Do income shocks play a determining role in children's schooling in the presence of credit market imperfection in Cameroon?

doi:10.21203/rs.3.rs-1974790/v1

Download PDF

Research Article

Do income shocks play a determining role in children's schooling in the presence of credit market imperfection in Cameroon?

https://doi.org/10.21203/rs.3.rs-1974790/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

This paper examines the determinants of children's schooling under imperfect credit market conditions in Cameroon, with a particular focus on the role of monetary and non-monetary shocks. Households that are more vulnerable to shocks have a greater incentive to build up a buffer stock to protect themselves against unforeseen negative shocks, and keeping children out of school may be part of such a strategy. The study uses microeconomic data from the fourth Cameroonian Household Survey (ECAM IV) conducted in 2014 by the National Institute of Statistics (INS) and an instrumental variable Probit model to demonstrate its point. The results show that uncertainty about household income as measured by transitory income and declining household income decrease the probability of children attending school in Cameroon. The same is true for increasing household size. Nevertheless, access to the credit market is a factor of household resilience to shocks.

Classification JEL: E24, I21, I31, O15.

schooling

monetary and non-monetary shocks

access to credit

risk management strategies

The renewed interest of development actors in education and financial development is based on the fact that they contribute to the economic growth of countries (Mankiw et al., 1992). These determinants of growth are closely linked. Indeed, perfect credit markets facilitate investment in human capital because in such a context, the investment decision is determined solely by its profitability. On the other hand, in the presence of imperfect or deficient credit markets, the decision to invest in education is no longer determined solely by its profitability, but also by the resources available to the household (Becker and Tomes, 1986). These credit market imperfections are much more prevalent in developing countries in general and in African countries south of the Sahara in particular, which are otherwise characterised by low and rather unpredictable incomes. This income unpredictability is most pronounced among agricultural households - particularly in rural areas - which often represent the majority of the population and are subject to multiple negative shocks that significantly affect their income (Kinda, 2010; Björkman, 2013). These shocks can be climatic hazards, a sudden drop in income, the death of a parent or parents, unemployment and/or illness of the head of the household. Moreover, the contingency or redress mechanisms such as access to credit markets, insurance and social security systems to compensate for them are limited and often non-existent. Households then tend to resort to other alternative mechanisms as response instruments.

The uncertainty induced by these shocks thus affects the human capital investment of school-age individuals. It limits access to education, but also favours early exit from the education system because households generally cannot borrow so that a sudden drop in income leads to a high marginal utility for current consumption; which de facto will result in a rather high opportunity cost of education in terms of loss of marginal utility. This loss of marginal utility of education, with associated crowding out processes in terms of investment in human capital and future earnings, has led to the development of an extensive economic literature on the subject. Indeed, the first works on investment in human capital accumulation date back to Becker (1965 and 1967) and Schultz (1988). But it was in the early 2000s that studies dealing with the link between income shocks and children's schooling developed in developing countries. They consider that income volatility due to adverse external events leads to behavioural changes, with households directing their spending towards current consumption goods rather than human capital investment. Several empirical studies have used this framework, and have identified income shock as a factor affecting children's schooling (De Janvry et al., 2006 ; Duryea et al., 2007 ; Kazianga, 2012; Hyder et al., 2015).

However, this thesis is challenged by a number of researchers such as Escobal et al (2005), Vasquez and Bohara (2010) and Dumas (2010) who conclude that there is no link between income shocks and children's schooling. Following these authors, other factors have been put forward as possible explanations for children's schooling. These factors can be grouped under two main headings. The first viewpoint is in the vein of the theory of educational demand initiated by Becker (1967). This author emphasises that, for a household, the choice of whether or not to send a child to school results from a trade-off between the expected returns to education and its total cost. The total cost of education includes direct costs (school fees, books, uniforms and the distance between home and school), indirect costs (foregone income when the child goes to school) and the opportunity cost of the wage or salary that the child foregoes when studying. Indeed, if the household perceives that the returns to education are too low relative to the cost, then it will invest relatively little in education. The low returns to education thus lead households to low enrolment of children.

The second viewpoint highlights the impact of various constraints that affect children's schooling. A first set of constraints results from the imperfect functioning of credit and insurance markets (Jacoby and Skoufias, 1997; Guarcello et al., 2003; Gitter and Barham, 2007; Gubert al., 2010). He points out that in the presence of credit and insurance constraints, households do not have the option of borrowing to finance their offspring's schooling. Indeed, credit constraints do not allow parents to transfer income from the future to the present. As a result, efforts to increase educational opportunities for the poorest households can have little success. A second set focuses on poverty-related constraints (Nembot Ndeffo et al., 2013; Keita, 2014). Their analysis focuses on the interaction between household living conditions and children's schooling. It shows that in families with limited financial resources, children often do not have access to education. Poverty then makes it difficult for disadvantaged households to invest in human capital, especially financially.

Empirical studies attribute a resource dilution phenomenon to education (Chernichovsky, 1985; Sawada, 2003; Cogneau and Jedwab, 2012). This suggests that households make trade-offs between their different children when allocating resources for schooling. Indeed, when a child in a household has siblings who do not attend school, the probability of schooling increases. The intra-household allocation of resources is then characterised by a negative relationship between the number of individuals in the household and school attendance. Consideration of all these explanatory factors sheds light on the policies to be adopted to increase the schooling of children in Cameroon.

The purpose of this article is to contribute to the identification of the determinants of children's schooling in Cameroon in a situation of credit market imperfection. The aim is to examine the influence of different household vulnerability factors, and not only income shocks, which have long been considered the dominant factor.

Certain stylised facts therefore motivate a study on the negative effect of income shocks on children's schooling and the importance of credit and insurance markets in mitigating the effects of these shocks. First, according to Ecam IV data, the adjusted net enrolment rate in primary education is estimated at 85.1% in 2014. Still far from universal enrolment, this rate has remained unchanged since 2009. Overall, the country has a predominantly young population: more than two out of five Cameroonians (44%) are under the age of 15, which puts the dependency ratio at 83.4%. Secondly, since 2008, the country has suffered, with a certain acuteness, the aftershocks of the global economic and financial crisis, the consequences of which on the standard of living of the population are still perceptible. According to the NSI (2014), poverty is thus significant. Of a population estimated at nearly 21.6 million in 2014, 37.5% live below the poverty line, which is about the same proportion as in 2007 (39.9%). Rural populations are particularly affected, since they account for 90.4% of households in poverty in Cameroon. Furthermore, indicators of the poor credit market also point to a worrying situation: more than half of the population (65%) is excluded from the formal financial system (Demirgüç-Kunt et al., 2018). As a result, the proportion of households that have received bank credit remains very low (2.8%). Finally, the occurrence of adverse external events with negative consequences on income increases the opportunity cost of children's time; this is likely to reduce their time in school (Meka'a and Ewondo Mbebi, 2015).

The rest of the paper is organised as follows. Section 2 presents the literature review. Section 3 describes the methodology. Section 4 discusses the results of the study while Section 5 is the conclusion.

Many studies have focused on the issue of education in the development process. These studies have undergone a certain evolution in recent years, thanks to the development of economic theories and models. Their analyses have taken various directions. Some base the debate on economic and demographic shocks to analyse the schooling of children. Others, on the other hand, are more interested in credit market failures and returns to education on the one hand, and in household income poverty and intra-household resource allocation on the other.

Income shocks to households are vulnerability factors that influence children's schooling. Indeed, when shocks occur, households' resilience is low because they have few risk management instruments at their disposal (Ewondo Mbebi, 2018). They have few assets and limited recourse to savings. Households are therefore forced to adapt their behaviour by investing less in education to smooth out their consumption. Reducing investment in education is thus a strategy that poor households use to minimise the risk of disruption to the flow of resources, in order to reduce the potential impact of any shocks (Kazianga, 2012). Several empirical studies have used this framework to confirm the negative impact of income shocks on children's schooling. Microeconomic data from the Progresa experiment are used to show that idiosyncratic (unemployment or illness of the head of household) and covariant (natural disasters) shocks cause manages to adopt risk management strategies by withdrawing their children from school for the Mexican labour market (De Janvry, 2006). While from a bivariate probit model, it is shown that an income shock represented by the loss of employment of the head of household significantly reduces the schooling of children, particularly girls over the age of 16 (Duryea et al., 2007). Estimating data from rural areas, the conclusion that individual or community shocks negatively influence children's schooling is drawn in the case of Malawi (Hyder et al., 2015). However, in various studies, the correlation between income shocks and children's enrolment is weak or even insignificant because households may continue to send their children to cheaper schools (Vasquez and Bohara, 2010; Dumas, 2010; Basab and Mohamed-Ihsan, 2011). They will also be able to reduce the direct costs of schooling without taking their children out of the education system (Escobal, Saavedra and Suárez, 2005).

Household income poverty is also an important source of inefficiency in human capital investment. Indeed, in households with limited financial resources, the problem of schooling is posed to families mainly in terms of costs. Poor households do not have productive assets or sufficient financial means to bear the costs of education. Their inability to cover educational needs prevents them from investing in human capital. This line of thinking shows that education is an investment. Investment in human capital is considered to be the sum of the direct and indirect costs of education. These costs seem small but are not. They are further exacerbated when a household member reaches a high level of education, as the higher the level, the more education becomes a luxury good. Yet households have low incomes that do not allow them to meet basic needs. Their state of poverty forces them not only to limit access to education, but also the length of schooling of those members who have already had the chance to access education (Lincove, 2009; Nembot-Ndeffo et al., 2013).

The return to education also determines the investment in education. Indeed, the problem of schooling is posed to households mainly in terms of the balance between the expected return to education and the opportunity cost of work (Ewondo Mbebi, 2018). Education no longer appears to be an effective means of increasing the chances of accessing the labour market. It is no longer perceived by households as a guarantee of access to employment. Education therefore no longer plays the role of selection or filter for the labour market.

Access to credit by households is another factor that explains the schooling of children. Indeed, when the credit market functions perfectly, households invest more in education since they can easily lend or borrow on this market (Edmonds, 2006; Guarcello et al., 2010). The development of credit and social insurance markets thus appears to be a means of transferring part of the costs of schooling to the generation that has benefited from it. This easy access to credit and social security by households increases the schooling of children. Indeed, the receipt of a pension provides a secure future income that can be used as collateral for a loan (Edmonds, 2006). This possibility of borrowing against future income (in this case the retirement pension) allows households to finance the education of its members. Access to credit is, in any case, a form of self-insurance, but also a safety net that households use to increase the level of schooling of children. In contrast, in the presence of an imperfect credit market, even altruistic households will benefit from offspring transfers. Faulty credit markets mean that households cannot protect themselves against the risk of income loss. They are unable to make an inter-temporal trade-off regarding their resources. Under these conditions, the difficulty in accessing credit leads households to limit the schooling of children, which could have noticeable long-term repercussions on their human capital stock.

Resource dilution also plays an important role in the decision to invest in human capital. Indeed, in large families, parents arbitrate for the allocation of resources between the different children in the household: having siblings who do not attend school in a family increases the probability of the child attending school (Sawada, 2003). A number of studies question this phenomenon of resource dilution. Using data from the Income Distribution Survey, the positive relationship between the number of individuals in the household and their demand for education is established in Botswana (Chernichovsky, 1985). This relationship is justified by the low demand for labour by each member of the household given the large number of people available, which will in fact reduce the opportunity cost of education. By estimating a sequential schooling decision model, Sawada and Lokshin (2001) show that the number of older sisters and older brothers increases the primary and secondary education of the youngest children in the household in Pakistan, respectively. Estimating generalized hierarchical models, Diagne (2010) shows that children who have reached the end of primary or secondary school serve as tutors for the youngest children attending primary school in the household in Senegal. The existence of children attending school in a family has a de facto spillover effect on the schooling of the youngest children in the household.

3.1. Data sources and variables

The data used come from the fourth Cameroonian Household Survey (ECAM IV), the most recent survey conducted in 2014 by the National Institute of Statistics. This survey, which covers the entire national territory, was not specifically designed for children's schooling. However, given that the main objective was to assess the relevance of public programmes and policies implemented in the fight against poverty, a section on children's schooling was reserved. In order to have fairly homogeneous strata with regard to children's schooling and for the purposes of harmonising the stratification, the country's two main cities, Yaoundé and Douala, were considered as autonomous urban strata. Each of the ten regions was divided into two strata (urban and rural), i.e. a total of 22 strata, 12 of which were urban and 10 rural. The sample design used was a two-stage stratified random sample. At the first stage of each stratum, enumeration areas (EAs) were drawn in proportion to their size to take into account existing disparities. In the second stage, a sample of households was drawn from each DZ selected in the first stage. The advantage of this sampling plan is that the mapping and enumeration operations in each DZ are carried out at the same time as the survey itself. Initially, a number of 12,847 households was identified according to their standard of living. This number was revised downwards to take account of non-respondents. In the sample allocation, 12 households per DZ were assigned for Yaoundé and Douala and 18 households per DZ in the other strata. The exploratory sample finally included 10,303 households for a total of 4,660 individuals in both urban and rural areas. Among these individuals, a global sample of 9195 children aged between 6 and 12 who had attended a primary school or not during the 2013/2014 school year was selected. This age range is defined according to the official school age in Cameroon. However, in practice, there are individuals over the age of 12 who are still in primary school. These were not included in our analyses because they represent a small proportion.

In this article, we consider two types of income shocks: monetary shocks and non-monetary shocks. For non-monetary shocks, we define two variables. The first variable takes the value 1 if the child has lost at least one parent and 0 otherwise. The second variable takes the value 1 if the head of the child's household was ill and 0 otherwise. As for monetary shocks, they are also captured by two different variables: the first one refers to the transitional income while the second one expresses the evolution of the income compared to the year preceding the survey.

With respect to the estimation of the first monetary shock variable, it consists of a decomposition of income into transitory and permanent income. Conceptually, the income of household i is defined as:

As for the second monetary shock variable, we capture it by the question: "how has the income from the main job of the head of the household evolved in relation to the 12 months preceding the last 12 months (annual assessment) of the survey? This variable has four modalities: increased, decreased, unchanged and not concerned. The 'unchanged' modality represents households that experienced no monetary shock, while the 'increased' and 'decreased' modalities express households that experienced a positive or negative shock respectively. For this measure, households not affected were excluded.

In addition to monetary and non-monetary shocks, we introduce the variable 'access to credit' which expresses the imperfection of the credit market. This variable is binary and takes the value 1 if the household had access to credit and 0 otherwise. Given the data collected in Ecam IV, the other variables can be grouped into three main categories: household characteristics, child characteristics and those related to the family environment. The characteristics of the child are: age, sex and status in relation to the head of the household. Being a child of the head of the household, for example, can lead to a differential allocation of time compared to other children in the household (Meka'a and Ewondo Mbebi, 2015).

In terms of household characteristics, we have the size of the household measured by the number of people living in the household, the sector of activity of the head of the household, the level of education, the gender and the age of the head of the household. In principle, the probability of children attending school decreases with the number of children in the household. Finally, with regard to factors related to the family environment, the geographical location of households or individuals is particularly important for two reasons: on the one hand, its absence may bias the effect of non-geographical characteristics on children's schooling. On the other hand, in an agricultural country like Cameroon, geographical location is an important characteristic determining the standard of living of households.

3.2. Empirical strategy

The estimation strategy consists of two steps. The first step estimates equation (1) by a Tobit model to extract the household's transitory income, which is the first measure of monetary shocks. The use of a Tobit model is justified by the fact that income is always positive or zero. The second step consists in integrating the estimated transitory income from the first step among the explanatory variables of the child's schooling equation. In addition, the other monetary and non-monetary shock variables are also included in the child's schooling equation. This equation is as follows:

In a simple way, a binary Probit model is sufficient to estimate equation (3). However, such a procedure does not allow us to correct the possible endogeneity problem of the intra-household allocation of resources for education. Indeed, in the event of an income shock affecting a household, it is very likely that the schooling of one child in this household affects that of the other children, given that schooling decisions are made simultaneously. In addition, it is likely that household income includes the income earned by a child who is not in school but working. In this case, household income and schooling are mutually explanatory. Therefore, income needs to be instrumented in order to identify its real effect as well as those of other variables that influence children's schooling. To this end, we define new instruments for the analysis. These are the possession of certain durable goods (television, refrigerator, air conditioner, car, computer, washing machine, telephone, etc.) owned by the household and the characteristics of the dwelling (floor and wall materials, presence of running water, presence of electricity, etc.). These variables are strongly correlated with income but do not explain the schooling of a child. Indeed, the choice of these instrumental variables was made according to what is generally adopted in the literature (Couralet, 2003). However, before presenting the different results, it is necessary to present some statistics on the children concerned in this study, on their household and on their family environment. These statistics are presented in the following Table 1:

Table 1: Descriptive statistics

Variables	Means	Standard deviations
Child's characteristics
Gender of the child (male)	0.50	0.50
Age of the child	8.81	1.98
Child of the head of household (yes)	0.75	0.43
Household characteristics
Holding assets
Access to credit (yes)	0.03	0.16
Ownership of land (yes)	0.22	0.61
Number of cars	0.01	6.32
Number of air conditioners	0.04	18.68
Number of refrigerators	0.83	2.79
Monetary and non-monetary shocks
Death of at least one parent of the child (yes)	0.08	0.27
The head of the child's household was ill (yes)	0.36	0.48
Change in household income (not concerned)	0.39	0.49
Unchanged	0.16	0.37
Increased	0.42	0.49
Decreased	0.03	0.16
Environmental characteristics
Residence area (urban)	0.53	0.50
Rainfall	7.54	13.48
Temperature	32.6	44.71
Other characteristics
Income	38603.72	106753.3
Gender of head of household (female)	0.30	0.45
Age of head of household	43.48	15.79
Size of household (number of persons)	4.47	3.12
Education level of head of household (no level)	0.21	0.41
Primary	0.33	0.47
Secondary	0.36	0.48
Higher	0.10	0.31
Access to drinking water (yes)	0.98	0.25
Access to electricity (yes)	0.98	0.32
Quality of soil (tiles)	0.04	3.49
Sector of activity (agriculture)	0.42	0.49
Industrys	0.16	0.36
Trade	0.17	0.37
Services	0.25	0.43
Sources : authors

Table 1 shows that our sample consists of 50% boys. The average age of the children is 9 years, which means that the majority of the children in the sample are almost in the last year of primary school. Furthermore, children whose heads of household are their biological parents represent about 75%. As regards the characteristics of the head of household, we note in general that the average age of the heads of household is around 44 years and that 30% of heads of household are women. In addition, only 10% of heads of household have attained a higher level of education. Thus, the majority of heads of households who must be at the beginning of their professional career have not only a low level of education, but also low incomes which may increase their vulnerability to shocks. With regard to the shocks experienced by the household, the statistics show that only 42% of heads of household have not experienced any variation in their income, compared with 36% and 8%, respectively, for heads of household who have been ill and those who have lost at least one spouse. With regard to other household characteristics, Table 1 shows that 98% of households have access to water and electricity, 22% own land and 53% live in urban areas. It is also noted that the average household size is around 5, which could increase the vulnerability of the household to unexpected adverse events. An analysis looking at asset ownership shows that only 3% of households have access to credit. This last statistic seems to illustrate the imperfection of the credit market in Cameroon, where households find it difficult to obtain credit when exposed to adverse external events. In terms of ownership of other physical assets, there is an average of one refrigerator per household, while an air conditioner and a car are owned by only a handful of households (4% and 1%, respectively).

This section is structured in three points: first, we present the estimation of the transitional household income. Second, we test the validity of the transitional household income instruments and third, we present the results of the estimation of children's schooling in Cameroon.

As mentioned above, the transitional household income is the first measure of monetary shocks in addition to the change in income relative to the year before the survey. Table 2 below summarises the results of estimates of this transitional income using the Tobit model. Indeed, the transitional income represents the residual from the regression of the logarithm of the current household income on the determinants of this current income as specified by Eq. (2). These determinants are the socio-economic variables, the variables informing on the demographic structure, the rainfall and the average temperature of the region of residence of the household.

Table 2

Transitional household income estimates
LR chi2(17)	5933.36	Number of observations				12570
Prob > chi2	0.000	R² ajusted				0.51
Dependent variable (logarithm of current household income)				Coefs	Probabilités
Education level of head of household (no level)				-	-
Primary				0.288***	0.000
Secondary				0.437***	0.000
Higher				0.808*	0.102
Area of residence (urban)				0.110***	0.000
Rainfall				0.151**	0.035
Square of rainfall				-0.233*	0.077
Temperature				0.015	0.294
Square of temperature				-0.794***	0.000
Sex of head of household (female)				-0.588***	0.000
Age of head of household				0.008**	0.028
Square of age of head of household				-0.441***	0.000
Informal sector (yes)				-0.413***	0.000
Sector of activity (agriculture)				-	-
Industrys				0.402***	0.000
Trade				0.461***	0.000
Services				0.563**	0.049
Constant				2.550***	0.000
Sources : Authors			P < 0.1 P < 0.05 **P < 0.01

Table 2 shows that the level of education, age, gender, sector of activity and informality of the head of household significantly determine household income. Similarly, the area of residence of the household, rainfall and high temperature significantly determine household income. Specifically, we note that household income increases with the level of education and the age of the head of the household. Similarly, the fact that the household lives in an urban area or that the head of the household works in a sector other than agriculture positively affects household income. On the other hand, environmental variables such as high rainfall and high temperature, represented respectively by the squared values of these variables, negatively affect household income. This would be due to the fact that the majority of Cameroonian households reside in rural areas and potentially live from agriculture. Finally, the estimate of transitory income is determined by the residuals of the estimate presented in Table 2.

We can now turn to the results of the effects of monetary and non-monetary shocks on children's schooling in Cameroon. However, as we mentioned above, the transitional income endogenous to children's schooling must be instrumented. Table 3 below presents the validity test of the instruments used in this work.

Table 3

Validity tests of the transitional household income instruments
Prob > F	0.000	Number of observations			10303
R²	0.62	R² ajusted			0.59
Dependent variable (Transitional household income)			Coefs	Probability
Number of cars			0.115**	0.042
Number of refrigerators			0.215***	0.001
Number of air conditioners			0.311**	0.029
Presence of clean water			0.111***	0.000
Presence of electricity			0.091**	0.032
Quality of the floor (tiles)			0.421**	0.037
Possession of land (yes)			0.005*	0.081
Constant			0.535***	0.000
Sources : Authors			P < 0.1 P < 0.05 **P < 0.01

We can see that the possession of durable goods and the characteristics of the dwelling significantly affect the transitional income of households since all the probabilities are lower than 5%. In fact, these variables are strongly correlated with transitional income without determining the schooling of children. These variables are thus good instruments of transitional income. Table 4 below presents the estimates of the effects of monetary and non-monetary shocks on children's schooling. This table presents both the signs of the parameters as well as the marginal effects.

Table 4

Estimates of the child enrolment equation
		Coefficients (probability)		Marginals effects (probability)
Monetary and non-monetary shocks
Transitional household income		*-0.123 (0.065)**		-0.032 (0.037)**
Death of at least one parent of the child (yes)		-0.328 (0.241)		-0.056 (0.127)
The head of the child's household was ill (yes)		0.258 (0.335)		0.161 (0.240)
Change in household income (not concerned)		-		-
Unchanged		-0.041 (0.216)		-0.069 (0.174)
Increased		1.163* (0.000)**		0.349* (0.000)**
Decreased		-0.406* (0.000)**		-0.021* (0.000)**
Child's characteristics
Gender of the child (male)		0.017 (0.280)		0.481 (0.159)
Age of the child		*0.145 (0.079)**		0.002* (0.000)**
Child of the head of household (yes)		0.130 (0.022)**		0.011* (0.000)**
Household characteristics
Access to credit (yes)		0.305* (0.002)**		0.118* (0.000)**
Education level of head of household (no level)		-		-
Primary		0.133 (0.019)**		0.027* (0.000)**
Secondary		0.924* (0.000)**		0.029* (0.000)**
Higher		0.017 (0.049)**		0.061 (0.044)**
Size of household (number of persons)		*-0.155 (0.055)**		-0.028* (0.000)**
Residence area (urban)		0.271* (0.000)**		0.044 (0.033)**
Gender of head of household (female)		-0.216 (0.131)		*-0.091 (0.091)**
Constante		2.273* (0.000)**		-
Numbre of observations = 12570	Wald chi2(16) = 26667.38		Prob > chi2 = 0.0000
Sources : authors		P < 0.1 P < 0.05 **P < 0.01

Table 4 shows that the estimated model is statistically significant since the probability of the Wald statistic being zero is rejected. As for the estimation results, columns two and three of Table 4 present the parameter coefficients of the instrumental variable Probit model and the marginal effects respectively. The result is that monetary shocks significantly affect the probability of children attending school in Cameroon, unlike non-monetary shocks. While positive shocks such as an increase in household income positively affect the probability of children attending school, negative shocks such as a decrease in household income negatively affect the probability of children attending school. Indeed, the probability of enrolment decreases by 3.2% when the magnitude of the transitional income varies by one unit. This negative effect of transitional income on the probability of enrolment can be explained by the fact that not only are the vast majority of households poor and generally have a negative transitional income, but also that access to credit remains difficult for most of them. Similarly, the probability of enrolment decreases by 2.1% when there is a decrease in household income. However, the effects of idiosyncratic shocks such as the death of at least one of the child's parents are not significant. Indeed, we find that the probability of schooling decreases non-significantly by 5.6% when at least one of the child's parents dies. These results corroborate those obtained by Hyder et al. (2012), who show that negative economic shocks have significant negative impacts on children's schooling and educational attainment, with the estimated effects of community shocks being larger and more widespread than the estimated effects of idiosyncratic shocks. In contrast to negative shocks, we find that the probability of a child's enrolment increases by 34.9% when there is an increase in household income.

Regarding the other results, the probability of schooling increases by 11.8% when the household has access to credit. This means that the use of credit can be a means of household resilience to shocks. Access to credit markets by parents allows children to stay in school by protecting them from shocks that affect their families (Beegle et al., 2003; Guarcello et al., 2010). Furthermore, it is observed that the higher the level of education of the head of the household, the higher the probability of schooling. For example, we find that the probability of enrolment increases by 2.7% when the head of the household has a primary education compared to the head of the household who has no education. This result can be justified by the fact that an increase in the level of education of the head of the household can enable the latter to have a decent job that generates a high income. Similarly, the fact that the household lives in an urban area increases the probability of children attending school by 4.4%. This increase in schooling shows that urban households are less sensitive to shocks than rural households. Such a result is not surprising given that rural households can at any time suffer an agricultural shock that would not only jeopardise their production, but also limit the investment in their children's human capital. Indeed, an agricultural shock affects crop yields and thus household income and food availability, forcing the household to reduce current consumption. To guard against unforeseen negative shocks, households have an incentive to build up a buffer stock by varying the amount of schooling their children receive or by withdrawing their children from the education system in order to have a labour force that can bring in additional income (Kinda, 2010; Kazianga, 2012; Ewondo Mbebi, 2018). Household size also matters. The simulations in Table 4 show that it reduces the probability of schooling by 2.8%. Indeed, at a constant level of income, a large household size implies a low income per person and, by induced effect, a reduction in the investment in schooling of certain members of the household; this would increase their vulnerability to unexpected negative events. Sawada (2003) arrives at similar results by showing that the response of schooling to an income shock is systematically greater for some children than others in the household and that there may be resource competition between siblings within the same household. Human capital investment and school allocation decisions within the household appear to be affected by a need for self-insurance arrangements under binding credit constraints.

The objective of this paper is to examine the determinants of children's schooling in a situation of credit market imperfection in Cameroon, with an emphasis on the role of monetary and non-monetary shocks. The aim is to show that the importance of monetary and non-monetary shocks in explaining children's schooling is only determinant in the presence of credit market imperfection. In theory, shocks to households are vulnerability factors that influence children's schooling. When these shocks occur, household resilience is low because they have few risk management instruments at their disposal, few assets in their possession and limited recourse to savings (Ewondo Mbebi, 2018). Households are therefore forced to adapt their behaviour by investing less in education to smooth out their consumption. Reducing investment in education is thus a strategy that poor households use to minimise the risk of disruption to the flow of resources, in order to reduce the potential impact of any shocks (Kazianga 2012).

Empirically, this study mobilises microeconomic data from the fourth Cameroonian Household Survey (Ecam IV) conducted in 2014 by the National Institute of Statistics (INS). Given that transitory income is endogenous to children's schooling since it is highly likely that the schooling of one child in the household affects that of other children and that household income incorporates the income brought in by a child who is not in school but working, we mobilize an instrumental variable Probit model to assess the effects of monetary and non-monetary shocks on children's schooling. The results confirm that monetary shocks, notably transitory income and the decrease in household income, significantly affect the probability of children's schooling. The same is true for household size and area of residence, with the results being more significant in rural areas. On the other hand, non-monetary shocks have no significant effect. However, the results show that access to credit markets by parents is a factor of resilience to shocks and allows children to remain in the school system by protecting them against these shocks. These results suggest that, in order to limit the magnitude of monetary and non-monetary shocks on children's schooling in Cameroon, it would be wise to facilitate access to credit by Cameroonian households, particularly the most vulnerable, which are generally large households and those in rural areas. This can be done by setting up insurance systems and other mechanisms such as conditional transfers to give households means of smoothing their consumption other than withdrawing their child from school. Indeed, access to credit is an important source of human capital financing to mitigate the negative effect of shocks on children's education (Kazianga, 2012).

Competing Interest

The authors have no conflicts of interest.

Funding declaration

This study has not received any funding.

Basab D., and A. Mohamed-Ihsan (2011), « Income Shocks reduce human capital investments: evidence from five East European countries », World Bank, Working Paper n°5926.
Becker, G.S. (1967), « Human Capital and the Personal Distribution of Income », Columbia University Press.
Becker, G.S. (1965), « A Theory of Allocation of Time », Economic Journal (75).
Becker, G., and N. Tomes (1986), « Human Capital and the Rise and Fall of Families », Journal of Labor Economics, 4 (3), p.1–39.
Beegle, K., D. Rajeev and G. Roberta (2003), « Child Labor, Income Shocks, and Access to Credit», World Bank Policy Research Working Paper, n°3075.
Björkman, M.N. (2013), « Income Shocks and Gender Gaps in Education: Evidence from Uganda», Journal of Development Economics, 105, P. 237–253.
Chernichovsky, D. (1985), « Socioeconomic and Demographic Aspects of School Enrolment and Attendance in Rural Botswana », Economic Development and Cultural Change, 33 (2), p. 319–332.
Cogneau, D., and R. Jedwab (2012), « Family Income and Child Outcomes: The 1990 Cocoa Price Shock in Côte d’Ivoire », Economic Development and Cultural Change, 60 (3), p. 507–534.
Couralet, P. E. (2003). « Le travail des enfants en Amérique latine », Document de travail, Questions sociales, emploi et migrations, n° 3, Paris: OCDE.
De Janvry, A., F. Finan, E. Sadoulet and R. Vakis (2006), « Can Conditional Cash Transfert Serve as Safety Nets in Keeping Children at School and from Working when Exposed to Shocks? », Journal of Development Economics, 79(2), p. 349–373.
Demirgüç-Kunt, A., L. Klapper, D. Singer, S. Ansar and J. Hess (2018), The Global Findex Database 2017: Measuring Financial Inclusion and the Fintech Revolution, World Bank, Washington.
Diagne, A. (2010). « Pourquoi les enfants africains quittent-ils l’école ? Un modèle hiérarchique multinomial des abandons dans l’éducation primaire au Sénégal », L'Actualité économique, 86(3), p. 319–354.
Dumas, C. (2010), « Chocs et dynamiques des choix d'allocation du temps des enfants au Sénégal », Cahiers du Cedimes, 3 (2).
Duryea, S., D. Lam and D. Levison (2007), « Effects of Economic Shocks on Children’s Employment and Schooling in Brazil », Journal of Development Economics, 84 (1), p.188–214.
Edmonds, E. V. (2006), « Child labor and schooling responses to anticipated income in South Africa », Journal of development economics, 81(2), p. 386–414.
Escobal, J., J. Saavedra and P. Suárez (2005), « Economic Shocks and Changes in School Attendance Levels and Education Expenditure in Peru », Working Paper No. 13. Young Lives: An International Study of Childhood Poverty. Save the Children UK. London.
Ewondo Mbebi, O. (2018), « Determinants of Child Labor in Cameroon: The Role of Area of Residence and Gender », Revue d’Economie du Développement, 26 (1), P.5–52.
Gitter, S. R., and B. Barham (2007), « Credit, Natural Disasters, Coffee, and Educational Attainment in Rural Honduras », World Development Economics, 35 (3), p. 498–511.
Guarcello, L., F. Mealli and F. C. Rosati (2010), « Household vulnerability and child labor: The effect of shocks, credit rationing, and insurance », Journal of Population Economics, 23 (1), p.169–198.
Gubert, F., A-S. Robilliard and J-N. Senne (2010), « Impact des chocs économiques et démographiques sur la scolarisation en milieu rural », In: Gastineau Bénédicte (ed.), Gubert Flore (ed.), Robilliard Anne-Sophie (ed.), Roubaud François (ed.), Ravelomantsoa P.G. (préf.), Chataigner J.M. (préf.) Madagascar face au défi des objectifs du millénaire pour le développement. Marseille: IRD, 2010, p. 157–185.
Hyder, A.J., R. Behrman and H. P. Kohler (2015), « Negative Economic Shocks and Child Schooling: Evidence from Rural Malawi », Development Southern Africa, 32(4), p.458–476.
Jacoby, H. and E. Skoufias (1997), « Risk, Financial Markets, and Human Capital in a Developing Country », Review of Economic Studies, vol.64, n°3, pp. 311–335.
Kazianga, H. (2012), « Income Risk and Household Schooling Decisions in Burkina Faso », World Development, 40 (8), p. 1647–1662.
Kazianga, H. and C. Udry (2006), « Consumption Smoothing? Livestock, Insurance and Drought in Rural Burkina Faso », Journal of Development Economics, 79(2), p. 413–446.
Keita, M. (2014), « Pauvreté et arbitrage entre scolarisation et travail des enfants au Mali », Working Papers N ⁰ 201418, CERDI.
Kinda, T. (2010), « Choc de revenu et éducation des enfants en présence d’imperfection du marché de crédit: le cas du Malawi », Recherche Économiques de Louvain, 76 (4), p. 391–411.
Lincove, J.A. (2009), « Determinants of schooling for boys and girls in Nigeria under a policy of free primary education », Economics of Education Review, 28(4), p.474–484.
Meka’a, C. B. and O. Ewondo-Mbebi (2015), « Child labour: An entierly poverty related problem? the effects of household economic situiation on child labour in Cameroon in 2007», Travail et Emploi, 143 (3), p. 5–19.
Mankiw, N.G., D. Romer and D. Weil (1992), « A Contribution to the Empirics of Economic Growth », Quarterly Journal of Economics, 107 (2), p. 407–437.
Nembot-Ndeffo, L., M. Makoudem-Tene and J. R. Tagne-Kuelah (2013), « Analyse des déterminants monétaires et non monétaires de l’accessibilité à l’éducation au Cameroun », Revue d’Economie du Développement, (27): 91–125.
NIS (2014), Report of the fourth Cameroonian household survey, NIS, Yaoundé.
Paxson, C., H. (1992), « Using Weather Variability to Estimate the Response of Savings to Transitory income in Thailand », American Economic Review, 82(1), p.15–33.
Sawada, Y. (2003), « Income Risks, Gender, and Human Capital Investment in a Developing Country », CIRJE Working Paper n°198.
Sawada, Y. and M. Lokshin (2001), « Household Schooling Decisions in Rural Pakistan », World Bank Policy Research, Working Paper, n°2541.
Vásquez, W.F. and A.K. Bohara (2010), « Household shocks, child labor, and child schooling », Latin American Research Review, 45 (3), p. 165–186.
Schulz, T. P. (1988), « Education Investment and Returns », In H. Chenery and T. N. Srinivasan (Eds.), Handbook of Development Economics, p. 643 – 630.

Basab D., and A. Mohamed-Ihsan (2011), « Income Shocks reduce human capital investments: evidence from five East European countries », World Bank, Working Paper n°5926.
Becker, G.S. (1967), « Human Capital and the Personal Distribution of Income », Columbia University Press.
Becker, G.S. (1965), « A Theory of Allocation of Time », Economic Journal (75).
Becker, G., and N. Tomes (1986), « Human Capital and the Rise and Fall of Families », Journal of Labor Economics, 4 (3), p.1–39.
Beegle, K., D. Rajeev and G. Roberta (2003), « Child Labor, Income Shocks, and Access to Credit», World Bank Policy Research Working Paper, n°3075.
Björkman, M.N. (2013), « Income Shocks and Gender Gaps in Education: Evidence from Uganda», Journal of Development Economics, 105, P. 237–253.
Chernichovsky, D. (1985), « Socioeconomic and Demographic Aspects of School Enrolment and Attendance in Rural Botswana », Economic Development and Cultural Change, 33 (2), p. 319–332.
Cogneau, D., and R. Jedwab (2012), « Family Income and Child Outcomes: The 1990 Cocoa Price Shock in Côte d’Ivoire », Economic Development and Cultural Change, 60 (3), p. 507–534.
Couralet, P. E. (2003). « Le travail des enfants en Amérique latine », Document de travail, Questions sociales, emploi et migrations, n° 3, Paris: OCDE.
De Janvry, A., F. Finan, E. Sadoulet and R. Vakis (2006), « Can Conditional Cash Transfert Serve as Safety Nets in Keeping Children at School and from Working when Exposed to Shocks? », Journal of Development Economics, 79(2), p. 349–373.
Demirgüç-Kunt, A., L. Klapper, D. Singer, S. Ansar and J. Hess (2018), The Global Findex Database 2017: Measuring Financial Inclusion and the Fintech Revolution, World Bank, Washington.
Diagne, A. (2010). « Pourquoi les enfants africains quittent-ils l’école ? Un modèle hiérarchique multinomial des abandons dans l’éducation primaire au Sénégal », L'Actualité économique, 86(3), p. 319–354.
Dumas, C. (2010), « Chocs et dynamiques des choix d'allocation du temps des enfants au Sénégal », Cahiers du Cedimes, 3 (2).
Duryea, S., D. Lam and D. Levison (2007), « Effects of Economic Shocks on Children’s Employment and Schooling in Brazil », Journal of Development Economics, 84 (1), p.188–214.
Edmonds, E. V. (2006), « Child labor and schooling responses to anticipated income in South Africa », Journal of development economics, 81(2), p. 386–414.
Escobal, J., J. Saavedra and P. Suárez (2005), « Economic Shocks and Changes in School Attendance Levels and Education Expenditure in Peru », Working Paper No. 13. Young Lives: An International Study of Childhood Poverty. Save the Children UK. London.
Ewondo Mbebi, O. (2018), « Determinants of Child Labor in Cameroon: The Role of Area of Residence and Gender », Revue d’Economie du Développement, 26 (1), P.5–52.
Gitter, S. R., and B. Barham (2007), « Credit, Natural Disasters, Coffee, and Educational Attainment in Rural Honduras », World Development Economics, 35 (3), p. 498–511.
Guarcello, L., F. Mealli and F. C. Rosati (2010), « Household vulnerability and child labor: The effect of shocks, credit rationing, and insurance », Journal of Population Economics, 23 (1), p.169–198.
Gubert, F., A-S. Robilliard and J-N. Senne (2010), « Impact des chocs économiques et démographiques sur la scolarisation en milieu rural », In: Gastineau Bénédicte (ed.), Gubert Flore (ed.), Robilliard Anne-Sophie (ed.), Roubaud François (ed.), Ravelomantsoa P.G. (préf.), Chataigner J.M. (préf.) Madagascar face au défi des objectifs du millénaire pour le développement. Marseille: IRD, 2010, p. 157–185.
Hyder, A.J., R. Behrman and H. P. Kohler (2015), « Negative Economic Shocks and Child Schooling: Evidence from Rural Malawi », Development Southern Africa, 32(4), p.458–476.
Jacoby, H. and E. Skoufias (1997), « Risk, Financial Markets, and Human Capital in a Developing Country », Review of Economic Studies, vol.64, n°3, pp. 311–335.
Kazianga, H. (2012), « Income Risk and Household Schooling Decisions in Burkina Faso », World Development, 40 (8), p. 1647–1662.
Kazianga, H. and C. Udry (2006), « Consumption Smoothing? Livestock, Insurance and Drought in Rural Burkina Faso », Journal of Development Economics, 79(2), p. 413–446.
Keita, M. (2014), « Pauvreté et arbitrage entre scolarisation et travail des enfants au Mali », Working Papers N ⁰ 201418, CERDI.
Kinda, T. (2010), « Choc de revenu et éducation des enfants en présence d’imperfection du marché de crédit: le cas du Malawi », Recherche Économiques de Louvain, 76 (4), p. 391–411.
Lincove, J.A. (2009), « Determinants of schooling for boys and girls in Nigeria under a policy of free primary education », Economics of Education Review, 28(4), p.474–484.
Meka’a, C. B. and O. Ewondo-Mbebi (2015), « Child labour: An entierly poverty related problem? the effects of household economic situiation on child labour in Cameroon in 2007», Travail et Emploi, 143 (3), p. 5–19.
Mankiw, N.G., D. Romer and D. Weil (1992), « A Contribution to the Empirics of Economic Growth », Quarterly Journal of Economics, 107 (2), p. 407–437.
Nembot-Ndeffo, L., M. Makoudem-Tene and J. R. Tagne-Kuelah (2013), « Analyse des déterminants monétaires et non monétaires de l’accessibilité à l’éducation au Cameroun », Revue d’Economie du Développement, (27): 91–125.
NIS (2014), Report of the fourth Cameroonian household survey, NIS, Yaoundé.
Paxson, C., H. (1992), « Using Weather Variability to Estimate the Response of Savings to Transitory income in Thailand », American Economic Review, 82(1), p.15–33.
Sawada, Y. (2003), « Income Risks, Gender, and Human Capital Investment in a Developing Country », CIRJE Working Paper n°198.
Sawada, Y. and M. Lokshin (2001), « Household Schooling Decisions in Rural Pakistan », World Bank Policy Research, Working Paper, n°2541.
Vásquez, W.F. and A.K. Bohara (2010), « Household shocks, child labor, and child schooling », Latin American Research Review, 45 (3), p. 165–186.
Schulz, T. P. (1988), « Education Investment and Returns », In H. Chenery and T. N. Srinivasan (Eds.), Handbook of Development Economics, p. 643 – 630.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Do income shocks play a determining role in children's schooling in the presence of credit market imperfection in Cameroon?

Status:

Version 1

Abstract

1. Introduction

2. Literature Review

3. Methodologie

3.1. Data sources and variables

3.2. Empirical strategy

4. Results

5. Conclusion

Declarations

References

Additional Declarations

Status:

Version 1