a) Source of Data
The data used in this study came from the Afrobarometer which is a network of researchers working in 36 African countries. The main objective here, is to collect primary data on governance and living conditions. In each country, there is a national partner who is in charge of data collection. There are three regional partners that assist the national partners, namely : Center for Democratic Development (CDD) in Ghana, Institute for Development Studies (IDS) of University of Nairobi in Kenya and Institute for Empirical Research in Economic Policy (IEREP) in Benin.Two institutions ensure training for network members. Michigan State in USA University and University of Cape Town in South Africa surpervised the surveys. The countries involved are
Southern Africa: Botswana, Lesotho, Malawi, Namibia, South Africa, Zambia, Zimbabwe, Mozambique, Madagascar, Mauritius, Swaziland, Losotho.
West Africa: Ghana, Mali, Nigeria, Cape Verde, Senegal, Burkina Faso, Liberia, Niger, Togo, Sierra Leone, Guinea, Ivory Coast.
East Africa: Tanzania, Uganda, Kenya,
Central Africa: Cameroon, Gabon, Burundi, São Tomé and Príncipe.
North Africa: Algeria, Egypt, Moroco, Tunisia, Sudan.
Each survey is usually conducted using a representative national random sample that fulfils certain criteria listed below. The survey sample is conceived as a cross-section which is representative of all eligible voting-age people in a selected nation. The pu rpose of this study is to provide an equal and known chance permitting to select each individual adult citizen for the survey. We shall attain our objective by implementing a sampling design with sampling probability that is proportional to the population size in such a way as to guarantee, as far and as practicable as possible. Most geographically larger entities have a greater likelihood of being adequately represented in the survey sample.
The representative base normally consists of all individuals 18 years and older. The normal procedure is to exclude people living in institutional settings such as students in dormitories, patients in hospitals and individuals in prisons or convalescent homes. In some cases, we also need to exclude individuals living in zones considered unreachable because of conflict or insecurity. Such exclusions are noted in the technical background report that comes with each dataset. The typical sample Size is 1,200 or 2,400. A random sample of 1,200 enables inferences to be drawn on national populations of individuals with a sampling margin of error of ± 2.8 per cent at the 95 percent reliability level. At a sample size of 2,400, the sampling error is reduced to ± 2.0 per cent at the 95 percent confidence level.
The sample is random, clustered, stratified to several degrees. Specifically, the survey is first of all stratified by the main population' social features, including the country's sub-sector (state, province, region, etc.) and place of residence (urban or rural). The geographic stratification decreases the chance that particular language or ethnic communities will be omitted from the survey sample. Sometimes, Afrobarometer deliberately outperforms some of a country's politically significant population to ensure that the small sub-sample size is sufficiently high to be analytically valid. Over-samples will be noted in the technical information report that accompanies each dataset. The sampling is then carried out in four or five steps: In the first stage: In rural areas only, secondary sampling units (SSU) are drawn. SSUs are not used in urban areas. At the second stage: The second stage is the random selection of primary sampling units (PSUs). In the third stage: The sampling departure points (SSPs) are selected randomly. In the fourth stage: Interviewers randomly select households. Fifth stage: finally, an individual respondent within each household is randomly selected.
Table 1 : Variables used
Variable
|
Survey question
|
Measure
|
Bribe for |
treatment at |
public clinic |
or hospital
|
How many times have you had to pay a bribe, give a gift or do a service to a medical staff member or the staff of a public hospital in order to receive the health services you wanted?
Never 0
Once or twice 1
A few times 2
Often 3
No conctact 4
Don't Know 5
|
0: if never
1: if once or twice, a few times, often
No contact and don't know are excluded
|
Education
|
What is your highest level of education?
|
1: No formal education (reference variable)
2: primary education
3: secondary education, first cycle,
4 : secondary education, second cycle,
4: Post secondary education
|
Age
|
How old are you?
|
Log (age)
|
Religion
|
What is your religion?
|
Catholic
Protestant
Muslim
Others
|
Area of residence
|
Wthat is your area of residence?
|
0 Rural
1 Urban
|
Gender
|
Respondent's gender?
|
0 Male
1 Female
|
Professional situation
|
Do you have a job that pays a cash income?
|
0 No job (unemployment)
1 Yes, part time
2 Yes full time
3 Student
|
Non monatary poverty
|
Which of these things do you personally own/use?
1 if Radio 0 Otherwise
1 if Television 0 Otherwise
1 if Motor vehicle or motorcycle, 0 Otherwise
1if Mobile phone 0 Otherwise
1 if Internet access, 0 otherwise
1 ifelectricity connection at home , 0 Otherwise
1 if water connection at home, 0 Otherwise
1 if Toilet in the household, Otherwise
|
Thing= Xi
|
b) Education and bribes in public hospitals : Econometric model
In this study, we want to establish a link between the level of education and the payment of bribes in public hospitals with an Instrumental Variable Probit. The dependent variable is binary and takes the value 1 if the patient has paid the bribe in the public hospital at least once and zero otherwise. The question is whether the probability of paying bribes in African public hospitals increases with education level.
Among the control variables, there is an indicator of non-monetary poverty (see table above). This variable is used to assess the influence of poverty level on the probability of paying bribes. Several studies show that poverty has a significant influence on corruption. In fact, the influence of consumer revenue on the corruption risk is non-linear. This is due to the fact that the vulnerability to corruption rises with income for the lowest quintile of the population due to an excludability effect, but decreases with household income beyond a certain threshold. This indicates the fact that although the risk of corruption falls with income, the lowest income groups have a much lower chance of paying bribes because they cannot or are unwilling to do so (Justesen et al., 2014. Hamelin, 2020). Yet, according to Gundlach & Paldam (2009), corruption is a poverty-related illness that decreases when states develop. The causality is therefore mainly related to low income. In this study, we used a non-monetary poverty indicator to assess the influence of poverty on corruption in Africa. The main motivation for paying bribes is to obtain access to basic public services, such as security, education, justice, connection to the public water network, connection to the public electricity network, official price reductions, civil service recruitment, permits and licences, or legal enforcement of contracts. Bribes related to the access to services on a routine basis are operationally similar to the provision of protection against adverse shocks to the access to public services. In other words, the type of corruption potentially inherent in these situations is extortionist in nature, not collusive as is the case, for example, in situations where officials pay bribes to avoid tariffs and regulations or otherwise access informal or extra-legal treatment. It is also 'corruption without theft', as defined by Shleifer & Vishny (1993), because bureaucrats demand bribes to enable access to public goods and services for which households have already paid taxes and charges. Bribery at street level is thus unilateral in the sense that it only gives the individual bureaucrat an advantage as compared to a corruption-free situation. As in any standard insurance model, willingness to pay - which in our context is equivalent to a willingness to pay bribes increases in terms of risk and income. The non-monetary poverty variable is therefore endogenous.
We also observe that countries in the sample are not geographically located in the same area and do not have the same cultures. Therefore, dummy variables will be used for the following regions: West Africa, East Africa, North Africa, Southern Africa and Central Africa. In fact, two theories have shown the importance of geographical position and legal origin on the quality of institutions and thus of corruption. The theory of law and finance of La Porta et al (1999) emphasized legal origins as a determinant of the quality of institutions. This approach highlights in particular the differences between the French civil law code and the English common law code. The second theory developed by Acemoglu, Johnson & Robinson (2001) emphasized colonial origins as a determinant of institutional quality. Because of climatic factors, settlers developed two types of institutions in former colonies, the good ones where climatic factors were favourable to them as in the United States.
The patient's characteristics such as age, gender, religion, place of residence and household size, and employment status were also used. Age was used to determine whether younger patients pay more bribes than older patients. It is generally assumed that the incidence of corruption is relatively high in cities because of the presence of public services. This study will test this hypothesis in the health sector. The study will also identify the type of religion that causes more corruption in public hospitals and assess the influence of the institutional sector on the probability of paying bribes in the public hospitals.
c) Descriptive analysis
The descriptive statistics presented in Table 2 shows that more than half of the African population (61.38%) is being cared for in public hospitals. Of these patients, 7.99% pay bribes once and or twice, 3.95% pay several times and 2.71% pay often (table 2). In the case of public hospitals, the incidence of corruption is estimated at 14.65%. In regard to the education level analysis, Figure 1 shows that there is an inverted U-shape between corruption in public hospitals and the education level of patients. The probability of paying bribes in public hospitals increases with the level of education up to a certain threshold (the first years of university studies) and then starts to decrease. The probability that a postgraduate will pay bribes in public hospitals is 6.09%, almost the same probability is observed at the level for an illiterate patient who has no formal education (5.98%). This probability reaches the maximum for patients who have completed one or two years of university studies. It should be noted that secondary education graduates are the most numerous in our sample, 36.14% compared to 28.9% for those with primary education and only 15.79% for those with higher education. 18.97% of Africans have not received any formal training. With regard to non-monetary poverty, 42.34% of Africans have less than half of the assets listed in the measurement of monetary poverty, 64.8% have five and (see specification of these assets in Table 1).
Table 2 : Descriptive statistics
|
|
Freq.
|
Percent
|
Cum.
|
Pay bribe for treatment at public hospital
|
Missing
|
742
|
1.38
|
1.38
|
Never
|
29,153
|
54.05
|
55.43
|
Once or Twice
|
2,471
|
4.58
|
60.01
|
A Few times
|
1,486
|
2.76
|
62.76
|
Often
|
930
|
1.72
|
64.49
|
No contact
|
19,01
|
35.25
|
99.73
|
Don't know
|
142
|
0.26
|
100.00
|
Education
|
No formal education
|
10,223
|
18.97
|
18.97
|
Primary
|
15,574
|
28.90
|
47.87
|
Secondary
|
19,474
|
36.14
|
84.00
|
Post-secondary
|
8,509
|
15.79
|
99.79
|
Don't know
|
111
|
0.21
|
100.00
|
Religion
|
Catholic
|
18540
|
35.16
|
35.16
|
Protestant
|
13333
|
25.28
|
60.45
|
Muslim
|
15623
|
29.63
|
90.08
|
Other
|
5122
|
9.71
|
99.79
|
Don't know
|
106
|
0.20
|
100.00
|
Residence
|
Urban
|
22689
|
42.06
|
42.06
|
Rural
|
31,246
|
57.93
|
98.89
|
Gender
|
Male
|
26,801
|
49.69
|
49.69
|
Female
|
27,134
|
50.31
|
100.00
|
Professional situation
|
Missing
|
34
|
0.06
|
0.06
|
Do not look for a job s
|
20,221
|
37.49
|
37.55
|
Unemployment
|
12,503
|
23.18
|
60.74
|
Yes, part time
|
6,424
|
11.91
|
72.65
|
Yes, full time
|
14,543
|
26.96
|
99.61
|
Don't know
|
210
|
0.39
|
100.00
|
Index of non monetary poverty
|
0
|
2
|
0.00
|
0.00
|
1
|
506
|
0.94
|
0.94
|
2
|
3,492
|
6.47
|
7.42
|
3
|
7,408
|
13.74
|
21.15
|
4
|
11,427
|
21.19
|
42.34
|
5
|
12,116
|
22.46
|
64.80
|
6
|
12,463
|
23.11
|
87.91
|
7
|
6,266
|
11.62
|
99.53
|
8
|
255
|
0.47
|
100.00
|
Source : Authors
Below, the author presents the rate of corruption by institutional sector in Africa. Among the sectors selected, justice is the most corrupt (31.46%), followed by the police (29.92%), water and electricity (21.46%), transport (19.82%), health (16.57%) and education (14.65%). In fact, African leaders of the justice institutional sector are not under the control of civil society, and not even of the executive power. They have a power that they exercise over other citizens, even though they are not themselves good examples in society. The situation is the same with the police, which have coercive power over citizens, i.e. they exercise the authority of the state over citizens. As far as health care is concerned, corruption in public hospitals is usually the consequence of queues. In fact, the physicians assigned to public hospitals usually have two jobs, one public and the other private. They spend more time in private hospitals where they are often well paid compared to the salaries they receive in public hospitals. Because of this rationally, but opportunistic behaviour, patients are often forced to pay bribes for their consultations. The phenomenon is further worsened with the qualification of physicians in specific areas. In fact, in some African countries, there are fewer specialists such as dentists, urologists and ophthalmologists. In addition to this problem of qualified doctors, the lack of public hospitals in some large African cities makes the supply of health care largely insufficient with respect of demand.
Table 3 : Paiement of bribes by institutional sector
|
services
|
Freq.
|
Percent
|
Freq.
|
Percent
|
Rate of Corruption
|
Public School
|
Missing
|
446
|
0.83
|
|
|
14,65%
|
Never
|
19689
|
36.51
|
19689
|
85,35%
|
Once or Twice
|
1844
|
3.42
|
1844
|
7,99%
|
A Few times
|
911
|
1.69
|
911
|
3,95%
|
Often
|
624
|
1.16
|
624
|
2,71%
|
No contact
|
30137
|
55.88
|
|
|
Don't know
|
284
|
0.52
|
|
|
Public Hospital
|
Missing
|
742
|
1.38
|
|
|
16,57%
|
Never
|
29153
|
54.05
|
29153
|
83,43%
|
Once or Twice
|
2471
|
4.58
|
2471
|
7,07%
|
A Few times
|
1486
|
2.76
|
1486
|
4,25%
|
Often
|
930
|
1.72
|
930
|
2,66%
|
No contact
|
19,01
|
35.25
|
|
|
Don't know
|
143
|
0.26
|
|
|
Transport
|
Missing
|
439
|
0.81
|
|
|
19,82%
|
Never
|
20418
|
37.86
|
20418
|
80,18%
|
Once or Twice
|
3075
|
5.70
|
3075
|
12,08%
|
A Few times
|
1279
|
2.37
|
1279
|
5,02%
|
Often
|
692
|
1.28
|
692
|
2,72%
|
No contact
|
27848
|
51.63
|
|
|
Don't know
|
184
|
0.34
|
|
|
Water and electricity
|
Missing
|
265
|
0.49
|
|
|
21,46%
|
Never
|
9888
|
18.33
|
9888
|
78,54%
|
Once or Twice
|
1304
|
2.42
|
1304
|
10,36%
|
A Few times
|
827
|
1.53
|
827
|
6,57%
|
Often
|
570
|
1.06
|
570
|
4,53%
|
No contact
|
40914
|
75.86
|
|
|
Don't know
|
167
|
0.31
|
|
|
Police
|
Missing
|
262
|
0.49
|
|
|
29,92%
|
Never
|
8474
|
15.71
|
8474
|
70,08%
|
Once or Twice
|
1804
|
3.34
|
1804
|
14,92%
|
A Few times
|
991
|
1.84
|
991
|
8,20%
|
Often
|
823
|
1.53
|
823
|
6,81%
|
No contact
|
41444
|
76.84
|
|
|
Don't know
|
137
|
0.25
|
|
|
Justice
|
Missing
|
126
|
0.23
|
|
|
31,46%
|
Never
|
4162
|
7.72
|
4162
|
68,54%
|
Once or Twice
|
944
|
1.75
|
944
|
15,55%
|
A Few times
|
581
|
1.08
|
581
|
9,57%
|
Often
|
385
|
0.71
|
385
|
6,34%
|
No contact
|
47583
|
88.22
|
|
|
Don't know
|
154
|
0.29
|
|
|
Source : Authors