4.2. Discussions
The following table (
Table 5) presents the results of the estimations of the binomial regression model, as well as those of the odds ratios and the marginal effects.
The value pseudo
McFadden of the estimated model (0.6889) suggests that the model with explanatory variables adjusted to 68.89 % is better with the data than the model which is constant. Moreover, the statistical feature
of the likelihood ratio
is significant at 1%, which means that the model predictor is generally appropriate (see
Table 5). At least one coefficient estimated in the model is significantly different from zero.
The odds ratios or relative risk ratios are a measure of association, which measures the link between the characteristic of the company or the individual and the occurrence of the event (Occupation during confinement).
If the odds ratio (OR) = 1, the event
and the characteristic
are independent. If OR > 1 (respectively OR < 1), the link between
and
is positive (respectively, negative). Thus, in
Table 1 above, we observe that several variables negatively influence the chances of having a job during confinement. Generally, the analysis of the odds ratio in this model shows that the individual characteristics of the elderly living in the central region and being remunerated by the dividend (shareholders) or by cash work are significantly and negatively linked to employment during the confinement period. Indeed, we can observe that the OR of age, the central region, shareholder, and cash work are lower than 1. Indeed, this result shows that the fact of being remunerated as a shareholder in cash work and to be in the central region decreases, respectively, by 0.45, 0.11, and 0.29 the chances of holding informal employment during confinement (see
Table 5). That the central region has been the most affected by the COVID-19 pandemic can explain this.
The rise in prices is significantly and positively linked to employment during containment. The results show rising prices increase the chances of having a job in the informal sector. Indeed, many individuals have taken advantage of this price increase to engage in commercial activities such as the sale of “mufflers” taking advantage of the fact that goods are scarce to start a business during the lockdown. This work is in line with the work of [
24,
34,
35] which shows that changes in mobility have changed the working relationships of millions of people around the world, thus leading to the restructuring of jobs.
Not having COVID-19 significantly decreases the chances of having an informal job by 0.07. Indeed, most people who work in the formal must all be negative for COVID-19. Therefore, most of those who have taken the test and declared negative are much more in the formal sector than in the informal one [
25,
26] for whom distancing is more difficult for informal workers than for formal workers, in particular due to space constraints, overcrowding, and an incapacity to meet basic needs.
Regarding satisfaction with barrier measures, we observe that most people who think that the time restrictions are “very bad” and who do not have enough materials for hand disinfection are significantly and positively linked to informal employment during containment. Indeed, the hourly restrictions and the absence of hand disinfection products respectively increase by 10.33% and 56.80% the chances of having an informal job during confinement. Companies with time restrictions and poor management of hand hygiene can explain this will push actors into the informal sector (for example, selling gels, or jobs with fewer time restrictions). These works [
16,
29] reveal that the lockdown due to COVID-19 and other government restrictions has economically affected informal sector workers in Kumasi (Ghana).
The analysis of the marginal effects on having an informal job during confinement presents the following results. About barrier measures, we can observe that the closure of schools and the wearing of compulsory masks reduce the probability of 5.7% and 4.32% of having a job during confinement. This measure seems particularly more restrictive for informal workers who have to work in these difficult conditions, particularly unsanitary conditions, lack of water and even fewer means of protection [
16].
The price increase significantly increases the probability of having a job during containment by 12.1%. Likewise, that wages are not late increases by 13.6% the probability of having a job during the lockdown. Large businesses are more likely to employ than medium-sized businesses during containment. Indeed, the results show that large enterprises have a 54.68% chance of employing in the informal sector than medium-sized enterprises with a 42% chance.
These results are consistent with the work of [
31] which shows that the informal sector is no longer a resilient sector that creates jobs and is a refuge for unemployed workers. On the contrary, it is the large companies that employ individuals, no doubt for the sake of continuity of production. Unless markets are reopened, UPIs use as few employees as possible.
Table 6.
Results of stopping or suspension of use during confinement.
Table 6.
Results of stopping or suspension of use during confinement.
Work during Confinement | Logit Coeff. | Margin Effects | Odds Ratio |
---|
Mode of remuneration | | |
Shareholder | −8.6893 ** | −0.4054 *** | 0.0001 ** |
| (4.3229) | (0.1090) | (0.0007) |
Commission | −3.783 | −0.2646 * | 0.02272972 |
| () | (0.1182) | (0.0675) |
Social distance | | |
Good | −5.1972 ** | −0.3207 *** | 0.0055 ** |
| (2.2080) | (0.0598) | (0.0122) |
School closing | | |
Very bad | −7.9375 ** | −0.4220 *** | 0.0003 ** |
| (3.8772) | (0.1121) | (0.0013) |
Time restrictions | | |
Bad | −9.9729 *** | −0.5618 *** | 0.00004 *** |
| (3.7438) | (0.0849) | (0.0001) |
Public transport restrictions | | |
Bad | 9.855 | 0.4067 ** | 19,055.25 |
| (13.77084) | (0.1934) | (262,406.8) |
Good | 2.4369 | 0.1940 * | 11.438 |
| (1.725) | (0.1106) | (19.7307) |
Mask Obligatory | | |
wear Good | −3.6343 * | −0.2319 | 0.0264 * |
| (1.9125) | (3.5073) | (0.0504) |
Extension of measures | | |
Somewhat strongly agree | 9.3418 *** | 0.4929 *** | 11,405.5 *** |
| (3.2598) | (0.0527) | (37,180.3) |
Not agree | 6.1284 ** | 0.3273 | 458.7219 ** |
| (3.0606) | (0.0970) | (1403.96) |
Hardening measures | | |
Somewhat agree | −4.7615 ** | −0.2838 *** | 0.0085 ** |
| (2.0426) | (0.0531) | (0.0174) |
Extension and hardening | | |
Somewhat agree | −4.7097 * | −0.2908 *** | 0.009 * |
| (2.4364) | (0.0780) | (0.0219) |
Difficulty in supply | | |
No | 7.3563 ** | 0.3792 *** | 1566.04 ** |
| (3.2859) | (0.0460) | (5145.95) |
Shortage of inputs | | |
No | −5.5257 ** | −0.4133 *** | 0.0039 ** |
| (2.6531) | (0.1009) | (0.0105) |
Customer loss | | |
No | 2.6224 | 0.2076 * | 13.769 |
| (2.0766) | (0.1195) | (28.59) |
Salary delay | | |
Don’t know | 10.111 * | 0.2807 *** | 24,624.8 * |
| (4.7163) | (0.0696) | (116,138.8) |
Lower salary | | |
Don’t know | −17.755 ** | −0.5535878 | 1.94 × 10−8 ** |
| (7.1930) | (0.8203) | (1.40 × 10−7) |
There are two ways to evaluate the overall quality of the estimated model. We base the first on its likelihood, the second on the probabilities predicted by the model.
One of the first indicators found in the literature is pseudo
McFadden’s measures how well the model with explanatory variables fits the data compared to the model without explanatory variables. This indicator increases with the (log) likelihood
lnL of the model.
The value pseudo
of McFadden’s estimated model (0.5702) suggests that the model with explanatory variables fitted to 57.02% better the data than the model with constant. Moreover, the statistical feature
of the likelihood ratio
is significant at the 1% means that the model predictor is appropriate (see
Table 6). At least one coefficient in the model is significantly different from zero.
I often present the estimation results of the logit model as the odds ratio (OR) also called the relative risk ratio. OR is a measure of association, which captures the link between the characteristic of the company or the company manager and the occurrence of the event (Stop or suspension during confinement). If OR = 1, the event and the characteristic are independents. If OR > 1 (respectively, OR < 1), the link between and is positive (respectively, negative).
The analysis of the odds ratio in
Table 6, shows that, overall, the suspension of work is significantly linked to six confinement measures. Indeed, social distancing, the closure of schools, and time restrictions and the wearing of the compulsory mask decreased respectively by 0.005; 0.0003; 0.0004 and 0.02, the chances of being suspended while in containment. Compliance with the barrier measures decreed by the government to protect individuals at work can explain this and, therefore, it does not frown upon them and has less chance of being suspended.
Extensions of these measures by the government increase the chances of being suspended from work by 11,405. I justify this since the barrier measures have considerably slowed down economic activity. As a result, extending them will put a lot of workers under arrest. In addition, the lack of inputs and the fact of being a shareholder reduces the chances of being suspended from work, respectively, by 0.003 and 0.0001. Finally, people with fewer difficulties in obtaining food products and catering services have a 1566.04% chance of being suspended during confinement. It is acceptable since it forces most businesses and people in the informal sector who are struggling during lockdown to continue their activities, despite certain restrictions. This situation can cause an amplification of the crisis.
These results are in the same direction as those of [
13] which highlights the role of institutional shortcomings, dangerous working conditions, poor law enforcement, and non-participation of informal workers in decision-making on amplification of the crisis for informal actors in Nigeria.
The analysis of the marginal effects first shows that the fact of being a shareholder, social distancing, the closure of schools, and bad time restrictions decrease respectively by 40%; 32.07%; 42.2%; 56.1% the probability of being suspended from informal work. Then, the extension of the measures increases the probability of being suspended by 49.29%.
These results are consistent with the work of [
5] which shows the significant effect of the pandemic on informal employment globally since the latter makes up most workers in Sub-Saharan Africa.
Effect of loss of employment during confinement (See
Table 7)
Table 7.
Results of loss and termination or suspension of employment during confinement.
Table 7.
Results of loss and termination or suspension of employment during confinement.
Loss of Employment 1 | Logit | Odds Ratio | Marginal Effects |
---|
Age | −4.1116 *** | 0.0163 *** | −0.3373 *** |
| (1.3402) | (0.0219) | (0.0985) |
Region | | | |
Coastline | −3.532 * | 0.0292 * | −0.1619 ** |
| (2.0468) | (0.0598) | (0.0767) |
North-West | 3.5827 * | 35.970 * | 0.3033 ** |
| (1.8446) | (66.35) | (0.1392) |
School closure | | |
Bad | −2.4658 * | 0.0849 * | −0.1753 ** |
| (1.4034) | (0.1192) | (0.0788) |
Restrictions of public transport | | |
poor | 3.3614 * | 28.83 * | 0.2774 * |
| (1850) | (53.34) | (0.1985) |
movement restriction persons | | |
Very bad | 4342 | 0.0130 | −0.2154 ** |
| (3.9787) | (0.0517) | (0.108) |
Stress pandemic | | |
Less stressed | −1.7422 * | 0.1751 * | −0.1378 * |
| (1.0513) | (0.1841) | (0.0755) |
Price level | | |
Rising | −6.2221 * | 0.0019 * | −0.5240 *** |
| (3.5807) | (0.0071) | (0.1605) |
Decreased activity | | |
No | −5.5735 * | 0.0037 * | −0.2328 *** |
| (3. 3608) | (0.0239) | (0.0555) |
Shortage of inputs | | |
No | −4.075 *** | 0.0169 *** | −0.3509 *** |
| (1.4086) | (0.0239) | (0.0873) |
Customer loss | | |
Don’t know | −4.9248 *** | 0.0072 *** | −0.3129 *** |
| (1.8433) | (0.0133) | (0.0836) |
Salary delay | | |
Don’t know | −4.5818 * | 0.0102 * | −0.3316 ** |
| (2.4768) | (0.0253) | (0.1069) |
The value pseudo
McFadden’s estimated model (0.5470) suggests that the model with explanatory variables adjust by 54.7% the data better than the model with constant. Moreover, the Statistical feature
of the likelihood ratio
is significant at 1% means that the model prediction is appropriate (see
Table 7 and
Appendix A:
Figure A1,
Figure A2 and
Figure A3). At least one coefficient estimated in the model is significantly different from zero.
Analysis shows that age is significantly and negatively related to the probability of losing a job in the informal sector during confinement. Specifically, age decreases by 0.0163% the chances of losing one’s job during confinement. This means that being older significantly reduces the risk of losing your job in the informal sector. Likewise, being in the littoral region significantly reduces the chances of losing a job by 0.0292 (see
Table 7). This can be defensible that there is a high concentration of informal sector activities on the coast, and specifically in the city of Douala, the economic capital. The chances of running out of business or losing a job, especially in the informal sector that characterizes this city, are therefore quite low.
Being in the Northwest region significantly and negatively increases by 35.70 the chances of losing a job in the informal sector during confinement. This is due to the security situation in the area; with the pandemic, activities will be more difficult. As a result, it is more likely to have a complete lack of informal activity in this part of the country. Regarding barrier measures, the poor restriction of public transport significantly increases the chances of losing a job during confinement by 28.83%.
Being less stressed during the pandemic reduces the risk of losing your job in the informal sector by 0.17. Likewise, rising prices do not increase the chances of losing a job; it decreases by 0.0019 the chances of losing a job during confinement. Failure to drop-in activities decreases the odds of losing an informal job during lockdown by 0.0037 [
18] estimates confirmed these results that the job losses of informal workers amount to 22.6% compared to formal workers who have only 3.6% of losses. This implies giving more consideration to informal workers.
The analysis of marginal effects presents similar observations. Indeed, age decreases by 33.73% the probability of losing an informal job during confinement. Likewise, those who found the measure to close schools bad have a 17.53% less chance of losing an informal job during confinement. Restrictions on public transport increase by 27.24% the probability of losing a job during confinement. Likewise, being in the northwest region increases the chances of losing a job while in lockdown by 30%.
People who have less stress and those who think there is a rise in prices have a lower probability of losing their job during lockdown (13.78% and 52.4%, respectively). Likewise, access to inputs significantly decreases (35%) the probability of losing a job in the informal sector during confinement. These results are consistent with the work of [
26,
29].