Legislating and implementing welfare policy reforms: What works politically in Africa and why?

Lead Research Organisation: University of Cape Town
Department Name: Faculty of Humanities

Abstract

This project falls primarily under the previous Call (in 2010) on "Inequality and Development".

Social assistance programmes - including 'social' pensions, conditional and unconditional cash transfers, and workfare through public employment programmes - have attracted considerable attention as a mechanism for reducing poverty and inequality, and stimulating development in the global South. Little attention has been paid, however, to the politics of programmes that "just give money to the poor". This research project will analyse how and why social assistance programmes are adopted in different parts of Africa, drawing comparisons with Brazil and India. The central question is "what works and what doesn't work politically?", i.e. what makes reforms politically feasible and sustainable?

The research will examine, for selected countries across Southern, East and West Africa, each stage of the policy-making and implementation process. How do ideas get onto the policy agenda? What shapes elite and public opinion on reforms? Under what circumstances do reforms become salient electorally, or in competition between or within political parties? What influence do civil society organisations, aid donors or international agencies exert over policy-making? How are reforms affected by legislative, executive or bureaucratic procedures within the state? In short, what factors favour and what factors impede policy reforms? The goal is to understand why policies are not adopted as much as why they are and the final form that they take.

The research will pay particular attention to the ways in which socio-economic inequalities and ethnic or racial differences affect the politics of welfare reforms. Economic inequalities ironically make it fiscally easier to reduce poverty gaps through cash transfers. Politically salient ethnic or racial differences are generally understood to impede programmatic policy-making. The combination of the two (as in South Africa) might make it politically easier to introduce reforms, if visible and effective poverty reduction is expected to reduce social and political tensions. Through a comparison of different countries, this research will inform an understanding of the political mechanisms through which socio-economic inequalities and regional or ethnic differences affect the political feasibility of welfare reforms.

This research will be conducted through a combination of existing research conducted through UCT, supplemented with focused new research in selected countries. Existing research foundations include a review of policy across the region, case-study research on reforms in South Africa, surveys of public opinion (through Afrobarometer and other research), legislative processes (through the African Legislatures Project), cross-national studies of public expenditure on health care programmes, and the relationships between civil society and public policy-making. New empirical research will focus precisely on elite and public opinion on welfare policy reforms, political processes with respect to welfare policy-making in state and civil society, and the determinants of expenditure on welfare programmes in the selected countries.

The project will contribute new understanding of how poverty alleviation can be delivered in a range of settings, including how the design of policies affects their political feasibility and the efficacy of interventions at different stages of the policy-making process.

Whilst building on the strengths of a major research institution in the global South, the project will also help to build capacity there.

Planned Impact

Policy-makers and practitioners must be integral to the research project. Probing the 'black box' of policy-making and implementation requires access to the inside knowledge of legislators, ministers, civil servants, civil society activists and donor organizations. As importantly, the project aims to provide a roadmap to past experiences in a range of African countries that can guide reformers in their future reform efforts. This research project will entail the following modes of engagement with practitioners and policy-makers, to ensure that the research is thorough and makes the desired impact.

1. A two-day colloquium in Cape Town will be held in the first six months of the project, to bring together a mix of researchers, policy-makers, donors and civil society activists from nine Southern African countries, to discuss and revise plans for the research programme.

2. A two-day colloquium in Nairobi will be held at the beginning of the second year of the project, to bring together researchers, policy-makers, donors and civil society activists from across East Africa to discuss the research that has already been conducted, and to discuss and revise plans for research in the East African countries.

3. A small advisory group of policy-makers, donors and civil society activists will be formed to guide the research project.

4. Five practitioners and policy-makers will be included among the visiting research fellows who will spend short periods of time at UCT, reflecting on their experiences, and advising and working with the UCT-based research team.

5. A three-day conference will be held in the final year of the project, bringing together researchers, policy-makers, donors and civil society activists to discuss research findings, locate these in global experiences of policy-making, and consider the implications. This conference will include dedicated sessions on the use of the research results in the policy-making process.

6. Policy briefs will be written and disseminated, to practitioners, policy-makers and news media in the selected countries, on both the national case-studies and cross-cutting themes.

7. The database of summary research findings will be available in a practitioner-friendly form.

8. Practitioners and policy-makers will be encouraged to participate in a short course on the politics of policy making in Africa to be included in the CSSR/Afrobarometer Summer Schools, held in Cape Town from 2013 for junior researchers and professionals from across Africa.

9. Dedicated pages will be established on the CSSR website, with links to the websites of the ALP and Afrobarometer.

Working with practitioners and policy-makers will be the responsibility of the PI together with Dr Haarmann. Dr Haarmann has considerable experience in working in and with civil society organizations, as well as with policy-makers, in both South Africa and Namibia.

It is anticipated that the "impact" dimension to the project will account for about £50,000 of expenditure (i.e. 13% of the total budget). This will comprise 10% of the PI's time on the project, 10% of the researcher's time, 20% of the project manager's time, 30% of Dr Haarmann's time, 50% of the workshop and dissemination expenses, and the expense of bringing five practitioners or policy-makers to Cape Town as visiting research fellows.
 
Description I have just drafted a Working Paper based on fieldwork in Zanzibar (which is a semi-autonomous territory within the United Republic of Tanzania), analyzing how and why Zanzibar introduced earlier this year a universal, non-contributory old-age pension programme. In making sense of the field research - comprising mostly interviews with political leaders, state bureaucrats and civil society activists - I have been struck by how rapidly the study of the politics of social policy in Africa has changed over the past five years. When we applied for the ESRC/DfID award, this 'field' was very poorly researched: Whilst there were many studies of the effects of cash transfer programmes, and some of their implementation, there was almost nothing on how and why different programmes were introduced in different places at different times (and, as importantly, why other programmes were not introduced in these places and times) - as was noted in reviews (Hickey, 2008; Niño-Zarazúa et al., 2011; Seekings, 2013). When we began our research, there were no clear case-studies to serve as models, and it was even unclear precisely what detailed questions we should be asking. Now, five years later, the field has been transformed.

We have contributed very significantly to that transformation. Our original goal was to produce a substantial set of working papers documenting the politics of social protection in 14 'Anglophone' African countries (including Mozambique, which had become at least semi-Anglophone in practice), all but one (Ghana) in East and Southern Africa. In the course of our research we recognized that we needed to disaggregate these country-studies. Many of our Working Papers focus on specific periods or programme's in each country: For example, we have separate studies of the pre-2008, 2011-14 and post-2014 governments in Zambia, and for South Africa we have separate studies of disability grants and child support grants. As of September 2016 we have published twenty-one Working Papers and a further twenty are scheduled to be published by the end of the year. We anticipate a further ten or so arising out of the research in the course of 2017. The complete list is set out, by country, as Appendix 1.

We have not been the only researchers contributing to the transformation of this field. The DfID-funded Effective States and Inclusive Development (ESID) programme (based at the University of Manchester) has conducted research in a smaller number of African countries, including two that were outside of our ambit (Ethiopia and Rwanda) and two that we also covered (Uganda and Kenya). Our research covers a larger set of countries and (often) longer historical period than ESID's, but our research programmes are clearly complimentary. Indeed, we are collaborating on a special journal issue on the politics of social protection in Africa, drawing on their and our research.

The work done by ESID and us has transformed this field by providing, for the first time, detailed documentation of the politics of social protection policy-making (and in some cases implementation) in a set of 'developing' countries across Africa. Many of our working papers are very substantial, running to 20,000 words or more. We have now, for the first time, a rich evidential basis for the comparative study of the politics of policy-making, enabling us to identify and assess recurrent patterns. This empirical 'dataset' is not in the form of a formal database, as we originally imagined, but comprises a set of rich, detailed case-studies.

Our recent study of Zanzibar is broadly typical. It points to economic, social and cultural factors that favoured reform, including especially the combination of (1) advanced deagrarianisation that meant that poverty among elderly was not redressed through kin or subsistence agriculture and (2) normative traditions of assistance to the poor (through both religious charity and state programmes). It provides a careful periodisation of the reform process between the initial appearance of a possible universal old-age pension in the research agenda in 2009 to the implementation of the reform in 2016. It shows how external actors - international agencies and non-government (or civil society) organisations - interacted with the Zanzibari state. It explores contestation within the state, examining how different ministries within the state interacted with each other and how reformers triumphed over skeptics. It notes the political benefits of reform to political parties in a very competitive political system.

Almost any one of our recent Working Papers reveals the questions that we now know we need to ask in varied cases: What normative or policy foundations existing in the territory prior to recent reforms? How did policy ideas get onto the research agenda? What roles did which non-government organisations or international agencies play in setting the policy-making agenda? How much space did bureaucrats have to drive or shape policy? How, if at all, did political competition within or between political parties contribute to the reform? Did the ideology or norms of political leaders shape the reforms? How were concerns over affordability overcome? What was the relationship between different parts of the state? How were the positions adopted by policy-makers shaped by underlying shifts in the nature of society? Through research we have reshaped the research agenda.

In broad terms, our research and the ESID research are very similar. In the details of our analysis, however, we can see significant differences. These reflect in part differences between our 'samples' of cases. In Africa, ESID's research on the politics of social protection has focused most closely on Ethiopia, Rwanda and Uganda, and in less detail on Zambia also. Ethiopia and Rwanda were not part of our sample. Ethiopia and Rwanda are authoritarian developmental regimes, and Uganda has elements of this. In contrast, most of the countries in our sample are more competitive democracies (notwithstanding differences in the conduct and quality of that 'democracy'). In several of our cases there have been full or partial changes in government (Kenya, Malawi, Zambia, Zimbabwe, Lesotho) and in others long-incumbent parties have faced significant electoral pressure (South Africa, Botswana, Tanzania and Zanzibar). ESID's emphasis on the underlying 'political settlement' in each country makes sense given the enduring character of the authoritarian developmental regimes. Our analysis attaches more importance to the dynamics of political competition: normative differences between political leaders and parties, the incentives provided by competitive elections for contrasting policy preferences, and the opportunities for reform provided by changes in political leadership.

In analyzing the politics of policy reform in diverse cases, we have shed light on both common features of diverse cases and the variation between them. Our first major finding concerns the comparative specificity of Africa, i.e. 'How is Africa different?'. We have substantiated our initial hypothesis that Africa - or at least Anglophone East and Southern Africa - is distinctive not so much in terms of its limited social insurance but in terms of its unusually extensive social assistance programmes (defined as inclusive of public employment or workfare programmes and some categories of food aid provided in kind). Across most of Africa, social insurance programmes are small in terms of both expenditure and coverage, but they are not distinctive controlling for the level of development (with the exception of a few middle-income countries which have not expanded their social insurance programmes as much as a cross-regional model would predict.

This distinctive pattern has deep historical roots, in many cases in the colonial period. In South Africa and Mauritius - which are often considered exceptional within Africa - the key periods of reform predated the 1960s. In these countries social insurance programmes expanded much more slowly and narrowly than would be expected given their industrialisation and urbanisation, whilst social assistance programmes had grown into global outliers long before the 1990s. In other cases, social protection policies have their origins in drought relief measures introduced on a minimal basis by late colonial states and more widely by post-colonial states. Across much of Africa, drought relief comprised a mix of workfare programmes for households with able-bodied adults and emergency relief for households without labour (for example, comprising elderly people or widows looking after children). We have documented the first major instance of post-colonial drought relief, in Botswana at the moment of independence (in 1966), when the new state - in conjunction with the equally new World Food Programme - mounted a massive programme of drought relief that was to evolve over time into the system of pensions, grants and workfare that make up the contemporary welfare regime in Botswana.

The case of Botswana is revealing also in terms of the normative foundations of the post-colonial welfare state. There was broad normative congruence between the outgoing colonial state and the new post-colonial state over the shape of state interventions (although the latter were much more generous than the former, expanding the scale of state interventions): Policies should respect the primacy of the family and kin, and the priority of household agricultural production; public policy should entail a broadly agrarian response to poverty, strengthening peasant agriculture and society; social protection programmes should be limited to moments or cases where household agricultural production and kinship failed to prevent extreme poverty. This general attitude seems to have characterised the supposedly 'socialist' leadership of (for example) Tanzania in the 1970s as much as the more conservative leadership of Botswana at the same time. Political elites across much (but not all) of Africa continue to be wary of excessive state intervention, which is often denigrated in terms of 'handouts' that encourage 'dependency' and laziness.

Political elites, often reluctantly, introduce and expand social protection programmes because of underlying shifts in the nature of the economy and society. 'Deagrarianisation' works through two distinct mechanisms. First, drought exposes temporary limits to agrarian responses to poverty. The scale of drought relief - whether in Botswana in the 1960s or Zimbabwe in the 2000s - is often massive, with as much as one half of the population dependent on food aid for a period of time. Secondly, the decline in or limits to redistribution between kin exposes specific categories of people - or 'vulnerable groups', as international agencies have come to call them - to extreme poverty: especially the elderly and poor families with children (often poor because they are single-parent families). Some cases are characterised by only the second mechanism: For example, Mauritius or Zanzibar. In most cases, the two are linked, in that large-scale drought reveals the need for ongoing, programmatic reforms.

Whilst we observe patterns across our sample of cases, our sample cannot be considered as entirely 'representative' of 'Africa'. Firstly, our set of cases is entirely Anglophone (especially given difficulties in completing our research in Mozambique). Secondly, our set of cases comprises countries (or territories) with competitive multi-party systems (in contrast to the ESID sample). Thirdly, we have only one case of a majority-Muslim territory (Zanzibar) - although approximately one-half of the total population of Africa is Muslim. We have no cases in the Sahel or north of the Sahara. Whilst our analysis of available country-level quantitative data suggests that the kind of welfare regime that predominates across Anglophone East and Southern Africa also predominates, albeit with some important differences, in North Africa, we have not conducted detailed case-studies in the Sahel, Horn of Africa or North Africa.

Even within our sample we can observe also variation, in terms of the scale or shape of reforms. In general, the Southern African countries have more wide-ranging social protection systems, with programmes for distinct categories of people (especially the elderly and children). Even within Southern Africa, however, there is clear variation. Both Botswana and South Africa have extensive welfare states, but the former is more conservative in terms of design (for example, limiting assistance to children to orphans, rather than targeting all poor children, and emphasizing workfare more emphatically) and generosity (with far more parsimonious benefits). Zimbabwe is a clear 'laggard' in terms of public policy, although non-government organisations are heavily involved on frequent and widespread drought relief. In East Africa, Zanzibar resembles Mauritius more than it does Kenya or Uganda, which are slowly expanding programmes, and especially mainland Tanzania, which is another marked laggard. In many cases there have been shifts over time, in all but one case involving expansion. The one exception (Zimbabwe, in the 2010s) has seen de facto programmatic contraction at the same time as a notional government commitment to expansion. There has also been variation in the pace of change, with successive governments accelerating or decelerating expansion. In the most striking case - Zimbabwe - there was a dramatic initial expansion under the 2009-13 Government of National Unity. In Zambia and Malawi, changes in government have had clear and direct effects on the reform process.

This variation is shaped but rarely (if ever) driven by processes of policy diffusion and the role of aid donors (such as DfID), international agencies (such as Unicef or the ILO) and non-government organisations (such as HelpAge International). Whilst aid donors and international agencies are often imagined to have considerable power, our case-studies suggest that their power has primarily entailed putting the idea of policy reforms on the agenda. Sometimes this was done through the initiation of experimental programmes. It often entailed playing a major role in drafting bold national statements about social protection policy. In other cases it involved reports on the feasibility - meaning the fiscal and organisational feasibility - of reform. In all of these respects, the powers of these external actors were bounded. Donors and agencies have often failed to persuade national governments either to extend (or 'scale up') experimental programmes across the whole country or to accept full financial responsibility for the programmes. In the extreme case of Zambia, pilot programmes operated for ten years before the national government began to expand them, despite the use of reformist discourse in planning documents. In Uganda, similarly, donors' initial efforts were unsuccessful. Later, however, bureaucrats bought into the reforms and seem to have become effective advocates of scaling up pilot programmes - at least, once the president gave his blessing. These cases point to the fact that donor power is easily overestimated. In Lesotho, our research suggests that donors were important in the recent introduction of a child grant, but not in the earlier introduction of old-age pensions. In Zanzibar, outside organisations - HelpAge International and Unicef - played important roles in shaping the process, but the key contestation was within the state.

In general, successful reform require that the state takes 'ownership' of the reforms. If external actors overplay their hand, they undermine the likelihood of local ownership of the process. The Zanzibar case is instructive as a case where the idea of a universal pension was placed on the agenda and then promoted for five years by external actors (especially HelpAge), but the reform was effected because the Zanzibari state saw it as their initiative with external actors providing assistance. In many cases, local ownership entails a clear political 'champion' of reform. But political champions can fail (as in Namibia, where former Lutheran Bishop Kameeta championed a basic income grant but, after being appointed as Minister of Poverty Eradication and Welfare, failed to get the President's support). And they are not necessary, as the Zanzibari case shows: even in the absence of a clear political champion, a powerful bureaucratic coalition can drive the reform process as long as political leaders provide sufficient support. Our case-studies provide important insights into the relationships in general between donors, bureaucrats and politicians. Our findings parallel research by ESID on Rwanda and Ethiopia, where the expansion of social protection programmes was driven more by the governments' developmental commitment, not donor influence.

Variation between countries, and between governments in a country, also reflect ideological differences. Governments and other actors make choices when they are aware of alternative models or possibilities, and in light of their ideologies and beliefs as well as their calculation of interests. Whilst political elites across much of Africa are generally ambivalent about 'handouts', they often also recognise collective responsibilities for certain categories of deserving poor, including even working-age adults and their dependents in times of drought. Within what might be considered a general conservative liberalism there are important variations. The political leaders of Botswana - beginning with the founding President Seretse Khama (1966-80) and continuing under his successors from the Botswana Democratic Party (BDP) - developed a benignly conservative welfare doctrine that justified public support along conservative lines. This was less radical than the more social democratic views associated with most of the Mauritian political elite, or reformers such as the South African Minister for Social Development Zola Skweyiya (1999-2009), or the more liberation-theological views of Bishop Kameeta in Namibia, or the opposition parties in Botswana itself (which have eroded the BDP's hegemony). But it is more benign than the neo-liberalism advocated by, for example, former Zambian Minister of Finance Peter Magande. In Malawi, the Mutharika brothers (who served as president prior to 2012 and since 2014) had clearly different policy preferences to Joyce Banda (president from 2012-14). Our work on these normative and ideological dimensions of public policy help to fill a conspicuous gap in the existing literatures on African politics.

This leads to our fourth substantive finding. We show that elections have played a significant role in driving policy reform in a number of cases. We began by collecting manifestos and studying campaign speeches with a view to assess the political importance of welfare policy reforms. Then we began to study what actually happens during elections. Our first such case-study was in Malawi, where welfare policy ('handouts') featured very prominently in the elections in May 2014. We have also examined how electoral competition shaped welfare state-building in Botswana between independence and 2014, the importance of elections in Zambia in 2011, 2014 and 2016, and whether in Zimbabwe the inclusion of the opposition party in a coalition government (in 2009) and then the exclusion of the same party (in 2013) made any difference in policy-making and implementation. Elections in Zanzibar in 2009 and 2015/16 framed the reform process. We monitored the elections in Uganda (in February 2016) and Zambia (August 2016), and we are monitoring also the consequences of elections in Namibia in 2014 (following which the new president appointed Bishop Kameeta as Minister of Poverty Eradication) and Mauritius (where the winning coalition promised to increase substantially the old age pension).
Exploitation Route We applied for a no-cost extension because we want to carry the findings forward, through the completion of the case-studies that are underway and synthesis of findings, envisaged workshops with other researchers (including ESID) and policy-makers in state and civil society. For technical reasons our application was declined, although we were granted a five-month extension to conclude ongoing contractual obligations. We had hoped for a long no-cost extension, and had to cancel some of our intended activities. This was, of course, very disappointing.

We are continuing to complete and publish the results of our research. As indicated above, we have published to date 21 Working Papers and have a further 20 in the immediate pipeline and a further ten expected in early 2017 (see Appendix). This should bring the total output of the project to about fifty Working Papers, which will entail an unprecedented set of documentation and analysis not only for Africa but for anywhere in the global South.

A selection of our papers (on Ghana, Malawi, Botswana, Kenya and Lesotho) are being included in a proposed special issue of the journal Development and Change. The special issue would publish papers from our programme together with some from the ESID programme. The journal accepted an initial proposal, and the papers are currently being finalised. Other papers are under submission to other scholarly journals.

We anticipate also completing at least two scholarly monographs during 2017. The first will examine the development of the 'exceptional' welfare states in South Africa, Mauritius and Botswana in the second half of the twentieth century. The second will examine the contemporary politics of social protection reforms in East and Southern Africa.

Finally, we are also planning to hold a workshop with practitioners from states and civil society, in mid 2017, with funding from other sources. This will provide an important opportunity for dissemination as well as discussion.
Sectors Communities and Social Services/Policy,Government, Democracy and Justice
URL http://www.cssr.uct.ac.za/liwpr