higher rates among young women. For youth in the labor force, unemployment is increasingly common especially in urban areas and among those with higher education (20 percent). Those employed tend to be relegated to low-skill, informal sector jobs that are characterized by insecurity, seasonality, and low returns. Having attained considerably higher education levels than previous generations, youth struggle more than ever to find work outside their families’ farms or to access the better employment opportunities they seek in urban areas. Effectively addressing the youth employment challenge requires a nuanced understanding of their labor market vulnerabilities and socio-economic characteristics. “Youth” are not a homogeneous group: for example, it can include a 16-year old boy working on the family farm in a remote rural area and a 24-year old single mother eking out an existence vending on the streets of a regional urban center. Since their circumstances are markedly different, they require tailored policy interventions that reflect characteristics such as gender, area of residence, and education level. This study aims to identify different profiles of youth experiencing labor market vulnerability and inform the development of tailored and effective employment promotion interventions. Using an advanced statistical method (Latent Class Analysis), it generates groups of youth who are likely to share similar labor market vulnerabilities (e.g., unemployment, temporary employment, unpaid employment) and are thus expected to benefit from similar types of assistance. It then describes dominant socio-demographic characteristics of each group to guide the design of well- targeted employment support policies and programs. While recognizing the essential role demand plays in improving labor market outcomes, the analysis focuses primarily on supply-side constraints and corresponding policies. Building on similar analyses conducted in advanced economies, the present study is the first of its kind in Sub-Saharan Africa. It consists of five sections including this introduction. Section 2 provides background on the Cameroonian labor market and its stark rural-urban divide. Section 3 describes the framework and the statistical clustering methodology. Section 4 presents the results, including a description of the identified clusters according to labor market vulnerability indicators and corresponding demographic and socio-economic characteristics. Finally, section 5 presents conclusions along with policy recommendations. 2The 2010-2020 Document de Stratégies pour la Croissance et l'Emploi (DSCE) aimed to reduce the underemployment rate from 76 percent to under 50 percent by 2020. 3 2 Country Context: The Cameroonian Labor Market 2.1 Labor Market Overview Poverty and inequality are increasing in Cameroon due to non-inclusive growth, rapid demographic change and conflict. Several years of robust growth have had insufficient impact on the poverty rate, which remained at 38 percent in 2014.3 Due to population growth, however, the number of people living in poverty has increased substantially over the past two decades, especially in the fragile North and Far North regions which account for 56 percent of the poor (see Annex 2). In addition to worsening regional inequality, the gap between rural and urban areas is also widening measurably: while the poverty rate decreased substantially in urban areas (from 18 percent in 2001 to 9 percent in 2014), it increased in rural areas (from 52 percent in 2001 to 57 percent in 2014). High labor force participation (74 percent) and low unemployment (4 percent) rates mask two important challenges. First, unemployment rates vary substantially across socio- demographic characteristics. As shown in Figure 1 below, unemployment is primarily an urban concern: 8 percent of urban workers are unemployed (10 percent in Yaoundé and 9 percent in Douala), compared to 1 percent of rural workers. Or, 83 percent of the unemployed live in urban areas. Unemployment is also feminized, with women being much more likely to be out of work than men, especially when using the broader definition of unemployment that includes those involuntarily out of the labor force4 (17 percent vs. 5 percent). Finally, those with higher education experience the highest rates of unemployment (14 percent vs. 1 percent among those with no formal education). Figure 1. Unemployment rates, by socio-demographic characteristics 13.7% 8.0% 5.6% 4.7% 4.1% 3.5% 1.9% 1.2% 0.7% Urban Rural Male Female No Education Primary Secondary Higher Total Area Gender Education Unemployment Second, underemployment is prevalent among those who are employed. A third of those employed (28 percent) are visibly underemployed, involuntarily working fewer than 40 hours a week, across socio-demographic groups. An even higher share (58 percent) are invisibly underemployed, meaning that they earn less than the national minimum wage of 36 270 CFAF per month. Figure 2 below shows substantial pay gaps, particularly by gender and educational attainment. There are substantial returns to education: the share of those invisibly underemployed is four times lower among those with higher education than among those with no education (22 percent vs. 82 percent). 3 National poverty rate, compared to 40.2 percent in 2001 and 39.9 percent in 2007. 4 Unemployment+ includes individuals who are seeking work (undemployed) or no longer seeking work for involuntary reasons (involuntarily inactive). Quoted reasons include: lack of employment, lack of required qualifications, does not know how to seek an employment, refusal of the spouse, respect of tradition, or other involuntary reason. 4 Women are considerably more likely to be invisibly underemployed (35 percent vs. 23 percent) than men, earning half as much (0.57 times) as men overall. The gender pay gap disappears, however, among workers with higher education. Figure 2. Underemployment rates, by socio-demographic characteristics 82.2% 73.8% 71.4% 64.8% 57.9% 47.2% 44.8% 39.3% 37.5% 34.9% 30.1% 29.1% 28.7% 28.4% 27.4% 23.1% 22.0% 18.8% Urban Rural Male Female No Education Primary Secondary Higher Total Area Gender Education Visible Underemployment Invisible Underemployment Overall underemployment is thus rising, despite its recognition as a top policy priority to promote inclusive growth and reduce poverty. Based on ECAM4, the underemployment rate increased from 68.7 percent in 2010 to 77 percent in 2014, with rates even higher among youth. This issue has long been recognized by Government, which identified reducing underemployment as a key priority under its 2010-2020 Growth and Employment Strategy Document (Document de Stratégie de Croissance et d’Emploi, DSCE). The DSCE aimed to reduce underemployment from 75.8 percent to less than 50 percent by 2020 through the creation of formal sector jobs. While the most recent data available is from 2014, the trend suggests that underemployment continued to increase rather than decline over the past decade. Widespread underemployment is closely linked to informality, with 84 percent of Cameroonian workers operating in the informal sector.5 The formal sector is largely located in Yaoundé and Douala and is highly concentrated: less than 1 percent of all registered firms generate 68 percent of all revenue. A 2016 World Bank report6 identifies several key challenges to increased formal sector employment, including a weak business environment (high taxes, corruption, limited access to credit, unfair competition), a mismatch between generalist skills at university (mainly for public sector) and skills demanded, and lack of entrepreneurial and technical skills for informal sector workers. These challenges also explain the increasing rates of unemployment and inactivity among youth with higher education who are reluctant to accept the low-paid, typically insecure, jobs available in the informal sector. Finally, there is a stark urban-rural divide in the labor market. As shown in Figure 3 below, approximately 90 percent of rural workers earn less than minimum wage compared to 10 percent of urban workers. There is a marked rural concentration (~70 percent) in the low-skilled, low-paid primary sector, while more lucrative service sector jobs (wages up to five times higher than primary sector jobs) are concentrated in urban areas. This points to two very different economies and labor markets: while rural areas remain focused on subsistence agriculture, urban areas are increasingly 5 83.7 percent based on authors’ analysis of ECAM4 data, compared to 87.9 percent reported by government. 6 Sosale & Majgaard (2016) 5 modernizing and diversifying. Indeed, according to a 2018 report, cities in Cameroon are at least 1.8 times more productive than rural areas (World Bank, 2018). While cities’ tradable sectors are important for generating better quality jobs, it’s critical at least in the short term to improve productivity and expand employment opportunities in the rural areas where a majority of Cameroonian (especially the poorest and most in need of assistance) continue to reside. Figure 3. Productivity estimates, by area of residence and employment sector 80 Minimum Wage Urban Rural 70 Primary 60 Share of employment in area (in %) 50 Services 40 30 Commerce 20 Industry Industry 10 Services Primary Commerce 0 0 10,000 20,000 30,000 40,000 50,000 60,000 Median monthly earnings (in CFAF) 2.2 Youth-Specific Vulnerabilities The Cameroonian workforce is young, with 59 percent of workers aged between 15 and 35, despite relatively lower labor force participation rates. The high share of youth is directly linked to the rapid population growth over the past couple of decades. However, the share of population aged 15 to 35 as compared to that aged 15 to 64 (and hence counted as part of the workforce) is even higher (66 percent). The gap is partly explained by youth still in school (60 percent of those aged 15- 35) and partly by economic inactivity: labor market participation is 25 percent higher among non- youth (ages 36-64) than youth (ages 15-35). As explored in more detail in the rest of the study, labor market exclusion and vulnerability are particularly acute among young Cameroonians, who have higher education attainment and aspirations than their parents but have access to limited employment opportunities. Youth are twice as likely to be inactive than non-youth (20 percent vs. 10 percent), either involuntarily or voluntarily. Among youth, as shown in Figures 4 and 5 below, the widest gaps are along gender and education lines. First, inactivity rates are more than double for young women than young men (25 percent vs. 11 percent), mainly driven by involuntary inactivity. The main reasons reported include a perceived lack of employment or of required qualifications, refusal of the spouse, or respect of tradition. Second, there’s a strong correlation between education and the likelihood of being inactive, however it goes in opposite directions for voluntary or involuntary inactivity. Comparing the two extremes, those with no education are much more likely to be out of the labor force involuntarily (15 percent vs. 2 percent), whereas those with higher education more often exit the labor force by choice (19 percent vs. 9 percent). The high rates of voluntary inactivity are particularly stricking when comparing with those among non-youth, which hover at 5-6 percent regardless of education level. 6 Figure 4. Rate of involuntary inactivity, by age group 14.9% 10.8% 7.8% 7.2% 6.8% 6.7% 6.3% 6.3% 5.4% 4.8% 4.3% 4.2% 3.5% 3.4% 2.4% 2.2% 1.8% 0.9% Urban Rural Male Female No Education Primary Secondary Higher Total Area Gender Education Youth (15-35) Non-Youth (36-64) Figure 5. Rate of voluntary inactivity, by age group 19.0% 14.3% 13.9% 13.8% 11.7% 9.5% 9.3% 9.0% 8.8% 8.4% 7.6% 5.8% 5.7% 5.6% 5.3% 5.0% 3.9% 3.5% Urban Rural Male Female No Education Primary Secondary Higher Total Area Gender Education Youth (15-35) Non-Youth (36-64) For youth in the labor force, unemployment is on the rise particularly in urban areas and among those with higher education. The youth unemployment rate has been steadily increasing over the past decade, from 4.6 percent in 2007 to 6.3 percent in 2014. As shown in Figure 6 below, overall unemployment is three times higher for youth compared to non-youth (6 percent vs. 2 percent), and this pattern holds across socio-demographic characteristics. The issue is particularly salient among those with higher education (19 percent vs. 3 percent) and in urban areas (11 percent vs. 2 percent). The unemployment rate among urban youth with university degrees is a staggering 20 percent, compared to 3 percent among non-youth with the same characteristics. Most unemployed youth are typically holding out for public sector jobs given high salaries and job security as well as the mismatch between their generalist skills and those required by the private sector, which has also lagged behind in terms of job creation for the increasingly well educated workforce. 7 Figure 6. Unemployment rate, by age group 19.0% 10.7% 6.9% 6.9% 5.8% 4.7% 3.7% 3.3% 2.6% 2.5% 1.9% 1.8% 1.7% 1.6% 1.0% 1.0% 0.4% 0.3% Urban Rural Male Female No Education Primary Secondary Higher Total Area Gender Education Youth (15-35) Non-Youth (36-64) Almost a third of young workers are visibly underemployed, or work for fewer than 40 hours per week. While still higher among youth than non-youth, the gap in terms of visible underemployment is narrower than for other indicators. A notable exception is, again, among those with higher education, with youth almost twice as likely as non-youth to be visibly underemployed (44 percent vs. 25 percent). An even clearer gap emerges when looking at temporary employment, which is twice as common among youth than non-youth (28 percent vs. 15 percent). As shown in Figure 8, population segments most likely to work for limited periods of time include youth with no education (43 percent) and young women (35 percent). While more likely to be inactive or unemployed than other youth in general, when they’re employed, youth with higher education are the least likely to work temporarily (4 percent). Figure 7. Rate of visible underemployment, by age group 44.0% 36.4% 32.5% 30.5% 30.2% 30.0% 30.0% 29.8% 29.4% 28.1% 27.3% 25.7% 24.8% 24.6% 22.5% 20.4% 20.4% 17.3% Urban Rural Male Female No Education Primary Secondary Higher Total Area Gender Education Youth (15-35) Non-Youth (36-64) 8 Figure 8. Rate of temporary employment, by age group 42.9% 36.4% 31.5% 29.1% 27.7% 26.5% 25.9% 24.4% 21.1% 20.0% 16.8% 15.7% 15.1% 13.6% 11.8% 7.6% 7.6% 4.0% Urban Rural Male Female No Education Primary Secondary Higher Total Area Gender Education Youth (15-35) Non-Youth (36-64) Invisible underemployment is equally common, with a high share of youth engaged in unpaid work. Two thirds of youth (60 percent) earn less than the minimum wage, compared to 56 percent of non-youth. On average, youth earn 40 percent less than their older counterparts. The differences are driven by sector of activity: public sector jobs, which are typically held by non-youth, pay 2 times more than formal private sector jobs and 4 times more than the informal sector (see Annex 2). Largely excluded from the public sector, youth are relegated to lower-paying sectors and activities within those sectors. As shown in Table 9 below, the relatively high share of youth in unpaid employment (22 percent vs. 6 percent) further explains the pay gap. Unpaid employment, in the form of household aid or apprentice, is especially common for youth in rural areas (31 percent) and among women (29 percent). Figure 9. Employment status, by age group 39.8 49.7 51.6 47.4 52.1 56.7 67.5 63.1 77.8 72.2 4.4 2.5 9.7 2.9 6.8 2.0 1.4 1.1 16.3 22.3 4.9 3.9 0.8 28.6 6.1 31.2 2.8 43.9 38.7 2.0 31.9 11.8 24.5 9.2 30.0 21.5 16.5 10.5 10.3 12.5 15-34 35-64 15-34 35-64 15-34 35-64 15-34 35-64 15-34 35-64 Total Urban Rural Male Female Waged Employment Unpaid Employment Employer Self-Employment Finally, widespread informality underlies all aspects of labor market vulnerability presented above. Similar shares of youth and non-youth work in the informal sector (85 percent vs. 83 percent). As Figure 10 illustrates, significant differences between youth and non-youth emerge only among those with higher education (46 percent vs 29 percent), driven by low access to public sector jobs among youth. 9 Figure 10. Informality and sector, by age group 97.9% 97.1% 94.4% 92.5% 92.4% 92.1% 89.2% 88.0% 84.4% 82.9% 81.0% 80.6% 77.0% 72.5% 70.5% 68.9% 46.2% 29.4% Urban Rural Male Female No Education Primary Secondary Higher Total Area Gender Education Youth (15-35) Non-Youth (36-64) 3 Youth Profiles: Methodology The remainder of the study deepens the analysis of labor market vulnerability among youth to (i) identify overlapping vulnerabilities and create groups experiencing similar challenges in the labor market, and (ii) describe the socio-demographic characteristics, or “profile”, these groups. While the previous section presented one labor market aspect at a time, designing and targeting employment support interventions also requires knowledge about the combined characteristics of people experiencing labor market vulnerability. The analysis employs an innovative statistical approach, Latent Class Analysis (LCA), previously applied in upper middle-income contexts such as Turkey, Croatia, Poland, Greece, and Romania.7 To our knowledge, the present study is the first of its kind in the Africa region, therefore opening the possibility of analogous studies in other low or lower middle-income countries. 3.1 Reference Population The reference population – the focus of the current analysis – comprises youth, ages 15 to 35 who are out of school. It is a subset of the Cameroonian population of working age, i.e. 15 through 64 years old, and includes both: (i) youth who are inactive, or are “detached” from the labor market, and (ii) youth who are unemployed or experiencing a range of vulnerabilities and, therefore, have a “weak-attachment” to the labor market. As shown in Table 1, the reference population includes 5,399,576 individuals, or 71 percent of youth and 25 percent of the entire population. Reflecting the substantial labor market differences in rural and urban areas documented in Section 2, separate analyses are conducted by place of residence. The reference population is split roughly equally between rural and urban areas (2,921,778 vs. 2,477,798 individuals) despite an overall majority of the population living in rural areas. This is because the urban population is 7 Report example and sources. In this context, Portraits of Labor Market Exclusion-2—a joint study between the European Commission (EC), the World Bank, and the Organization for Economic Cooperation and Development (OECD)—aims to inform employment support, activation, and social inclusion policy making, through an improved understanding of labor-market barriers. Covering 12 countries, the study builds on the previous joint EC and World Bank study to map the diversity of profiles of individuals who are out of work in six countries (Sundaram, et al., 2014)and other analyses that characterize individuals with labor market difficulties (World Bank and OECD, 2016; Ferré, Immervoll, & Sinnott, 2013; Immervoll, 2013). 10 substantially younger than the rural population (youth make up 41 percent vs. 30 percent), due to the relatively higher rates of migration among young people seeking education and employment opportunities in towns and cities. Table 1. Reference population, national and by region (population in thousand, share in %) Samples National Urban Rural Category Pop. Share Pop. Share Pop. Share Total 21,770,519 8,806,211 12,964,308 Working age (15-64) 11,411,622 5,250,420 6,161,202 Youth (15-35) 7,580,589 100.00 3,630,549 100.00 3,950,041 100.00 In school 2,181,014 28.77 1,152,751 31.75 1,028,263 26.03 Reference Pop. 5,399,576 71.23 2,477,798 68.25 2,921,778 73.97 Employed 4,078,332 53.80 1,733,598 47.75 2,344,734 59.36 Unemployed 224,119 2.96 185,039 5.10 39,080 0.99 Inactive 1,097,125 14.47 559,161 15.40 537,964 13.62 Source: Authors' calculation on ECAM4 (2014) 3.2 Labor Market Vulnerability Indicators The descriptive analysis identified seven main labor market vulnerability indicators among out-of-school youth in Cameroon. As discussed above, these indicators indicate either detachment or weak-attachment to the labor market. The weak-attachment indicators can overlap (e.g., those working informally will typically have a higher risk of being unpaid). To ensure consistency across indicators, or “symptoms”, and simplify analysis, all indicators used in the analysis are binary. Three indicators are used for labor market detachment: 1. Involuntary inactivity (INI): individuals not working or not seeking work because there is no employment, do not think he/she can obtain employment without qualification, do not know how to seek employment, face refusal from the spouse, or other involuntary reasons; 2. Voluntary inactivity (VOI): individuals not working or not seeking work because they’re waiting for a reply to a job application, do not need/want to work, do not consider themselves of working age, or other voluntary reasons; 3. Unemployment (UNE): individuals not working but seeking employment opportunities. Four indicators are used for weak-attachment to the labor market: 4. Visible underemployment (VIS): individuals working less than 40 hours a week, also referred to as “time-related underemployment”; 5. Temporary employment (TEM): individuals working less than a third of the calendar year, typically employed by the day or to work on specific, time-bound tasks; 6. Unpaid employment (UNP): workers not remunerated for their economic activities; 7. Informal sector employment (INF): individuals working for enterprises which are not formally registered or without a contract. As shown in Table 2, weak-attachment symptoms are more common than detachment symptoms. Informal employment is by far the most widespread symptom, affecting 63 percent of 11 the national reference population (48 percent urban vs. 75 percent rural)8. Unpaid employment is also frequent, especially in rural areas where 24 percent of youth are unpaid. Among detachment symptoms, voluntary inactivity is the most common and especially so in urban areas (14 percent vs. 10 percent). Only 9 percent of the reference population have “good jobs” and don’t experience any symptoms of labor market vulnerability. Table 2. Statistics by symptoms (% of reference population) Samples National Urban Rural Category Pop. Share Pop. Share Pop. Share Reference Pop. 5,399,576 2,477,798 2,921,778 None 461,400 8.55 372,377 15.03 89,022 3.05 INI 464,805 8.61 205,250 8.28 259,556 8.88 VOI 632,320 11.71 353,911 14.28 278,408 9.53 UNE 224,119 4.15 185,039 7.47 39,080 1.34 VIS 554,847 10.28 248,964 10.05 305,882 10.47 UNP 843,467 15.62 137,989 5.57 705,478 24.15 INF 3,396,699 62.91 1,193,925 48.18 2,202,774 75.39 TEM 551,627 10.22 322,420 13.01 229,206 7.84 Source: Authors’ calculation on ECAM4 (2014) 3.3 Methodology Latent Class Analysis (LCA), a statistical clustering method, is used to identify subgroups of youth with similar patterns of labor market vulnerability. The methodology enables the characterization of a categorical latent, or unobserved, variable (in this case, labor market vulnerability) by analyzing the relationship between several observed variables (the indicators, or “symptoms”, defined above). It allows the statistical segmentation of the reference population into groups based on the likelihood that individuals experience similar symptoms. In contrast to traditional regression analysis, which identifies the effect of one indicator while assuming all the other indicators stay constant, LCA exploits the interrelations of the various symptoms to generate accurate, real-life vulnerability profiles. BOX 1. A Short Introduction to Latent Class Analysis (LCA) 9 Let us consider 3 broad binary labor market vulnerability symptoms: 1) underemployment, 2) informality, 3) lack of social security. Each indicator takes value 1 when the worker faces the vulnerability and 0 otherwise. With a sample of N individuals, the response pattern depends on the values taken by the four items (Table B1). Given the presence of 3 indicators and 2 possible values for each indicator, the number of possible response patterns is 2x2x2 = 8. Table B1. Response pattern of labor-market vulnerability symptoms Individual Under- Lack of Social Response Informality ID employment Security Pattern 1 1 1 0 101 8 A social security variable was included in the initially tested methodology given its relevance to the study of labor-market vulnerabilities, yet it was highly correlated to the informality variable and it was therefore discarded, see the section pair-wise correlation coefficient in Annex 2. 9 Adapted from Fernandez et al. (2016). 12 2 1 1 0 100 3 0 1 1 011 4 0 0 1 001 The Latent Class Analysis (LCA) is used for searching the most frequent and similar response patterns among the data and group them in homogeneous clusters (or classes). For instance, in the above example, individuals 1 and 2 suffer from underemployment and informality. Therefore, a frequent response pattern could be y = (11.), where the dot represents lack of social security and leaves open the possibility for further heterogeneity among the entire target population. For instance, if the LCA algorithm finds the most frequent response patterns among the sample to be y1 = (1,1,0) and y2 = (0,1,0), it would identify the existence of two latent classes , one for each of the most frequent response patterns. Then, workers with different response patterns would be allocated to either class 1 or class 2, but with different probabilities, depending on the overlap between the responses. The LCA has multiple advantages relative to other clustering techniques, including: (i) it does not require pre-specifying the number of clusters, (ii) it provides probabilities of membership to groups, which reduces the potential classification error, and (iii) it deals more easily with common survey issues, such as missing data or complex measurements (Fernandez et al., 2016). The selection and parametrization of the model begins with the implementation of a baseline model which only requires the list of indicators, or “symptoms”. The model is then tested by incrementing the number of classes and generating a series of goodness-of-fit and misspecification statistics to enable researchers to select the optimal number of classes, or groups. Once the number of classes is selected, the final model is refined through the inclusion of modelling restrictions and covariates. As the descriptive analysis in the previous section, the LCA relies on the most recent nationally-representative household survey. The Fourth Cameroon Household Survey (Quatrième Enquête Camerounaise Auprès des Ménages, ECAM4) was conducted by the National Insitute of Statistics in 2014. The survey is part of government’s effort to regularly measure living conditions indicato rs and to monitor progress towards the national development and growth strategy. ECAM4 is the fourth of its kind, following previous waves in 1996, 2001 and 2007. It was conducted on a sample of 12,897 households across 1,024 clusters nationwide. 4 Youth Profiles: Results 4.1 Urban Results In the urban sample, the reference population is segmented into six distinct classes of varying sizes by the LCA. The model finds three labor market detachment classes and three weak- attachment classes, accounting for 30 and 70 percent of the reference population, respectively. The detachment classes have no overlapping symptoms as their attributes are exclusive: an individual is either involuntarily inactive, voluntarily inactive, or unemployed. The weak-attachment classes, on the other hand, may share the same symptoms but with differenct incidences or probabilities, depending on the combination of labor market vulnerabilities experienced by different groups of employed youth. Table 3 below shows the size of each group (or class) along with the associated likelihood of experiencing each symptom. 13 Table 3. LCA Estimates – Urban Groups Detachment Classes Weak-attachment Classes Class-membership 1 2 3 4 5 6 Probabilities 13.97 8.23 7.42 35.09 19.44 15.85 Population Size 346,187 203,869 183,867 869,566 481,695 392,613 Symptoms Probabilities (in %) INI 0.00 100.00 0.00 0.00 0.00 0.00 VOI 100.00 0.00 0.00 0.00 0.00 0.00 UNE 0.00 0.00 100.00 0.00 0.00 0.00 VIS 0.00 0.00 0.00 6.66 24.73 15.76 UNP 0.00 0.00 0.00 1.29 25.63 0.25 INF 0.00 0.00 0.00 82.50 85.95 23.28 TEM 0.00 0.00 0.00 8.43 34.24 15.61 Source: Authors' calculation on ECAM4 (2014) The largest of the six identified groups, comprising of a third of the urban reference population (35 percent), includes informal workers unlikely to experience other symptoms. Individuals in Group 4 have an 83 percent probability of being informal sector workers, 8 percent to be in temporary employment, and 7 percent to work part-time. The second largest group (Group 5), accounting for 19 percent of urban youth, appears to be the most vulnerable: not only are individuals in this group even more likely that those in Group 4 to be informal sector workers, but they’re also substantially more likely to experience a range of other vulnerabilities, including being unpaid, visibly underemployed, and in temporary employment. Finally, the third largest and last weak-attachment group, accounting for 16 percent of urban youth, includes those experiencing no or little vulnerability. The three detachment groups include the remaining third of out-of-school urban youth, most of whom are outside the labor force. Unlike with the weak-attachment classes which reflected a combination of symptoms, the detachment classes are characterized by single, mutually exclusive, symptoms. Most of the urban youth who are out of work are involuntarily inactive (14 percent, Group 1), followed by those who are voluntarily inactive (8 percent, Group 2), and the unemployed (7 percent, Group 3). These figures emphasize the importance of going beyond traditional measures of youth marginalization to account for those who are either discouraged or prevented from looking for employment and who, combined, represent 1 in 5 out-of-school youth in urban areas. The boxes below provide short descriptions of each of the six groups, including associated demographic and socio-economic characteristics. The groups have been named according to their most salient characteristics. However, this naming is subjective and simplistic in nature and requires a closer look at the mix of labor market vulnerabilities faced by each group as well as the fuller list of individual and household characteristics that are also pertinent for the design and tailoring of employment support policies and interventions. 14 The Urban Detachment and Weak-attachment Classes Group 1: Well-Educated, Unmarried Women Who Are Voluntarily Inactive Class Size Demography 69% female, mainly (40%) between 25-29 years old, largely 13.97% of the reference single (60 percent) population Human Capital 75% with secondary/higher education, 50% with 346,187 individuals vocational training or apprenticeships, and 1/2 without any working experience Labor-Market Attributes Geography 37% in Yaoundé and 31% in Douala Voluntary Inactive 100% Wealth and Income 8% in poverty, 67% without savings Group 2: Poor, Low-Skilled, Married Women Who Are Involuntarily Inactive (over-represented among Muslim households in the Far North region) Class Size Demography 86% female, 64% between 20-29 years old, 63% married and 8.23% of the reference largely Muslim (41%) population Human Capital 55% with no or with primary education, 31% with vocational 203,869 individuals training or apprenticeships, and 52% without any working experience Labor-Market Attributes Geography 24% in Douala, 20% in Yaoundé and 17% in the Far-North Involuntary Inactive 100% Wealth and Income 10% in poverty, 74% without savings 15 Group 3: Well-Educated, Single Women Who Are Unemployed Class Size Demography 64% female, 70% older than 25 years, 66% single 7.42% of the reference Human Capital 84% with secondary or higher education, 60% with population vocational training or apprenticeships, 27% without any 183,867 individuals experience and 41% with more than 10 years of experience Geography 43% in Yaoundé and 36% in Douala Labor-Market Attributes Wealth and Income 5% in poverty, 65% without savings, d Unemployment 100% Group 4: Educated, Older Men Who Work in the Informal Sector Demography 66% male, 30% in small households of 1-2 members, Class Size 45% between 30-35 years old 35.09% of the reference Human Capital 92% with primary and secondary education, 60% with population vocational training or apprenticeships, 58% with more 869,566 individuals than 10 years of experience Geography 34% in Douala and 29% in Yaoundé Labor-Market Attributes Employment 56% self-employed and 20% qualified employee, 83% in Informality 82.50% private non-ag enterprises, 42% worked since younger than 15 years Wealth and Income 7% in poverty, 63% without saving. 16 Group 5: Young, Single Women Who are Employed but Experiencing Most Vulnerability Demography 64% female, 31% in households larger than 7, 52% younger Class Size than 25, 61% single 19.44% of the reference Human Capital 66% with secondary and higher education, 61% with population vocational training or apprenticeships 481,695 individuals Geography 32% in Yaoundé and 31% in Douala Employment 44% own-account workers, 33% contributing family Labor-Market Attributes workers, 41% in services, in private enterprises, 44% worked Informality 85.95% since younger than 15 Temporary Employment Wealth and Income 9% in poverty, 62% without savings 34.24% Unpaid Employment 25.62% Visible Underemployment 24.73% Group 6: Older, Well-Educated Men With Low Vulnerability Demography 57% male, 31% in small (1-2 members) households, 60% is Class Size 30-35 years old and 34% is 25-29 years old, 49% married 15.85% of the reference Human Capital 55% with higher education, 45% with secondary education, population 79% with vocational training or apprenticeships, 66% with 392,613 individuals more than 10 years of experience Geography 36% in Yaounde and 39% in Douala Labor-Market Attributes Employment 54% qualified employee and 20% managers, 67% in services, Informality 23.28% 18% in the administration, 32% in private formal enterprises. Visible Underemployment Wealth and Income 2% in poverty, 49% without savings 15.76% Temporary Employment 15.61% 17 4.2 Rural Results In the rural sample, the reference population is segmented into seven distinct classes by the LCA. The model identifies three detachment classes and four weak-attachment classes, accounting for 20 and 80 percent of the out-of-school youth, respectively. As with the urban sample, the detachment classes are standalone in terms of vulnerability symptoms, while the weak-attachment classes combine different symptoms at varying probabilities. Table 4 below shows the size of each group (or class) along with the associated likelihood of experiencing each symptom. Table 4. LCA Estimates – Rural Groups Detachment Classes Weak-attachment Classes Class-membership 1 2 3 4 5 6 7 Probabilities 9.43 8.86 1.34 44.57 25.58 5.66 4.57 Population Size 275,569 258,765 39,080 1,302,201 747,418 165,236 133,510 Symptoms Probabilities (in %) INI 0.00 100.00 0.00 0.00 0.00 0.00 0.00 VOI 100.00 0.00 0.00 0.00 0.00 0.00 0.00 UNE 0.00 0.00 100.00 0.00 0.00 0.00 0.00 VIS 0.00 0.00 0.00 14.61 5.75 29.24 13.41 UNP 0.00 0.00 0.00 2.55 61.07 42.49 0.79 INF 0.00 0.00 0.00 100.00 100.00 91.91 13.55 TEM 0.00 0.00 0.00 7.62 3.01 34.37 15.85 Source: author's calculation on ECAM4 (2014) The largest of the seven identified groups, accounting for almost half (45 percent) of all out- of-school rural youth, includes informal sector workers who are somewhat likely to be visibly underemployed or in temporary employment. The second largest group (26 percent, Group 5) includes informal sector workers who are, instead, likely to be unpaid. Far smaller groups experience either multifold labor market vulnerability (6 percent, Group 6) or no to little vulnerability (5 percent, Group 7). A smaller share of the rural sample is detached from the labor market, compared to the urban sample, with virtually all those who are out of work being inactive. Only 1 percent (Group 3) of out-of-school youth in rural areas are unemployed. If out of the labor force, there’s an equal chance that rural youth are either voluntarily inactive (9 percent, Group 1) or involuntarily inactive (9 percent, Group 2). More detailed descriptions of each group are presented in the boxes below. 18 The Rural Detachment and Weak-attachment Classes Group 1: Unmarried Women with Some Education Who Are Voluntarily Inactive Class Size Demography 64% female, 88% between 15-29 years old, 62% single 9.43 of the reference Human Capital 61% without or with primary education, 23% with vocational population training or apprenticeships, 69% without any working 275,569 individuals experience Geography 24% in Far-North, 17% in North-West, 14% in Center Labor-Market Attributes Wealth and Income 46% in poverty, 77% without savings, household rearing Voluntary Inactive 100% poultry/cattle (36%) and cultivating land (59%) Group 2: Uneducated, Married Women Who Are Involuntarily Inactive Class Size Demography 87% female, equally distributed ages, 65% married, 51% Muslim 8.86 of the reference Human Capital 43% without education, 14% with vocational training or population apprenticeships, 71% without any working experience 258,765 individuals Geography 20% in Adamaoua, 17% in Far-North, 13% in Center Wealth and Income 49% in poverty, 85% without savings, household rearing Labor-Market Attributes poultry/cattle (36%) and cultivating land (59%) Involuntary Inactive 100% 19 Group 3: Individuals with Some Education and Some Experience Who Are Unemployed Demography 55% female, 78% between 20-29 years old, 54% married, Class Size and largely Protestant (36%) 1.34 of the reference Human Capital 50% with secondary and 39% with primary education, 49% with population vocational training or apprenticeships, 36% without any working 39,080 individuals experience Geography 22% in Center, 21% in North-West, 15% in Adamaoua Labor-Market Attributes Wealth and Income 30% in poverty, 75% without savings, household rearing Unemployment 100% poultry/cattle (27%) and cultivating land (59%) Group 4: Married Individuals Engaged in Subsistance Agriculture Class Size Demography 47% female, 80% between 25-35 years old, 72% married 44.57 of the reference Human Capital 26% without education, 26% with vocational training or population apprenticeships, 80% with more than 10 years of experience 1,302,201 individuals Geography 21% in Far-North, 20% in North, 12 in North-West Employment 89% self-employed, 62% in agriculture and informal, Labor-Market Attributes 56% worked since younger than 15 Informality 100% Wealth and Income 49% in poverty, 82% without savings, household rearing Visible Underemployment poultry/cattle (44%) and cultivating land (74%) 14.61% 20 Group 5: Poor, Married Women with Limited Education Who Are Unpaid Family Workers (mainly in the North and Far North regions) Demography 73% women, 45% in households with more than 7 members, ages equally distributed, 60% married, large share of Muslim (34%) Class Size Human Capital 50% without education, 39% with primary education, 14% with 25.58 of the reference vocational training or apprenticeships, with more than 2 years of population experience 747,418 individuals Geography 44% in Far-North, 22% in North Employment 66% contributing family workers, 32% self-employed, 85% in Labor-Market Attributes agriculture and informal, 68% worked since younger than 15 Informality 100% Wealth and Income 68% in poverty, 87% without savings, household rearing Unpaid Employment 61% poultry/cattle (57%) and cultivating land (85%) Group 6: Unmarried Women with Some Education Experiencing Multifold Vulnerability (mainly in the Far North and Anglophone regions) Class Size Demography 60% women, 43% in households with more than 7 members, 5.66 of the reference 52% aged 15-19 and 29% aged 20-24, 61% single population Human Capital 41% with primary and 49% with secondary education, 165,236 individuals 34% with vocational training or apprenticeships, with more than 2 years of experience Labor-Market Attributes Geography 22% in Far-North, 19% in North-West, 14% in South-West Informality 91.90% Employment 55% contributing family workers, 35% self-employed, 54% in Unpaid Employment primary sector, 37% in non-agricultural-informal-private sector, 42.50% 61% worked since younger than 15 Temporary Employment Wealth and Income 45% in poverty, 79% without savings, household rearing 34.80% poultry/cattle (44%) and cultivating land (77%) Visible Underemployment 29.20% 21 Group 7: Well-Educated, Married Men with Low Vulnerability Demography 32% female, 79% in households with less than 7 members, 34% aged Class Size 25-29 and 58% aged 30-35, 62% married, largely Protestant (37%) 4.57 of the reference Human Capital 33% with higher education, 45% with secondary education, 76% population with vocational training or apprenticeships, with more than 2 years 133,510 individuals of experience Geography 18% in North-West, 14% in South-West Labor-Market Attributes Employment 82% qualified employee, 24% managers, 18% laborers, 70% in Temporary Employment services, 47% in public administration, 16 % in private and formal 15.80% Wealth and Income 19% in poverty, 61% without savings, household rearing Informality 13.50% poultry/cattle (30%) and cultivating land (47%) Visible Underemployment 13.40% 22 4.3 Priority Classes Among the six urban and seven rural groups identified in the target population, four groups are singled out for prioritization for employment support interventions. They consist of two groups of individuals with weak attachment to the labor market – Group 4, Urban (Educated, Older Men Who Work in the Informal Sector) and Group 4, Rural (Married Individuals Engaged in Subsistence Agriculture) – together with two groups of individuals detached from the labor market – Group 1, Urban (Well-Educated, Unmarried Women Who Are Voluntarily Inactive) and Group 3, Urban (Well- Educated, Single Women Who Are Unemployed). These groups were selected based on their size as well as potential for improving human capital utilization. The prioritized weak-attachment groups are the largest in their sub-samples (35 and 45 percent, respectively) and represent the archetypal underemployed, informal sector worker in either urban or rural areas. The detached groups are smaller in size, yet indicative of the growing issue of unemployment and inactivity among well- educated youth in urban areas. Given broad similarities between the groups, we analyze Groups 4 urban and rural (informal sector workers) together, and Group 1 and Group 3 urban (out-of-work women) together. The largest groups in the urban and rural sub-samples are made of relatively older youth who are underemployed in the informal sector. Taken together, Group 4, Urban and Group 4, Rural, comprise 40 percent of the youth workforce, or 2,171,767 individuals. The groups experience a similar pattern of labor market vulnerability: the majority are employed in the informal sector (83 percent for the urban group and 100 percent for the rural group), with relatively low chances of being time-based underemployed (7 percent, urban; 15 percent, rural) or temporarily employed (9 percent, urban; 8 percent, rural). Both groups also consist of older youth: more than 80 percent are aged over 25 (82 percent, urban; 80 percent, rural) and almost half are aged over 30 (45 percent, urban; 47 percent, rural). Despite these similarities in terms of labor market vulnerability, there are important differences with respect to socio-demographic characteristics and types of employment between the two groups. While both groups are majority male, the share of women is almost equal to that of men in the rural group (66 percent male, urban; 53 percent, rural). Individuals in the rural group are also significantly more likely to be married (52 percent, urban; 72 percent, rural) and to live in larger households (70 percent households with 3 or more members, urban; 87 percent, rural). The rural group is much more likely to be poor (7 percent, urban; 49 percent, rural) and to lack any education (8 percent, urban; 26 percent, rural). Finally, the nature of the work is markedly different between the two groups and their locations: more individuals in the rural group are own account workers (56 percent, urban; 89 percent, rural; 20 percent of the urban group work as qualified employees) and concentrated in a single sector (the urban group is split evenly across commerce, industry and services, while 65 percent of the rural group works in the primary sector). The other two priority groups highlight the growing-concern issue of labor market detachment among well-educated youth in urban areas. Groups 1 and 3 include 21 percent of the urban sub-sample, or 530,054 individuals. Although the two groups are distinct in terms of labor market status – Group 1 includes those voluntarily inactive and Group 3 the unemployed – they are remarkably similar in terms of socio-economic characteristics. Both groups are two-thirds women (69 percent, inactive; 64 percent, unemployed) and unmarried (60 percent, inactive; 66 percent, unemployed). About half of the individuals in either group have secondary or higher education (75 23 percent, inactive; 84 percent, unemployed) but no working experience (49 percent, inactive; 52 percent, unemployed). Poverty rates are low in both groups (8 percent, inactive; 5 percent, unemployed). In addition, four groups are identified as priorities for labor market interventions targeting women. They include two weak-attachment groups comprising large shares of women employed as unpaid family workers – Group 5, Urban (Young, Single Women Who are Employed but Experiencing Most Vulnerability) and Group 5, Rural (Poor, Married Women with Limited Education Who Are Unpaid Family Workers) – and two detachment majority-women groups of involuntarily inactive individuals – Group 2, Urban (Poor, Low-Skilled, Married Women Who Are Involuntarily Inactive) and Group 2, Rural (Uneducated, Married Women Who Are Involuntarily Inactive). The two majority-women weak-attachment groups are equally likely to work in the informal sector as the groups discussed above, yet they face additional vulnerability. Taken together, Group 5, Urban and Group 5, Rural, comprise 45 percent of the youth workforce, or 1,229,113 individuals. They differ between Group 4, Urban and Group 4, Rural along two key socio- demographic lines: they’re majority women (64 percent vs. 34 percent for the urban groups, and 73 percent vs. 47 percent for the rural groups) and substantially younger (52 percent under 25 vs. 18 percent, urban; and 58 percent vs. 20 percent, rural). The consequences in terms of labor market vulnerability are significant: while having similar changes of working in the informal sector (86 percent vs. 84 percent, urban; and 100 percent, rural), the younger, majority-women groups are also often unpaid (26 percent vs. 5 percent, urban; and 61 percent vs. 0 percent, rural). This is driven by the high share of family aid workers (33 percent, urban; 66 percent, rural). Besides the two detachment groups discussed above, two majority-women groups include those who are involuntarily inactive in either rural or urban areas. The two groups of involuntarily inactive youth make up 8-9 percent of both the urban and rural sub-samples, or 944,329 individuals in total, and have the highest shares of women among all groups (86 percent, urban; 87 percent, rural). The most salient difference between those involuntarily and voluntarily inactive is marital status (63 percent vs. 39 percent, urban; 65 percent vs. 35 percent, rural). Nevertheless, two other important correlates of the type of inactivity emerge: those who are inactive involuntarily tend to have lower educational levels (45 percent attained secondary or higher education vs. 75 percent, urban; 22 percent vs. 39 percent, rural) and larger proportions of the involuntarily inactive groups are Muslim (41 percent vs. 15 percent, urban; 51 percent vs. 31 percent, rural). Finally, the three remaining groups are not prioritized for labor market interventions. Group 3, Rural (Individuals with Some Education and Some Experience Who Are Unemployed) includes a negligible share of the population (1 percent). Group 6, Urban (Older, Well-Educated Men With Low Vulnerability) and Group 7, Rural (Well-Educated, Married Men with Low Vulnerability) are also not prioritized because they include well-educated, older youth who experience little to no labor market vulnerability. 24 5 Conclusion and Recommendations Out-of-school Cameroonian youth, despite comprising two-thirds of the workforce, are disadvantaged in the labor market relatively to older workers. Youth (ages 15-35) are twice as like as non-youth to have exited the labor force (20 percent vs. 10 percent), with particularly high inactivity rates among young women (27 percent). In addition, unemployment is three times higher for youth compared to non-youth (6 percent vs. 2 percent), reaching a staggering 20 percent among urban youth with university degrees. When employed, young workers are more likely to be underemployed: 28 percent of youth are in temporary employment (vs. 15 percent among non-youth) and 22 percent work for free (vs. 6 percent of non-youth). Unpaid employment, in the form of household aid or apprentice, is especially common for youth in rural areas (31 percent) and among women (29 percent). “Youth”, however, are a heterogeneous group with distinct labor market challenges and socio-demographic characteristics. Using an advanced statistical technique called Latent Class Analysis, we identify six youth profiles in urban areas and seven profiles in rural areas. ➢ In urban areas, a third of youth (35 percent, Group 4) work in the informal sector without experiencing other vulnerability symptoms. This group includes mostly older (ages 30-35) men who have completed primary or secondary education and accumulated a few years of work experience. Women, on the other hand, typically face a range of other vulnerabilities – such as unpaid or temporary employment – in addition to informality (19 percent, Group 5) even when they’ve achieved similar education levels. They also make up the majority of all three groups detached from the labor market, tending to be involuntarily inactive (8 percent, Group 2) if they’re married – particularly among Muslim households in the north, and either unemployed (7 percent, Group 3) or discouraged (14 percent, Group 1) if they’re unmarried and with secondary education or above. Only 16 percent (Group 6) of all Cameroonian youth living in urban areas, mainly older ones with higher education, are in good quality employment. ➢ In rural areas, almost half (45 percent, Group 4) of youth are engaged in subsistence agriculture. This group includes both men and women, who tend to be married and with over 10 years of work experience, having either left school early or never attended it. Another quarter of rural youth (26 percent, Group 5) are women working as unpaid family support, mainly in the northern regions of the country. Women are, again, overrepresented among inactive youth – split evenly along marital status between voluntary (9 percent, Group 1) and involuntary (9 percent, Group 2) employment. Unemployment is virtually inexistent in rural areas (1 percent, Group 3), and only 5 percent of youth (Group 7) experience no or low labor market vulnerability. Three main recommendations for policy development stem from the analysis: ➢ The stark differences between rural and urban areas require different policy and programmatic responses. Low productivity, associated with the continued predominance of household farming, is particularly salient in rural areas. To improve productivity as well as make agriculture more attractive to youth and curb the pace of urbanization, interventions should provide young people with the resources (e.g., business skills and grants) to build 25 sustainable micro-enterprises producing, transforming, or trading agricultural products. On the other hand, urban areas are confronted with rising unemployment in addition to persistent informality and low productivity. While more youth are completing secondary and higher education, formal sector job remain out of reach – either because of a mismatch in skills or because the supply of jobs is inadequate. ➢ Employment promotion interventions should be tailored to the needs and characteristics of specific youth groups. For example, intermediation services and improved access to jobs-related information would be useful for highly-qualified youth. In contrast, entrepreneurship support would be more appropriate for the majority of youth likely to remain confined to the informal sector for the foreseeable future. Given the incidence of involuntary inactivity among married women in rural areas, complementary interventions to engage husbands and community leaders may be needed to promote social norms change and encourage women’s economic participation. ➢ A gender lens should be applied across employment promotion policies and programs. The analysis unambiguously showed that young women face higher labor market vulnerability than men across indicators. If married, women have a high likelihood of being economically inactive and, if unmarried, they tend to work either as unpaid family support or in the most precarious forms of employment. Supporting women’s productive employment through gender-responsive measures is important not just to improve their autonomy and living standards, but also to slow down demographic growth, improve children’s outcomes, and reduce the intergenerational transmission of poverty. 26 6 Bibliography Schultz, T. P. (2002). Wage gains associated with height as a form of health human capital. Yale Economic Growth Center Discussion Paper(841). Parr, N. J. (1995). Pre-Marital Fertility in Liberia. Journal of Biosocial Science, 27(1). Schmidhuber, J., Bruinsma, J., & Boedeker, G. (2009). Capital requirements for agriculture in developing countries to 2050. In FAO Expert Meeting on “How to Feed the World in 2050.”. Rome: FAO. Pratap, S., & Quintin, E. (2006). The Informal Sector in Developing Countries: Output, Assets and Employment. WIDER Working Paper Series(RP2006-130). Rubiano-Matulevich, E., & Viollaz, M. (2019). Gender Differences in Time Use: Allocating Time between the Market and the Household. Policy Research Working Paper. Liberia Institute of Statistics and Geo‐Information Services (LISGIS). (2010). Report on the Liberia Labour Force Survey 2010. Monrovia: ILO & Liberia Institute of Statistics and Geo‐ Information Services (LISGIS). Vandeninden, F., & Ovadiya, M. (2017). PORTRAITS OF LABOR MARKET EXCLUSION 2.0: Country Policy Paper (CPP) for Croatia. Washington, DC: The World Bank. Hagenaars, J., & McCutcheon, A. L. (2002). Applied Latent Class Analysis. Cambridge: Cambridge University Press. Schreiber, J. (2016). Latent Class Analysis: An example for reporting results. Research in Social and Administrative Pharmacy. NIS. (2014). Fourth Cameroon Household Survey: ECAM 4. Yaoundé: National Institute of Statistics of Cameroon. Nylund, K., Asparouhov, T., & Muthen, B. (2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: a Monte Carlo simulation study. Structural Equation Modeling, 14(4), 535-569. Vermunt, J. K., & Magidson, J. (2005). Structural equation models: Mixture models. In B. Everitt, & D. Howell, Encyclopedia of Statistics in Behavioral Science (pp. 1922-1927). Wiley. Wurpts, I. C., & Geiser, C. (2014). Is adding more indicators to a latent class analysis beneficial or detrimental? Results of a Monte-Carlo study. Frontiers in psychology, 5(920). Vermunt, J. K., & Magidson, J. (2005). Latent Gold 4.0: A general program for the analysis of categorical data. Tilburg University. World Bank. (2018). Cameroon City Competitiveness Diagnostic. Washington DC: The World Bank Group. Sosale, S., & Majgaard, K. (2016). Fostering Skills in Cameroon Inclusive Workforce: Development, Competitiveness, and Growth. Washington: The World Bank. Sundaram, R., Hoerning, N., De Andrade Falcao, N., Millan, C., & Tokman, Z. M. (2014). Portraits of Labor Market Exclusion. Washington, DC: World Bank. 27 World Bank and OECD. (2016). Joint Methodology Paper. World Bank and OECD (Organisation for Economic Co-operation and Development). Ferré, C., Immervoll, H., & Sinnott, E. (2013). Profiling of People with No or Limited Labor-market Attachment, Latvia: Who is Unemployed, Inactive of Needy? Assessing Post-Crisis Policy Options. Washington, DC: World Bank. Immervoll, H. (2013). Profiles of Vulnerable People in Turkey. Washington, DC: World Bank. Fernandez, R., Immervoll, H., Pacifico, D., & Thévenot, C. (2016). Faces of Joblessness: Characterising Employment Barriers to Inform Policy. OECD Social, Employment and Migration Working Papers(192). 28 Annex 1. Latent Class Analysis Methodology The Latent Class Analysis (LCA) segments the reference population by assigning to the units of analysis different probabilities of membership to an unobserved or latent categorical variable. The LCA exploits interrelations and relationship between the labor market attributes selected in comparison to a traditional regression analysis, in which the constant relationship between the predictor and the other selected attributes or covariates is assumed (Vandeninden & Ovadiya, 2017). The latent or un-observed variable is often viewed as the missing and antecedent characteristic causing or inducing the spurious relationships among the observable attributes (Hagenaars & McCutcheon, 2002). In other applications it is instead viewed as an “intervening” variable which completely explains the associations among the among the observable attributes. The latent variable is in this study defined as the “intervening” grouping variable which is able to explain the relationships among the chosen labor-market attributes. LCA provides a segmentation of the reference population by maximizing the between-cluster differences and minimize the within-cluster differences (Schreiber, 2016). The technique does not automatically provide the adequate number of classes to use but requires a study of the optimal number of clusters that best represent the reference population and its respective observed attributes. The selection and parametrization of the model begins with the implementation of a baseline model. The initial baseline model is tested through the use of labor-market attributes as predictors of class-membership to a multinomial latent variable. This is done by incrementing the number of classes from 2 classes to 12 classes. The goodness-of-fit statistics are then assessed across the tested models in order to identify the correct number of classes to use (Table A1, Table A2). These statistics include the G2, the log-likelihood statistics, the Akake Information Criterion (AIC), the Bayes Information Criterion (BIC) and the degrees of freedom (Nylund, Asparouhov, & Muthen, 2007). The log-likelihood reflects the logarithm of the likelihood ratio which compares the prediction of two models which differ in the number of restrictions made and it is assessed through its p-value and the chi-squared statistic, which is by convention denoted by the G² (G-squared). The BIC is the Bayes Information Criterion and the AIC is the Akake Information Criterion, which are statistics or “parsimony indeces” that penalize the number of factors included in the model. Table A1. Baseline comparison goodness-of-fit statistics across models, URBAN sample Classes N LL LL Ratio G-sq DF AIC BIC 2 6313 -14,475.83 0.00 1,780.90 11.00 28,973.66 29,047.91 3 6313 -13,945.72 0.00 720.68 23.00 27,937.45 28,092.70 4 6313 -13,847.30 0.00 523.83 31.00 27,756.59 27,965.85 5 6313 -13,614.30 1.00 57.83 31.00 27,290.59 27,499.85 6 6313 -13,586.88 1.00 2.99 43.00 27,259.75 27,550.02 7 6313 -13,587.34 1.00 3.91 49.00 27,272.67 27,603.44 8 6313 -13,588.35 1.00 5.94 45.00 27,266.70 27,570.47 9 6313 -13,586.47 1.00 2.18 64.00 27,300.94 27,732.97 10 6313 -13,586.55 1.00 2.34 69.00 27,311.10 27,776.88 11 6313 -13,585.52 1.00 0.27 73.00 27,317.03 27,809.81 12 6313 -13,585.74 1.00 0.72 79.00 27,329.48 27,862.76 Source: Authors’ calculation on ECAM4 (2014) 29 Table A2. Baseline comparison goodness-of-fit statistics across models, RURAL sample Classes N LL LL Ratio G-sq DF AIC BIC 2 5259 -10,802.93 0.00 1,204.08 13.00 21,631.87 21,717.25 3 5259 -10,719.81 0.00 1,037.84 20.00 21,479.63 21,610.98 4 5259 -10,702.58 0.00 1,003.38 31.00 21,467.17 21,670.77 5 5259 -10,701.43 0.00 1,001.07 31.00 21,464.86 21,668.46 6 5259 -10,285.98 0.00 170.16 32.00 20,635.95 20,846.12 7 5259 -10,203.41 1.00 5.03 53.00 20,512.82 20,860.91 8 5259 -10,589.69 0.00 777.59 58.00 21,295.38 21,676.31 9 5259 -10,592.39 0.00 783.00 63.00 21,310.78 21,724.55 10 5259 -10,201.35 1.00 0.91 49.00 20,500.70 20,822.52 11 5259 -10,280.76 0.00 159.74 56.00 20,673.53 21,041.32 12 5259 -10,201.20 1.00 0.62 73.00 20,548.41 21,027.85 Source: Authors' calculation on ECAM4 (2014) The estimates are then assessed for identifying the restrictions to be made. More specifically, the weak-attachment attributes are not allowed in detachment classes. Once the identifying restrictions are implemented, the goodness-of-fit statistics are once more compared (Table A3, Table A4) to validate the efficiency of the method. These restrictions are implemented to increase parsimony and stability to the specification (Vermunt & Magidson, 2005) and are in every case validated by their ability to better represent the reference population and the labor-market attributes. Once the efficiency of the restrictions is validated, the active covariates are entered into the specification and the efficiency of the specification is once more assessed (Table A3, Table A4). The introduction of covariates which have a significant effect on class-membership probabilities can further reduce miss-classification errors and improve the predictive power of the model (Wurpts & Geiser, 2014). A series of standard variables used for the analysis of the labor-market are here used and are the sex, age-group and education of youth. The introduction of the active covariates to the constrained solution has reduced the goodness-of-fit statistics across the samples studied, hence indicating a more fitting and efficient solution. A classification statistic is also implemented to provide an additional validation of the predictive and classifying effectiveness of the model. The modal allocation assigns each unit to its respective class with the highest posterior probability. The misclassification error can be measured by cross-classifying the modal assignment with the probability-based assignment (Vermunt & Magidson, 2005). The misclassification error stands as a measure of predictive effectiveness for the specification used. A statistic higher than 30% highlights that the specification is significantly not able to discriminate between classes (Fernandez et. al, 2016). Table A3. Goodness-of-fit statistics across models, URBAN sample Model N LL DF AIC BIC Unconstrained 6313 -13586.88 43 27259.75 27550.02 Constrained 6313 -13586.83 16 27205.66 27313.67 Complete 6269 -12446.63 52 24997.26 25347.92 Source: Authors' calculation, ECAM4 (2014) 30 Table A4. Goodness-of-fit statistics, RURAL sample Model N LL DF AIC BIC Unconstrained 5259 -10203.41 53 20512.82 20860.91 Constrained 5259 -10205.31 21 20452.62 20590.54 Complete 5237 -9257.61 62 18639.22 19046.15 Source: Authors’ calculation, ECAM4 (2014) Once the complete model is finalized and estimated, the resulting probabilities of class- membership are used for predicting the classes of the sample. The designation is in turn used for the characterization, which is undertaken by computing two-way statistics summarizing the relationship between the group-membership of the sample and the respective socio-economic characteristics. The latter are arbitrarily chosen and based on the knowledge of covariates deemed as suitable for understanding intra-cluster dynamics and beneficial for policy-making purposes as in the case of targeting decisions, for example. The individual features adopted for the characterization are broadly represented by groups of indicators which can respectively describe the demographic, human capital, geographic, employment and income profile of the classes. 31 Annex 2. Additional Results Table A5. Poverty rate by region, in % Region Poverty Rates Douala 1.38 Yaounde 1.78 Adamaoua 6.81 Centre 5.18 Est 3.11 Extreme-North 35.81 Littoral 1.41 North 19.99 North-West 13.29 West 5.06 South 3.01 South-West 3.17 source: author's calculation, ECAM4 (2014) Table A6. Male and female earnings and gender pay gap Ref. Period Male Female M-F Gap Hourly 518.366 283.9245 45.23% Monthly 79971.32 35750.3 55.30% Yearly 889675.2 384739 56.76% source: author's calculation, ECAM4 (2014) Table A7. Labor force participation by youth and non-youth, in % Youth Youth LFP Non-Youth (with students) (w/o students) Active 10.04 34.57 20.32 Inactive 89.96 65.43 79.68 source: author's calculation, ECAM4 (2014) Table A8. Earnings by institutional sector by reference period, working age (15-64), CFAF % of % of % of Institutional Sector Hourly Monthly Yearly Admin. Admin. Admin. Public 1,344 1.00 151,274 1.00 1,740,893 1.00 Private, Formal 627 2.14 125,939 1.20 1,358,129 1.28 Informal, Not Agriculture 354 3.80 57,968 2.61 637,596 2.73 Informal, Agriculture 305 4.41 37,630 4.02 413,057 4.21 source: author's calculation, ECAM4 (2014) 32 Table A9. Hourly earnings by institutional sector and economic sector of employment (15-34) Male Female Mean SD Sample Mean SD Sample Inst. Category Administration 1,289 3,778 122 1,490 2,862 89 Publicly funded/Org. int. 605 562 56 329 404 20 Private Formal 406 427 174 443 361 80 Private Non-Agr. - Informal 387 613 1,188 247 357 860 Association 250 321 16 281 242 7 Private Agr. - Informal 188 279 517 134 396 530 Sector Services 491 1,564 681 468 1,167 523 Industry 444 671 481 212 243 278 Commerce 365 534 361 272 422 300 Primary 228 452 583 135 427 545 Figure A1. Hourly and annual earnings of youth (15-34) by subgroup and compared to working-age (15-64) earnings 700 1,200,000 600 1,000,000 817,294 500 696,426 800,000 580 Annual earnings Hourly earnings 661,105 624,001 613,890 400 463 488,837 429 600,000 402 300 372 343,229 321,206 370,641 324 400,000 200 236 171,976 244 264 100 200,000 127 0 0 Urban Rural Male Female No Yes Paid Employer Self-Empl. Total Area Gender Poverty Employment Status Earnings (15-64) Youth earnings (15-34) 33 Pair-Wise Correlation Coefficients Table A9. Correlation matrix between indicators – urban sample Indicators INI VOI UNE VIS UNP INF TEM SC INI 1.00 -0.12 -0.09 -0.10 -0.07 -0.29 -0.12 0.36 VOI -0.12 1.00 -0.12 -0.14 -0.10 -0.39 -0.16 0.48 UNE -0.09 -0.12 1.00 -0.09 -0.07 -0.27 -0.11 0.34 VIS -0.10 -0.14 -0.09 1.00 0.05 0.17 0.14 -0.19 UNP -0.07 -0.10 -0.07 0.05 1.00 0.17 0.12 -0.20 INF -0.29 -0.39 -0.27 0.17 0.17 1.00 0.16 -0.78 TEM -0.12 -0.16 -0.11 0.14 0.12 0.16 1.00 -0.25 SC 0.36 0.48 0.34 -0.19 -0.20 -0.78 -0.25 1.00 Source: author's calculation on ECAM4 (2014). Table A11. Correlation matrix between indicators – rural sample Indicators INI VOI UNE VIS UNP INF TEM SC INI 1.00 -0.10 -0.04 -0.11 -0.18 -0.55 -0.09 0.58 VOI -0.10 1.00 -0.04 -0.11 -0.18 -0.57 -0.09 0.61 UNE -0.04 -0.04 1.00 -0.04 -0.07 -0.20 -0.03 0.22 VIS -0.11 -0.11 -0.04 1.00 -0.01 0.14 0.11 -0.14 UNP -0.18 -0.18 -0.07 -0.01 1.00 0.30 0.04 -0.30 INF -0.55 -0.57 -0.20 0.14 0.30 1.00 0.08 -0.93 TEM -0.09 -0.09 -0.03 0.11 0.04 0.08 1.00 -0.14 SC 0.58 0.61 0.22 -0.14 -0.30 -0.93 -0.14 1.00 Source: author's calculation on ECAM4 (2014). Urban Characterization Table A12. Characterisation of classes by demographic variables Detachment Weak-attachment Classes Classes Variables 1 2 3 4 5 6 Sex Female 0.69 0.86 0.64 0.34 0.64 0.43 Male 0.31 0.14 0.36 0.66 0.36 0.57 Household size 1-2 0.14 0.10 0.19 0.30 0.19 0.31 3-6 0.48 0.53 0.53 0.48 0.51 0.51 7-10 0.25 0.26 0.22 0.17 0.23 0.16 >10 0.13 0.11 0.06 0.05 0.08 0.02 Age Group 15/19 0.14 0.15 0.07 0.00 0.19 0.00 20/24 0.27 0.32 0.24 0.18 0.33 0.06 25/29 0.40 0.32 0.43 0.37 0.30 0.34 30/35 0.18 0.22 0.27 0.45 0.18 0.60 34 Marital status Married 0.39 0.63 0.33 0.52 0.38 0.49 Single 0.60 0.35 0.66 0.46 0.61 0.51 Widowed/Divorced 0.01 0.02 0.01 0.03 0.02 0.01 Residence 12 months ago Different Location 0.15 0.15 0.12 0.16 0.16 0.18 Same Location 0.85 0.85 0.88 0.84 0.84 0.82 Religion Animism 0.01 0.00 0.01 0.00 0.01 0.00 Catholic 0.52 0.32 0.54 0.47 0.49 0.55 Muslim 0.15 0.41 0.09 0.21 0.16 0.06 None 0.01 0.02 0.01 0.03 0.03 0.02 Other Christian 0.07 0.05 0.06 0.06 0.08 0.08 Other Religion 0.04 0.02 0.04 0.03 0.04 0.04 Protestant 0.21 0.17 0.24 0.19 0.20 0.25 Source: author's calculation on ECAM4 (2014). Table A13. Characterisation of classes by demographic human capital variables Detachment Classes Weak-attachment Classes Variables 1 2 3 4 5 6 Education Higher 0.20 0.05 0.33 0.00 0.16 0.55 No Education 0.07 0.24 0.03 0.08 0.04 0.00 Primary 0.17 0.31 0.13 0.37 0.30 0.00 Secondary 0.55 0.40 0.51 0.55 0.50 0.45 Technical or vocational training No 0.50 0.69 0.40 0.40 0.39 0.21 Yes 0.50 0.31 0.60 0.60 0.61 0.79 Working experience 0 0.49 0.52 0.27 0.01 0.08 0.03 1-2 0.04 0.04 0.04 0.04 0.11 0.04 10-15 0.17 0.13 0.24 0.33 0.25 0.28 16-20 0.09 0.09 0.12 0.22 0.12 0.24 3-5 0.09 0.08 0.15 0.10 0.19 0.10 6-9 0.09 0.09 0.14 0.16 0.18 0.17 >20 0.02 0.05 0.05 0.13 0.06 0.14 Source: author's calculation on ECAM4 (2014). 35 Table A14. Characterisation of classes by geographic variable Weak-attachment Detachment Classes Classes Variables 1 2 3 4 5 6 Region Adamaoua 0.04 0.06 0.01 0.03 0.03 0.02 Center 0.01 0.01 0.01 0.01 0.01 0.01 Douala 0.31 0.24 0.36 0.34 0.31 0.39 East 0.01 0.04 0.01 0.02 0.02 0.01 Far-North 0.05 0.17 0.02 0.06 0.04 0.02 Littoral 0.01 0.02 0.01 0.02 0.02 0.02 North 0.04 0.11 0.03 0.04 0.04 0.02 North-West 0.04 0.03 0.03 0.05 0.06 0.04 South 0.01 0.02 0.01 0.02 0.02 0.02 South-West 0.05 0.04 0.06 0.07 0.08 0.06 West 0.05 0.08 0.04 0.06 0.06 0.05 Yaounde 0.37 0.20 0.43 0.29 0.32 0.36 Source: author’s calculation on ECAM4 (2014). Table A15. Characterisation of classes by employment variables Weak-attachment Detachment Classes Classes Variables 1 2 3 4 5 6 Socio-Professional Category Employer 0.06 0.03 0.02 Executive / Managers 0.01 0.01 0.20 Family Aid 0.05 0.33 0.05 Labourer 0.12 0.10 0.11 Own account worker 0.56 0.44 0.07 Qualified employee 0.20 0.10 0.54 Industry Commerce 0.31 0.27 0.15 Industry 0.28 0.25 0.16 Primary 0.05 0.07 0.02 Services 0.36 0.41 0.67 Institutional sector Administration 0.01 0.02 0.18 Association 0.01 0.00 0.01 Household 0.03 0.04 0.04 Private (formal) 0.07 0.04 0.32 Private Agr. (informal) 0.04 0.06 0.01 Private Not Agr. (informal) 0.83 0.82 0.37 Pub. Entity/funded 0.02 0.01 0.06 Worked since age <15 No 0.77 0.78 0.73 0.58 0.56 0.63 Yes 0.23 0.22 0.27 0.42 0.44 0.37 Source: author’s calculation on ECAM4 (2014). 36 Table A16. Characterisation of classes by wealth and income variables Weak-attachment Detachment Classes Classes Variables 1 2 3 4 5 6 Poverty Yes 0.08 0.10 0.05 0.07 0.09 0.02 No 0.92 0.90 0.95 0.93 0.91 0.98 Savings quintiles 0 0.67 0.74 0.65 0.63 0.62 0.49 1 0.07 0.05 0.08 0.09 0.09 0.09 2 0.07 0.05 0.05 0.08 0.08 0.09 3 0.04 0.05 0.04 0.06 0.06 0.09 4 0.06 0.06 0.06 0.08 0.06 0.13 5 0.11 0.05 0.12 0.06 0.08 0.11 At least there a household member rear poultry/cattle No 0.90 0.88 0.88 0.90 0.88 0.92 Yes 0.10 0.12 0.12 0.10 0.12 0.08 Possession of financial assets (equities, property or bonds) No 0.99 1.00 0.98 0.99 0.99 0.98 Yes 0.01 0.00 0.02 0.01 0.01 0.02 A household member owns exploited land No 0.79 0.81 0.75 0.79 0.76 0.80 Yes 0.21 0.19 0.25 0.21 0.24 0.20 Source: author’s calculation on ECAM4 (2014). Rural Characterization Table A17. Characterisation of classes by demographic variables Detachment Classes Weak-attachment Classes Variables 1 2 3 4 5 6 7 Sex Female 0.64 0.87 0.55 0.47 0.73 0.60 0.32 Male 0.36 0.13 0.45 0.53 0.27 0.40 0.68 Household size 1-2 0.11 0.09 0.10 0.13 0.07 0.10 0.32 3-6 0.43 0.46 0.50 0.56 0.48 0.46 0.47 7-10 0.23 0.27 0.29 0.22 0.29 0.29 0.17 >10 0.23 0.18 0.12 0.09 0.16 0.14 0.04 Age Group 15/19 0.27 0.24 0.13 0.00 0.29 0.52 0.00 20/24 0.28 0.33 0.42 0.20 0.29 0.29 0.08 25/29 0.33 0.26 0.36 0.33 0.21 0.19 0.34 30/35 0.11 0.17 0.09 0.47 0.21 0.01 0.58 Marital status Married 0.35 0.65 0.46 0.72 0.60 0.36 0.63 Single 0.62 0.33 0.54 0.24 0.37 0.61 0.35 37 Widowed/Divorced 0.03 0.02 0.00 0.05 0.03 0.03 0.02 Residence 12 months ago Different Location 0.08 0.09 0.09 0.07 0.05 0.10 0.19 Same Location 0.92 0.91 0.91 0.93 0.95 0.90 0.81 Religion Animism 0.02 0.00 0.00 0.03 0.09 0.05 0.01 Catholic 0.32 0.22 0.34 0.26 0.19 0.29 0.35 Muslim 0.31 0.51 0.12 0.26 0.34 0.25 0.14 None 0.02 0.01 0.01 0.04 0.06 0.01 0.02 Other Christian 0.06 0.07 0.11 0.08 0.05 0.06 0.08 Other Religion 0.04 0.03 0.05 0.03 0.02 0.02 0.03 Protestant 0.24 0.16 0.36 0.30 0.26 0.31 0.37 Source: author’s calculation on ECAM4 (2014). Table A18. Characterisation of classes by demographic human capital variables Detachment Classes Weak-attachment Classes Variables 1 2 3 4 5 6 7 Education Higher 0.10 0.01 0.04 0.00 0.00 0.08 0.33 No Education 0.27 0.43 0.07 0.26 0.50 0.02 0.03 Primary 0.34 0.35 0.39 0.48 0.39 0.41 0.19 Secondary 0.29 0.21 0.50 0.26 0.11 0.49 0.45 Technical or vocational training No 0.77 0.86 0.51 0.74 0.86 0.66 0.24 Yes 0.23 0.14 0.49 0.26 0.14 0.34 0.76 Working experience 0 0.69 0.71 0.36 0.01 0.04 0.11 0.02 1-2 0.02 0.02 0.01 0.02 0.04 0.07 0.01 10-15 0.10 0.12 0.31 0.33 0.31 0.23 0.29 16-20 0.05 0.03 0.05 0.29 0.17 0.07 0.22 3-5 0.05 0.04 0.14 0.06 0.17 0.27 0.16 6-9 0.07 0.07 0.11 0.11 0.22 0.23 0.15 >20 0.02 0.02 0.02 0.18 0.05 0.01 0.14 Source: author’s calculation on ECAM4 (2014). Table A19. Characterisation of classes by geographic variable Detachment Classes Weak-attachment Classes Variables 1 2 3 4 5 6 71 Region Adamaoua 0.12 0.20 0.15 0.07 0.04 0.05 0.10 Center 0.14 0.13 0.22 0.08 0.09 0.10 0.11 East 0.04 0.12 0.07 0.06 0.04 0.07 0.09 Far-North 0.24 0.17 0.07 0.21 0.44 0.22 0.07 Littoral 0.02 0.03 0.05 0.03 0.01 0.02 0.05 North 0.09 0.12 0.03 0.20 0.22 0.08 0.07 North-West 0.17 0.08 0.21 0.12 0.05 0.19 0.18 South 0.04 0.06 0.05 0.05 0.01 0.04 0.09 38 South-West 0.10 0.06 0.06 0.11 0.03 0.14 0.14 West 0.04 0.03 0.08 0.08 0.07 0.09 0.10 Source: author’s calculation on ECAM4 (2014). Table A20. Characterisation of classes by employment variables Detachment Classes Weak-attachment Classes Variables 1 2 3 4 5 6 7 Socio-Professional Category Employer 0.02 0.00 0.02 0.00 Executive / Managers 0.00 0.00 0.00 0.24 Family Aid 0.03 0.66 0.55 0.02 Labourer 0.04 0.01 0.06 0.18 Own account worker 0.89 0.32 0.35 0.04 Qualified employee 0.03 0.01 0.02 0.52 Industry Commerce 0.11 0.04 0.11 0.06 Industry 0.13 0.06 0.18 0.12 Primary 0.65 0.86 0.54 0.13 Services 0.11 0.04 0.17 0.70 Institutional sector Administration 0.00 0.00 0.02 0.47 Association 0.00 0.00 0.00 0.01 Household 0.02 0.01 0.05 0.01 Private (formal) 0.00 0.00 0.01 0.16 Private Agr. (informal) 0.62 0.85 0.54 0.04 Private Not Agr. (informal) 0.35 0.14 0.37 0.20 Pub. entity/funded 0.01 0.00 0.00 0.12 Worked since age <15 No 0.81 0.86 0.77 0.44 0.32 0.39 0.59 Yes 0.19 0.14 0.23 0.56 0.68 0.61 0.41 Source: author's calculation on ECAM4 (2014). Table A21. Characterisation of classes by wealth and income variables Detachment Classes Weak-attachment Classes Variables 1 2 3 4 5 6 7 Poverty Yes 0.46 0.49 0.30 0.49 0.68 0.45 0.19 No 0.54 0.51 0.70 0.51 0.32 0.55 0.81 Savings quintiles 0 0.77 0.85 0.75 0.82 0.87 0.79 0.61 1 0.05 0.02 0.04 0.04 0.04 0.06 0.03 2 0.03 0.03 0.10 0.03 0.03 0.06 0.07 3 0.04 0.03 0.05 0.04 0.03 0.03 0.07 4 0.06 0.03 0.02 0.04 0.02 0.04 0.14 5 0.06 0.04 0.05 0.02 0.02 0.03 0.08 At least there a household member rear poultry/cattle 39 No 0.64 0.63 0.73 0.56 0.43 0.56 0.70 Yes 0.36 0.37 0.27 0.44 0.57 0.44 0.30 Possession of financial assets (equities, property or bonds) No 0.98 0.99 0.99 0.99 1.00 0.99 0.98 Yes 0.02 0.01 0.01 0.01 0.00 0.01 0.02 A household member owns exploited land No 0.41 0.41 0.41 0.26 0.15 0.23 0.53 Yes 0.59 0.59 0.59 0.74 0.85 0.77 0.47 Source: author's calculation on ECAM4 (2014). 40