The political economy of public expenditures in agriculture: Applications of concepts to Mozambique Citation Mogues, Tewodaj; and do Rosario, D. The political economy of public expenditures in agriculture: Applications of concepts to Mozambique. South African Journal of Economics 84(1): 20 - 39. http://dx.doi.org/10.1111/saje.12076 DOI http://dx.doi.org/10.1111/saje.12076 Access through IFPRI e- brary http://ebrary.ifpri.org/cdm/singleitem/collection/p15738coll5/id/4814 Terms of Use IFPRI uploaded the final published version to the institutional repository and is made available under the copyright law of the United States (Title 17, United States Code) which governs "fair use" or the making of photocopies or other reproductions of copyrighted material. Photocopy or reproduction is not to be “used for any purpose other than private study, scholarship, or research.” Permitted Re-Use Permitted re-use of this open access article is determined by the author’s choice of user license. e-Access to IFPRI Research IFPRI has made this article openly available. If this access benefits you or your community share your story with ifpri-km@cgiar.org. http://dx.doi.org/10.1111/saje.12076 http://ebrary.ifpri.org/cdm/singleitem/collection/p15738coll5/id/4814 http://creativecommons.org/licenses/by-nc-sa/3.0/ mailto:ifpri-km@cgiar.org THE POLITICAL ECONOMY OF PUBLIC EXPENDITURES IN AGRICULTURE: APPLICATIONS OF CONCEPTS TO MOZAMBIQUE TEWODAJ MOGUES* AND DOMINGOS DO ROSARIO † Abstract This paper undertakes an investigation of agricultural public investments in Mozambique, drawing on insights from qualitative field interviews conducted in Mozambique, secondary data analysis and examination of the existing empirical literature, and by situating these insights within a political economy conceptual framework. We explore the driving factors behind the amount and allocation of public funds to agriculture, and behind the differential attention that various types of public investments receive in the process of making decisions on resource allocation. Agricultural public investments are more likely to be made that have two key features: higher attributability to politicians and donors of the output of public spending, and a shorter lag time between expenditures incurred and outputs produced. Evidence on geographical targeting of agricultural public funds corresponds more closely with theories suggesting that resources are used to sway communities opposed to the ruling party, rather than to reward political supporters. Examination of the effect of actors’ and organisations’ incentives and constraints on resource allocation in agriculture points to the importance of not treating “government,” “the ruling party” and other institutions as monolithic bodies; the paper instead highlights how differentiated interests within seemingly coherent institutions drive what gets public expenditure attention in the agricultural sector. JEL Classification: O13, D72, H50 Keywords: Mozambique, agricultural public expenditures, public investments, political economy 1. INTRODUCTION The agricultural sector plays a crucial role in Mozambique’s economy and for the welfare of its people (MINAG, 2011). As a member of the community of African states, Mozambique has committed itself to spurring growth and development in the sector through the continent’s currently central initiative for agriculture, the Comprehensive Africa Agriculture Development Programme (CAADP). A decade ago, the African Union * Corresponding author: Senior Research Fellow, Development Strategy and Governance Division (DSG), International Food Policy Research Institute (IFPRI), 2033 K St NW, Washington, DC, USA. E-mail: t.mogues@cgiar.org † Department of Political Science and Public Administration, Universidade Eduardo Mondlane. An earlier version of this research paper was produced with financial support by the Mozambique office of the United States Agency for International Development provided to IFPRI’s Mozambique Strategy Support Programme (MozSSP). The authors are grateful to this financial assistance. Thanks also go to two anonymous referees of this journal for their written comments. Remarks on early drafts were helpful that were provided by Samuel Benin, and by two discussants and other participants of a workshop held on 18 October 2012 by IFPRI’s Mozambique Strategy Support Programme, during which early results of this work were presented. Sileshi Woldeyohannes provided research assistance. Any errors are solely those of the authors. South African Journal of Economics Vol. ••:•• •• 2015 © 2015 Economic Society of South Africa. doi: 10.1111/saje.12076 1 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. doi: 10.1111/saje.12076 20 South African Journal of Economics Vol. 84:1 March 2016 South African Journal of Economics and the New Partnership for Africa’s Development (AU/NEPAD) convened in Maputo and launched CAADP with the objective of accelerating agricultural growth in the continent. In connection to this, with the aim of achieving at least 6% annual growth in the agricultural sector, heads of state of Africa agreed and signed on to commit 10% of their national budget to the development of the sector (AU/NEPAD, 2003). The key elements of CAADP reflect the general NEPAD principles that are believed to add value to countries’ processes, including changing the quality of policy development and implementation, building of partnerships, dialogue, peer review and mutual accountability at all levels, and exploitation of regional complementarities. Mozambique has, at the time of this writing, finalised its CAADP National Agricultural Investment Plan (or Programa Nacional de Investimento do Sector Agrário – PNISA), and has embarked on orienting its agricultural public investments in line with CAADP’s process and goals. The PNISA is based on the recently launched Strategic Plan for Agricultural Development (PEDSA), which lays out the country’s vision of transforming the agricultural sector from a predominantly subsistence farming industry to a competitive and sustainable sector that would contribute to food security and raise incomes of rural households (MINAG, 2011). In light of the fact that public investments in agriculture in Mozambique can have profound effects on productivity in the sector, it is important to understand what factors drive decisions on these public investments, so that one can gain insights into the possible enabling and inhibiting factors for policymakers to undertake the needed agricultural public investments. The analysis in this paper is based on a political economy conceptual framework, qualitative field data, secondary data, and examination of the pertinent theoretical and empirical literature. We use a framework of drivers of public investment decisions that has as its elements (i) the budget process, (ii) characteristics of publicly provided goods and services, and (iii) incentives, influences and constraints of public and private actors. The fieldwork consists of in-depth semi-structured interviews that took place from March to June 2011 with a selection of 17 key informants from government, donors, the private sector, civil society, academic researchers and experts. In the next section, the conceptual framework is presented of the driving forces behind the magnitude and composition of public funds allocated in and for the sector. This is then used in Section 3 to interpret the public resource allocation process in Mozambique, which corresponds to the first element in the framework. Section 4 discusses the empirical evidence with regard to the framework’s second element, namely the role that the characteristics of different agricultural public investments play in driving resource allocation. The third framework element – the incentive structure and constraints of key actors, such as donors and private sector agents – is empirically explored in Section 5. The final section concludes, including by reflecting on how the CAADP process – now a central element in agricultural investment decision making in Mozambique – may influence public expenditures for agriculture. While the latter is a question worthy of proper study in its own right, only some reflections on this will be offered in the concluding remarks. 2. CONCEPTUAL FRAMEWORK ON THE DETERMINANTS OF PUBLIC INVESTMENT DECISIONS The conceptual framework structures the different influencing factors on public investment decision making, on the basis of the theoretical literature and empirical South African Journal of Economics Vol. ••:•• •• 20152 © 2015 Economic Society of South Africa. 21South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. evidence on these factors. There are four core dimensions along which the drivers of agricultural public investments are organised. The first dimension is the budget process: the most direct, but not always most complete, way in which the resource allocation process can be understood is by an examination of the formal process through which planning, budgeting, execution and monitoring for agricultural public sector interventions take place. Understanding this formal process, while certainly an important start, will only provide partial insights on the key question at hand. This is both because the actual process of planning, budgeting, execution and reporting often diverges from actual processes in many countries, but also because the description of the stages and steps of writing planning documents, drafting budget proposals, etc. leave unanswered how the content of the planning documents in fact come about, and what forces drive negotiations on budgets. A second dimension, pertaining to agents’ incentives, begins to fill some of these gaps left in the examination of the budget process. This dimension is concerned with the roles, incentives, interests and constraints of key actors, such as policymakers, donors, bureaucrats, small farmers, agricultural industry, etc. This of course is an expansive topic, given the multiplicity of key actors, and the complexity of their incentives and scope for action (see Mogues (2015) for a review of the literature on this topic). Therefore, the empirical exploration of the case of Mozambique that this paper provides can only offer an initial look at this important dimension. The third dimension – investments’ characteristics – focuses on how features of different types of agricultural public services, subsidies, infrastructure and goods will influence whether and to what extent resources will be allocated to creating them. This examination, as that of the other dimensions, takes a political economy angle in seeking to understand how investments’ characteristics – such as the time lag with which investments will show benefits, and the visibility of these investments – interface with policymakers’ opportunities to be associated with the fruits of the investments and thus get credit for their creation (see Mogues (2015) for a more detailed conceptual treatment of this dimension). We subsequently examine the three above-described elements within the Mozambican context by analysing the findings from the fieldwork, as well as by drawing on insights from the existing literature on Mozambique within this framework and interpreting the literature’s results within this light. 3. THE BUDGET PROCESS AND ITS INFLUENCE ON AGRICULTURAL INVESTMENTS Pertaining to the first element in the framework, we first reflect on the planning and budgeting process in Mozambique in both its de jure and de facto forms, and consider its influence on public resource allocation to agriculture. 3.1 Formal Planning and Budgeting Process Fig. 1 provides a simplified diagrammatic illustration of the formal planning and budget process in Mozambique. The upper half of the illustration represents that part of the process that reflects planning, budgeting and reporting at the national, cross-sectoral level. The bottom half shows the equivalent procedures in the agricultural sector. The left part of the diagram (in the oval shape) captures the subnational elements of the planning 3South African Journal of Economics Vol. ••:•• •• 2015 © 2015 Economic Society of South Africa. 22 South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. and budgeting process, both the cross-sectoral and the agriculture sector-specific dimensions. At the higher level of the formal planning process, there is an overarching guiding framework that lays out the strategies for development, as a whole (upper part of diagram) or in the agricultural sector (lower part). At the cross-sectoral level, these are the government’s five-year plan, the PQG, and the poverty reduction strategy paper (PARPA – Plano de Acção para a Redução da Pobreza Absoluta, i.e. action plan for the reduction of absolute poverty), both covering close to the same five-year time period. For agriculture, the guiding framework is the PEDSA, which has a longer, 10-year time frame. These broad multi-year strategy documents are to guide shorter term and more specific activity plans, such as the PES (Plano Económico e Social, i.e. economic and social plan) at the cross-sectoral level and PAAO (Plano Annual de Actividades e Orçamento; i.e. annual plan of activities and budget) in agriculture. The PAAO receives inputs from activities proposed by district and provincial agricultural offices, and in turn is meant to feed into the development of the cross-sectoral PES. Table 1 summarises the key documents of the planning and budgeting process, and classifies them, analogous to Fig. 1, into cross- sectoral and agriculture sector-specific documents. Complementary to Table 1, Fig. 2 highlights the time dimension to which selected key strategy and budgeting documents apply. In this formal structure, the budgeting process is intended to be closely coordinated with the planning process. The ministry of agriculture submits the annual sectoral budget proposal to the ministry of finance. One of the manifestations of the planning- budgeting coordination is supposed to be reflected, for example, in the way that the overall magnitude of the proposed budget is informed by the PAAO, and the PAAO is to detail what outputs and outcomes are to be achieved with the proposed budget. On Figure 1. The planning and budget process in Mozambique South African Journal of Economics Vol. ••:•• •• 20154 © 2015 Economic Society of South Africa. 23South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. the basis of the agricultural (and other sectors’) budget proposals, the ministry of finance develops an aggregate budget (OE; Orçamento do Estado, i.e. government budget). Analogous to the sectoral level, at the cross-sectoral level the PES is supposed to be well synchronised with the OE. The OE determines what the ministry of agriculture’s approved budget will be. From here, execution of the agricultural budget proceeds in each spending unit on the basis of the so-called expenditure tables, which show the authorised levels of expenditures. Table 1. Key documents and reports in the agricultural planning and budgeting process Document Description Instrument Time frame Lead agency Not specific to agriculture PARPA I PRSP Planning 5 years (2001-2005) MPD PARPA II PRSP Planning 4 years (2006-2009) MPD PARPA III PRSP Planning 6 years (2011-2016) MPD PQG Government five-year plan (drawn from ruling party’s platform) Planning 5 years (2010-14) Government/ruling party CFMP MTFF (yearly rolling) Planning 3 years MPD PES Economic and social plan Planning 1 year MPD PAF Performance assessment framework Planning 1 year GBS donors OE National budget Budgeting 1 year MF BdPES Report on PES execution Reporting 1 year MPD CGE Financial report Reporting 1 year MF TA report Report by the auditor general Reporting 1 year TA REOE Budget execution report Reporting 1⁄4 year MF JR Joint review on budget support Monitoring 1 year GBS donors Specific to agriculture PEDSA Strategy document (in alignment with CAADP) Planning 10 years (2011-2020) MINAG PNISA Investment plan (for implementation of PEDSA) Budgeting 5 years (2013-2017) MINAG PAPA Strategy document (to face the food price crisis) Planning 3 years (2008-2011) MINAG ProAgri II Strategy document (for sector budget support) Planning 8 years (2006-2011) MINAG ProAgri I Strategy document (for sector budget support) Planning 6 years (1998-2006) MINAG PAAO Agricultural activity plan Planning 1 year MINAG PTAO Agricultural activity plan Planning 1⁄4 year MINAG APAR Performance report Reporting 1 year MINAG FMR Financial management report Reporting 1⁄4 year MINAG JR ag Review of agricultural sector Monitoring 1 year GBS donors Notes: Acronyms are as follows: APAR: annual performance assessment report; BdPES: Balanço do PES (annual report on execution of the previous year’s PES); CFMP: Cenário Fiscal de Médio Prazo (medium-term fiscal framework); CGE: Conta Geral do Estado (general public accounts); FMR: financial management report; GBS: general budget support; JR: joint review; MF: Ministry of Finance; MINAG: MINistry of AGriculture; MPD: Ministry of Planning and Development; MTFF: medium-term fiscal framework; OE: Orçamento do Estado (government budget); PAAO: Plano Annual de Actividades e Orçamento (annual plan of activities and budget); PAF: performance assessment framework; PAPA: Plano de Acção para a Produção de Alimentos (food production action plan); PARPA: Plano de Acção para a Redução da Pobreza Absoluta (action plan for the reduction of absolute poverty); PEDSA: Plano Estratégico de Desenvolvimento Agrário (strategy and plan for agricultural development); PES: Plano Económico e Social (economic and social plan); PNISA: Programa Nacional de Investimento do Sector Agrário (national programme for agricultural sector investment); PQG: Programa Quinquenal do Governo (government five-year plan); PROAGRI: PROgramme for AGRIcultural sector strategy and expenditures; PRSP: Poverty Reduction Strategy Paper; PTAO: Plano Trimestral de Actividades e Orçamento (quarterly activity plans); REOE: Relatório de Execução do Orçamento do Estado (quarterly financial report); TA: Tribunal Administrativo (administrative court). 5South African Journal of Economics Vol. ••:•• •• 2015 © 2015 Economic Society of South Africa. 24 South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. This then represents a highly simplified description of how the formal planning and budgeting process leads to the realisation of agricultural expenditures, in the star shape in Fig. 1. It is already apparent, as mentioned earlier, that such a description, even if it were more detailed, does not satisfy. It leaves crucial questions open on how, for example, the priorities in the PEDSA get determined; whether in fact these priorities ever find themselves properly reflected in expenditures, and if not why not; what forces actually determine the budget ceiling for agricultural ministries and agencies; and so forth. As a start, then, it is important to consider how the actual budget and planning process may diverge from its formal structure. The next subsection addresses this issue. 3.2 De Facto Dynamics in the Planning and Budgeting Process and Its Influence on Investment Decisions The diagram in Fig. 1 reveals, but probably does not even do full justice to, the complexity that underlies the vertical and horizontal governmental structure of Mozambique. The multiplicity of layers and cross-links in the planning and budgeting process – between higher and lower tiers of government, between sectoral and cross- sectoral agencies, and the matrix relationship that these two dimensions represent – has implications for the smooth functioning of the process as well as for the de facto influence of the various layers on final resource allocations in the sector. An example of how both functionality of process and influence of components within the process can be compromised in the complex institutional landscape can be seen in the case of the district agricultural activity plans. Despite the intention for the planning process to be well coordinated with the budgeting process, this coordination has often not taken place as intended. For example, PAAO had been treated as a “wish-list” for various agencies and units for agriculture, the sum total of which would far exceed what ultimately becomes the available budget. Even if the expressed activities were all needed and relevant to the responsibilities of these agencies and offices, the lack of correspondence between the PAAO and the total approved budget compromises the PAAO’s relevance in influencing expenditure allocations within the sector (World Bank, 2011a). There also remains some tension between the way that planning and budgeting systems operated in the past, vs. in more recent years. Prior to 2006, the vertical coordination of planning tools proceeded generally speaking at a sectoral level. The Figure 2. Timeline of main cross-sectoral and agriculture-specific strategies South African Journal of Economics Vol. ••:•• •• 20156 © 2015 Economic Society of South Africa. 25South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. ministry of agriculture, for example, had a relatively strong role in coordinating the inputs of the agricultural directorates at the district and province levels. After 2006, there was a shift of vertical coordination towards the central (as opposed to line, or sectoral) ministries. For example, the national ministry of planning and development would provide ceilings to its provincial analogues, the provincial directorates for planning and finance. These ceilings cover the entire province’s budget, including all sectoral activities, such as agriculture, health, education, etc. Within the matrix system of dual subordination (Hodges and Tibana, 2004) – whereby a local sectoral agency is subordinate both to the higher tier sectoral agency, as well as to the local cross-sectoral leadership – this shift after 2006 then represents a strengthening of the role of local cross-sectoral bodies and a weakening of the role of national sectoral ministries. For any given sector, like agriculture, this also corresponds to a potential shift of some weight away from national agricultural priorities and towards local priorities in decision making about the allocation of public resources. As with the agriculture-specific plan, the cross-sectoral plan’s relevance for ultimate resource allocation in the agricultural sector is also compromised in the de facto planning and budgeting process – albeit for different reasons. The information concerning agriculture captured in the PES is usually not sufficiently focused on activities that specific public agencies and institutions that are (to be) funded by the budget plan undertake. Rather, the agriculture section of the PES concerns itself with activities of typically private agents – farmers, processors, etc. – that are not “public spending units” in the sense described earlier (World Bank, 2011a). By providing an overview and forecasts of agricultural production in the country, for example, its pertinence as a direct planning tool for the spending units of the government and as budgeting input for the ministry of finance is highly limited. These examples – shift of powers between horizontal and vertical processes; weaknesses in planning documents on capturing agricultural sector-related information rendering the part of the process less relevant than these documents represent; and local plans sidelined in their importance by their efforts to maximise access to funds – make an important illustration: Any reference to the planning, budgeting and execution process in order to understand the forces that shape public expenditures in agriculture should not give short shrift to the way this process plays out in actuality – even if it may take more effort to discern the de facto budget process than the task of mapping out the formal process. 4. HOW DO THE CHARACTERISTICS OF PUBLIC GOODS AFFECT THE RESOURCES ALLOCATED TO CREATING THEM? As explained in the general framework, the decision by policymakers on whether and to what extent to invest in a given public good, service or infrastructure can be influenced by the characteristics of these goods themselves. 4.1 Ability to Attribute Agricultural Public Investments to Public Decision Makers’ Efforts In this respect, one of the important features of goods and infrastructure is the extent to which public expenditures incurred in creating them can in fact be attributed to the policymakers who are responsible for these investments. Such attribution matters to policymakers because the success with which they can rally citizens’ support depends on 7South African Journal of Economics Vol. ••:•• •• 2015 © 2015 Economic Society of South Africa. 26 South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. the citizens’ perception of politicians’ performance in service provision and public goods delivery. This perception, in turn, is driven not only by actual performance along this metric, but also by citizens’ knowledge about the actual links between the goods and services on the one hand, and politicians’ efforts and decisions on the other. Considering different services that government actors are responsible for, some types of these services are more easily traceable to the public officials’ efforts than are other types of services. Thus, all else equal, officials are more inclined to give greater attention and allocate more resources to services and investments that are more easily traceable to their (i.e. the officials’) efforts, than they would to services and investments that are harder for citizens to attribute to their efforts. The empirical literature on agricultural investments in Mozambique does not examine head-on the question of how attributability affects investments. However, there are some suggestive features in the existing evidence concerning the ways that attributability affects policy decisions in resource allocation in and for agriculture. World Bank (2011a) makes a distinction between, on the one hand, the “core functions” of the state in agriculture, and on the other hand “promotional activities.” Core functions here refer to activities that have a clear rationale for public sector involvement, based on the public goods character and the existence of positive externalities in providing these goods or services. Examples are investments in agricultural research, extension, disease control, and the establishment of appropriate regulation of private activity such as setting quality standards for certified seeds. On the other hand, promotional activities are defined as those with limited or no public goods characteristics and externalities, and those that the private sector could alternatively carry out given the appropriate enabling environment for private agents. These include the distribution of subsidised agricultural inputs and credit, and direct state involvement in productive activities, such as running agricultural processing and storage facilities. This distinction includes the assessment that promotional activities are time-bound interventions, the impact of which are particular and can be monitored, and the results of which are visible in the short term. In contrast, core functions are generally ongoing, routine activities, the results of which may be medium to longer term and thus less attributable to actions of policymakers. As such, World Bank (2011a) suggests that promotional activities in Mozambique lend themselves more than do core functions to deployment for gaining political credit, given their more pronounced visibility and attributability, although their societal contribution to agricultural development may be more limited. Indeed, the study finds that there is underinvestment in core functions, in favour of higher spending prioritisation of promotional activities in the agricultural budget in Mozambique. This assessment has also been made at earlier times, such as in a joint review of development cooperation between the Government of Mozambique and the European Commission (EC, 2006), which concluded that during the year of examination none of the four indicators of the delivery of agricultural core services (access to extension, total number of farms with vaccinations carried out, area benefited from irrigation systems built and/or rehabilitated with public funds, and percentage of land applications processed) had reached the target. Some of the promotional activities are financed by the OIIL (Orcamento de Investimento de Iniciativa Local; or Local initiative investment budget) budget allocation, which has helped in expanding the acceptance and influence of the ruling party, Frelimo, throughout the rural areas, which were traditionally the support base of the major opposition, Renamo (see also more on this below). South African Journal of Economics Vol. ••:•• •• 20158 © 2015 Economic Society of South Africa. 27South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. The choice of investments within the irrigation sector can be examined in this light as well. Public expenditures on irrigation overwhelmingly go to large-scale projects; the share of expenditures to the latter ranged from 81% to 92% between 2002 and 2007 (World Bank, 2011a). This cannot necessarily be explained by a superior contribution of large irrigation. Evidence points to underperformance of irrigation projects in general, but also to a lopsided performance among irrigation scheme types. A recent analysis of Mozambique’s infrastructure finds that unit costs of small-scale irrigation projects are about two thirds of those of large irrigation projects. At the same time, the rates of return to small irrigation investments are similar to that of other countries in the Southern Africa region, while the returns to large-scale irrigation expansion are low relative to those in the region (Domínguez-Torres and Briceńo-Garmendia, 2011). Existing small-scale schemes are also able to be utilised to a greater extent by farmers than medium and large schemes, with 51%, 24% and 35% of small-, medium- and large-scale facilities operational, respectively (World Bank, 2011a). While the rationale for the strong favour in public spending towards large schemes in Mozambique despite their lower performance has not been well documented, a review that summarises the key findings from more than 30 sectoral- and subsectoral-level agricultural public expenditure reviews indicates that large-scale “pet projects” with high visibility may prevail in agriculture, although their economic relevance cannot be supported by technical evaluation (World Bank, 2011b). It is not only the government that may orient public expenditures towards those areas that bring political gains through relatively high attributability and visibility. Some existing studies on Mozambique suggest that similar incentives are also present within the donor community. According to Hodges and Tibana (2004), the wish to “show the flag” has constrained donors’ willingness to channel funds through sectoral or general budget support, as opposed to their favoured mechanism of providing funding for distinct projects. The study implies that donors could tend to choose development projects that are easier to mark for home audiences, such as bodies of accountability of public funds including the parliament of the donor country. Furthermore, each donor may have different bilateral conditions, administrative and reporting requirements, and legal requirements, which could discourage the government from conceiving important projects that require joint funding and coordinated planning (Killick et al., 2005). There are, however, other possible reasons for the limited desire on the part of donor agencies to channel resources through budget support in Mozambique. Such pooled arrangements require adequate trust of efficient management of public resources on the part of the government, precisely because donors’ ability to control the use of their funds is strongly constrained under pooled arrangements. If such trust is undermined by observed poor quality of the government’s public financial management, or low execution rates for example due to unwieldy processes of tendering for investments – as were expressed in earlier periods, see e.g. Batley et al. (2006) – inclination towards projectised aid as opposed to budget support may have little to do with better attributability features of the former. 4.2 Time Lag of Investments from Allocation of Funds to Generating Outputs In addition to attributability and visibility, the temporal features of agricultural investments are also important in determining the nature of agricultural public spending in Mozambique. One important feature is the time gap that may exist between the initial 9South African Journal of Economics Vol. ••:•• •• 2015 © 2015 Economic Society of South Africa. 28 South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. incurrence of public expenses and the realisation of services or infrastructure provided. Such a time gap will influence whether the service or investment will in fact be made, or how much resources will go into making them. There are several reasons why this lag from expenditures to outputs or even outcomes matter. First, the longer the lag, the more likely that policymakers will not be in office by the time that the benefits of investments are realised. Changes in office may occur for the usual reasons, including electoral outcomes, changes in political appointments, etc. If a decision maker cannot reap the benefits of receiving credit for investments because of a long time span until these benefits come to fruition, this dampens the incentives to make these investments in the first place. The reverse holds with regard to expenditures that give quick results. Another reason why the time lag of investments influences whether and to what extent these investments will take place relates to the uncertainty dimension of time. The longer the time that passes between allocation of funds and realisation of outputs, the greater the uncertainty about the product from these funds, given possible fluctuations in the many other exogenous factors that can also affect outputs besides the specific investments under the control of the public official. World Bank (2011a) highlights that the particular modality of public investments in irrigation in Mozambique have been detrimentally affected by the political imperatives of both donors and government to show quick results. When properly developed from conceiving the idea to finalising the construction, and given the existing institutional arrangements and capacities, an irrigation scheme in Mozambique could take an average of about three years. However, this time frame has often been shortened given the political pressure to have something to show for in a short time period. The quality of technical and feasibility studies – or even their being undertaken at all – have been sacrificed in the process, with resultant problems in the functionality of the irrigation schemes. Another important type of agricultural investment in any country, research and development (R&D), has a notoriously long time lag between the time that an R&D endeavour is initiated and the time that, say, an improved seed variety for a crop has been developed. This itself may become a disincentive to commit adequate and stably flowing funds to agricultural R&D. In Mozambique, agricultural R&D investments are, by any standards, including in comparison to Africa-wide statistics, extremely low. According to World Bank (2011a), agricultural research spending intensity – the ratio of public spending on research to agricultural gross domestic product – is as low as a quarter of 1%.1 In Africa, this figure was 0.61% in 2008 (Stads and Beintema, 2012). This is still substantially less than the 2004 recommendation by the Inter-Academy Council for Africa of 1.5% intensity (IAC, 2004). While existing studies on Mozambique have not yet explicitly investigated the reasons for such staggeringly low levels of agricultural research investments, it is instructive to consider the time lag-related potential factors at play from the wider literature, as described above. 1 Various metrics on trends in agricultural research investments over time for recent years give different pictures. The number of full-time-equivalent researchers grew by more than 21⁄2 times from 2004 to 2008, although the amount of public spending in real terms has been about the same in these two time periods, and agricultural research spending intensity has actually declined from 0.55% to 0.41% in this four-year period (Flaherty et al., 2010). South African Journal of Economics Vol. ••:•• •• 201510 © 2015 Economic Society of South Africa. 29South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. Finally, the time frame over which results on the ground can be expected from different types of investments in and for the agricultural sector, and the impact this has on the willingness to undertake or continue these investments, is no more prominently on display in the Mozambican context than in the case of public expenditures through the sector budget support, PROAGRI, in its first and second phase. As described in more detail below, the first phase was characterised by strong focus on the kinds of investments – institutional development – that would not be able to bear fruit in the form of higher agricultural growth until a substantial number of years had elapsed. In this vein, although there are also other criticisms of the first-phase sector budget support, the lack of balance between process and results is the major critique to PROAGRI. Cabral et al. (2007) argue that after spending hundreds of millions of dollars on institutional capacity building of the agricultural sector, the lack of improvement in due time in agricultural service delivery and thus in farmer welfare has resulted in weakened government commitment to the programme, such that PROAGRI Phase II ended up reversing this institutional investment focus in favour of a much stronger emphasis in its design on direct service financing. 5. ACTORS AND THEIR INTERACTIONS, AS DRIVERS OF AGRICULTURAL PUBLIC INVESTMENTS The third element of this paper’s framework for understanding the drivers of agricultural public investments in Mozambique pertains to the incentives, roles and constraints of various actors and interest groups, public and private. Standard (micro-)economics concerns itself with the way that private agents – mainly, producers anywhere in the value chain, and consumers – make economic decisions, with the outcomes of these decisions, and importantly with the underlying incentives of the agents and their constraints that are driving these decisions. As argued by the well-established public choice and political economy literature, in order to understand the actions of public agents – including actions such as allocation of public funds in and for the agricultural sector – one needs to apply the same rigour in understanding public agents’ incentives and constraints as one does for private economic actors. This discussion embeds some of the existing evidence on Mozambique within the conceptual treatments of this broad question, thus shedding some light on drivers of resource allocation to the agricultural sector by employing this frame. 5.1 Political Parties: Inter-Party Competition and Agricultural Resource Allocation One element of the political economy of policymaking in Mozambique can be discussed against the backdrop of a wider literature on distributive politics. One of the key political economy frameworks of public resource allocation is concerned with whether politicians will be more prone to channel resources to their traditional base supporters, or whether they will find it politically more efficacious to transfer resources to segments of the population that are less closely aligned with them. The key argument underlying the latter hypothesis is that core supporters of a political party already are committed to vote for their own party and are unlikely to change their voting habits, whereas politically less committed citizens are more readily swayed by receipt of state resources (Lindbeck and Weibull, 1987; Dixit and Londregan 1996). The other hypothesis, namely that a governing party is more likely to tilt public resource allocation towards its core supporters, rests in part on the notion of a higher 11South African Journal of Economics Vol. ••:•• •• 2015 © 2015 Economic Society of South Africa. 30 South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. uncertainty associated with the political returns to economic investments going to less- known, uncommitted voters. If incumbent governments are risk-averse, they may choose constituency areas to invest resources in a way that yields less variable electoral results (Cox and McCubbins, 1986). Another argument supporting this “core constituency” hypothesis is that public goods and transfers to any given area potentially not only affect the choices made by voters between alternative parties, but also affect the likelihood of a citizen engaging in voting, as opposed to not voting. Considering this turnout effect of transfers, it is more effective to target more resources to localities with historic support for one’s own party, as one can more easily draw on local networks in such areas to increase turnout, and therefore increase votes for the party, than one could by investing in localities supporting the opposing party (Cox, 2009). Results from the fieldwork interviews in Mozambique conducted for this enquiry, as well as analysis of the pertinent literature, can be brought into this framework. Our main findings are consistent with the first hypothesis that resources will be directed to groups not already aligned with the governing party in order to expand political support among these groups. It should be said that the theories discussed above juxtapose core supporters of one party against the so-called swing voters, and not against core supporters of the other party. Our fieldwork findings, however, have only been able to identify how resources are allocated to pro-government vs. opposition supporting areas, without explicit distinction between clearly strong opposition supporters and “swing areas” that support the opposition party with a small margin. The governing party used various methods to gain stronger support among constituencies traditionally backing the opposition party, Renamo. Three dimensions featured in three key approaches, respectively, with the dimensions being ethnicity, geography and rural/urban areas. One approach was political patronage, specifically the assignment of senior government positions to leaders of ethnic organisations prevalent in the northern part of the country, in which Renamo has a traditional stronghold. The second approach relates to differentiated governance arrangements in urban vs. rural areas, from which Frelimo and Renamo derive their relative political strength, respectively (Reaud and Weimer, 2010). Initially, a 1994 local governance law passed before the first multiparty elections in the same year foresaw elected assemblies for all districts, both rural and urban. After Renamo performed more strongly than expected in the rural areas in the 1994 election, the constitution was amended in 1996, creating a bifurcated local government system, and this system was further elaborated in a new set of laws in 1997. According to the new arrangements, only urban municipalities would have assemblies and mayors elected by the residents in the respective jurisdictions. Rural districts, on the other hand, would be governed under a deconcentrated arrangement, with district administrators appointed by central government, and these local administrators in turn appointing the members to the two district-level bodies: councils and institutions of participation and community consultation.2 With this reorganisation of local governance, urban areas, broadly more strongly leaning to the ruling party, gained 2 Subnational governance at the intermediate level between districts/municipalities and central government is a mix between deconcentration and a weak form of decentralisation. The president of Mozambique appoints the provincial governors. Provincial assemblies were introduced to the governance system in 2005, but it was not until 2009 that members of provincial assemblies were elected by citizens in the respective jurisdictions. South African Journal of Economics Vol. ••:•• •• 201512 © 2015 Economic Society of South Africa. 31South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. greater political autonomy from the centre, while rural districts, many of which tended to lend their support to the opposition party, were more closely controlled by central government. The urban–rural differentiated strategy included not only local governance restructuring as described above, but also public resource allocation in agriculture. A new type of transfer from central government to the districts called the Local Initiatives Investment Budget (OIIL) was introduced in 2006. Although eligible areas of spending were initially meant to be diverse, starting from the first year’s disbursement and even most strongly as of 2007, central guidelines called for a large share of these funds – approximately half – to be allocated for agricultural development (World Bank, 2011a). These funds, channelling agricultural and other resources to rural areas, boosted Frelimo’s popularity in rural areas. Fig. 3 shows that the comprehensive strategies of Frelimo, including although not limited to the approaches detailed here, have led to rapid political gains across the country. They transformed all their negative margins of 1994 (in five provinces) to positive margins by 2009 (seen in Fig. 3 by the fact that there are no data points in the lower left quadrant) and retained their prior positive margins (no data points in the lower right quadrant). Even more, they have experienced an increase in their electoral margins over Renamo in areas in which they were historically disadvantaged that was faster than the increase in areas of their historical strength. This can be seen in the fitted line in Fig. 3 Figure 3. Evolution over time of Frelimo’s margin in parliamentary elections, by province Notes: “Margin” here refers to the per cent of votes received by Frelimo minus Renamo’s per cent of votes. Therefore, for example, the data point for Cabo Delgado reflecting “35” along the horizontal axis means that the per cent of votes received by Frelimo in 1994 in Cabo Delgado (57.8 %) exceeds Renamo’s analogous figure (22.7 %) by 35 percentage points. Source: Authors’ calculation based on statistics in Table 2. 13South African Journal of Economics Vol. ••:•• •• 2015 © 2015 Economic Society of South Africa. 32 South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. converging to the 45° line. For example, as the data underlying this graph in Table 2 show, in Sofala Frelimo lagged the most behind Renamo in 1994, namely by 62 points. By 2009, they turned the strongly negative margin to a high a positive one, of 30%. In contrast, in their strongest province in 1994 – Gaza – the margin improvement over the 15 years was only from 81% in 1994 to 97%. Table 2. Election results by province and year (% of votes) Province Parliamentary elections Presidential elections Frelimo Renamo Other Margin Frelimo Renamo Other Margin 1994 Cabo Delgado 57.76 22.67 19.57 35.09 67.92 18.63 13.45 49.29 Gaza 81.41 2.68 15.91 78.72 94.92 1.86 3.22 93.06 Inhambane 59.47 12.93 27.59 46.54 78.44 10.36 11.20 68.08 Manica 27.06 57.88 15.07 −30.82 33.61 50.42 15.97 −16.81 Maputo City 78.84 9.00 12.16 69.84 87.10 8.74 4.16 78.36 Maputo Province 77.56 6.95 15.49 70.61 89.94 6.17 3.89 83.77 Nampula 30.56 48.43 21.01 −17.87 37.15 42.80 20.05 −5.65 Niassa 46.26 32.76 20.98 13.50 56.53 26.90 16.57 29.63 Sofala 14.23 76.00 9.76 −61.77 17.53 73.48 8.99 −55.95 Tete 30.98 49.09 19.93 −18.11 40.43 42.37 17.20 −1.94 Zambezia 31.40 51.99 16.61 −20.58 38.38 47.54 14.08 −9.16 Mozambique 44.33 37.78 17.89 6.55 53.30 33.73 12.97 19.57 1999 Cabo Delgado 60.98 26.59 12.43 34.39 66.50 33.50 0.00 33.00 Gaza 86.90 3.51 9.59 83.38 95.20 4.80 0.00 90.40 Inhambane 61.89 20.55 17.56 41.34 71.10 28.90 0.00 42.20 Manica 31.70 56.89 11.41 −25.18 34.00 66.00 0.00 −32.00 Maputo City 82.63 13.55 3.82 69.08 86.70 13.30 0.00 73.40 Maputo Province 84.66 9.54 5.81 75.12 90.10 9.90 0.00 80.20 Nampula 39.11 43.87 17.02 −4.76 44.20 55.80 0.00 −11.60 Niassa 39.83 45.79 14.38 −5.96 43.00 56.90 0.10 −13.90 Sofala 19.31 70.43 10.26 −51.12 20.10 79.90 0.00 −59.80 Tete 37.06 49.16 13.78 −12.10 40.40 59.60 0.00 −19.20 Zambezia 25.93 59.12 14.95 −33.19 29.60 70.40 0.00 −40.80 Mozambique 48.54 38.81 12.65 9.73 52.29 47.71 0.00 4.58 2004 Cabo Delgado 75.60 17.30 7.10 58.30 77.60 18.00 4.40 59.60 Gaza 94.00 1.70 4.30 92.30 96.40 1.80 1.80 94.60 Inhambane 77.20 9.90 12.90 67.30 83.60 10.60 5.80 73.00 Manica 46.00 45.40 8.60 0.60 47.40 47.50 5.10 −0.10 Maputo City 81.90 13.60 4.50 68.30 85.20 12.90 1.90 72.30 Maputo Province 85.80 8.80 5.40 77.00 89.30 8.60 2.10 80.70 Nampula 48.30 40.00 11.70 8.30 49.80 43.90 6.30 5.90 Niassa 66.40 27.00 6.60 39.40 67.30 28.30 4.40 39.00 Sofala 25.60 64.90 9.50 −39.30 26.00 68.50 5.50 −42.50 Tete 74.00 21.10 4.90 52.90 74.20 22.20 3.60 52.00 Zambezia 36.30 52.70 11.00 −16.40 37.50 56.80 5.70 −19.30 Mozambique 62.14 29.78 8.08 32.36 63.74 31.74 4.52 32.00 2009 Cabo Delgado 80.85 14.90 4.25 65.95 Gaza 96.89 1.41 1.70 95.48 Inhambane 83.75 6.88 9.37 76.87 Manica 70.99 24.97 4.04 46.02 Maputo City 76.55 5.38 18.07 71.17 Maputo Province 88.39 7.76 3.85 80.63 Nampula 65.66 28.43 5.91 37.23 Niassa 81.28 13.11 5.61 68.17 Sofala 50.77 23.12 26.11 27.65 Tete 87.24 10.32 2.44 76.92 Zambezia 53.64 40.84 5.52 12.80 Mozambique 74.48 17.82 7.70 56.66 75.01 16.41 8.58 58.60 Notes: Electoral results by province for the 2009 presidential elections were not available to the authors. Source: Authors’ compilation and calculation based on data from the Electoral Institute for Sustainable Democracy in Africa (EISA); the African Elections Database; and Secretariado Técnico da Administração Eleitoral (STAE, 2002 and 2006). South African Journal of Economics Vol. ••:•• •• 201514 © 2015 Economic Society of South Africa. 33South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. A third approach was that of geographical targeting of public resources in agriculture. Studies point to several instances in which the government channelled public expenditures in agriculture to those provinces, districts and municipalities in which it previously lost elections. One example given is a big push by the government to rehabilitate four large sugar estates. Two of these estates were in locations where Renamo was either politically dominant or had strong strands of support (Buur et al., 2011). The sugar farm public investments in the Renamo strongholds were actively used on campaign platforms in the 1999 elections, with strongly improved results for Frelimo. Another case of targeting of Renamo areas for public investments in agriculture occurred in the context of the GPZ (Zambezi Development Authority), a large-scale programme to support production of crops and the forestry and fishery subsectors. All priority districts receiving support through the GPZ had elected Renamo in 1999 by a wide margin. Finally, the food production action plan (PAPA), an agricultural initiative launched in 2008 to counter the effects of the food price crisis, targeted as priority areas locations that went to Renamo in the 2004 elections (see Table 2 for more detail on the election results). 5.2 Donors: The Influence of International Aid on Public Expenditures in Agriculture Agriculture is a prominent sector for external development assistance in Mozambique. Of the 16 donor agencies providing direct budget support to the country in the mid-2000s, 12 have been engaged in this sector (Killick et al., 2005), with the health sector being the only one with a greater number of the direct budget support donors involved. The first phase of the main vehicle for coordinated donor funding of agriculture starting in the 2000s, the sector budget support, covered about 60% of the budget of the ministry of agriculture (Killick et al., 2005), and total aid to agriculture including projectised aid represented a bit more, some 66% of the ministry’s budget (de Renzio and Hanlon, 2007). Thus, a key element in the framework on the determinants of public expenditure decision making in agriculture is the role of donor engagement in Mozambique. Hanlon (2004) suggests that the donor community has substantial sway over both expenditure and other policies of the government, in part by trading off insisting on good governance of public resources against seeing through the implementation of their preferred policies, in favour of the latter. Hanlon (2004) gives one example of donors not immediately reacting to the misuse of agricultural funds for the sake of seeing their preferences met on wider ideological directions. Other work, in contrast, proposes that too much is made of donor organisations’ influence over allocative decisions in agriculture. Buur et al. (2011) give the example of large-scale investments as well as non-expenditure policy decisions in the sugar subsector. According to this study, the government and the ruling party had strong influence over the pace and timing of the rehabilitation of the sugar industry. Given that the broad direction of policies on this commodity, including reform and partial privatisation, was not contrary to the preferences of donors, the impetus for the policies could be misconstrued as having been implemented solely at the instigation of aid agencies and international investors. Rather, the study suggests, the government’s own realisation of the employment and export revenue potential of the industry and of private investors’ past track record in financing social services for farmers and workers on the sugar estates drove it to pursue investments necessary to revive sugar production and processing. As mentioned earlier, a significant initiative to channel international development assistance to agriculture in the country has been sector budget support, referred to in 15South African Journal of Economics Vol. ••:•• •• 2015 © 2015 Economic Society of South Africa. 34 South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. Mozambique as PROAGRI. PROAGRI unfolded in two phases: The first phase was initiated in 1998, and the second, PROAGRI II, was established in 2004 and came to a close in 2011. Cabral (2009) discusses the two phases of this sector budget support, and their achievements and failures. In terms of PROAGRI’s influence on agricultural public expenditures in Mozambique, interestingly the study proposes that there has been a crowding out of domestic government spending on the sector by the external funding offered through the budget support. While there have been substantial inflows of donor support through PROAGRI, these have not affected the basic metrics of overall spending on the sector, such as agricultural public spending as a share of national income or as a share of total public spending. Tracing domestic against external public financial flows to the sector, increases in the latter have been met by decreases in the former. While aggregate investments in the sector appear not to have been affected by the donor infusion through PROAGRI, at least according to the analysis3 in Cabral (2009), the allocation of agricultural expenditures have been, according to the same paper: As part of the heavy focus on institutional strengthening especially in PROAGRI’s first phase, the role of subnational agricultural offices has been strengthened, and this has made itself felt in a greater share of overall agricultural expenditures going to the province level than before the use of PROAGRI as a channel for donor support. The same institutional capacity building focus of the first phase of the sector budget support had another consequential allocative effect. Sector spending became much more skewed towards institutional development of agricultural public agencies, and away from the financing of the provision of agricultural services, inputs and public goods. Investments in institutional development included, for example, improvements of staff quantity and quality, establishments of management information systems, introduction and financing of processes in planning and budgeting such as the annual activity planning (PAAO) exercises in which local, provincial and central agencies participate, and so forth. Such institutional development investments took up about 45% of all agricultural spending on average in the period from 1999 to 2005. The allocative effect of donor engagement via PROAGRI on institutional development vs. direct service delivery came about not only given the underlying principles of PROAGRI that institutional capacity is a prerequisite for coordinated, coherent, and therefore successful agricultural policymaking. Another driving factor for this allocative effect has been the larger view about the role of the state underpinning PROAGRI and held by the donor community financing this facility. This view posited that government’s proper role in the agricultural sector is in providing a guiding regulatory framework for the private operators – farmers, processors, retailers of agricultural inputs and outputs, agribusiness, etc. – to be able to function in an environment characterised by sound property rights and limited distortion of incentives. The view proposes that government should refrain or at least limit engaging in direct productive activity, or the provision of those goods and services that the private sector would be incentivised to offering if the business environment is sound. These principles embraced by international development 3 In reviewing the literature on the impacts of agricultural public investments in developing economies, Mogues (2015) also discuss empirical work on agricultural aid fungibility. These papers provide examples of analytical approaches for measuring such sectoral aid fungibility that go beyond descriptive analysis on the topic. South African Journal of Economics Vol. ••:•• •• 201516 © 2015 Economic Society of South Africa. 35South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. investors into PROAGRI contributed to the shrinking of the share of agricultural public spending on direct goods and services (Castel-Branco, 2008; Cabral, 2009). 5.3 Agricultural Industry: Overlapping and Intersecting Interests of Industry and Political Elites An important driver of the extent to which a commodity or activity receives public policy attention has to do with the way that the roles of different actors – in particular, those who make policy and those who are affected by policy – overlap. Overlapping can be construed in different ways: Actors can take on each other’s roles over time, or they can be embedded in each other’s circles or arenas. Both are illustrated in Mozambique’s sugar subsector. Sugar was certainly not the sole subsector that had suffered decline and decay in the course of the civil war. However, as discussed above, the government chose to select cane production and sugar processing for special and heightened policy attention after the peace agreement. Part of what facilitated this, as documented by Buur and Whitefield (2011), is the phenomenon of strongly overlapping roles among the ruling party officials, government bureaucrats and private industry operating in the sugar sector. For example, key policymakers charged with developing the sugar rehabilitation programme had previously worked in the private sugar industry, or senior public bureaucrats providing regulatory oversight of the private activities in the subsector may sit on the board of a sugar company. Such embeddedness deepens information in both directions about the other entity’s constraints and interests, as well as creates social relations across the divide, both of which have increased public investments and improved public policies in favour of the (private sector actors operating in) sugar commodity sector. One can also gain insights about the driving forces behind agricultural investments just as well from exploring the reasons for lack of investments as for their presence. Policies intended and carried out in Mozambique’s fishery sector pose an interesting example of how incentives and interests of different core actors combined to thwart a set of policies to establish and expand semi-industrial fishery activities in the country (Buur, 2012). From among the three tiers of the fishing industry – the industrial, semi-industrial and small-scale subsectors – the medium-tier, semi-industrial sector was planned to be promoted by the government because of its potential to expand the role of domestic industry participants in fishery, as high-tier, industrial fishery was already dominated by international companies. And on the other hand, semi-industrial fishery development was seen by the government as a way to modernise production and processing of this commodity, while the small-scale or artisanal fishery sector was viewed as insufficiently commercially oriented. Investment and policy support for semi-industrial segment of fishery was thus properly enshrined in official policy documents, such as the Master Plan for Fisheries established in 1994. If one were to consider the formal planning process, as described earlier, as an adequate determinant of public investment choices, in a case like this one would draw the wrong conclusions from observing the strong formal place that semi-industrial fishery received, including the above master plan document in the planning process, as the actual public support for the semi-industrial fishery sector soon faltered and ultimately failed (Buur, 2012). First, although generally the government and ruling party were behind this policy, it was ironically undermined by contrary interests of key elites within the ruling party. These ruling elites had been benefiting all along from their access to licences and concessions in the industrial fishery sector, and a move to strengthening the semi- 17South African Journal of Economics Vol. ••:•• •• 2015 © 2015 Economic Society of South Africa. 36 South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. industrial segment would have constituted a threat to their established sources of rent. Similarly, existing businesses operating large industrial vessels used poorly articulated legal language in the new regulations to access the licences and rights that were reserved for semi-industrial fleets. Finally, the public agency for small-scale fishing, under which the newly established semi-industrial fishery was supposed to fall, showed benign neglect towards this subsector, with the agency influenced by concerned artisanal fishermen whom it originally served. The initial failure of the policy in light of this combination of contrary interests was further accelerated when donors’ enthusiasm towards a faltering endeavour waned. Generally, this suggests that a monolithic treatment of “government” or “ruling party” can often be misleading, and instead broader governmental policy preferences need to be separated from economic or other interests of elites and groups within government – these different sets of interests may, or may not, coincide at different instances. 6. CONCLUSIONS, AND REFLECTIONS ON THE POTENTIAL INFLUENCE OF CAADP ON AGRICULTURAL PUBLIC SPENDING IN MOZAMBIQUE What influences agricultural public investment decision making in Mozambique, and what are the implications for getting policymakers to make the needed investments? Because some key investments needed for agricultural development are typically not easily attributable to specific individual efforts or they take long for their results to be manifested – both of which lower the incentive of policymakers to undertake them – finding innovative ways to increase the political returns to such investments will be critical. Theory, existing evidence, as well as fieldwork evidence from Mozambique were examined to address the research question. Evidence on geographical targeting of agricultural public funds corresponds more closely with theories suggesting that resources are used to sway communities opposed to the ruling party, rather than to reward political supporters. The effect of donors’ role in public expenditures has made itself felt not so much in the aggregate flows to agriculture, but in the composition of these flows: The donor community’s general stance on the proper role of the state led to a stronger orientation of expenditures towards institutional strengthening and information systems than to more direct public sector involvement in the sector. Examination of the effect of actors’ and organisations’ incentives and constraints on resource allocation in agriculture points to the importance of not treating “government,” “the ruling party” and other institutions as monolithic bodies; the paper instead highlights how differentiated interests within seemingly coherent institutions drive what gets public expenditure attention in the agricultural sector. A vivid example in the fisheries sector demonstrates this particularly poignantly, with the interests of ruling party elites in one fishing subsector trumping the broader interests of the ruling party-led government plan to develop another fishing subsector. While this paper is focused on Mozambique as a case study, the phenomena uncovered have strong relevance beyond this country. First, the generalisability of the country- specific findings can, to some extent, be derived from the study’s methodology, which includes situating the Mozambican findings in the more general theoretical literature – as, for example, in the discussion of geographical targeting. Second, there are several elements in the case analysis that intuitively extend to other countries. For example, the need to understand how differential intra-ruling party interests drive public resource South African Journal of Economics Vol. ••:•• •• 201518 © 2015 Economic Society of South Africa. 37South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. allocation is present in other contexts – such as those with dominant-party systems, and de facto or de jure one-party systems – where opposition actors carry limited influence to effect the flow of public resources in agriculture or other sectors. With Mozambique fully entrenched in the CAADP process, a key question is how this process may affect resource allocation to agriculture. Building on its principle of inclusive participation, CAADP can, at least in principle, bring the civil society and media closer to the activities and deliberations of policymakers involved in the agricultural sector while educating the public on the country’s investment needs in non-partisan ways. This has the potential of increasing the attributability of and thus political returns to key but underprovided public goods and services. However, little is known on how this may have played out in other African countries. Granted, the main stages of the CAADP process are being completed in African countries only in recent years; thus, there has been little opportunity to study this potential effect. One of the insights from the few extant studies of CAADP in other African contents is that the extent of CAADP’s influence has been rather mixed. A better understanding of how and in what way it has a bearing on investments is crucial for improving support to the sector in Mozambique. This is in light of the serious policy and resource attention given to the CAADP process, the goal to have this process guide most if not all public investments for the sector, and the existing evidence that suggests that it is certainly not a given that the CAADP process tends to unfold as planned to have the envisaged effects. By way of last remarks, we offer some areas for investigation in future research. The CAADP process’ impact can be examined along the lines of the political economy framework discussed in this paper. In relation to the first element of the framework, CAADP has become formally embedded in Mozambique’s planning and budgeting process. The country’s core 10-year agricultural strategy document, PEDSA, makes this fully explicit (MINAG, 2011). Also, as described earlier, one of the key features of CAADP is to propose certain guidelines on what should be the nature of the process through which plans are forged and investments are decided upon. One should explore, then, to what extent CAADP is ultimately affecting investments by shaping the formal and de facto planning and budgeting process in Mozambique. With regard to the second element of the framework, one may ask whether or not CAADP will be – consciously or unknowingly – increasing the attributability of agricultural public investments in Mozambique, thus making allocation of funds to them more incentive-compatible. For example, if the CAADP process will bring the civil society and media closer to the activities and deliberations of policymakers in the agricultural sector arena, it may be that the enhanced visibility of what public officials concerned with agriculture are doing will bring more to public attention that the new or improved agricultural programmes are the work of these public officials. This heightened attributability in turn may increase officials’ investments in such programmes. This is not a conclusive outcome of course, but a testable hypothesis. Finally, Mozambique’s CAADP process may also reconstitute the roles, incentives and power of different actors in various ways that may differentially affect the magnitude and orientation of public resources going to the sector. Studies on different countries show that CAADP enhanced the relative influence of technocrats vis-à-vis civil society, which may have implications on investment choices. Other relevant questions include, for example, whether the envisaged strengthening of government ownership vis-à-vis donor determination of agricultural policy has in fact been realised in Mozambique, and 19South African Journal of Economics Vol. ••:•• •• 2015 © 2015 Economic Society of South Africa. 38 South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa. whether CAADP’s objective to make investment planning and programme implementation more evidence-based would give a greater role for analysts to shape resource allocation. These are outstanding questions worthy of future empirical investigation in the Mozambican context. REFERENCES AFRICAN UNION/NEW PARTNERSHIP FOR AFRICA’S DEVELOPMENT (AU/NEPAD). (2003). Comprehensive Africa Agriculture Development Programme. Midrand, South Africa: AU/NEPAD. BATLEY, R., BJØRNESTAD, L. and CUMBI, A. (2006). Evaluation of General Budget Support: Mozambique Country Report. Joint Evaluation of General Budget Support 1994-2004 Report Series, Birmingham, UK. BUUR, L. (2012). Mozambique Synthesis Analysis: Between Pockets of Efficiency and Elite Capture. DIIS Working Paper 2012:01, Copenhagen, Denmark. ———, MONJANE, C. and BALOI, O. (2011). Strategic privatisation: Rehabilitating the Mozambican sugar industry. Review of African Political Economy, 38(128): 235-256. ——— and WHITEFIELD, L. (2011). Engaging in a Productive Sector: Comparison between Mozambique & Ghana. Danish Institute for International Studies (DIIS), Working Paper 2011: 22. CABRAL, L. (2009). Sector Budget Support in Practice, Desk Study: Agriculture Sector in Mozambique. Overseas Development Institute and Mokororo. London, U.K. ———, SHRIVASTAVA, A. and MUENDANE, C. (2007). Formulating and Implementing Sector-Wide Approaches in Agriculture and Rural Development: (PROAGRI) – Mozambique. Bonn, Germany: Global Donor Platform for Rural Development. CASTEL-BRANCO, N. (2008). Aid and Dependency and Development: A Question of Ownership? A Critical View. Working Paper No. 1, IESE, Maputo. COX, G. W. (2009). Swing voters, core voters, and distributive politics. In I. Shapiro, S. C. Stokes, E. J. Wood and A. S. Kirshner (eds), Political Representation. Cambridge: Cambridge University Press, 342-355. COX, G. W. and MCCUBBINS, M. D. (1986). Electoral politics as a redistributive game. Journal of Politics, 48(2): 370-389. DIXIT, A. and LONDREGAN, J. (1996). The determinants of success of special interests in redistributive politics. Journal of Politics, 58(4): 1132-1155. DOMÍNGUEZ-TORRES, C and BRICEŃO-GARMENDIA, C. (2011). Mozambique’s Infrastructure: A Continental Perspective. Policy Research Working Paper, WPS 5885, World Bank. EUROPEAN COMMISSION & GOVERNMENT OF MOZAMBIQUE (EC). (2006). EC-Mozambique Joint Annual Report 2006. Prepared by National Authorizing Office and European Commission Delegation in Mozambique. FLAHERTY, K., MAZUZE, F. and MAHANZULE, R. (2010). Mozambique – Recent Developments in Agricultural Research. Washington, DC: ASTI – Agricultural Science and Technology Indicators. HODGES, T. and TIBANA, R. (2004). The Political Economy of Budget in Mozambique. Oxford Policy Management. Oxford, U.K. HANLON, J. (2004). Do donors promote corruption? The case of Mozambique. Third World Quarterly, 25(4): 747- 763. INTER-ACADEMY COUNCIL (IAC). (2004). Realizing the Promise and Potential of African Agriculture. Inter-Academy Council. Amsterdam, Netherlands. KILLICK, T., CASTEL-BRANCO, N. and GERSTER, R. (2005). Perfect Partners? The Performance of Programme Aid Partners in Mozambique 2004. A Report to the Programme Aid Partners and Government of Mozambique. LINDBECK, A. and WEIBULL, J. W. (1987). Balanced-budget redistribution as the outcome of political competition. Public Choice, 52: 273-297. MINAG. (2011). Plano Estratégico para o Desenvolvimento do Sector Agrário – PEDSA, 2011-2020. Ministério da Agricultura, Repúplica de Moçambique. Aprovado na v Sessão do Conselho de Ministros, Maio de 2011. MOGUES, T. (2015). What Determines Public Expenditure Allocations? A Review of Theories, and Implications for Agricultural Public Investments. European Journal of Development Research. REAUD, B. A. and WEIMER, B. (2010). Comparative Assessment of Decentralisation in Africa: Mozambique Desk Study. Paper prepared by ARD Inc. for USAID. de RENZIO, P. and HANLON, J. (2007). Contested Sovereignty in Mozambique: Dilemmas on Aid Dependency. Managing Aid Dependency Project. GEG Working Paper 2007/25, Oxford University. STADS, G.-J. and BEINTEMA, N. (2012). Agricultural R&D in Africa: Investment, Human Capacity and Policy Constraint. Improving Agricultural Knowledge and Innovation System. OECD Conference Proceedings, OECD Publishing. WORLD BANK. (2011a). Mozambique: Analysis of Public Expenditure in Agriculture, Volume 1: Core Analysis. Report No. 59918-MZ. World Bank, Washington, DC. ——— (2011b). How Do We Improve Public Expenditure in Agriculture: A Synthesis Report of the UK Department for International Development–World Bank Knowledge Partnership Project “Public Expenditures for Pro-Poor Agricultural Growth.” Joint Organizational Discussion Paper – Issue 3, World Bank, Washington, DC. South African Journal of Economics Vol. ••:•• •• 201520 © 2015 Economic Society of South Africa. 39South African Journal of Economics Vol. 84:1 March 2016 VC 2015 The Authors South African Journal of Economics published by John Wiley & Sons Ltd on behalf of Economic Society of South Africa.