Contract, Gender, and Job Satisfaction: Evidence from Benin

Abstract

This study analyzes the effects of job contract and salary on job satisfaction. Job satisfaction is measured using a subjective approach. As a result, workers themselves have reported their satisfaction levels. The study thus pursues two objectives: (i) to evaluate the effect of the job contract on satisfaction and (ii) to test the existence of a differentiated satisfaction according to gender. Using the data from the School-to-Work Transition Survey (SWTS) database conducted in Benin, we carried out econometric analyses based on the estimation of an ordered probit and a simple probit. Our results confirm the validity of our two hypotheses and also show that nonwage benefits improve worker satisfaction.

1. Introduction

Job quality is increasingly analyzed in the economic literature and is closely linked to that of job satisfaction. This question goes beyond the neoclassical model in which wages are the only variable that gives satisfaction to the worker. In other words, the economic agent participates in the labor market because the salary provides utility and the latter increases as the salary increases. Indeed, several studies have shown that a worker’s satisfaction does not depend only on his salary [1, 2, 3, 4, 5]. These authors have shown, for example, that nonmonetary factors such as holidays with pay, vocational training, and union membership are also factors that contribute to the job quality and thus may influence the worker’s level of satisfaction. Several measurement approaches are used to analyze job quality. Basically, these methods can be classified into two groups. The first group offers a subjective analysis where workers are made evaluators of their job quality levels. This method uses the levels of satisfaction that workers report as an indicator of job quality. It is therefore possible to directly analyze workers’ responses to questions about the degree of importance given to safety, wages, personal fulfillment, or work schedules [6]. The second group relies on the use of data to construct job quality indices [7, 8, 9]. This approach is described as objective in the literature.

Work is a particular good that has attributes that provide different benefits to the worker. These attributes are the type of work contract, hours of work, salary, and extra-salary benefits. A standard employment relationship is governed by contract. But in the context of developing countries where the informal sector is an important part of economic activities, the employment contract is often nonexistent most of the time. And yet, the majority of benefits and job arrangements are governed by the employment contract. It is therefore of great importance in job satisfaction. Job satisfaction can vary significantly according to some characteristics of the worker such as sex. By the way, several studies have concluded that women are generally more satisfied than men [10, 11, 12, 13]. For Lévy-Garboua et al. [14], satisfaction expresses an experienced preference of current employment against available job alternatives. Clark [15] suggests that women in general have fewer requirements than men, which explains their higher level of job satisfaction.

This study takes a labor analysis approach that goes beyond the neoclassical framework because it is based on the assumption that only salary does not provide satisfaction. In this sense, this study examines the effect of the employment contract on job satisfaction on the one hand and the effect of gender on job satisfaction on the other hand. The rest of the study is organized in four sections. The second section presents the review of the literature, the third presents the methodology, the fourth presents the results and the discussions, and the last concludes the study.

2. Literature review

We summarize the literature with emphasis on three important aspects: (i) the role of nonwage aspects in the labor supply, (ii) the relationship between work contract and satisfaction, and (iii) the link between heterogeneity of preferences and job satisfaction.

2.1 Beyond the neoclassical model

The salary is no longer the only element of satisfaction of the worker. The unemployed say they are less happy than the workers, but this difference in satisfaction is not only explainable by wages. Beyond wages, there are other attributes that provide usefulness and satisfaction to workers.

2.1.1 The Kunze and Suppa labor supply model: the role of nonwage aspects

The model begins with the Lancaster Demand Features approach [16]. The idea of Lancaster [17] is that it is not the goods that produce utility to the consumer but the characteristics that the goods possess (see, e.g., the works of [18]). The good can have more than one characteristic, and the same characteristics can be obtained from more than one good. It is assumed that the characteristics are objective and measurable, for example, a meal provides (i) a certain amount of calorie, (ii) a nutritional composition, and (iii) esthetic characteristics. The technical relationship that transforms goods into their characteristics is called consumer technology. Utility or preference is supposed to classify characteristics, while all goods are only classified indirectly using the characteristics they possess. Formally, Lancaster’s consumer program is as follows:

$$Max\mathrm{U}\left(\mathrm{z}\right)$$ E1

$$\mathrm{z}$$

$$\text{subject to}:\mathrm{px}\le \mathrm{k}$$

$$\text{with}:\mathrm{z}=\mathrm{\Phi x}$$

$$\mathrm{z},\mathrm{x}\ge 0$$

where U (z) is a standard neoclassical utility function defined in the feature space (C-space). The budget constraint px ≤ k is defined on the property space (G-space). The system of equations z = Φx represents the transformation between C-space and G-space where the matrix Φ describes the consumption technology of the economy. Consumer choices can be either studied in the space of goods (as in traditional analysis) or in the space of the characteristics of goods.

To construct a model that analyzes the relationship between the characteristics of a job and the number of hours of work chosen by the individual, we consider N + 1 goods, that is to say x = (L1, L2, ..., Ln, ...., LN, C). C represents the consumption of good, and each of the other N goods, Ln, represents a specific amount of time used in the consumption of these respective goods. As a result, each Ln is called activity in the following, and the total amount of time used for these activities is normalized to unity. It is further assumed that each activity provides K characteristics, while the consumption of good has a single characteristic. Consumption technology Φ is a format matrix (K + 1) * (N + 1). The labor supply produces any of the K activity-related characteristics, but none of these produces the characteristics of consumption. Consequently, each element of the matrix Φ, denoted φkn ≥ 0, determines the quantity of characteristics k provided by the use of a unit of time for the activity n. There are therefore several activities that provide different combinations of features. The approach developed here is based on two activities: the work activity providing a remuneration, a certain amount of selected characteristics; and a residual activity possibly including leisure, volunteering that produce the same selected quantity of characteristics. From the perspective of an empirical analysis of the couple’s labor supply, we interpret C as the consumption of the family in the following, and we denote L1, L2, and (1-L1) and (1-L2) as the times of each spouse in the work activity and in the residual activity. The vector x can be written as follows:

$$\mathrm{x}=\left({\mathrm{L}}_1,1-{\mathrm{L}}_1,{\mathrm{L}}_2,1-{\mathrm{L}}_2,\mathrm{C}\right)$$ E2

By limiting the theoretical analysis to a characteristic of work by spouse and normalizing some coefficients to the unit, the consumption technology Φ is finally of the following form:

$$\left(\begin{array}{l}{\phi }_{1}1000\\ 00{\phi }_{2}10\\ 00001\end{array}\right)$$ E3

The parameters φ_i (i = 1, 2) measure the productivity of the work with respect to the residual activity. Since (1-L1) is the combination of several distinct activities, we assume that φi ˂ 1 (i = 1, 2). This indicates that there is at least one activity that is more productive to achieve this characteristic. Using Eq. (3), the second constraint of Eq. (1) can be finally written as follows:

$$z=\mathrm{\Phi }{X}^{\text{'}}=\left(\begin{array}{l}{\phi }_1{L}_1+\left(1-L_1\right)\\ {\phi }_2{L}_2+\left(1-L_2\right)\\ C\end{array}\right)$$ E4

We start from a Cobb-Douglas utility function. Here is the problem:

$$\begin{array}{l}U\left(z\right)={\alpha }_{1}ln\left(z_1\right)+{\alpha }_{2}ln\left(z_2\right)+\left(1-{\alpha }_1-{\alpha }_2\right)ln\left(z_3\right)\end{array}$$ E5

$\text{subject\hspace{0.17em}to}C=w_1{L}_1+w_2{L}_2+R_1+R_2$

$z=\mathrm{\Phi }{X}^{\text{'}}$

where x and Φ are not given the previous equations. Ri (i = 1, 2) is the nonwage exogenous income of the individual. This income alone is not important, but what is it is the sum R = R1 + R2. The utility function is increasing in ${\mathrm{\phi }}_{\mathrm{i}}$ (i = 1, 2) which means that $U_{{\mathrm{\phi }}_{\mathrm{i}}}$ ˃ 0. The implication is that an improvement in quality translates into an increase in the utility level of the individual. Solving the problem gives:

$$L_i^{\ast }=\frac{1-{\alpha }_i}{1-{\phi }_i}-\frac{{\alpha }_i}{w_i}\left\{\frac{w_j}{1-{\phi }_j}+R\right\}$$ E6

with i = 1, 2; j = 1, 2 et i ≠ j.

Note that the labor supply function $L_i^{\ast }$ is decreasing with respect to the level of nonwage income R and also in relation to the income level of spouse $w_j$ . On the other hand, this function is increasing with respect to the level of income ${\mathrm{\phi }}_{\mathrm{i}}$ of the individual and in relation to the level of quality of work. The main implication of this discrete choice model is that it shows the importance of the nonpecuniary aspects of employment. This is because the elasticities of the labor supply are significantly different, given the characteristics of the work. This model can then be used as a basis to show that the worker’s satisfaction goes beyond the only salary aspect.

2.2 Hours, contract, and job satisfaction

There are several studies in the empirical literature that show that work hours have a negative effect on the job satisfaction level [12, 13, 15]. These studies, after controlling for work income bias, obtain negative effect of working hours on the job satisfaction level. This result can be explained by the fact that the work is in fact a disutility and that it is desired only for the benefits it provides. Dawson and Veliziotis [19] found that fixed-term workers report a lower level of subjective well-being than workers with indeterminate employment contracts. The explanation of these authors is that workers who have permanent work contracts have their state of well-being decreased mainly through their increased feeling of insecurity at work.

This result is in line with that found by Zanden [20]. The idea is that the longer people work in a job, the more they feel more secure and therefore postpone a higher level of well-being. It is in this sense that Chadi and Hetschko [21] examined the role of flexicurity with a view to compensating for job security for workers with temporary employment contracts. Green and Heywood [22] showed that while part-time work is an important negative determinant of job satisfaction from a safety at work perspective, it can be a positive contributor to other dimensions of job satisfaction. In a study of developing countries, Bóo et al. [23] showed that women’s job satisfaction level is not related to the partial nature of employment. On the other hand, their results indicate that part-time jobs have a negative effect on the level of job satisfaction of men. This indicates that men have a stronger preference for permanent jobs compared to women.

2.3 Heterogeneity of preferences and job satisfaction

Women have jobs that are, in general, less attractive than those available to men. Better still, women are less empowered and more supervised and have little access to career development opportunities relative to men [24, 25]. However, women have a better attitude and appreciation of their jobs than men [12, 13, 26]. According to Hodson [10], there are two possible explanations for this paradox. The first is that women do not evaluate the same aspects of employment. The second is that women see themselves more in the role of housework rather than worker. This means that when they work outside the home, they feel extra satisfaction. According to Clark [15], the difference in job satisfaction is due to the fact that women generally have fewer requirements than men and therefore may have to postpone a higher level of satisfaction than men.

3. Methodology

This study uses data from the School-to-Work Transition Survey (SWTS) database. The survey collected data on the Beninese labor market, focusing on the age group 15–29 in 2012. The survey is representative of the target population at the national level.

3.1 Models of job satisfaction

This study adopted two measures of job satisfaction. The first is the level of overall job satisfaction reported by the workers themselves, and the second is the satisfaction of the same workers in relation to their employment contract.

3.1.1 Equation of the overall level of job satisfaction

The job satisfaction variable is multinomial and has four modalities. These terms range from “very dissatisfied” to “very satisfied” to “somewhat dissatisfied” and “somewhat satisfied.” A level of satisfaction reported can be considered as the achievement of a certain random variable having a probability distribution. In this sense, we can model the level of satisfaction as Kifle and Kler [27] did:

$$\mathrm{y}\ast =\mathrm{x\beta }+\mathrm{\mu }$$ E7

where y* is a latent variable indicating the unobservable level of job satisfaction. X is a matrix of sociodemographic variables, β a parameter vector, and μ the error term. This equation cannot be estimated unless we have a categorical variable whose terms are ordered randomly. Let y be this categorical variable with a set of threshold points Zi. The conditional probability of any observation can be expressed as follows:

$$Pr\left(\mathrm{y}=\mathrm{i}/\mathrm{x}\right)=Pr\left(\mathrm{Zi}-1\le \mathrm{x\beta }˂\mathrm{Zi}\right)$$

$$=Pr\left(\mathrm{Zi}-1\le \mathrm{y}\ast +\mathrm{\mu }˂\mathrm{Zi}\right)$$ E8

where i is in this case the level of job satisfaction reported on a scale of 1–4:

$$Pr\left(\mathrm{y}=\mathrm{i}/\mathrm{x}\right)=Pr\left(\mathrm{Zi}-1-\mathrm{x\beta }\le \mathrm{\mu }˂\mathrm{Zi}-\mathrm{x\beta }\right)$$

$$=Pr\left(\mathrm{\mu }˂\mathrm{Zi}–\mathrm{x\beta }\right)–Pr\left(\mathrm{\mu }\le \mathrm{Zi}-1–\mathrm{x\beta }\right)$$

$$=\mathrm{\Phi }\left(\mathrm{\mu }˂\mathrm{Zi}–\mathrm{x\beta }\right)–\mathrm{\Phi }\left(\mathrm{\mu }\le \mathrm{Zi}-1–\mathrm{x\beta }\right)$$ E9

where Φ (.) is the standard cumulative distribution function.

The probability that an employee will choose a level of satisfaction i given the explanatory variables (x) is the difference between the cumulative distribution function of the normal law evaluated at the point of i (Zi) minus the vector of explanatory variables multiplied by their respective coefficients and the cumulative distribution function of the normal law evaluated in the previous point (Zi-1) minus all the explanatory variables multiplied by their respective coefficients.

It will therefore be estimated as the following equation:

$$\mathrm{yi}=\mathrm{xi\beta }+{\mathrm{\mu }}_{\mathrm{i}}$$ E10

The appropriate estimation technique to the estimation of Eq. (10) is the ordered probit, given the categorical and orderly nature of these modalities.

3.1.2 Equation of satisfaction with the employment contract

The satisfaction variable with respect to the employment contract provides information on the state of satisfaction of the work in relation to his employment contract. It is obvious that in labor markets in developing countries such as Benin, the issue of employment contract is of particular importance, particularly because of the high prevalence of informal sector activities, sector known for its lack of regulation and therefore its lack of respect for the labor code. It appears that in this context, the use of job satisfaction is a relevant measure of job satisfaction. But one could think to remain in a so-called objective process and systematically consider a worker having no employment contract as dissatisfied or less satisfied compared to another who has a contract. But the reality is that it is not always obvious that someone with a part-time contract has less well-being than someone with a full-time contract. Instead of remaining in such an approach and defining criteria of external satisfaction to the worker, we prefer the approach that makes the worker his own evaluator of his level of satisfaction. The variable we use for this purpose seems relevant to us since it takes 1 when the worker declares to be satisfied with respect to his employment contract and 0 otherwise.

The equation of worker satisfaction with respect to his employment contract can be estimated by the probit method given the dichotomous nature of the satisfaction variable. An equation will therefore be estimated of the following general form:

$$\mathrm{Si}=\mathrm{Bi\Sigma }+\mathrm{\epsilon i}$$ E11

where Si is the satisfaction status with respect to the employment contract, Bi is the matrix of the explanatory variables, and εi is the error term.

4. Results and discussions

Table 1 presents the ordered probit estimation results of Eq. (10), while the second model presents the probit results of Eq. (11). These two estimates take satisfaction in employment as explained variable. The goal of two regressions is to test the robustness of our results.

Table 1 presents the estimation results of two equations, which are variable in explaining the level of job satisfaction. The difference between the two equations lies in the measurement of the type of work contract. Indeed, in the first estimate, we measured the contract type by the contract type variable. This variable is dichotomous and takes 1 when the worker is under an undetermined contract (CDI) and 0 if under a fixed-term contract (CDD). In the second equation, the type of work contract was measured by the variable duration of the contract which is multinomial with three modalities. The first category takes into account contracts with a duration of less than 12 months, the second category takes into account contracts from 12 to 36 months, and the last modality contracts longer than 36 months. Despite this difference, the two equations produced the same results.

4.1 Age and job satisfaction

It appears that age has a negative effect on job satisfaction. This means that as the worker gets older, he feels less and less satisfied at work. If we use age as a proxy for work experience, this result means that the most experienced workers are less and less satisfied. We tested the nonlinear relationship assumption, but the results indicate that there is no effect of age on the level of satisfaction. This result could be surprising but can still be explained.

The first explanation is that the data we use is collected on a sample of young workers only. The age range is from 15 to 29 years old. This may not be able to bring out the good effect of age since in this age group, workers are often still far from retirement age or have not yet had enough experience in the profession to assess all facets of their expectations vis-à-vis employment. A second explanation is that as mentioned by Franěk and Večeřa [28], older workers may feel disappointed. Their expectations become limited and experience increases the pressure by factors such as technological developments and adaptations in the work. Clark et al. [29] found that the relationship between age and job satisfaction can be synthesized by a U-shaped curve. Our result does not follow the same direction as [29] but does not contradict it either. Indeed, a U-shaped relationship means that the level of satisfaction drops with age first and then increases. This is the first negative effect we obtained in this study because Clark et al.’s [29] study focused on workers aged 15–60 years and older.

4.2 Work contract and job satisfaction

The permanent employment contract is associated with a higher level of satisfaction compared to the fixed-term contract. This result seems obvious in the sense that a part-time worker is less satisfied than a full-time worker. This difference is mainly due to the job security that the worker feels and that is increased in the case of permanent work contracts. Our result is in line with what is generally found in the literature. For example, Green and Heywood [22], Chadi and Hetschko [21], and Zanden [20] have all shown that there is a significant difference between the satisfaction levels of part-time and full-time workers and that, in addition, a part-time contract is a factor that lowers the level of job satisfaction. The contract of employment as a document defining the essentials of labor relations is the one that gives a status to the work. It helps to make him feel a job security. This has the effect of impacting his level of well-being at work. But in a labor market where the rate of informality is high, not all workers are necessarily entitled to the formal employment contract.

We obtain the same effect when we consider the results of Eq. (2) where the type of contract of employment was measured in duration. Thus, we note that compared to the contract of less than 12 months, the other longer terms of contract are associated with higher levels of satisfaction. It can then be stated that not only is the type of work contract a determinant of the level of satisfaction, but also that workers on short-term contracts are less satisfied than those on long-term contracts.

4.3 Education and job satisfaction

Overall, it appears that compared to workers who have no level of education, other workers report a lower level of satisfaction. This result is explained by the fact that as the level of education increases, the worker has more and more high expectations of employment. By combining this state of affairs with the age group of our study (15–29 years), we can easily understand the effect obtained. By taking the example of an individual with a university level, it is obvious that he enters the labor market at an age that is such that around 29 years old, he is still very little experienced. It is established that early-career workers expect a lot from employment, which leads them to postpone low satisfaction levels. Our results do not agree with those of Pisani [30] who finds that a high level of education is associated with a better state of job satisfaction. According to Bender and Heywood [31] who analyzed the satisfaction of high-skilled workers by gender, women are more satisfied than men in nonacademic occupations.

4.4 Perceived discrimination and job satisfaction

The analysis of discrimination was introduced in economics by Becker [32]. It consists of treating differently two people with identical productive characteristics but having a different observable nonproductive characteristic. Becker considers that some employers, colleagues, or customers have a particular taste for discrimination. As a result, members of the discriminated group receive a lower salary to compensate for this inconvenience. Arrow [33] and Phelps [34] developed statistical discrimination as an alternative to Becker’s vision. It is based on beliefs. If the employee belongs to the employer’s group, the latter knows its true productivity. If the employee belongs to the discriminated group, which the employer does not know, the latter thinks that he is on average less productive than his own group and then pays him according to this average and not according to his real productivity. We see that the effect of the discrimination passes through another factor which may be the salary, career development possibilities, etc. In this study, we examine the effect of discrimination perceived by the worker on his level of job satisfaction. Our results in this regard indicate that when the level of job satisfaction is high, then workers perceive an absence of discrimination, that is, equal opportunities for men and women. This result is also that obtained by Madera et al. [35]. Indeed, these authors have shown that perceived discrimination is related to a low level of satisfaction and a greater desire for job rotation at the worker level.

4.5 Financial situation, salary, and job satisfaction

The neoclassical theory of the labor market establishes that salary is the only variable that determines the level of job satisfaction. In this sense, there is a positive relationship between salary levels and satisfaction. Given the availability of statistics, we used the financial status variable rather than the salary variable. This choice is justified by two reasons. The first is that we assume that the wage is closely related to the worker’s financial situation. The second is that the study borrows approach of subjective evaluation. Thus, the financial situation variable used is an evaluation made by the workers themselves. As a result, it seems to us to be a relevant variable in our analysis.

The results indicate that in both equations, as the worker has a better perception of his financial situation, he feels more satisfied at work. In the empirical literature, the relationship between salary level and satisfaction is often studied. But even if the determinants of wages are well known, the effect of wages on the level of job satisfaction is less known. According to results obtained by Judge et al. [36], the wage level is positively correlated with the level of job satisfaction. This is in line with the effect we have achieved. On the other hand, authors such as Young et al. [37] found that the increase in wages has a negative effect on the level of worker satisfaction. The explanation of these authors is that wages evolve with experience. Thus, the more the worker becomes experienced and gains better and better, he sets even more new goals. Thus, his quest to always reach higher wage levels is an explanation that bases such a salary effect on satisfaction.

4.6 Desire to change jobs and job satisfaction

The variable desire to change jobs as an explanatory variable could possibly be the basis of a possible endogeneity bias if the unobservable factors that determine this desire are the same ones that base the state of satisfaction. But estimating models with and without this variable produced the same overall results. We note that the desire to change jobs has a negative influence on job satisfaction. It is obvious that when a worker wants to change jobs, either he is not satisfied with his current job or he has a better opportunity. In both cases, this leads to a low level of job satisfaction (Table 2).

	(1)	(2)
Variables	Job satisfaction	Job satisfaction
Age	−0.058*; (0.031)	−0.141***; (0.048)
Sex (reference: woman)	−0.266; (0.262)	−0.163; (0.424)
Marital status (married or not)	0.343; (0.223)	0.132; (0.368)
Want to change job	−1.237***; (0.244)	−2.131***; (0.318)
Type of contract (full time or part time)	0.378*; (0.207)	—
Contract duration (reference: less than 12 months)
12–36 months	—	0.785*; (0.406)
36 months and more	—	0.772**; (0.368)
Financial status (reference: difficult)
Rather difficult	0.766***; (0.293)	1.030**; (0.402)
In the national average	1.133***; (0.307)	1.259***; (0.420)
Rather easy	1.258**; (0.532)	3.134***; (0.969)
Easy	2.483***; (0.827)	8.676***; (0.688)
Education (reference: no level)
Primary	−0.919***; (0.305)	−2.252***; (0.528)
Vocational	−0.658; (0.409)	−0.875; (0.708)
Secondary	−0.331; (0.294)	−1.754***; (0.547)
Higher vocational	−1.241***; (0.427)	−2.871***; (0.670)
University	−1.116***; (0.375)	−1.801***; (0.628)
Postgraduate	−0.970**; (0.428)	−2.599***; (0.738)
Discrimination (reference: more chances for men)
Equal opportunities	0.508**; (0.252)	1.049***; (0.399)
More chances for women	−0.103; (0.355)	−0.864; (0.558)
Want to work more	0.207; (0.237)	—
Observations 1	379	179

Table 1.

Determinants of the level of job satisfaction.

¹

The number of observations has risen to 179 because of the contract duration variable which has a small number. As a result, we did not include in the second estimate the variable want to work more so as not to lose more degrees of freedom.

Source: Estimate from the SWTS database.

Variables	Coefficients	Margins
Age	0.019; (0.029)	0.005; (0.008)
Sex	0.494**; (0.244)	0.143**; (0.069)
Marital status	−0.283; (0.243)	−0.082; (0.069)
Paid vacation	0.760***; (0.276)	0.22***; (0.077)
Insurance (reference: do not know)
Unlikely	0.244; (0.603)	0.083; (0.199)
Probable but not certain	1.007*; (0.583)	0.357*; (0.19)
Very probable	1.575***; (0.603)	0.518***; (0.191)
Observations	394	394

Table 2.

Gender and employment contract satisfaction.

4.7 Gender and satisfaction with the contract

In contrast to the result obtained with regard to job satisfaction, it is clear here that sex has a significant influence on job satisfaction. Indeed, we note that women are more satisfied with the employment contract than men. Empirical work often shows that women are more satisfied than men in most aspects of job satisfaction. Kifle and Desta [11] showed, for example, that men are more satisfied with hours of work and career opportunities, while women are more satisfied with relationships with their colleagues and with their contribution to the society. In order to test the validity of the effect we obtained and to test its robustness, we proceeded to tests of comparison of proportions according to the sex on the two measures of satisfaction which we adopted. To do this, we recoded the global satisfaction variable. Thus, we considered all those who reported satisfaction levels 1 and 2 as satisfied and who were satisfied with those who reported levels 3 and 4. This allowed us to have a binary variable just like the other variable with the satisfaction with the contract to test sex ratio comparisons. The results indicate that, in general, women report being more satisfied than men. This result is in line with what is often found in the empirical literature.

4.8 Extra-salary benefits and satisfaction with the employment contract

Nonwage benefits are an important part of workers’ evaluation of their jobs. The Kunze and Suppa [16] model has helped to show the importance of nonwage aspects in the labor supply and more generally in the decision to participate in the labor market. Our results indicate that workers have access to these kinds of benefits and they are satisfied. This is a result that is easily understood.

4.9 Job retention insurance and satisfaction with the contract

A worker who does not have the assurance of keeping his job is in a situation of insecurity. Such a situation must have a consequence on his level of satisfaction. In examining the effect of such a variable, we have noticed that as the worker is assured of keeping his job, he is satisfied with his employment contract. And better, the results indicate that the effect is gradually increasing until the worker reaches the point where he has the certainty of keeping his job.

5. Conclusion

This paper examined the effects of work contract and gender on job satisfaction in Benin. Satisfaction was measured using two variables: overall satisfaction level and satisfaction with the work contract. The effects were obtained by estimating two models. The results indicate that the level of job satisfaction is sensitive to the type of contract. This result was obtained by evaluating the work contract by its type (permanent contract or fixed-term contract) and by its duration. In both cases, the results indicate the same effect. Thus, workers with a long-term work contract are associated with high levels of satisfaction. In terms of satisfaction with the employment contract, it appears that the job retention insurance improves the level of satisfaction of the worker. In terms of gender, we noted a lack of difference in effect in estimating the equation of the overall level of satisfaction. But there is a difference of effect when one considers the satisfaction with the contract of employment. This difference is favorable to women. In other words, women are more satisfied with their employment contracts compared to men. This effect has been confirmed by a set of proportional comparison tests that go in the same direction. In addition to these results, we also found that nonwage benefits improve worker satisfaction.