A. Estimated results of the labour force participation models

For model 1, the percentage correctly predicted by this model is approximately 83 per cent, and the area under the receiver operating characteristic curve[1] is 0.9, indicating an excellent, powerful and accurate performance of the model.

Results of this model are displayed in the first two columns of table 3. The results are substantially in accordance with expectations in the economic theory. All coefficients are statistically significant at a 99 or 95 per cent confidence level, except that of the married category, urban/rural variable, second education category and unemployment rate at the governorate level. In other words, being married results in being no more nor less likely to be in the labour force than never marrying. Moreover, living in urban or rural areas has no statistical effect on the probability of being in the labour market.[2] As for education, a person who has a primary degree does not differ from an illiterate one in the probability of participation. It is worth noting that these variables were insignificant in all estimated logit models.

As long as the coefficients of logit models are not readily interpretable, the discussion depends on the reported odds ratios. Regarding the key question of this paper, the results show that being a person with a disability significantly decreases the odds of being in the labour force by 13 per cent, holding all other factors constant.

As for the other variables, not surprisingly, being a female is associated with lower probabilities of participation (the odds ratio is 0.053). This may highlight the conventional household roles or the existing barriers that can impede female labour force participation, confirmed throughout the related literature on the gender gap in labour market. For age, there is an increasing pattern in the odds ratios along the first three categories compared with the age cohort (15–19), then a decreasing pattern appears. Hence, as individuals are getting older, the probability of participation increases until they reach the age cohort (40–49), at which point it starts to decline.[3]

Concerning marital status, individuals who are divorced/widowed are approximately 17 per cent less likely to be in the labour force, compared with the control group “never married”. As for wealth, there is a decreasing pattern in the corresponding odds of participation along the richer wealth quintiles compared with the poorest one. For example, going from the poorest quantile to the richest one is associated with a decrease in the odds of being in the labour market by approximately 31 per cent.

Education is especially noteworthy since it is a powerful factor affecting the probability of participation in the labour market. As results reveal, there is an increase in the odds ratios at higher levels of education, implying that a higher education level corresponds significantly with a higher probability of participation in the labour market. The likelihood of being in the labour force among individuals with tertiary education is approximately multiplied by 5 compared with individuals who are illiterate. On the other hand, being in preparatory or general secondary education significantly reduces the likelihood of being in the labour force compared with being illiterate. This may be due to the fact that once children enrol in school, they are more likely to continue education rather than enter the labour market. In sum, investment in education seems to have a positive influence on the individual’s participation in the labour market.

Furthermore, the results reveal that individuals who are the heads of households are three times more likely to participate in the labour market than those who are non-heads of households. Moreover, the greater the household size, the higher the probability of the person entering the labour market.

According to the literature, the influence of disability on the probability of participation may depend on the magnitude of other socioeconomic factors. Therefore, to investigate how disability interacts with these variables, interaction terms are constructed between disability status and each of the other socioeconomic variables, following El-Saadani and Metwally (2019). The results of model 2 show that only the interactions between disability and gender, and disability and status of the individual in the household, are statistically significant at p-value<0.01. Moreover, the likelihood ratio test (p-value=0.0000) indicates that the model with one or both interaction terms fitted significantly better than the baseline model. This means that the relationship between participation and disability probably depends on gender and household status of the individual. For both persons with and without disabilities, being a female reduces the odds of participation in the labour market (the odds ratios are 0.1 and 0.05, respectively). On the other hand, among males, disability significantly decreases the likelihood of participation in the labour market by approximately 43 per cent. However, being a woman with disabilities significantly increases the odds of participation by approximately 20 per cent compared with women without disabilities. These estimates indicate that the penalty for disability is substantial for males compared with females. Such a result is not surprising since it is even confirmed by the raw data, which indicate that the participation rate of females in the labour market with disabilities (23.7 per cent) is higher than that of their peers without disabilities (20.7 per cent). It is also confirmed by the descriptive analysis conducted by Sieverding and Hassan (2019).

When we insert the other interaction term, the results reveal that being the head of household increases the odds of participation in the labour market for the person with a disability (the odds rations are 2.2) and more for the person without a disability (the odds ratios are 3.3). On the other hand, among males who are non-heads of households, disability significantly decreases the probability of their participation by 27 per cent.

The result that disability reduces the likelihood of participation entails a question concerning the impact of disability types on the odds of participating in the labour market. Hence, rather than using one dummy variable denoting the presence of a disability, four dummies are created for disability in mobility, hearing, seeing and, for the final dummy, in cognition, self-care, or communication.

The results of model 3, presented in table 3, show that having a disability in mobility, cognition, self-care or communication significantly reduces the likelihood of participation. Having a mobility disability is associated with a decrease in the likelihood of participation by approximately 28 per cent, compared with other persons with disabilities or persons without disabilities. Furthermore, disability in cognition, self-care or communication tends to decrease the odds of being in the labour force by approximately 13 per cent. Of note is the fact that the variables denoting a disability in hearing and seeing are not statistically significant. This means that these types of disability do not have a statistical impact on the probability of participation in the labour market. It is worth noting that the results of the other variables are generally robust since they remain substantially unchanged over different models.

Source: Calculated by the authors using ELMPS (2018).