Investigating the remuneration of health workers in the DR Congo: implications for the health workforce and the health system in a fragile setting

Abstract

The financial remuneration of health workers (HWs) is a key concern to address human resources for health challenges. In low-income settings, the exploration of the sources of income available to HWs, their determinants and the livelihoods strategies that those remunerations entail are essential to gain a better understanding of the motivation of the workers and the effects on their performance and on service provision. This is even more relevant in a setting such as the DR Congo, characterized by the inability of the state to provide public services via a well-supported and financed public workforce. Based on a quantitative survey of 1771 HWs in four provinces of the DR Congo, this article looks at the level and the relative importance of each revenue. It finds that Congolese HWs earn their living from a variety of sources and enact different strategies for their financial survival. The main income is represented by the share of user fees for those employed in facilities, and per diems and top-ups from external agencies for those in Health Zone Management Teams (in both cases, with the exception of doctors), while governmental allowances are less relevant. The determinants at individual and facility level of the total income are also modelled, revealing that the distribution of most revenues systematically favours those working in already favourable conditions (urban facilities, administrative positions and positions of authority within facilities). This may impact negatively on the motivation and performance of HWs and on their distribution patters. Finally, our analysis highlights that, as health financing and health workforce reforms modify the livelihood opportunities of HWs, their design and implementation go beyond technical aspects and are unavoidably political. A better consideration of these issues is necessary to propose contextually grounded and politically savvy approaches to reform in the DR Congo.

Democratic Republic of the Congo, financial incentives, health worker motivation, human resources for health, livelihood strategies, remuneration, user fees

Key Messages

• This research shows how Congolese health workers earn their living from a variety of sources and enact different strategies for their financial survival.

• Doctors are the only cadre to rely on governmental allowances as their main income source, while for other cadres the main revenue is represented by user fees for those working in facilities, and per diems and top-ups for those in Health Zone Management Teams.

• The complex and fragmented remuneration structure of Congolese health workers is likely to affect their motivation and performance, as well as their livelihood strategies, and should be taken into account for health system and health workforce reforms.

Introduction

Human resources for health (HRH) are essential for the provision of healthcare services, but the recruitment, retention and motivation of health workers (HWs) remain a challenge ( Campbell et al. 2013 ). While successful strategies to address HRH issues include both financial and non-financial incentives ( Buchan et al. 2000 ; Willis-Shattuck et al. 2008 ), HW payments remain a core concern, as a precondition for motivation ( Franco et al. 2002 ; Chandler et al. 2009 ). In many low-income settings, the informality of the health sector ( Ensor and Witter 2001 ) and the low salary levels ( McCoy et al. 2008 ; Hernandez-Pena et al. 2013 ) mean that HWs revenues are often not limited to formal incomes (e.g. salaries, governmental allowances and regulated private practice which are prevalent in high-income settings) as HWs face the need to enact a variety of formal and informal coping strategies to ensure their subsistence and the livelihoods of their families ( Ferrinho and Van Lerberghe 2002 ; Van Lerberghe et al. 2002 ; Muula and Maseko 2006 ; Akwataghibe et al. 2013 ). However, in most of the empirical literature, the analysis of HWs remuneration is usually limited to certain incomes, or to the separate study of each revenue, while there is little attention to the entire set of financial incentives available to HWs and their consequences on motivation and performance ( Bertone and Witter 2015 ). Additionally, as pointed out by ( Durham et al. 2015 ), the focus of the debate is typically on HRH in the public sector, while in many settings, and especially in post-conflict and fragile ones, the health workforce goes beyond the state employees and often includes large numbers of unregulated HWs in other segments of the healthcare arena, such as the private (for-profit and non-for-profit) and informal sectors.

In this article, we explore the financial remunerations of HWs in the Democratic Republic of the Congo (DRC). We look at the entire set of potential revenues earned by HWs working in both public and private facilities, as well as those within local health administrations. Incomes include not only official salaries and allowances, but also contributions from external partners, as well as payments from patients, from private practice and non-health-related activities. As shown elsewhere ( Bertone and Witter 2015 ), although rarely undertaken, the analysis of the level and composition of the HWs ‘complex remuneration’, the exploration of the differences between and within cadres and of the drivers of these differences, is essential to gain a better understanding of the motivation of HWs and of their livelihood strategies, which impact on performance and service provision. This is even truer in a rarely studied setting such as the DRC. The DRC is a huge, diverse country, with a complex history of state fragility. One of the historical legacies is the difficulty of the central government to ensure the provision of public services and to govern and control its influential and heterogeneous peripheries ( Herbst and Mills 2009 ). The social, economic, cultural and political landscape that history defined affect the healthcare providers whether civil servants or not, and is essential to understand the Congolese health system, or rather the ‘healthcare service delivery arena’ in its broadest definition. In particular, the difficulties of the state to provide public services via a well-supported and financed public workforce may be reflected in the ‘predation’ strategies of the civil servants ( Rackley 2006 ; Trefon 2009 ).

Given the picture of public underfunding of HRH, in this article we aim, first, to describe what are the components of the income of HWs across DRC and the level of those incomes. Secondly, we analyse the drivers at individual, facility and provincial level of the variation in the HWs’ total income. The analysis of the HWs remunerations allows us to discuss the possible livelihood strategies that the remuneration structure entails and their potential consequences on HWs motivation and performance. Additionally, it provides insights into the organization of the healthcare arena, which are critical to devise tailored actions to address both HRH and broader health system issues.

Health system organization

The Congolese health system is organized around 11 Provinces (soon to become 26 provincial health divisions) and 516 Health Zones (‘districts’ elsewhere). Each Zone includes at least one hospital, and is divided into Health Areas covering about 10 000 habitants and usually including one health centre ( MSP 2010 ). Despite the recent stabilization and economic growth, and the restoration of bilateral and multilateral aid, health indicators in the DRC are extremely weak with maternal mortality estimated at 846 deaths per 100 000 live births and under 5 mortality at 104 deaths per 1000 live births ( MPSMRM MSP & ICF International 2014 ). Health financing is still characterized by low public expenditure (12% of the total health expenditure), high out-of-pocket payments (47%) and the fragmented yet relevant contribution of external aid (37%) ( MSP 2011a ).

Because of the minimal state involvement in health financing and service provision, the private and especially non-for-profit faith-based sector is particularly prominent ( De Lamalle 2004 ; Murru and Pavignani 2012 ). Non-for-profit facilities often hold an agreement with health authorities which recognize them as the service provider in a defined health area, in the absence of a public facility. Additionally, except for a limited number of free preventative services usually funded by external programs (such as vaccination), patients are normally charged user fees for the services they use, whether in public or private for- and non-for-profit facilities. The revenues earned from user fees are spent to purchase drugs, fund recurring costs, small investments and equipment. A portion of the collected fees is paid to the higher level health administration (‘ financement ascendant ’), a practice which has been deemed illegal by the Ministry of Public Health (MoPH), although it is still often happening. The funds that remain are rightfully shared among the staff as a part of their remuneration.

HWs in the Congolese health system

There is no fully reliable source about the number of HWs in the Congolese health system. The MoPH estimates them at 127 716 including clinical and administrative cadres ( MSP 2013 ), but this number is likely to be underestimated given the problems of the HRH information system and includes only public HWs. One of the main challenges for HRH is the uneven distribution of HWs across the country, with a ‘plethora’ of HWs in urban and higher level facilities and a chronic understaffing of rural and primary health centres ( MSP 2011b ; Bertone and Lurton 2015 ). Low education standards, high absenteeism and presence of ‘ghost workers’, and overall low performance of HWs are other key issues ( MSP 2011b ). In the context of the ongoing decentralization process which entails the devolution of HR tasks to the provincial level, substantial reforms in HRH management, including recruitment, deployment, payment and career path have been long discussed by government and donors alike. These reforms have been proposed both in the framework of a civil service reform, as well as specifically for the health workforce, with the creation of a paper-based or electronic management system. While some pilots have begun, no country-wide decisions have yet been taken.

Currently, the employment status of HWs reflects the multi-sectoral organization of the health system ( Durham et al. 2015 ). Some of the HWs are publicly employed civil servants, who may be deployed to work in public or private for-profit or non-for-profit facilities. However, only a proportion of them is included in the MoPH payroll and receives a governmental allowance. A recent study ( World Bank 2014 ) found that 32% of HWs in the civil service receive a salary and 81% are paid a risk allowance. Despite its name, the risk allowance ( prime de risque ) is a governmental payment to which all publicly-employed HWs are entitled, regardless of their area of posting. HWs holding a civil servant status and deployed by the central level coexist with HWs in public or private facilities who are locally recruited ( MSP 2011b ). These workers are not included in the payroll and do not receive salary or risk allowance, but, as all HWs, they can receive a number of other payments, further described in the following sections, paid by the facility (e.g. share of user fees), by external agencies (top-ups and per diems), or earned from activities outside of the facility (private practice or non-health activities).

Methods

The research questions and methodology for this work are based on those delineated in Bertone and Witter (2015 ) and complement a study carried out in Sierra Leone ( Bertone and Lagarde ; Bertone 2015 ).

Data collection

Our analysis is based on a quantitative survey carried out in four provinces of the DRC: Bandundu, Equateur (southern part), Katanga and Sud Kivu. In total, 288 health centres ( Centres de Santé and Centres de Santé de Référence ), 32 hospitals ( Hôpitaux Généraux de Référence ) and 30 Health Zone Management Teams (HZMT, Equipes Cadres de Zone ) were surveyed. The provinces were purposefully selected among those where a health support project of the World Bank operated until 2014 (Bandundu, Equateur and Katanga) and among those where World Bank projects are envisaged in the future (all four provinces). Within each province, eight Health Zones (HZ, Zones de Santé ) were selected (Supplementary Figure 1). The HZ were sampled with probability proportional to size (PPS) sampling, where the size was defined as the number of health facilities in the HZ. As a result selected HZ include some zones supported by the World Bank project and some which are not, except for Equateur where, for logistic reasons, all HZ included in the survey were supported by the World Bank project. This may have led to selection bias in Equateur, as some of our results suggest. Subsequently, ten Health Areas were sampled in each HZ, using the same PPS weighting. In each of the selected HZ, the HZMT and the hospital were surveyed, as well as the health centre of each of the sampled Health Areas. ¹ The survey covered HWs working with the HZMT, in hospitals and health centres. Within each facility, all qualified HWs present on the day of the survey were interviewed, for a maximum of 12 HWs per facility. For the HZMT, the Zonal Medical Officer (ZMO, Médecin Chef de Zone ), the zonal administrator and 2 supervising nurses were included. A total of 1771 HWs were interviewed (Supplementary Table 1). Ethics approval for this research was granted by the Kinshasa School of Public Health (KSPH). The survey was programmed on tablets and carried out in October–November 2014 by four trained teams, each composed of 10 enumerators and 1 supervisor from the KSPH.

Questions to individual HWs regarded initially the sociodemographic characteristics of the respondent, education and current post, title and cadre. Based on contextual knowledge and on the existing literature on DRC and other sub-Saharan settings ( Roenen et al. 1997 ; McCoy et al. 2008 ; Fox et al. 2014 ; Bertone 2015 ), a second set of questions included nine categories of potential income: (1) salary, whether paid by the government or the private employer; (2) risk allowance paid by the government; (3) top-up payments paid by external sources (NGOs, donors or programs), whether fixed or based on a measure of performance; (4) ‘per diems’ payments; incomes realized within the facility such as (5) share of user fees and (6) paid overtime; (7) informal incomes received within the facility, including gifts and under-the-table payments from patients, as well as sale of drugs or services to patients; (8) private practice and (9) income generating activities outside of the health sector. Different recall periods were used for each type of revenue to maximize the probability to get data on irregular types of income (Supplementary Table 2). We used ‘unfolding brackets’ to capture the most sensitive incomes, without the problems found with other indirect questioning techniques ( Akwataghibe et al. 2013 ; Bertone 2014 ). Under the ‘unfolding brackets’ technique, if a respondent does not want to give a precise estimate for a quantitative question, she is asked if the quantity lies within a pre-specified bracket of values. The question is then pushed further presenting the respondent with another bracket of values, determined by her previous answer. Data collected result in a bracket of values within which the quantity of interest lies. Unfolding brackets have proved successful in reducing non-response rates for sensitive quantitative questions ( Heeringa et al. 1995 ; Börsch-Supan and Jürges 2005 ). In our questionnaire, we designed the bracket progression so that the fourth (and last) level of question would include 10 USD-wide brackets for values under 60 USD (or their equivalent in Congolese Francs), while brackets were wider for higher values (the highest closed bracket being 150-200 USD, followed by 201 USD or more).

Data analysis

We first standardized all reported payments into monthly averages to reconstruct the monthly incomes of HWs. For data resulting from unfolding bracket questions, the value of that income component was taken as a random point assuming uniform distribution within the last bracket answered. Total income distribution being highly skewed, we report standardized income of each cadre category using mean distribution of different incomes and median value of overall income. We then modelled the determinants of receiving each income using logistic regressions, and the determinants of the amount received, using standard linear regressions on log-transformed amounts received. All models use a standard specification including, at individual level, gender and age and HW role in the structure. For HWs in facilities, we also included their position within the facility (authority or not), type of facility, facility ownership, the existence of a motivation scheme for workers (fixed or performance-based), rural or urban location and province. For HWs working in HZMTs, we only included the location (urban/rural) and the existence of a motivation scheme for HZMT staff. Thirty-eight models were run to estimate separately the probability of receiving an income and the income amounts for each income, and for HWs in facilities and in HZMTs. Complete results are reported in Bertone and Lurton (2015 ). In this article, we focus on the results obtained by modelling the amount of total income received by each cadre of HW.

Findings

The total remuneration of HWs and its components

The analysis of the survey data reveals a very high variation in total income of HWs and stark differences in total income between cadres with doctors earning seven times more than nurses, and those in administrative positions in facilities earning 1.7-fold the nurses’ income ( Table 1 ). The latter result seems to be driven by the fact that administrators are more likely to receive higher incomes from user fees. Indeed, it is possible that, by being responsible of the redistribution of the user fees among the staff, they tend to appropriate a higher share of the facility revenues. This points to the potential tensions that may exist at facility-level around the fee sharing. Moreover, our results show that HZMT staff earns more than their counterparts of the same cadre working at facility level, mostly thanks to the higher payments received from external organizations (per diems and top-ups). Importantly, the analysis also highlights stark differences in income within cadres, with few individuals earning very high total revenues (as shown by the maximum values in Table 1 ). Revenues from non-health activities seem to be the main driver of the variation of total income between individuals of the same cadre.

Table 1.

Total monthly revenues (median and maximum), by cadre

Cadre	Median (USD)	Max. (USD)
Health facility staff
Doctor	785	4815
Administrator	166	1396
Lab technician	122	1445
Pharmacist/pharm technician	111	704
Nurse	101	2908
Other	90	2142
HZMT staff
Zonal Medical Officer (ZMO)	887	2450
Supervising Nurse (HZMT)	228	1716
Zonal administrator	167	1822

When looking at the components of the total income, our findings confirm that HWs rely on a variety of revenue sources to make ends meet ( Table 2 ). Among those working in facilities, salary and risk allowance represent the main source of income only for doctors, for whom they account for a mere 46% of the total income. For those working in the facility pharmacy, in the laboratory, and for nurses and administrators, the relative importance of salary and risk allowance decrease to 22, 22, 19 and 18% of the total income, respectively. In contrast, for those same cadres, the importance of the user fees shared among facility staff raises to 37, 38, 35 and 48% of the income, thus representing the most important source of revenue. About 39% of the HWs included in the sample had received at least one top-up payment from external organizations (maximum number of top-up accumulated was 3), while 62% received at least one per diem payment in the last three months (highest number of per diem payments received in the last three months was 6). However, despite being perceived by many actors at central level to account for a high proportion of HWs income, top-ups and per diems represent less than one fifth of it, ranging from 17% for nurses to 9% for doctors. While income from non-health activities is a key component of the total remuneration (more than 10% for all categories), revenues generated through private practice account for less than 5% of remuneration for all categories. The relative little importance of private practice incomes could suggest that the lines between public and private activities are indeed extremely blurred, making private practice hardly distinguishable from the user fees charged to patients within the facility. Additionally, 6% of nurses’ income comes from informal revenues, while for all other cadres informal incomes account for a smaller proportion of total income. Due to the sensitive nature of this question, we consider this a lower-bound estimate for informal revenues. However, it seems that an even higher measure of informal revenues would not change the overall picture. Paid overtime was found almost irrelevant, with a few exceptions for some urban hospitals and higher cadre HWs.

Table 2.

Total income (median USD) and contribution to total income of each component (USD and %), by cadre

Cadre	Total income	Salary + risk allowance	Top-up + per diems + overtime	Share of user fees	Private practice	Non-health activities	Informal revenues
Health facility staff
Doctor	785 USD	358 USD	72 USD	219 USD	41 USD	75 USD	20 USD
		(46%)	(9%)	(28%)	(5%)	(10%)	(2%)
Administrator	166 USD	29 USD	18 USD	80 USD	3 USD	31 USD	6 USD
		(18%)	(11%)	(48%)	(2%)	(19%)	(4%)
Lab technician	122 USD	27 USD	19 USD	46 USD	3 USD	21 USD	6 USD
		(22%)	(15%)	(38%)	(3%)	(17%)	(5%)
Pharmacist/pharm technician	111 USD	25 USD	13 USD	41 USD	2 USD	27 USD	3 USD
		(22%)	(12%)	(37%)	(2%)	(24%)	(3%)
Nurse	101 USD	19 USD	17 USD	35 USD	3 USD	20 USD	6 USD
		(19%)	(17%)	(35%)	(3%)	(20%)	(6%)
Other	90 USD	13 USD	14 USD	31 USD	1 USD	24 USD	6 USD
		(15%)	(16%)	(35%)	(1%)	(27%)	(6%)
HZMT staff
Zonal Medical Officer (ZMO)	887 USD	507 USD	154 USD	73 USD	47 USD	95 USD	10 USD
		(57%)	(18%)	(8%)	(5%)	(11%)	(1%)
Supervising nurse	228 USD	57 USD	83 USD	31 USD	7 USD	40 USD	10 USD
		(24%)	(36%)	(14%)	(3%)	(18%)	(5%)
Zonal administrator	167 USD	27 USD	55 USD	45 USD	1 USD	31 USD	8 USD
		(16%)	(33%)	(27%)	(1%)	(18%)	(5%)

Main source of income for each cadre of HW in bold in the table.

The survey results highlight a different situation in the relative importance of each income components for HWs working at HZMTs. While for ZMOs the main source of income are their salary and risk allowance, which account for more than half of their monthly revenues, more than a third of the income of supervising nurses and administrators comes from payments made by external organizations in the form of top-ups (whether fixed or performance-based) and per diem payments for the various activities that they carry out, including training (both as trainees and trainers), supervision, meetings and reporting, and coordination of public health campaigns. Private practice and non-health activities are relatively less important than for HWs in facilities, and similarly the relevance of the revenues from user fees diminishes—although HWs in HZMTs still earn from this component thanks to the contribution exerted from the facilities (‘ financement ascendant ’).

HWs livelihood strategies

The findings above reveal the fragmentation of HWs income and the fact that HWs rely on number of income sources and activities to ensure their livelihoods. The main source of income for 38% of the HWs sampled within facilities are the user fees shared at facility level, which confirms the high reliance of the Congolese health system on out-of-pocket payments. As salaries are paid to a small proportion of staff and no budget contributions are made to facilities, facility managers and administrators in both public or private sector must ensure the revenues of their facilities, and their staff incomes, in an ‘entrepreneurial’ way, by maximizing the fees they legally charge. Importantly, for about 26% of HWs, health is (financially) a secondary activity, as revenues from non-health activities, such as agriculture or farming, small business, trading, and rental of property, vehicle or motorbike, are higher than all others. The relative minor importance of incomes from health activities for those workers may result in increased moonlighting and absenteeism or, in case non-health activities are carried out by others on behalf of the HW, may be detrimental to their commitment to health work.

HWs employed in HZMTs show a heavy reliance on the contribution of external organizations, which can be accumulated. The survey revealed that 60% of HZMT workers had had at least one per diem payment in the past three months (for a maximum of 5), and 93% had received at least one monthly top-up (for a maximum of 4 cumulative top-ups). Anecdotal evidence points to the fact that external organizations are often unaware of the amounts paid by others and HWs can take advantage of the lack of coordination to maximize their income. Activities for which per diems and top-ups are paid are often outside of the routine tasks of HZMT and refer to the particular mission of each organization. It is possible that HZMT staff would choose the activities which allow them to increase their revenues, skewing away from their routine responsibilities and the priorities defined at central level.

Determinants of total income

The diversification and fragmentation of the overall income seems to be driven by differences between HWs which relate to individual, facility and provincial features. Looking at results by cadre ( Table 3 ), it appears that having a position of authority (e.g. being nurse in-charge, or director of a ward or a hospital) allows nurses to earn on average 20 USD more. Higher total income is correlated for most cadres with working in urban areas, as well as with working in bigger facilities, such as hospitals and referral health centres (except for pharmacists/pharmacy technicians). For example, nurses working in referral health centres earn 10 USD more than those in health centres, and 10 USD less than those in hospitals. Finally, working in a privately owned facility (compared to a public or faith-based one) has a significant effect in determining higher income for doctors, nurses, administrators and lab technicians. Interestingly, however, the presence of external organizations supporting the facility has no effect on total income.

Table 3.

Determinants of logarithm of total income amount by cadre

	Health facility staff												HZMT staff
	Doctor		Administrator		Lab Technician		Pharmacist/Pharm Tech		Nurse		Other		Zonal Medical Officer (ZMO)		Supervising Nurse		Zonal Admin.
	Estimate		Estimate		Estimate		Estimate		Estimate		Estimate		Estimate		Estimate		Estimate
	(SE)		(SE)		(SE)		(SE)		(SE)		(SE)		(SE)		(SE)		(SE)
(Intercept)	5.20088	***	3.2996386	***	2.906021	***	2.89194	***	3.431149	***	4.226523	***	−4.92457	***	4.83366	***	2.84499
	(0.76514)		(0.6352639)		(0.444069)		(0.65236)		(0.124132)		(0.358109)		(0.56167)		(0.78243)		(0.97138)
Male	0.42554		0.5985739	*	0.382758	.	−0.24249		0.150676	**	0.206790		n/a		0.50740		−0.37654
	(0.31742)		(0.2663948)		(0.213222)		(0.25173)		(0.053824)		(0.139454)				(0.37410)		(0.47355)
Age	0.01306		−0.0009823		0.013420	.	0.01556		0.010988	***	−0.007315		0.03857	**	0.01855		0.03800
	(0.01602)		(0.0121145)		(0.007631)		(0.01198)		(0.002355)		(0.005426)		(0.01173)		(0.01301)		(0.01996)
Authority w/in facility	−0.37946		n/a		n/a		n/a		0.445332	***	n/a		n/a		n/a		n/a
	(0.23947)								(0.059201)
Urban location	−0.10269		0.5705933	*	0.385310	.	0.44896		0.414219	***	0.543819	**	0.56025	**	−0.14098		0.29194
	(0.23198)		(0.2833799)		(0.209298)		(0.27067)		(0.064360)		(0.199797)		(0.16542)		(0.26409)		(0.39465)
External support	0.09149		0.0030003		0.355368	.	0.12425		0.115795	.	0.048189		12.7139		−0.86683	*	0.72477
	(0.29403)		(0.2820389)		(0.200466)		(0.31959)		(0.068115)		(0.154004)		(49.7977)		(0.37115)		(0.60557)
Ownership of facility (public, private or faith-based)
Private facility	−0.07257		1.2920184	.	0.456010		0.68791		0.319508	*	0.069843		n/a		n/a		n/a
	(0.36888)		(0.6514790)		(0.288132)		(0.66298)		(0.126106)		(0.322735)
Public facility	−0.32985		0.1190185		−0.030626		0.32474		−0.063741		−0.163562		n/a		n/a		n/a
	(0.23559)		(0.2654759)		(0.211477)		(0.23606)		(0.061205)		(0.195536)
Type of facility (CS, CSR, HGR)
CSR	0.04242		0.9378347	*	0.252650		0.23275		0.261529	***	0.534922	**	n/a		n/a		n/a
	(0.31997)		(0.3505423)		(0.239101)		(0.31274)		(0.064286)		(0.167283)
HGR	0.09946		1.0812147	**	0.852332	***	−0.03726		0.513514	***	0.803312	***	n/a		n/a		n/a
	(0.28979)		(0.3244033)		(0.223350)		(0.27502)		(0.068025)		(0.226660)
Province (Equateur, Katanga, S.Kivu, Bandundu)
Equateur	0.63749	.	−0.6736642	.	−0.202527		0.69987		−0.030772		−0.976262	***	−0.01695		−0.06613		0.34066
	(0.35291)		(0.3495732)		(0.305004)		(0.41906)		(0.071755)		(0.278816)		(0.20890)		(0.30559)		(0.48100)
Katanga	0.88203	*	0.4946461		0.833392	***	1.22316	***	0.585765	***	0.480185	*	0.10196		0.98750	**	−0.40622
	(0.33424)		(0.3725344)		(0.233039)		(0.34659)		(0.073626)		(0.207614)		(0.22947)		(0.32917)		(0.67141)
S. Kivu	1.10851	**	0.9085166	*	0.240276		0.93712	*	0.411955	***	0.124827		0.64329	*	−0.12757		0.95272
	(0.35738)		(0.3462614)		(0.250078)		(0.39254)		(0.071737)		(0.209106)		(0.23451)		(0.31672)		(0.45531)

Gender coefficient could not be determined for ZMOs as there were only males in the sample.

Significance codes:

***0.001

**0.01

*0.05

. 0.1

At provincial level, differences in the socio-economic conditions, political and military stability, and health system features are reflected in the individuals’ income. For example, Sud Kivu is characterized by insecurity and therefore by the massive influx of humanitarian actors, while Katanga is a generally wealthier region with higher levels of formal and informal privatization of healthcare delivery. Overall, HWs income is higher in Katanga and Sud Kivu provinces (differences are driven by higher revenues from per diems, top-ups and user fees in both provinces, and from private practice and non-health activities in Katanga). For those working within HZMT, differences in total income by cadre are less evident and the province of employment seems to have the most important effects, in particular for those in administrative positions. It appears that ZHMT administrators in Sud Kivu earn more than their counterpart in other provinces.

Discussion

Our findings shed light over the remunerations and livelihoods of HWs of different cadres and working in different positions across the DRC. Given the quasi-complete absence of salary and risk allowance payments (with a relevant exception for doctors), Congolese HWs earn their living from a variety of sources and enact different strategies for their financial survival, even more so than in other contexts ( McCoy et al. 2008 ; Akwataghibe et al. 2013 ; Bertone 2015 ). In some of the literature analysing the role of the civil service in the DRC, the workers’ livelihood strategies have been labeled ‘governance by predation’ as state agents put in place rather sophisticated methods to obtain money from the population ( Rackley 2006 ; Trefon 2009 ). Indeed, the saying (and practice) for low-level functionaries, police and army since Mobutu’s time has been ‘ population baza bilanga na bino ’ (the population is your provider) ( Rackley 2006 ). In the health sector, in the absence of other viable options from within the public health system, HWs—in an economically rational way—resort to extreme income diversification and, in particular, revenues from patients (if working in facilities) or payments from external organizations (if working in HZMTs) and activities outside of the health sector as their main sources of income. In addition, facility administrators seem to be able to exploit their position to accumulate higher revenues at the expenses of their colleagues, a rather poor indicator of governance and transparency in the fee-sharing system.

Because of the limitation of our study to quantitative income data, we have no possibility of triangulating our results with data on the activities effectively carried out by HWs, and limited qualitative information exists to help us further explore the actual implementation of the livelihood strategies, understand the HWs views on them, and reflect on how rational and cognizant HWs are in their enactment. For example, our data do not allow us to explore the intrinsic value attached to certain income components beyond their face monetary value, while Fox et al. (2014) found that in Katanga receiving a salary is seen by HWs as a recognition of their role as agents of the state. Similarly, Bertone and Lagarde found that the non-financial features of income components (such as payment regularity, reliability, ease of access, etc.) play a key role in defining the income use patterns and the livelihoods of primary HWs in Sierra Leone ( Bertone and Lagarde ). However, the evidence available points out to the fact that the need to enact such strategies is likely to have consequences on HWs motivation and performance. In terms of performance, it could impact on time spent within the facility, activities chosen to undertake and patients’ referral decisions. Regarding the latter, Coolen (2011 ) describes how HWs are often reluctant to refer to higher level facilities in order to keep potential revenues at their level. Little attention is paid to health priorities and needs of the local populations, and even less to the ministerial policies—individual ‘strategies have replaced policy’, as Trefon (2009 :13) notes with reference to civil servants. Secondly, the level and composition of the HWs income and the extreme individual variations may act as demotivation factors. While some of the variation may be justifiable (e.g., higher cadres earn more) or determined by personal circumstances (e.g. property ownership, entrepreneurship, etc.), some differences, such as the strong rural–urban divide and the differences between provinces reveal profound inequalities, which may deincentivize HWs. Additionally, inequalities, and in particular those in favour of urban workers, are of high policy relevance also because they contribute to an imbalanced distribution of HWs, characterized by rural facilities lacking the basic number of qualified personnel, and overstaffed urban facilities, where HWs reportedly take monthly shifts ( World Bank 2014 ).

Looking broadly at the health system organization, our findings confirm the absence of the state and the de facto privatization of the healthcare delivery ( De Lamalle 2004 ). As noted by Murru and Pavignani (2012 ), the public health sector seems to be ‘informalized from within’, as both public and private for-profit and non-for-profit sectors respond to their own dynamics and needs, which results in a pervasive ‘commodization’ of the healthcare provision. This situation has relevant implications for both HRH and health sector reform. Undoubtedly, a better comprehension of the actual livelihood strategies of HWs is a precondition for the design of an effective incentive package, which could contribute to addressing HWs performance as well as distribution issues. This understanding is also essential when considering health system financing reforms, such as the introduction of fee exemptions. Indeed, the relevance of user fees within HWs income calls for caution, in donors, NGOs and government circles alike, when proposing free healthcare policies in order to ensure that the loss of a vital income for facilities and HWs does not correspond to a further informalization of healthcare with the introduction of higher and more widespread under-the-table fees. Similarly, the design of PBF schemes must take into consideration income levels and composition to make sure that performance bonuses are sufficient to compensate the increased effort and must ensure that the scheme’s design does not include elements that may lead to perverse incentives and demotivation, such as the reduction of the overall income ( Huillery and Seban 2015 ). We find that in facilities, top-ups (including PBF bonuses) are relatively marginal in the HWs remuneration and constitute a weak incentive compared to other sources of income. On the other hand, the composition of the income of HZMT staff highlights the detrimental effects of the proliferation of aid and donor-led initiatives. Addressing this situation will require a true effort by aid agencies to forgo their own (vertical) priorities and visibility needs to ensure the harmonization of their programs. As a possible solution, a proposal has been made for the creation of a harmonized ‘ contrat unique ’ (single contract) between the health authorities and all donors, which is being discussed for implementation, although for the moment is envisaged at provincial level only.

Trefon (2010 ) noted that, as public service delivery in the DRC depends on individual strategies often based on personal opportunism, it seems unconceivable that institutional change could be introduced and/or supported by civil servants themselves, as they would lose out if such reforms were successful. Indeed, the complex web of political economy dynamics, embedded in the patronage (in relation to HWs hiring) and predation (in relation to their subsistence) structures in place, inevitably impacts on the likelihood of the reforms’ success. The consequence is the irrelevance of technocratic approaches to reform of civil service and HRH and, in parallel, the salience of purely political and sociological considerations ( Putzel et al. 2008 ). HRH reforms, it emerges from our analysis, are not only political themselves as they touch the distribution of power between actors at central and decentralized level, but are also entrenched in a complex sociopolitical and cultural environment, as they modify the subsistence and livelihood opportunities of HWs that have emerged and stabilized overtime in the context of the DRC. Similarly to other civil servants, some HWs may lose out from potential changes. It is likely that those who are advantaged by the current system are also those in more powerful positions (e.g. urban, higher cadres, some provinces), who are unlikely to allow changes to the status quo .

Beyond the technical approaches to reform currently proposed in the DRC (e.g. HRH census, payroll cleaning, electronic payroll), a more contextually-adapted, politically-savvy and coordinated action of external agencies, grounded at both central and provincial level, is necessary in order to increase the chance of an actual and positive impact of reforms. HRH reforms should also move beyond the perspective of the central health administration and of national-level stakeholders. Rather than country-wide prescriptions, there is a need for flexible and locally adapted approaches—as exemplified by the cases described by Rackley (2006 ). While the high informalization poses obvious challenges for donors, it has been argued that in fragile settings more than elsewhere, donors should accept a minimalist role for the recovering state, alongside a clearer recognition and inclusion of non-state (including private) actors and a broad rethinking of the engagement strategies, in particular with reference to unrealistic time-frames and expectations ( Pavignani et al. 2013 ). Herbst and Mills (2009 ) provocatively suggest that, in the DRC, the international community should stop dealing with a central state designed for dysfunction, but should favour concrete partnerships at local level with power holders, whatever their official status. For the health sector (and perhaps more reasonably for donors) and in line with others ( Waldman 2006 ), we believe that this may partially translate in an increased involvement of, and coordinated investment in, actors at provincial level, in particular given the proposed creation of 26 smaller provinces (instead of the current 11) and the establishment of autonomous health authorities at that level. Meanwhile, to allow the emergence of coherent local administrations, external agencies should refrain from imposing diverging agendas, but rather foster the definition and implementation of aligned local health policies by coordinating their action among themselves and with the central-level administration.

Conclusions

Our study provides the first comprehensive mapping of HWs incomes in the DRC and allows exploring the income sources and levels, as well as their fragmentation. From a methodological perspective, despite the absence of a qualitative component to triangulate and further explore findings (which could to be subject of future research), this work contributes to developing data collection and analysis techniques and provides information on the complex remuneration of HWs, which are useful to build on for an emerging cross-country research agenda on the topic ( Bertone and Lagarde ; Bertone and Witter 2015 ). From a policy perspective, it provides new insights on a topic and country over which there is limited research. Results are of relevance for the DRC, but also for other contexts. The key policy implications are 2-fold. First, we found that HWs incomes come from a variety of sources and that the income composition may have effects on HWs motivation and performance. Building on this finding, policy makers designing incentive packages should not only consider official allowances, but should have an understanding of the HWs income opportunities from formal and informal sources alike. This is essential to address challenges of HWs motivation, performance and distribution, which are at the centre of the debates on HRH in Kinshasa and in many other sub-Saharan African countries. Second, policy implications pertain to the features of potentially successful HRH reforms and interventions in fragile and complex contexts, of which the DRC represents a perhaps extreme example. Our findings point to the fact that those settings require a profound understanding not only of the technical aspects, but also of the underlying social, cultural and political context and of the institutional roots of the key issues, as well as a capacity for tailored, locally-adapted approaches to reform.

1

The sampling methods are described in further details in the survey report ( Bertone and Lurton 2015)

Acknowledgements

The authors would like to thank Hadia Samaha, Rafael Nunga, Michel Muvudi, Nicolas de Borman, Hélène Barroy and Françoise Andre for their support and insights during field research. We are also grateful to Abraham Flaxman for methodological advice and to Jean-Benoit Falisse for feedback on an earlier draft. Many thanks to Jack Kokolomami, Pierre Akilimani, Steve Bwira, Dalau Nkamba and Jacques Zandibeni of the Ecole de Santé Publique in Kinshasa for their precious role in coordinating and supervising field data collection.

Funding

This work was supported by the World Bank (funded by Canada Development Cooperation). The opinions expressed in this paper are those of the authors, and do not necessarily reflect the views of the funders.

Conflict of interest statement . None declared.