Report No. 47455-MZ Mozambique Higher Fuel and Food Prices Impacts and Responses for Mozambique December 2008 Poverty Reduction and Economic Management, AFTP1 AFCS2 Africa Region Document of the World Bank ABBREVIATIONS AND ACRONYMS CGE Computable General Equilibrium model CNCS Conselho Nacional de Combate ao SIDA (AIDS National Council) CPI Consumer Price Index CTA Confederation of Economic Associations of Mozambique EU European Union GDP Gross Domestic Product GNP Gross National Product GoM Government of Mozambique HIPC Heavily Indebted Poor Countries ICA Investment Climate Assessment IFRS International Financial Reporting Standards IMF International Monetary Fund INE Instituto Nacional de Estatística (National Statistics Institute) INSS National Institute for Social Security (National Pension System) MDGs Millennium Development Goals MEC Ministry of Education and Culture MIC Ministry of Industry and Commerce MINAG Ministry of Agriculture MoF Ministry of Finance MoH Ministry of Health MOZAL Mozambique Aluminium Company MPD Ministry of Planning and Development MoPWH Ministry of Public Works and Housing MTEF Medium-Term Expenditure Framework MTFF Medium-Term Financial Framework MYR Mid Year Review NBR Net Benefit Ratio OE Orçamento do Estado (State Budget) PAP Programme Aid Partners PARPA Programa de Acção para a Redução da Pobreza Absoluta (Action Plan for the Reduction of Absolute Poverty) PCPI Poor Consumer Price Index PES Plano Económico e Social PMTCT Prevention of mother-to-child transmission PRSP Poverty Reduction Strategy Plan PSIA Poverty and Social Impact Analysis QUIBB Qualitative Indicators Survey SADC Southern African Development Community SISTAFE Integrated Financial Management System TA Technical Assistance VAT Value Added Tax iii TABLE OF CONTENTS EXECUTIVE SUMMARY................................................................................................................viii 1 Introduction........................................................................................................................... 1 2 International price changes and price transmission to Mozambique markets............... 2 3 Household-level analysis....................................................................................................... 5 3.1 Measuring the impact of higher food prices: the Net Benefit Ratio ................................... 5 3.2 Estimating the Net Benefit Ratios for Mozambique............................................................ 5 3.3 Estimating the impact of higher fuel prices...................................................................... 11 4 General Equilibrium Analysis of the Increase in Food and Fuel Prices........................ 13 4.1 Macroeconomic dimensions of the price shock ................................................................ 13 4.1.1 Balance of payments............................................................................................. 13 4.1.2 Macroeconomic aggregates .................................................................................. 14 4.1.3 Wages.................................................................................................................... 15 4.2 Impact simulations and results using a general equilibrium model of Mozambique ....... 17 4.2.1 Simulation descriptions......................................................................................... 18 4.2.2 Model results......................................................................................................... 19 5 Policy responses................................................................................................................... 23 5.1 Principles of a good policy response................................................................................ 23 5.1.1 Mitigating the impact of the price shocks by strengthening targeted safety nets . 25 5.1.2 Interventions to mitigate the domestic food prices through short-run trade policy measures or administrative action......................................................................................... 26 5.1.3 Interventions to mitigate the domestic fuel prices through short-run trade policy measures or administrative action......................................................................................... 27 5.2 Possible policy responses in Mozambique: An analysis of the fiscal cost and benefit incidence of alternative policy measures using first-order partial equilibrium analysis........ 28 5.2.1 Accelerate safety nets expansion .......................................................................... 28 5.2.2 Reduce import tariffs on milled grains to 2.5 percent .......................................... 33 5.2.3 Food price subsidies: maize, rice and wheat......................................................... 34 5.2.4 Fuel price subsidies............................................................................................... 35 5.2.5 Support to the semi-collective urban transport system......................................... 36 5.2.6 Summary of the impact of alternative policy measures using first-order partial equilibrium analysis.............................................................................................................. 37 5.3 Possible policy responses in Mozambique: An analysis of the impact of alternative policy responses using the CGE model............................................................................................... 38 5.3.1 Food and fuel subsidies......................................................................................... 40 5.3.2 Trade liberalization............................................................................................... 41 5.3.3 Agricultural technology ........................................................................................ 42 6 Conclusions.......................................................................................................................... 43 References.................................................................................................................................... 45 LIST OF FIGURES Figure 1: Changes in CPI and Poor-CPI in Maputo and Beira (December 2005 = 100).............. 4 Figure 2: Share of food in total household consumption............................................................... 7 Figure 3: Share of staple foods in household consumption: Maize, Rice, Wheat, Cassava .......... 8 iv Figure 4: Share of own food production in household food consumption ..................................... 9 Figure 5: Share of food in total household cash expenditures....................................................... 9 Figure 6: Net Benefit Ratios for Mozambique ............................................................................ 10 Figure 7: Share of fuel (Gasoline/Diesel) in total household consumption................................. 12 Figure 8: Share of public transport in total household consumption............................................ 12 Figure 9: Impact of food and fuel price shocks on poverty headcount........................................ 23 Figure 10: Assumed distribution of benefits from cash transfers safety nets program ................ 31 Figure 11: Distribution of benefits from removal of import tariffs on milled grains ................... 33 Figure 12: Distribution of benefits from a generalized fuel subsidy ............................................ 36 Figure 13: Distribution of benefits from a targeted subsidy to semi-collective transport ............ 37 LIST OF TABLES Table 1: Changes in international and domestic retail prices July 2005 to January 2009.............. 3 Table 2: Calculation of net benefit ratios for Mozambique.......................................................... 11 Table 3: Net benefit ratios by expenditure quintile for Mozambique.......................................... 11 Table 4: Trade in agriculture, food and petroleum products in 2006 ........................................... 13 Table 5: World price shocks simulated using CGE model of Mozambique................................. 18 Table 6: Macroeconomic results for world price shocks.............................................................. 20 Table 7: Sectoral production results for world price shocks......................................................... 21 Table 8: Factor price results for world price shocks..................................................................... 22 Table 9: Welfare and poverty results for world price shocks ....................................................... 22 Table 10: Summary overview of policy options........................................................................... 24 Table 11: Simulation of introducing a 20 percent price subsidy on wheat, maize, and rice ........ 34 Table 12: Summary table of simulated impact of policies responses using first-order partial equilibrium analysis1..................................................................................................................... 38 Table 13: Macroeconomic results for policy responses................................................................ 39 Table 14: Sectoral production results for policy responses .......................................................... 39 Table 15: Factor price results for policy responses ...................................................................... 40 Table 16: Welfare and poverty results for policy responses......................................................... 40 v PREFACE In response to the rapid increases in the food and fuel prices worldwide, in April 2008 the Government of Mozambique: (i) asked the UN team to come up with policy and program ideas in response to the higher food and fuel prices; (ii) set up an inter-ministerial Task-Force to come up with preliminary recommendations; and, (iii) asked the World Bank and IMF to organize a workshop on the role of subsidies in response to the higher prices. The workshop was held in Maputo in May 2008. The workshop was designed as a small forum for top policy makers to discuss policy options and implications with the Bank/IMF. This policy note summarizes the analysis and recommendations presented at the workshop. The first part of the policy note presents the analysis of the impact of the price increases in food and fuel at the macro- and micro level in Mozambique. Specifically, we disaggregate the impact for different types of households (by urban/rural, geographical, and income levels), and also estimate net impacts to households from buying and selling of agricultural products. In addition to this partial equilibrium analysis, we present the results of a general equilibrium model to account for economy-wide effects. The second part of the policy note focuses on policy response options for Mozambique. We discuss the possible use of trade policies, food subsidies, fuel subsidies, transport subsidies, and social protection measures, with examples from other countries. We then simulate the impact of specific policy options in Mozambique in terms of their fiscal costs, and the distribution of benefits across the population. The analysis was carried out by a team comprised of Antonio Nucifora (Team Leader, World Bank, AFTP1), Rui Benfica (World Bank, AFTP1), Felix Fischer (IMF), Channing Arndt (University of Copenhagen and DNEAP, Ministry of Planning and Development), James Thurlow (IFPRI), Nelson Maximiano (DNEAP, Ministry of Planning and Development), Jean Dupraz (UNICEF), and Patricia Fernandes (UNICEF). Numerous comments were received from participants in the workshop in May 2008 in Maputo. Valuable comments and advice were also received by the two peer reviewers, Harold Alderman (World Bank, AFTHD) and Hassan Zaman (World Bank, PRMPR), and also by Peter Nicholas (AFCS2), John Panzer (World Bank, AFTP1), Gregor Binkert (World Bank, AFTP1), Louise Fox (World Bank, AFTP1), and Young Kim (World Bank, AFTP1). Adelina Mucavele provided assistance in document preparation. Funding was provided by the World Bank. vi vii EXECUTIVE SUMMARY 1. The dramatic increases in world food and fuel prices during 2007 and early 2008 has put at risk Mozambique's considerable advances in poverty reduction during the past decade. Higher world food and fuel prices represent a negative terms-of-trade shock for Mozambique, since the country imports almost all of its fuel and is a net importer of food. However, the poverty impact of higher prices depends on a range of factors, including: (i) the structure of production and consumption at the household level; (ii) the extent of the agricultural supply response; (iii) the extent of export response; and (iv) the fuel intensity of the economy. On the one hand, higher agricultural prices may represent an opportunity to raise rural incomes, since about 80 percent of the labor force derives their livelihoods from agriculture and related activities. Conversely, many households rely on purchased food, particularly in urban areas, and so may be adversely affected by rising food prices. Moreover, higher fuel prices will also affect poverty due to fuel's economywide linkages, especially to Mozambique's burgeoning processing sectors. Finally, macroeconomic adjustments and public policy responses to accommodate the terms-of-trade shock will also affect household incomes. 2. Detailed price data show that the world price increases have been transmitted to domestic prices, with the increase in cost of living being slightly higher for the poor in Maputo. A poor persons' consumer price index (PCPI), developed for this analysis, indicates that the increase in the cost of the basket of commodities consumed by lower income households is similar to the increases registered for the average economywide basket. However, regional differences were observed. In Maputo, the price increases have been slightly faster for poor consumers, while the reverse is true in Beira. In both cities the difference between PCPI and CPI is not very large, at least for the period starting October 2007. 3. The first round of effects from Short term impact of higher food prices on the higher food prices is beneficial in welfare in Mozambique rural areas on average, while urban areas and the south rural areas are Net Benefit Ratios for Mozambique by Area of Residence losers, with Maputo/southern urban 20 poor particularly hit. First order 15 effects of the impact of the food price 10 shock have been estimated by 5 calculating the Net Benefit Ratios ) 15.3 10.1 %( 0 (NBRs).1 NBRs vary substantially Urban North Rural Center Rural South Rural BR -5 N -21.5 -1.3 across households by geographic -10 location and income group. Rural -15 households often benefit from their net -20 seller position, particularly those in the -25 middle of the income distribution. The Area of Residence net benefit ratio is 10 percent for the Source: Authors' calculations using the 2003 rural and -22 percent for the urban national household survey (INE, 2003). households. This means that, on 1The NBR proxies for a measure of the short run impact of food price changes on household welfare by calculating whether a household is a net consumer (buyer) or net producer (seller) of these food items. The NBR can be interpreted as the household elasticity of real income with respect to a food price change. viii average, a 10 percent food price increase would, in the short run, raise rural real incomes by 1.0 percent and reduce urban real incomes by 2.2 percent. The results also highlight sharp differences across rural and urban areas and regions within the country. The population in the urban south and center are most severely hurt by food price increase, followed by the urban north. Moreover, the population in the rural south would also be negatively affected. Based on the NBR, rural households in the north and the center benefit as a group from rising world prices. The poorest households in Maputo tend to lose the most in the short-run from food price increases, while middle-income groups in rural north and center would gain the most. 4. In terms of the higher fuel price, the short term impact on welfare at the household level is negative, independently of geographic location and level of income, with the richest households being affected the most. As expected, direct consumption of fuel in Mozambique is highly skewed towards the richest quintile, and is mostly concentrated in the South region and in Maputo city. Consumption of kerosene is slightly more evenly spread across the population as a percentage of household budgets. In absolute terms, however, most kerosene is consumed by the richest quintiles. Indirect fuel consumption for public transport services is highest in the middle three quintiles of the population, and particularly in Maputo city and in the South region. The richest quintile in Maputo city does not appear to use much public transport. In rural areas, use of public transport is slightly correlated with the level of income. Notably, the poorest quintile in the South region does not use much public transport. Indirect fuel consumption for transport of goods and services can be expected to be approximately proportional to the household's level of consumption (which we use as a proxy for household income). Hence, in absolute terms, richer household account for a much higher consumption than poorer households. An analysis of the distributional impact of changes in fuel prices in Mozambique was carried out in 2003 and found that a 20 percent increase in fuel prices is roughly equivalent to reduction of 1 percent in household income, across the population. 5. The long term and economy wide impact of higher world prices has been simulated using a comparative static Computable General Equilibrium (CGE) model of the Mozambican economy. CGE models provide a framework which is able to capture both first and second order impacts of price shocks. These models have a number of characteristics that make them suitable for analyzing external shocks, trade/tax policies, income distribution, and structural change. Four simulations have been run to analyze the impact of the price shocks. 6. The full impact of the fuel price increase is more than double that of the food prices. The simulations highlight that the decline in the terms-of-trade due to fuel price increases is more than double the decline due to food price increases. Macroeconomic impacts are commensurately larger. Compared with the food price shocks, the fuel shocks force a larger increase in the quantity of exports and a larger decrease in the quantity of imports in order to balance the external account. Due principally to these changes in trade flows, the decline in total absorption (or overall welfare) due to fuel prices (3.5 percent) is approximately double the decline registered under the food prices simulations. 7. The macroeconomic and poverty impacts of higher world food and fuel prices are negative and substantial. The simulation of the combined food and fuel price shocks that Mozambique actually received highlights that the effects are considerable. Terms-of-trade declined by more than 16 percent, and in order to balance the external account, exports increased by nearly 6 percent and imports declined by almost 10 percent. These shifts in production generate a decline in GDP of slightly more than 1 percent. All of these adjustments imply a ix reduction in the quantity of goods and services in the economy resulting in a reduction in absorption of more than 5 percent. Since recurrent government expenditure is assumed to be fixed and investment declines by only 1.2 percent, household consumption bears the bulk of the adjustment, declining by more than 7.0 percent. This is a substantial decline in a country where approximately half of all households are absolutely poor. 8. The negative welfare impact for rural households is less severe than for urban households in all simulations. Substantial home consumption amongst rural households provides considerable insulation from both fuel and food prices shocks. In addition, rural wages rise relative to urban wages, particularly in response to the higher food prices. As a result of these consumption and income impacts, rural households are less strongly affected than urban households in all simulations. The stimulating effect of improved agricultural terms-of-trade for the rural economy does not outweigh the negative impacts of the fuel shock, and Simulation of the impact of food and fuel price welfare declines for all households under shocks (various scenarios) on poverty levels the simulation of the combined food and Impact of the shocks on poverty headcount fuel price shocks. The degree of land 60 ownership is the primary factor ne)il yt 59 differentiating outcomes across quintiles er 58 pov in rural areas. tnu 57 eht 56 co 9. Poverty impacts are large, ow 55 eadh bel particularly in urban areas. The yt 54 sd 53 simulations highlight that the combined veroP ehol 52 shocks result in a four percentage point hous 51 increase in the national poverty of 50 Base year Fuel Food Food Combined headcount rate. The effect is much %( value (Fixed land) (Flexibl land) stronger in urban areas where the National Rural households Urban households poverty rate increases by eight percentage points. In fact, the combined Source: Results from the Mozambique CGE model simulation of the recent shocks sets the urban poverty rate above the rural rate. Fuel price increases are the principal driver of increased poverty in both rural and urban zones. As would be expected, the capacity to reallocate land reduces poverty with the effect being slightly stronger in rural zones. Recommended policy responses 10. Increasing world prices for food and fuel presents policy makers with a range of policy challenges. Allowing the pass through of increases in global prices to domestic prices will reduce real incomes of households, leading to an increase in poverty rates. On the other hand, attempting to prevent passing through higher prices by either subsidizing or reducing taxation can result in a substantial fiscal burden that needs to be financed, and will introduce distortions in consumption patterns, thereby exacerbating adverse balance of payments impacts (or reducing balance of payments benefits). For instance, not passing through the price increases would reduce the incentives to increase agricultural production, and exports, which is a great opportunity for Mozambique. Further, not passing through the international price increases to the domestic markets is often undesirable on equity grounds, since most of the benefits from the subsidies or lower taxation to shield the domestic prices, would accrue to the richer groups in the population. The best policy is to allow the domestic prices to adjust in line with international x prices, while protecting the poorest households by scaling up targeted social safety nets. A few good principles can guide policy makers to identify the best policies to respond to the crisis: (i) short-term policies should ideally support and definitely not undermine, long term priorities; (ii) policies should preserve incentives (e.g. labor, agricultural production); (iii) interventions should be targeted to the most vulnerable groups; (iv) the costs should be limited, and leakages outside national borders should be minimized; (v) the measures should be easy to implement and introduce; (vi) the measures should entail minimum management and/or governance concerns. 11. The best short term measure to alleviate the impact of the shock on Mozambican households is to scale-up the direct cash-transfers program. Social protection programs (preferably cash transfers) offer the best policy responses to mitigate the impact of the shocks for poor households in the short term. These measures support the purchasing power of the poor without distorting domestic incentives to produce more food, and without reducing the incomes of poor food sellers. In Mozambique, scaling up the cash-transfers program (Programa de Subsidio de Alimentos, PSA) would be fiscally manageable and reasonably well targeted. The simulation of an immediate scaling-up scenario of the PSA targeted cash transfers program from 4 percent in 2008 to 14 percent of population by 2010, suggests that the fiscal cost would increase from the current US$14 million, to about US$33 million in 2010, which would almost entirely reach the beneficiaries given the relatively low administration costs of the PSA. No data exist on the benefit incidence of the PSA program, but based on the experience with reasonably well targeted cash transfers schemes in other countries we can expect a leakage around 20 percent of funds to the non-poor, with the richest quintile getting less than 5 percent of the transfers and the poorest quintile receiving about 40 percent of the transfers, corresponding to an increase in incomes of about 17 percent for the poorest quintile. 12. Other recommended policy measures in the short term include reducing tariffs on imported milled grains, and scaling-up of the targeted lump-sum compensation to the semi- collective transport industry. Reducing tariffs on imported milled grain (wheat and maize flour) is also a good policy response. However, in order to avoid a collapse of the milling industry, it would be best to accompany the removal of the tariffs with a temporary direct transfer to the milling industry (possibly decreasing over time). Targeted lump-sum compensation to the semi-collective transport industry (i.e., to private mini-vans operators) is reasonably efficient. 13. The least preferable policy would be the provision of a blanket fuel subsidy (or fuel tax removal). The analysis presented in this study highlights that a generalized fuel subsidy would be extremely expensive and highly regressive. In order to examine the likely impact of a fuel subsidy in Mozambique, we have carried out a back-of-the-envelope simulation of introducing a 20 percent price subsidy on gasoline and 10 percent on diesel. The simulation used data from the 2006 national accounts. The fiscal costs of the subsidies would be extremely high, at around US$100 million per year (equivalent to 1.25 percent of GDP in 2007). Further, as discussed in Section 3, fuel is used mostly consumed by the richer part of the population, and as a result a blanket subsidy to fuel would be highly regressive. As expected, most of the benefits would accrue to the richest quintile, who would receive around 71 percent of the value of the subsidy. The poorest quintile of the population would only receive 3 percent of the benefits. In both cases, the subsidy would correspond to just a small increase in incomes, at 1 percent for poorest quintile and 3 percent for richest. xi Distribution of benefits from a direct cash transfer safety net program compared to the introduction of fuel a subsidy Assumed distribution of benefits from Distribution of benefits from cash transfers the fuel subsidy 70 80 Share of benefits 60 Assumed Shares 70 (National) of benefits 50 60 Share of benefits e Subsidy as % of (Maputo) 50 agt 40 total consumption Subsidy as % of total 40 cen 30 consumption (National) er egatnecr 30 Subsidy as % of total P 20 Pe consumption (Maputo) 20 10 10 0 0 Poorest 2 3 4 Richest Poorest 2 3 4 Richest Welfare quintiles Welfare quintiles Source: Authors calculations 14. Food subsidies to maize and rice are also not recommended, as they are expensive and introduce a dis-incentive to domestic production. However, they are better targeted to the poor compared to wheat subsidies, and particularly rice is well targeted to the areas most negatively affected by the higher food prices. 15. The analysis of policy responses using the CGE model highlight the importance of agricultural production response and points to difficult trade-offs between short-run mitigation measures and long-run growth. The results of the CGE simulations highlight the importance of export supply response with particular emphasis on the agricultural sector. In the long term, improved agricultural technology is the preferable policy response to higher world prices. As a result of the productivity gain and associated increase in agricultural production, the economy achieves substantial gains in exports and a reduction in imports. However, while improving agricultural productivity is most effective in addressing the adverse effects of higher food (and fuel) prices, expanding agricultural production in the short term will remain difficult despite improved agricultural terms-of-trade, and is unlikely to address the immediate impacts of the current food and energy crisis. Final remarks 16. The main recommendations are aligned with ongoing policy initiatives. In sum, in the short term the Government could: (i) accelerate the expansion of the PSA direct cash-transfers safety net program; (ii) continue to liberalize trade by reducing tariffs on imported milled grain (wheat and maize flour), perhaps with temporary direct assistance to the milling industry; and (iii) continue targeted compensation to private urban transport operators which were started in February 2008 (in exchange for keeping fares constant). These measures need to be complemented by medium term measures to improve agricultural productivity and increase domestic supply. Over the medium term it is also important to strengthen the safety nets system by channeling the beneficiary group towards the poorest and most vulnerable and strengthening the capacity to administer the program. Also, there is a need to develop long-term solutions for urban transport. We advise against general subsidies and trade restrictions. xii 17. The long term impact of the shocks will likely be smaller than simulated in this paper as the international prices of food and fuel have recently decreased. In recent months, international fuel prices have gone back to pre-crisis levels and food prices have decreased from their peak of mid-2008. In as far as the price increases may have been a temporary event, the main impact of the shocks in 2008 would be to slow down the pace of poverty reduction. However, if prices remain higher than they were before the shocks, it can be expected that there will be a permanent impact. This study illustrates the type of tools and analysis that can be used to assess the impact of a range of terms-of-trade shocks, and what policy responses might be best applied. 18. The recent commodity price volatility is not unusual and underlines the need for policies to allow the Government to implement a rapid and flexible response, adapted to the level of the shocks. The crisis holds an important general lesson for Government policy in response to external terms-of-trade shocks. Protecting vulnerable households from the impact of the external shocks by strengthening safety nets is the long run ideal response to external terms- of-trade shocks. The best policy measure is to provide cash or near-cash transfers that are conditional upon meeting a requirement (such as low income, location or occupation) or engaging in a mandated behavior (such as sending children to school). One drawback of cash transfer programs is that the fixed costs to set up and administer such a program are significant, such that these programs are not cost-effective for only a short term intervention (e.g. to respond to a single climatic event). An important advantage of cash transfer programs, however, is that they are flexible so that, once the fixed costs to establish a cash transfers program has been incurred it can be adapted to different shocks. The recent events have underlined that we live in a world where commodity prices exhibit substantial volatility and external shocks are a recurrent feature (even if their nature and timing are unknown). For instance, the recent global financial crisis is expected to have a negative impact on Mozambique, and again enhances the importance of social protection programs. It makes sense, therefore, to build-up a sizeable and cost-effective cash transfers program targeted to the poorest and most vulnerable households to provide the Government with a flexible response mechanism to a range of shocks. xiii HIGHER FUEL AND FOOD PRICES: IMPACTS AND RESPONSES FOR MOZAMBIQUE 1 INTRODUCTION 1. The dramatic increases in world food and fuel prices during 2007 and early 2008 may set back Mozambique's considerable advances in poverty reduction during the past decade. Mozambique has made tremendous strides in reducing poverty over the past 14 years, following the conclusion of the civil war in 1994. Household survey data indicate that the national poverty headcount fell from 69 to 54 percent during 1997-2003. Reduction in rural poverty has been even more pronounced, although the proportion of people who are poor in these areas remains higher than in urban centers. Given these trends and with the country still growing rapidly, it was expected that the next household survey due in 2009 would confirm that poverty has continued falling. The rapid increases in world agricultural and fuel prices over the past two years may set back at least some of these gains. 2. This study assesses the impact of higher fuel and food prices at both household and macroeconomic levels, and also considers policy options to mitigate some of the negative impacts of higher prices. Rising world prices certainly represents a negative terms-of-trade shock for Mozambique, since the country imports almost all of its fuel and is a net importer of food. However, the poverty impact of higher prices depends on a range of factors, including: (i) the structure of production and consumption at the household level; (ii) the extent of the agricultural supply response; (iii) the extent of export response; and (iv) the fuel intensity of the economy. On the one hand, higher agricultural prices may represent an opportunity to raise rural incomes, since about 80 percent of the labor force derives their livelihoods from agriculture and related activities. Conversely, many households rely on purchased food, particularly in urban areas, and so may be adversely affected by rising food prices. Moreover, higher fuel prices will also affect poverty due to fuel's economywide linkages, especially to Mozambique's burgeoning processing sectors. Finally, macroeconomic adjustments and public policy responses to accommodate the terms-of-trade shock will also affect household incomes. 3. In recent months international fuel prices have gone back to pre-crisis levels and food prices have partially fallen back; hence, the permanent impact of the shocks is expected to be much smaller than initially anticipated. In as far as the price shock has been a temporary event the main impact will be to temporarily slow down the pace of poverty reduction. At present food prices remain substantially higher than they were before the shocks, and therefore some of the impact will be permanent. More important, the analysis presented in this report holds valuable general lessons for Government policy to respond to external terms-of- trade shocks. This study illustrates the type of analysis that can be used to assess the impact of a range of terms-of-trade shock, and what policy responses might be best applied. 4. The report is structured in six sections. Section 2 presents information on the extent of international food and fuel price increases and their transmission to local markets in Mozambique. Section 3 presents household-level analysis focused on the first order impact of the food price increases. Section 4 complements previous sections by examining the impact of higher food and fuel prices within a general equilibrium framework. Section 5 discusses the likely impact of alternative policy options in the short and long term. A final section summarizes and concludes. 1 2 INTERNATIONAL PRICE CHANGES AND PRICE TRANSMISSION TO MOZAMBIQUE MARKETS 5. The Government has allowed the recent movements in world agricultural and fuel prices to pass through to domestic markets, by avoiding introducing trade distortions or subsidies while providing support to economic sectors most vulnerable to rising energy prices. Table 1 compares nominal fuel and staple food prices in international and Maputo retail markets. The increase in domestic fuel prices is consistent with the increase in international prices. While the pass-through has been lower for rice and maize, this may reflect the timing of the local harvest in cushioning the increase in local retail prices, and some additional pass-through has Box 1: Price transmission This Box briefly describes the relationships which regulate price transmission from international to domestic markets. The relationship between a commodity's international and domestic prices can be expressed as: where the consumer price (Pc) is equal to its `free-on-board' price (Pfob) multiplied by the exchange rate (e) and adjusted for insurance and freight (acif) and border taxes (). Added to this is a second term accounting for internal transport costs (tr) and commercialization mark-up (). In absolute terms, an increase in international prices is expected to translate into an even larger absolute increase in domestic prices. In percentage terms, however, domestic prices are not expected to vary as much as international prices. Depending on the relative weight of the two terms in the equation, a given percentage increase in world prices may translate into a smaller percentage increase in domestic retail prices. Notably this occurs if the second term accounts for a large share of the domestic retail price, which is typically the case. For example, Mozambique's petroleum pricing structure is such that the first term accounts for approximately only half of the retail price of fuel products. The `Law of One Price' suggests that without trade barriers and with costless trade and perfectly functioning markets, prices for the same commodity should be equal in all countries. Most empirical evidence on border price transmission, however, rejects the Law of One Price, finding that international and domestic prices do not converge in the short run (see Goldberg and Knetter, 1996 for an overview). This rejection is generally attributed to the violation of the assumption of costless trade and arbitrage, or to the existence of price rigidities and imperfect competition. Imperfect competition can exist both on the export-side, where firms may apply price discrimination, and on the import-side, where distribution mark-ups impede perfect price transmission. The role of imperfect competition is highlighted by the empirical observation that movements in exchange rates are not fully reflected in domestic prices. Estimated exchange rate pass-through to import prices is around 60 percent in the US (Campa and Goldberg, 2005). Smaller countries typically experience larger pass-through (Frankel, Parsley and Wei, 2005). Exchange rate pass-through to consumer prices tends to be even lower than to border prices (i.e., around 13-30 percent, see Campa and Goldberg, 2006). Pollard and Coughlin (2004) find that exchange rate pass-through is asymmetric, with firm's pricing policies reacting differently to appreciations than depreciations. Studies allowing for the existence of thresholds to arbitrage, where prices converge only if price differentials are above a certain threshold that makes arbitrage profitable, tend to find higher pass-through (Obstfeld and Taylor, 1997). The most recent literature has focused on the import-side and the fact that border prices seem to be more sensitive to exchange rate changes than consumer or retail prices (Goldberg and Campa, 2006; Frankel, Parsley and Wei, 2005). There are three possible explanations: (i) the existence of non-tradable goods in the CPI; (ii) the distribution sector reduces the foreign content value of imports, and (iii) imperfect competition in the distribution sector implies double marginalization, where distributors adjust profits and margins according to exchange rate fluctuations in order to expand their market share. Goldberg and Hellerstein (2006) model pass- through in a Bertrand oligopolistic setting with differentiated products, where retailers set prices according to the ones given by wholesalers. In this setting, firms adjust margins in response to exchange rate changes, reducing full pass-trough. 2 occurred over the most recent months. By contrast, wheat prices have increased much faster than international prices, since virtually all domestic consumption is from imports. More recently, as international prices have begun to fall, the domestic price of wheat has started to decrease, while the price of rice and maize have continued to increase, albeit at a slower pace. The retail price of fuel at the pump has begun to fall rapidly in line with international oil prices. 6. The slower than anticipated increase in domestic retail prices of staple foods may reflect some degree of imperfect competition. In a study of Mozambique, Cirera and Nhate (2008) quantify empirically the degree of price transmission from border to retail prices. They find that transmission of changes in cif unit values or trade taxes tends to be small or not statistically significant. This suggests that firms can use distribution margins to offset changes in import unit values and trade taxes, thus indicating evidence of imperfect competition.2 This may partly explain the slow increase in domestic retail prices of staple foods discussed above. Table 1: Changes in international and domestic retail prices July 2005 to January 2009 July July July July Sept January Change, 2005 2006 2007 2008 2008 2009 July 2006 ­ July 2008 (%) Average international price Rice, Thailand, 5% (US$/ton) 277.0 315.0 329.0 732.0 683.8 580.0 132.4 Maize (US$/ton) 108.0 114.0 147.0 265.0 233.8 172.8 132.5 Wheat, US, HRW (US$/ton) 144.0 202.0 238.0 328.0 295.6 239.1 62.4 Crude oil, spot (US$/barrel) 56.0 72.0 74.0 133.0 99.66 43.86 84.7 Average retail price in Maputo Rice (Meticais/kg) 9.3 11.4 14.5 19.3 19.0 20.5 68.8 Maize (Meticais/kg) 5.9 6.5 6.4 10.2 11.4 13.0 57.0 Wheat flour (Meticais/kg) 11.0 11.8 15.5 24.5 25.7 23.6 107.6 Gasoline (Meticais/liter) - 27.2 33.7 41.6 41.6 24.3 52.9 Diesel (Meticais/liter) - 27.2 27.5 35.4 35.4 28.1 29.9 Kerosene (Meticais/liter) - 16.5 20.3 28.7 28.7 19.5 74.1 Rice average retail prices (US$/ton) Beira 391.9 307.9 503.9 1,141.1 1129.3 - 270.6 Chimoio 419.9 488.1 596.9 1,190.9 1185.5 1129.5 144.0 Maputo 381.1 452.4 562.0 800.8 798.8 806.7 77.0 Nampula 373.2 460.7 418.6 1,020.7 1064.5 806.7 121.6 Pemba 512.3 555.6 542.6 1,120.3 1157.0 984.3 101.6 Tete 532.8 634.9 515.5 1,161.8 1058.3 984.3 83.0 Exchange rate (Meticais/US$) 24.4 25.2 25.8 24.1 24.2 25.4 - Source: The World Bank, Development Economics Prospects Group (http://decpg.worldbank.org); Ministry of Agriculture, Weekly Agriculture Market Bulletin (http://www.sima.minag.org.mz/index.htm); Ministry of Energy. 7. Broader measures of domestic price movements also indicate considerable price transmission. Figure 1 presents the consumer price index (CPI) for the major cities of Maputo and Beira.3 Fuel and food represent more than half of the consumption bundle in each city.4 Both 2Cirera and Nhate (2008) also find that changes in the exchange rate tend to be fully transmitted. However, in principle one would expect imperfect competition to affect similarly transmission of cif unit values or trade taxes, and exchange rate movements. 3Similar results to Beira were obtained for the northern city of Nampula. 3 graphs indicate rapid price rises from October 2007. The rate of growth in prices is considerably more rapid in Beira reflecting both a higher share of fuel and food in the basket and a more rapid increase in the price of the food basket. Figure 1: Changes in CPI and Poor-CPI in Maputo and Beira (December 2005 = 100) 150 Maputo CPI 140 Maputo Poor-CPI 130 Beira CPI ndexI ecir Beira Poor-CPI 120 P er 110 um nso C 100 90 80 40.c 50.b 50r. 50.n 05 50.t 50.c 60.b 60r. 60.n 06 60.t 60.c 70.b 70r. 70.n 07 70.t 70.c 80.b 80r. De Fe Ap ug. Ju A Oc De Fe Ap ug. Ju A Oc De Fe Ap ug. Ju A Oc De Fe Ap Source: Mozambique National Institute of Statistics (INE 2008) and authors' calculations 8. The changes in the cost of living have affected the whole population, but have been higher for the poor in Maputo. We compare the evolution of prices in the consumer price index (CPI) with a specific poor-consumer price index (PCPI), which focuses on the goods consumed by the poor. With respect to food, the weights in the CPI calculation were adjusted to reflect both the composition of the food basket of the poor and the weight of food in total consumption of the poor. With respect to non-food items, quality and units issues make adjustment more difficult. To obtain the poor consumers' non-food basket, non-food items manifestly not consumed by the poor, such as airline tickets and automobiles, were eliminated from the CPI non-food basket. The weights on the remaining non-food items were then scaled such that the sum of weights for non-food items equals the observed non-food consumption share of poor households. The adjustments to develop the PCPI in Maputo and Beira reflect observed consumption patterns of the poor in each city. In both cities, the PCPI, like the CPI, exhibits rapid growth since October 2007. In Maputo, the price increases have been slightly faster for poor consumers, while the reverse is true in Beira. In both cities the difference between PCPI and CPI is not very large, at least for the period starting October 2007. 9. Data published by the Ministry of Agriculture suggest that price movements have been transmitted to internal markets as well. Table 1 shows retail rice prices in selected local markets in July 2005 to January 2009. Retail prices rose substantially during this period throughout the country, both in port cities, such as Beira, and in inland markets, such as Tete and Chimoio. While percentage price increases are sometimes larger in inland markets, these were 4It should be noted that a price rise in one commodity can lead to a rise in another, even one that is not imported (such as cassava) via upward pressures from substitution as measured by cross price parameters. 4 often from lower initial price levels. Absolute price changes were more consistent across markets, reflecting the transaction cost wedge between local and border prices. Overall, domestic market price trends suggest that the food price increases have been widespread and will affect households throughout the country.5 It is also worth noting that relative prices across markets vary substantially from one month to the next. While this observation may say more about data than markets, it also underscores the structural challenges in Mozambique. In fact, even with a perfectly competitive market infrastructure, high internal transport costs can lead to periods in which the bands between local import and export parity prices are wide and internal markets are effectively delinked from the ports or from South Africa. 3 HOUSEHOLD-LEVEL ANALYSIS 3.1 Measuring the impact of higher food prices: the Net Benefit Ratio 10. The first order welfare impact of rising food prices depends on whether a household is a net consumer (buyer) or net producer (seller) of these food items. Typically, the urban poor are net consumers and are thus adversely affected by higher food prices. Effects on the rural poor are more varied, since they depend on the structure of consumption and household crop production and marketing. In a given country, regional differences can be expected and the average household net position may vary by crop. 11. The first order effects of the impact of a food price shock can be estimated by calculating the Net Benefit Ratio. In a seminal piece related to the estimation of the short-run welfare impact of price changes on household welfare, Deaton (1989) postulated that the first- order welfare effect of relative food price changes is proportional to the Net Benefit Ratio (NBR). This ratio is the difference between the consumption and production ratio. The `consumption ratio' is defined as the elasticity of the cost of living with respect to changes in price, which is driven by consumption shares. The `production ratio' is the elasticity of food sales to total household monetary income. The NBR proxies for a measure of the short run impact of food price changes on household welfare, and can be interpreted as the elasticity of real income with respect to a food price change. For net producers this elasticity is positive and for net consumers it is negative. The basic model used by Deaton can be represented as follows for a single household: where w is the change in welfare, p is the food price change, and PR and CR are the food production and consumption ratios respectively. The proxy used for the production ratio (PR) is the share of the value of agricultural sales and own production in total household income, while the proxy used for consumption (CR) is the share of the value of food purchases and own consumption in total household expenditures. 3.2 Estimating the Net Benefit Ratios for Mozambique 12. Food is very important in household consumption and expenditures in Mozambique, especially in rural areas. Consumption ratios are calculated using data from Mozambique's 2003 nationally representative household survey, which contains the most recent information on 5The main harvest month is in May/June. The high prices registered in July 2008 could also reflect a poor 2008 harvest. Reliable information on the quality of the harvest will not be available until the end of 2008. 5 Box 2: International experience: the impact a food price shock A brief literature review of the empirical application of the Net Benefit Ratio approach is provided by Zaman (2008). Deaton, in his work in Thailand, showed that, relative to either the poorer or wealthier rural households, it was middle-income farmers that benefited the most from an increase in food prices. These methods were subsequently applied by Barret and Dorosh (1996) using data from Madagascar, Budd (1993) in Cote d'Ivoire and Klytchnikova and Diop (2006) in Bangladesh. These techniques do not allow for any behavioral change on the part of producers/consumers (i.e., production and consumption patterns remain unchanged). The analysis thus illustrates the first order impact of the food price shock. The 2008 World Development Report (World Bank 2008) shows that, in four out of seven surveyed countries (Bolivia, Ethiopia, Bangladesh, Zambia), the rural poor are on average net consumers, while in three other countries (Cambodia, Madagascar, Vietnam) they are on average net producers (see Table 2). However, most empirical analyses suggest that the rural poor are net consumers (Weber et al., 1988; Christiansen and Demery, 2007) and therefore suffer from higher food prices. As discussed in Zaman (2008), this is because the rural poor are often constrained by small landholdings, low productivity, input costs and distance to markets. As a result, they are generally unable to produce the marketable surplus required to exceed their food expenditures. These expenditures are typically financed via the sale of household labor and engagement in other nonfarm activities. Net buyer/seller positions of eight lower-income countries Bolivia Ethiopia Bangladesh Zambia Cambodia Madagascar Vietnam 2002 2000 2001 1998 1999 2001 1998 Share of internationally traded staples 25.5 24.1 41.2 40.4 56.3 62.7 64.4 in food consumption of the poor (%) Distribution of poor (%) Urban (buyers) 50.9 22.3 14.9 30.0 8.4 17.9 6.1 Rural landless (buyers) 7.2 - 53.3 7.4 11.5 14.8 5.8 Smallholders net buyers 29.1 30.1 18.8 28.8 25.8 18.9 35.1 Smallholders self-sufficient 7.1 39.5 4.6 20.8 18.0 27.3 19.4 Smallholders net sellers 5.6 8.0 8.4 13.0 36.3 21.1 33.6 Source: The World Bank, World Development Report 2008 (www.worldbank.org) Second order adjustments, such as shifts in consumption (production) away from (to) commodities with relatively large price increases should dampen any negative first order impacts. A CGE model can be used to evaluate the long terms and economy wide impact of price changes. A CGE model was used in Indonesia to assess the impact of changing rice tariffs (SMERU 2003). It found that the overall welfare impact from the proposed change in rice tariffs was small in relation to average household income, but that this effect varied significantly across household groups. Specifically the increase in real wages for unskilled workers is insufficient to make up for the increased living costs for the poor. More recently Ivanic and Martin (2008) find that in many poor countries, the recent increases in prices of staple foods raise the real incomes of those selling food, many of whom are relatively poor, while hurting net food consumers, many of whom are also relatively poor. The overall impact on poverty depends upon the balance between these two effects, and can only be determined by looking at real-world data. Results using household data for ten observations on nine low-income countries show that the impacts of higher staple food prices on poverty differ substantially by commodity and by country, but, that poverty increases are considerably more frequent, and larger, than poverty reductions. 6 household incomes and expenditures (INE, 2003).6 As indicated above, a households' consumption ratio is determined by its expenditure shares. At the national level, the share of food in total household consumption is 60 percent. It is highest for rural households in the north and center regions of the country (about 70 percent) and lowest for the richest quintile in Maputo city (11 percent) (see Figure 2). Figure 2: Share of food in total household consumption URBAN AREAS RURAL AREAS 80 80 70 70 60 60 ) ) 50 %( %( 50 era e 40 arhS40 Sh30 30 20 20 10 10 0 0 1 2 3 4 5 1 2 3 4 5 Quintiles of Consumption pc Quintiles of Consumption pc Maputo City South Center North South Center North Source: Authors' calculations using the 2003 national household survey (INE, 2003). 13. The consumption of individual food staples varies substantially across regions and income groups. Maize has significant share of the poor's budget in the urban areas in north and center, and its consumption is slightly negatively correlated with income (Figure 3). Rice accounts for a significant share of the poor's budget in Maputo and urban areas in south and center, and in rural south, but again its consumption is only slightly (negatively) correlated with income. Wheat products are significant in Maputo city and in the south, but are not mainly consumed by the poor. In fact wheat is a luxury for north/central households even in urban areas. Cassava is important in the budget of poor urban households in the north, but not anywhere else. 14. Own production is important in Mozambique, accounting for three quarters of rural household food consumption. This suggests that rural households may be fairly insulated from variations in market prices. Indeed, even though food accounts for a larger share of total consumption of rural households, the share of purchased food in total food consumption is lower for rural households (25 percent) compared to urban households (81 percent). Higher dependence on marketed foods is also observed at higher income levels and in the southern region (see Figure 4). Home consumption is less prevalent for urban households in general, and in the capital city, Maputo, consumption of own-produced foods is virtually nonexistent. 6The survey was carried out from July 2002 to June 2003 and interviewed 8700 households with the reference period for consumption being one week. Households were visited at least three times during the reference period. More recent data on rural farm household net buyer status is shown in Boughton et al. (2007) and Tschirley and Abdula (2007). 7 Figure 3: Share of staple foods in household consumption: Maize, Rice, Wheat, Cassava Maize Shares in Urban Areas Maize Share in Rural Areas by Region and Wealth by Region and Wealth 10 10 8 8 entc 6 entc 6 erP 4 erP 4 2 2 0 0 1 2 3 4 5 1 2 3 4 5 Quintiles of pc Consumption Quintiles of pc Consumption Maputo-City South Centre North South Centre North Rice Shares in Urban Areas, by Region and Rice Shares in Rural Areas, by Region and Wealth Wealth 10 15 tnecr 8 10 6 tnecr 4 Pe Pe 5 2 0 0 1 2 3 4 5 1 2 3 4 5 Quintiles of pc Consumption Quintiles of pc Consumption Maputo-City South Centre North South Centre North Shares of Wheat/Products in Urban Areas Shares of Wheat/Products in Rural Areas by Region and Wealth by Region and Wealth 8 4 tnecr 64 tnecr 32 Pe 2 Pe 1 0 0 1 2 3 4 5 1 2 3 4 5 Quintiles of pc Consumption Quintiles of pc Consumption Maputo-City South Centre North South Centre North Share of Cassava in Urban Areas Shares of Cassava in Rural Areas by Region and Wealth by Region and Wealth 20 2.5 2 15 tnecr 1.5 10 tnecr 1 Pe Pe 5 0.5 0 0 1 2 3 4 5 1 2 3 4 5 Quintiles of pc Consumption Quintiles of pc Consumption Maputo-City South Centre North South Centre North 8 Figure 4: Share of own food production in household food consumption URBAN AREAS RURAL AREAS 100 100 80 80 ) ) %( 60 %( 60 e ers harS 40 40 Sh 20 20 0 0 1 2 3 4 5 1 2 3 4 5 Quintiles of pc Consumption Quintiles of pc Consumption Maputo City South Center North South Center North Source: Authors' calculations using the 2003 national household survey (INE, 2003). 15. Given the importance of own production in rural households' food consumption, it is reasonable to assume that they could be largely insulated from variations in market prices. Indeed, even though food accounts for a larger share of total consumption of rural households, the share of food in total cash expenditures is lower for rural households (32 percent) compared to urban households (42 percent). In fact, the share of food in total cash expenditures is higher for urban households in the poorest three quintiles compared to rural households, irrespective of their income quintile or geographic location (Figure 5). Figure 5: Share of food in total household cash expenditures URBAN AREAS RURAL AREAS 70 70 60 60 ) 50 ) 50 %( 40 %( 40 e e 30 30 harS 20 harS 20 10 10 0 0 1 2 3 4 5 1 2 3 4 5 Quintiles of pc Consumption Quintiles of pc Consumption Maputo City South Center North South Center North Source: Authors' calculations using the 2003 national household survey (INE, 2003). 16. The Net Benefit Ratios for Mozambique indicate rural areas on average are net beneficiaries from the higher food prices, while urban areas and the south rural areas are losers, with Maputo/southern urban poor particularly hit. The net benefit ratios for Mozambique have been calculated by subtracting the consumption ratio from the production ratio, and shown in Table 2 and Figure 6. Production ratios were derived using the 2003 household survey. Although the survey does not contain specific information on agricultural production, it has information on income from the sale of agricultural output and on own consumption levels. In the analyses that follow, agricultural production is proxied by sale of 9 agricultural items combined with own consumption/production. Thus the production ratio is the share of agricultural sales and own production in total household income (including own consumption). Similarly, the consumption ratio was calculated as total expenditure on all food items, including the value of own consumption, relative to total household expenditure. Figure 6: Net Benefit Ratios for Mozambique Net Benefit Ratios for Mozambique Urban and Rural Areas by Wealth by Area of Residence 20 15 10 10 ) 5 10 %( 0 ) 0 R 1 2 3 4 5 %( -10 Rural Urban NB R -5 NB -10 -20 -15 -22 -30 -20 Quintiles of pc Consumption -25 Area of Residence Rural Urban URBAN AREAS RURAL AREAS by Region and Wealth by Region and Wealth 0 20 1 2 3 4 5 -10 15 ) ) 10 %( -20 %( R R 5 -30 NB NB 0 -40 1 2 3 4 5 -5 -50 -10 Quintiles of pc Consumption Quintiles of pc Consumption M aputo City South Center North South Center North Source: Authors' calculations using the 2003 national household survey (INE, 2003). 17. Net benefit ratios (NBRs) vary substantially across households by geographic location and income group. For each geographic location and income group, it is important to distinguish between net sellers (positive NBR) and net buyers (negative NBR). As shown in Table 2, 74 percent of rural households are net food sellers, whereas 76 percent of urban households are net buyers. Accordingly, the net benefit ratio is 10 percent for the rural and -22 percent for the urban households. This means that, on average, a 10 percent food price increase would, in the short run, raise rural real incomes by 1.0 percent and reduce urban real incomes by 2.2 percent. Figure 6 and Table 2 also highlights sharp differences across rural and urban areas and regions within the country. The population in the urban south and center are most severely hurt by food price increase, followed by the urban north. Moreover, the population in the rural south would also be negatively affected. Based on the NBR, rural households in the north and the 10 center benefit as a group from rising world prices. Table 3 shows the NBR across national expenditure quintiles. The poorest households in Maputo tend to lose the most in the short-run from food price increases, while middle-income groups in rural north and center would gain the most. This is consistent with the findings of Deaton (1989). Table 2: Calculation of net benefit ratios for Mozambique Net position (% of households) Effect of 100 percent food price increase (%) Net food sellers Net food buyers Food / total Food / total Net benefit ratio (NBR>0) (NBR<0) income (PR) expenditure (CR) (NBR) Mozambique 58.5 41.5 65.0 64.5 0.5 Urban areas 23.7 76.3 28.3 49.7 -21.5 Rural areas 73.7 26.3 80.9 70.9 10.0 South region 38.4 61.6 45.4 55.2 -9.8 Urban areas 14.6 85.4 16.4 42.2 -25.8 Rural areas 51.2 48.8 60.8 62.1 -1.3 Maputo City 4.1 95.9 3.4 30.8 -27.4 Centre region 65.0 35.0 72.8 69.2 3.5 Urban areas 23.5 76.5 31.5 54.2 -22.6 Rural areas 75.5 24.5 83.2 73.0 10.1 North region 67.8 32.2 73.3 67.7 5.6 Urban areas 36.9 63.1 42.4 58.2 -15.8 Rural areas 82.1 17.9 87.5 72.1 15.4 Source: Own calculations using the 2003 national household survey (INE, 2003). Table 3: Net benefit ratios by expenditure quintile for Mozambique All Household expenditure quintiles Households Q1 Q2 Q3 Q4 Q5 Mozambique 0.5 -1.6 0.9 2.9 0.9 -0.8 Urban areas -21.5 -22.8 -20.6 -22.9 -22.4 -18.7 Maputo City -27.4 -39.9 -32.2 -28.4 -22.9 -12.0 South region -25.8 -21.6 -29.6 -32.0 -26.0 -18.6 Center region -22.6 -27.7 -17.2 -22.0 -26.6 -20.1 North region -15.8 -14.7 -15.5 -13.3 -15.4 -20.4 Rural areas 10.0 7.6 11.5 12.5 11.0 7.2 South region -1.3 3.6 -1.9 -0.9 -4.1 -5.2 Center region 10.1 5.1 11.5 14.0 11.5 8.6 North region 15.4 14.0 18.3 16.5 16.1 11.5 Source: Own calculations using the 2003 national household survey (INE, 2003). 3.3 Estimating the impact of higher fuel prices 18. The welfare impact of an increase in fuel prices is always negative at the household level. The first order impact of a price rise depends on the direct and indirect use of fuel by the household. Direct consumption consists of the household's expenditure on gasoline and/or diesel for fuelling vehicles and machinery, and/or kerosene for lighting and cooking. Indirect fuel consumption occurs in the form of use of transport services, for travel of the members of the households and/or through the purchase of goods and services which require transport. As 11 expected, direct consumption of fuel in Mozambique is highly skewed towards the richest quintile, and is mostly concentrated in the South region and in Maputo city (Figure 7). Consumption of kerosene is slightly more evenly spread across the population as a percentage of household budgets. In absolute terms, however, most kerosene is consumed by the richest quintiles. Indirect fuel consumption for public transport services is highest in the middle three quintiles of the population, and particularly in Maputo city and in the South region (Figure 8). The richest quintile in Maputo city does not appear to use much public transport. In rural areas, use of public transport is slightly correlated with the level of income. Notably, the poorest quintile in the South region does not use much public transport. Indirect fuel consumption for transport of goods and services can be expected to be approximately proportional to the household's level of consumption (which we use as a proxy for household income).7 Hence, in absolute terms, richer household account for a much higher consumption than poorer households. An analysis of the distributional impact of changes in fuel prices in Mozambique was carried out in 2003 (Nicholson, et al., 2003). The results highlight that a 20 percent increase in fuel prices is roughly equivalent to reduction of 1 percent in household income, across the population. Figure 7: Share of fuel (Gasoline/Diesel) in total household consumption Shares of Fuel in Urban Areas by Region and Wealth 2.5 2.0 t 1.5 cen er 1.0 P 0.5 0.0 1 2 3 4 5 Quintiles of pc Consumption Maputo-City South Centre North Figure 8: Share of public transport in total household consumption Shares of Transport in Urban Areas by Region and Wealth 7.0 6.0 5.0 tnecr 4.0 3.0 Pe 2.0 1.0 0.0 1 2 3 4 5 Quintiles of pc consumption Maputo-City South Center North 7This assumes that the share of transport costs is roughly a constant proportion of the price of goods and services. 12 4 GENERAL EQUILIBRIUM ANALYSIS OF THE INCREASE IN FOOD AND FUEL PRICES 19. A CGE model has been used to evaluate the long term and economy wide impact of price changes in Mozambican economy. The previous section examined rising food prices from a microeconomic perspective, using first order approximations. A limitation with using partial equilibrium analysis is that this does not take into account the economy-wide impacts of changes in relative prices. However, changes in terms-of-trade brought about by rising world prices are fundamentally a macroeconomic phenomena. It is thus indispensable to consider macroeconomic dimensions, including the balance of payments, the distribution of the shock across macroeconomic aggregates, and the implications for wages. We consider each of these factors in turn, before proceeding to use a CGE model to evaluate the long terms and economy wide impact of price changes (see Box 3). 4.1 Macroeconomic dimensions of the price shock 4.1.1 Balance of payments 20. Mozambique is a food deficit country and food price increases are therefore a negative terms-of-trade shock. Mozambique is a food deficit country, importing all of the wheat and three-quarters of the rice demanded internally (i.e., 470,000 and 320,000 tons annually). Maize is both imported by the south and exported from the north. Overall, the country is also a net importer of maize. Major agricultural exports include tobacco, cashew, cotton, sugar and fish. Overall, national accounts for 2006 indicates that the FOB value of agriculture and food exports amounted to about 90 percent of the CIF value of agriculture and food imports. Food price increases are therefore a negative terms-of-trade shock, despite being partially offset by rising prices for some agricultural exports. Table 4 provides basic information on trade for the year 2006. Table 4: Trade in agriculture, food and petroleum products in 2006 Value Share of total Share of (US$ million) exports/imports (%) total GDP (%) Value of imports 2,966 100.0 43.0 Agriculture and food 351 11.8 5.1 Staple grains and derived products 153 5.2 2.2 Petroleum and petro-chemicals 537 18.1 7.8 Petroleum products 361 12.2 5.2 Value of exports 1,971 100.0 28.5 Agriculture and food 318 16.1 4.6 Staple grains and derived products 4 0.2 0.1 Source: National accounts data with local currency converted to US$ using the average exchange rate for 2006. 21. In macroeconomic terms, the fuel price shock is likely to dominate. Fuel and petro- chemical imports amounted to 18 percent of imports in 2006 compared to 12 percent for agriculture and food (Table 4). Moreover, the recent price shock has centered on fuel and cereals, which represented 12 and 5 percent of imports respectively. Since Mozambique does not export fuel or petro-chemical products, there is no compensating rise in export prices to mitigate the 13 negative terms-of-trade shock.8 The impacts of rising fuel prices on the balance of payments can be considered through the following identity: where B are net financial flows, E are exports, Mo and Mn are fuel and non-fuel imports, P are world prices, and R are changes in international reserves. The identity indicates that increases in world oil prices (Po) must be accompanied by some combination of reduced fuel imports (Mo), reduced non-fuel imports (Mn), increased exports (E), increased foreign borrowing (B), or falling foreign reserves (R). Since fuel prices are expected to remain high, at least over the medium- term, foreign borrowing (B) and the use of foreign reserves (R) can only act as transition measures.9 Thus, the long-run solution involves exporting more and/or importing less. 4.1.2 Macroeconomic aggregates 22. The shocks entail a reduction in the total welfare of Mozambique. Exporting more and importing less involves a (sometimes painful) shift in the structure of the economy away from the production of non-tradeables (e.g., services) towards the production of tradeable goods, which are either exported or displace imports. Exporting more and importing less also involves a reduction in absorption, which is the measure of the total volume of goods and services available in the economy. Absorption is a measure of total welfare and can be depicted by rearranging the national income identity: where GDP is gross domestic product, M is imports, E is exports, C is household consumption, I is public and private investment, and G is recurrent government spending. With constant GDP, a decrease in imports and an increase in exports imply a reduction in absorption. In this case, Mozambique ­ already one of the poorest countries in the world ­ becomes even poorer. If the adjustments needed to reduce imports and increase exports cause GDP (or the rate of GDP growth) to decline, then absorption is further reduced (relative to trend). 23. The impact of the reduction in welfare is strongly influenced by policy, as resources can be shifted from public investment to support household consumption. Reduced absorption must be borne by consumption (C), investment (I), and/or government spending. The household-level analysis in Section 3 focused on changes in consumption. However, in the case of Mozambique, where foreign assistance represents half of government spending and almost all public investment, it would be possible to redirect foreign aid to subsidize food and fuel consumption. In this case, household consumption (C) would be preserved at the cost of reduced public investment (I) in education, health and other sectors. In other words, while absorption is likely to decline after a negative terms-of-trade shock, the distribution of reduced absorption across the macroeconomic aggregates of GDP is strongly influenced by policy. As shown in Section 2, policymakers in Mozambique have allowed higher world prices to be transmitted to domestic markets, without any large-scale efforts to insulate household consumption. 8Mozambique has recently started exporting natural gas. However, these exports are relatively small and most of the revenues accrue to foreign exploration companies. Pricing contracts are also fixed so that the natural gas price that Mozambique receives varies little with world prices for hydrocarbons. 9There is no evidence that Mozambique will benefit from a significant special dispensation of donor funding to cope with the current crisis. 14 Nevertheless, in the following section, we will consider the implications of policies to insulate domestic markets. 4.1.3 Wages 24. A rise in rural wages following the increase in food prices could mitigate and even reverse the negative first order impacts for food-deficit households. Rising world prices will alter Mozambique's structure of production, and bring changes in relative wages across sectors, which will contribute to determine the overall impact of the shocks for each sector of the economy. In general, a negative terms-of-trade shock favors the production of tradeables over non-tradeables. More specifically, production of commodities whose prices have risen, such as cereals, should increase due to enhanced profitability, while fuel-intensive producers' profitability should decline. In addition, structural changes may be driven by the shifting composition of absorption. The changing production structure will affect factor returns. For example, if rising world prices favor cereals and if cereals production uses land and unskilled labor intensively, then land rental rates and unskilled wages should rise relative to the market returns for capital and skilled labor. In this case, a rise in rural wages following a food price increase could mitigate and even reverse the negative first order impacts for food-deficit households. 25. Empirical studies show that the long run impact of rising food prices on rural poor can be positive after adjusting for changes in wage rates, but this is not always the case. The impact of higher food prices on wages has been examined empirically in other countries (Zaman 2008). Ravallion (1991, 2000) uses data from Bangladesh and India to argue that, while the rural poor are adversely affected in the short run by rising food prices, the long run impact can be neutral after adjusting for changes in wage rates. This result is due to the response of rural wages to the price of food grains (a second order or medium-term impact). However, the extent to which wages respond to changing food prices has been questioned by Rashid (2002). Using time series data from Bangladesh, the author argues that changes in rice prices since the 1980s have had negligible effect on agricultural wages. Christaensen and Demery (2006), using data from a number of African countries, extend this analysis by including the additional second-round effect of increased farm productivity. They conclude that policies leading to higher food prices are likely to increase poverty, even after accounting for wage and productivity effects. The inability of Rashid (2002) and Christaensen and Demery (2006) to isolate a wage effect could be due to a number of factors, however, including the magnitude of the food price shock, difficulties in measuring real rural wages, the technologies employed in response to the food price shock, the size of the agricultural sector relative to the rest of the economy, and the degree of labor mobility between agriculture and non-agriculture. However, even if wages catch up and fully mitigate the impact of the higher prices in the long term, empirical studies suggest that this catch up is lagged by about 2 years. Hence, the short term impact on households would be significant (and may lead to the adoption of coping mechanisms which have a permanent negative impact, such removing children from schooling and lowering child nutritional intake). 15 Box 3: The structure and behavioral specification of the Mozambique CGE model The Mozambique CGE model contains 51 activities/commodities, including 23 agricultural sectors. Five factors of production are identified: three types of labor (unskilled, semi-skilled and skilled), agricultural land, and capital. Segmented rural and urban labor markets distinguish between rural nonfarm and urban economies. Labor and agricultural land is assumed to be fully employed, while capital is immobile earning sector-specific returns. Within this structure and subject to macroeconomic constraints, producers in the model maximize profits under constant returns to scale, with the choice between factors governed by a constant elasticity of substitution (CES) function. Assuming that fixed rigid production technologies are relatively fixed over the medium-term, we assume low and uniform factor substitution elasticities (0.5). Factors are then combined with fixed-share intermediates using a Leontief specification, which captures the varying fuel-intensity of sectors. Under profit maximization, factors receive income where marginal revenue equals marginal cost based on endogenous relative prices. Thus, macroeconomic wage-effects in model are endogenously determined by sector-specific factor demands and economywide factor supply constraints. Structure of the Mozambican economy in 2003 SAM Sector or group name GDP Production Export Export Import Import- sectors share (%) share (%) share (%) intensity share (%) intensity Total GDP 100.0 100.0 100.0 11.5 100.0 23.5 1-21 Agriculture 25.9 15.4 20.3 14.9 3.3 5.2 1 Maize 3.5 1.9 0.2 1.0 0.3 4.1 2 Sorghum 1.1 0.6 0.0 0.0 0.0 0.0 3 Unshelled rice 0.7 0.4 0.0 0.0 0.0 0.0 4 Wheat 0.0 0.0 0.0 0.0 1.7 100.0 22 Mining 0.3 0.4 0.3 4.9 0.2 6.7 23-40 Manufacturing 13.7 20.5 54.5 35.4 70.1 56.9 32 Gasoline 0.0 0.0 0.0 0.0 0.8 100.0 33 Diesel 0.0 0.0 0.0 0.0 7.0 100.0 34 Other fuels 0.0 0.0 0.0 0.0 2.3 80.5 35 Other petro-chem. 0.4 0.6 0.3 6.1 10.0 74.5 37 Metals 5.2 5.8 48.0 94.8 5.3 71.2 41-43 Construction and 9.2 13.9 11.2 79.8 5.4 79.6 energy 44-51 Services 50.9 49.7 13.7 3.2 21.0 9.2 Source: Mozambique 2003 social accounting matrix (SAM) (McCool et al., forthcoming). Notes: `GDP' refers to value added at factor cost, and `Production' refers to total sales by domestic activities. `Export intensity' is the share of exports in domestic output, and `import penetration' is the share of imports in total domestic demand. As discussed earlier, higher fuel prices will reduce foreign exchange availability, forcing Mozambique to export more and import less. Accordingly, sectors with high trade shares (either a large share of production exported or a high degree of import competition) are expected to expand more than non-traded sectors. Fuel is especially crucial since it is entirely imported and does not have domestic or imported substitutes. The Mozambique model captures changes in international trade by allowing producers and consumers to shift between domestic and foreign markets depending on changes in the relative prices of imports, exports and domestic goods. Under a constant elasticity of transformation (CET) function, profit maximization drives producers to sell in markets where they achieve the highest returns based on domestic and export prices (where the latter is determined by the world price times the exchange rate adjusted for internal transaction costs). Conversely, under a CES Armington function, cost-minimization determines final and intermediate demand for imported and domestic goods based on relative prices (both of which include relevant taxes). Trade function elasticities are taken from the Global Trade and Analysis Project (Dimaranan, 2006). Under a small country assumption, Mozambique faces perfectly elastic world supply/demand at fixed world prices. 16 Box 3 (CONTINUED): The structure and behavioral specification of the Mozambique CGE model Various institutions are identified in the model, including enterprises, the government, and ten representative household groups (i.e., rural and urban households disaggregated across national income quintiles). Households and enterprises receive income in payment for producers' use of their factors of production, and then pay direct taxes to government (based on fixed tax rates) and save (based on marginal propensities to save). Enterprises pay their remaining income to households, which, in turn, use their income to consume commodities under a linear expenditure system (LES) of demand. Each household in the CGE model is then linked to its corresponding households in the 2002 household survey (INE, 2003). Under this expenditure-side micro-simulation module, changes in representative households' consumption and prices in the CGE model are passed down to the survey, where household consumption expenditures are recalculated. This new level of per capita expenditure for each survey household is compared to the official poverty line, and standard poverty measures are recalculated. The Mozambique model thus simultaneously accounts for wage- and price-effects in determining households' real incomes and poverty outcomes. The government receives income from sales and direct taxes and import tariffs, which it uses to purchase commodities in the form of government recurrent expenditure. The remaining income of government is (dis)saved. All domestic and foreign savings (i.e., foreign borrowing and assistance) are collected in a savings pool from which investment is financed. Here three closure rules are used to capture the macroeconomic dimensions of the price shock. First, government recurrent expenditure is fixed and the fiscal deficit (i.e., public savings or investment) adjusts to align revenues with total expenditures. Secondly, a savings-driven closure is assumed in order to balance the overall savings-investment account (i.e., household and enterprise savings rates are fixed and investment adjusts to changes in incomes and the fiscal deficit to ensure that the level of investment and savings are equal). Finally, for the current account it is assumed that the exchange rate adjusts to maintain a fixed level of foreign savings (i.e., the external balance is held fixed in foreign currency). Together these three closure rules allow the model to capture the balance of payments constraint and absorption trade-offs discussed earlier. 4.2 Impact simulations and results using a general equilibrium model of Mozambique 26. CGE models provide a framework which is able to capture both first and second order impacts of the prices shock. The impact of higher world prices is simulated using a comparative static computable general equilibrium (CGE) model of Mozambique. These models have a number of characteristics that make them suitable for analyzing external shocks, trade/tax policies, income distribution, and structural change. First, CGE models simulate the functioning of a market economy, including markets for labor, capital and commodities, and track how changes in economic conditions are mediated through price and quantity adjustments. Secondly, the structural nature of these models permits a decomposition of multiple shocks, such as simultaneous increases in fuel and food prices. Thirdly, CGE models respect economywide constraints, including the balance of payments and macroeconomic aggregates. Fourthly, CGE models contain detailed sectoral breakdowns allowing for differential price increases across commodities. Finally, these models provide a theoretically consistent framework for welfare and distributional analysis. The structure and behavioral specification of a CGE model determines its results. Brief information on the Mozambique CGE model used in this study is provided in Box 3 (Thurlow (2008) and McCool et al. (forthcoming) provide a detailed description of the CGE model and the 2003 social accounting matrix (SAM) to which it is calibrated). 17 4.2.1 Simulation descriptions 27. The model has been shocked with the price increases which took place in 2007-2008. This paper focuses on the impact of the world prices increases taking place between the second half of 2007 into 2008. However, the CGE model is calibrated to a 2003 base social accounting matrix (SAM), raising the issue of what magnitude price shock should be imposed on the model. For instance, oil prices rose more than threefold during 2003-2008 (from US$32 to greater than US$100 per barrel), but this increase did not occur all at once. Between 2003 and 2006, the world price for oil doubled to US$64 a barrel. This is responsible for the higher fuel import shares in Table 4, which is for 2006, compared to the Table in Box 3, which is for 2003 (i.e., from 10 to 12 percent). The intention of the modeling effort is to gain insights into the impacts of the recent price increases using available tools and data. For the purposes of the CGE model, tripling oil prices seems unrealistic. It was decided that the model should be shocked with only the 2007-2008 price increases. The thought experiment that is being undertaken is what would have happened in 2003 had fuel and food prices increased in similar proportions to the recent world price increases. The actual shocks applied are depicted in Table 5. The shocks applied tend to be somewhat smaller in magnitude than the price increases depicted in Table 1. Inflation explains a part of the difference. The shocks applied should reflect real price increases while the shocks in Table 1 reflect nominal price increases in USD. Also, the simulations were built on the expectation that commodity prices would come off of the peaks registered in the middle of 2008 (such as oil at US$145 per barrel), which has indeed occurred. Overall, the objective of the shocks is to reasonably capture the shift in international relative prices that occurred in late 2007 and first half of 2008. Table 5: World price shocks simulated using CGE model of Mozambique Agriculture and food price simulations Fuel price simulations Commodity Shock (%) Commodity Shock (%) Agricultural commodities Petroleum and petro-chemicals Maize 75 Gasoline 75 Sorghum 50 Diesel 75 Rice 75 Other fuels 75 Wheat 75 Other petro-chemicals 25 Pulses & groundnuts 50 Horticulture 25 Raw tobacco 25 Cotton 25 Livestock 25 Processed agricultural commodities Meat and fish products 40 Other processed foods 40 Grain flours 50 Processed sugar 40 Processed tobacco 25 Processed cotton 25 Note: Equivalent shocks are applied to world export and import prices. 28. Four simulations have been run to analyze the impact of the price shocks. The first simulation (`Fuel') uniquely shocks fuel prices. The second simulation (`Food - Fixed land') considers the shocks to agriculture and processed food prices under the assumption that land allocations between crops cannot be altered (i.e., a very short run scenario with similar 18 assumptions to the household survey analysis in Section 3). The third simulation (`Food - Flexible land') considers the shocks to agriculture and processed food prices assuming that farmers can reallocate land across crops (i.e., a stronger supply response). This implies a one to three year adjustment period. The fourth simulation (`Combination') combines the first and third simulations. 4.2.2 Model results 29. The simulations highlight that the impact of the fuel price increases is more than double that of the food price increases. The impacts of the fuel and food price shocks are depicted in Tables 6-9. Macroeconomic impacts are shown in Table 6. As suggested by the structure of imports presented earlier in this section, the fuel shocks generate more severe impacts on the overall terms-of-trade. The decline in the terms-of-trade due to fuel price increases is more than double the decline due to food price increases. Macroeconomic impacts are commensurately larger. Compared with the food price shocks, the fuel shocks force a larger increase in the quantity of exports and a larger decrease in the quantity of imports in order to balance the external account. Due principally to these changes in trade flows, the decline in total absorption (or overall welfare) under the `Fuel' simulation (3.5 percent) is approximately double the decline registered for either of the `Food' simulations. 30. While both the `Food' and `Fuel' simulations lead to a real depreciation of the currency, the food shock causes a nominal appreciation, while the fuel shock leads to a nominal depreciation. As emphasized above, the components of absorption are influenced by economic structure and macroeconomic closure rules. The heavy dependence of Mozambique on foreign savings implies that real investment depends in part on the nominal exchange rate. Depreciation (appreciation) of the nominal exchange rate increases (decreases) the local currency value of investment and can lead to a real increase (decrease) in investment under a savings- driven closure. While the `Food' and `Fuel' simulations lead to a real depreciation of the currency, the nominal currency value moves in opposite directions between the two sets of simulations. In the two `Food' simulations, the increases in world prices for agricultural and processed commodities automatically shift relative prices towards tradeable commodities. The relative price shift towards tradeables generated by the world price increases is in fact so strong that the nominal currency actually appreciates in order to reestablish external balance. By contrast, in the `Fuel' scenario, the world price increases do little to shift the price ratio between tradeable and non-tradeable sectors because both sectors use fuel as an intermediate input (and there is very little domestic production of fuel and petro-chemicals). As a result, a strong nominal depreciation is required to balance the external account. 31. Real household consumption in the `Fuel' scenario falls by more than three times the declines registered in the `Food' simulations. Principally as a result of opposing movements in the nominal exchange rate, real investment rises under the `Fuel' simulation (because foreign assistance lays greater claim to domestic resources due to the depreciated currency) and decreases in the two `Food' simulations (for the same reasons but in an opposite direction). Since real government consumption is fixed in real terms across all scenarios, the decline in absorption in the `Fuel' scenario is borne entirely by household consumption. And, household consumption must decline further to accommodate the rise in the real value of investment. Overall, real household consumption in the `Fuel' scenario falls by more than three times the declines registered in the two `Food' simulations due to a larger decline in absorption overall and differential movements in the components of absorption, particularly investment. The differences 19 between the `Fixed' and `Flexible' food simulations manifest themselves primarily through the production response. With flexible land, agricultural production can be reallocated towards export crops, particularly those whose world prices are rising, permitting a greater increase in exports than in the fixed land simulation. Furthermore, the export stimulus and import compression are achieved with a smaller decline in the real exchange rate. Table 6: Macroeconomic results for world price shocks Change from base year value (%) Fuel Food scenario Combined scenario Fixed land Flexible land scenario Quantities GDP -0.6 -0.5 -0.5 -1.2 Absorption (C+I+G) -3.5 -1.8 -1.8 -5.1 Consumption (C) -5.8 -1.9 -1.8 -7.3 Investment (I) 1.5 -2.5 -2.8 -1.2 Recurrent government (G) 0.0 0.0 0.0 0.0 Exports (E) 5.6 0.6 1.0 5.9 Imports (M) -6.4 -4.0 -3.7 -9.6 Prices Nominal exchange rate 4.5 -5.0 -5.6 -1.5 Real exchange rate 15.4 1.3 0.6 15.2 Terms-of-trade -12.9 -4.8 -4.8 -16.2 Source: Results from the Mozambique CGE model 32. The combined effects of the `Fuel' and `Food' scenarios, which are the actual shocks that Mozambique received, are considerable. The scenario `Combined' shows effects that are roughly the sum of the two preceding scenarios. Terms-of-trade decline by more than 16 percent, and in order to balance the external account, exports increase by nearly 6 percent and imports decline by almost 10 percent. These shifts in production generate a decline in GDP of slightly more than 1 percent. All of these adjustments imply a reduction in the quantity of goods and services in the economy resulting in a reduction in absorption of more than 5 percent. Since recurrent government expenditure is assumed to be fixed and investment declines by only 1.2 percent, household consumption bears the bulk of the adjustment, declining by more than 7.0 percent. This is a substantial decline in a country where approximately half of all households are absolutely poor (i.e., they experience difficulty meeting caloric needs). 33. The price shocks particularly favor export products that experience price increases. The implications of the world price shock for production are presented in Table 7. The table shows, in the first column, the share in value added of each sector depicted at base 2003 values. For ease of interpretation, most depicted sectors are aggregates of the sectors available in the 2003 SAM and employed in the CGE model. The columns under each simulation provide the percentage change in the real output of each sector relative to the base. Across all simulations, exporting and import competing sectors are favored. The food price shocks particularly favor export products that experience price increases. In the combined scenario, particularly strong growth is registered in `Export crops', led by tobacco and cotton, and `Processed products', led by processed cotton and processed sugar. Production of non-tradeables, such as root crops (which is dominated by cassava, the largest single crop in value added terms) and services (which represents about half of the economy) decline in all scenarios. These declining sectors free resources which permit the tradeables sectors, particularly the export sectors, to expand. 20 34. These results highlight the importance of export supply response with particular emphasis on the agricultural sector. Agriculture and derived products comprise the large bulk of the export response with particular emphasis on cashew, tobacco, cotton, sugar, and other processed foods. Exports from these sectors are projected to approximately double, although the increases take place from relatively small bases. Biofuels represent another export potential that is not modeled here but is considered in detail in Arndt et al. (2008). 35. A robust export response is crucial to avoid severe import compression. Even with the export response attained, the onus of adjustment is already taking place largely on the import side. This can be seen from the macroeconomic impacts in Table 6. Imports values are about double those of exports and the percentage decline in imports is greater in absolute value than the relative expansion of exports. While export responses tend to concentrate in specific sectors, imports decline across the board. Particularly large declines in imports are registered in products where domestic sectors compete strongly with imports, such as maize, grain milling, and meats. Table 7: Sectoral production results for world price shocks Base value- Change from base year value-added (%) added share Fuel Food scenario Combined (%) Scenario Fixed land Flexible land Scenario Agriculture 25.9 0.2 0.6 0.5 0.7 Cereal crops 5.3 -0.8 3.1 3.2 2.9 Roots crops 7.2 -0.9 0.3 -0.9 -1.8 Pulses and groundnuts 2.3 1.1 1.4 3.0 4.2 Horticulture 3.3 -1.2 0.6 -0.7 -1.7 Export crops 1.1 9.4 5.3 11.9 21.2 Livestock 1.7 -0.4 3.9 4.1 4.2 Forestry 2.7 -0.3 -1.1 -1.2 -1.9 Fishery 2.3 3.8 -7.2 -7.9 -5.9 Non-Agriculture 74.1 -0.2 -0.3 -0.3 -0.5 Industry 23.1 1.0 0.2 0.3 1.4 Mining 0.3 -0.1 -0.3 -0.4 -0.5 Manufacturing 13.7 1.0 1.7 2.1 3.2 Primary product processing 7.4 1.3 3.3 4.1 5.7 Other industry 9.1 1.1 -2.0 -2.3 -1.2 Electricity 1.9 1.7 -1.3 -1.5 0.2 Water 0.3 -2.2 -0.1 0.0 -2.1 Construction 7.0 1.0 -2.3 -2.6 -1.5 Services 50.9 -0.7 -0.5 -0.6 -1.4 Source: Results from the Mozambique CGE model 36. Changes in factor prices favor rural labor, especially unskilled labor, and returns to land, which fare better than urban labor and capital. Implications for factor prices are shown in Table 8. As discussed above, both shock vectors stimulate tradeable agriculture and processed foods. These sectors use unskilled (primarily rural) labor intensively though the stimulus to these sectors is much more pronounced in the `Food' simulations. In nearly all cases, urban wages decline more than rural wages. The exception is urban skilled labor in the `Fuel' scenario, which benefits from a fairly broad based expansion of traded non-agriculture. Relative to other factors, the food shocks favor unskilled rural labor and land. Under the `Fixed land' scenario, the returns to rural labor and land are lower than under `Flexible land' scenario. The relatively large differential impacts across factors in the food simulations carry over into the `Combined' 21 simulation, where rural labor, especially unskilled labor, fares better than urban labor and capital. There is also a pronounced positive impact on land returns. Table 8: Factor price results for world price shocks Change from base year factor prices (%) Fuel Food scenario Combined scenario Fixed land Flexible land Scenario Rural labor Skilled -5.2 2.9 3.3 -1.6 Semi-skilled -5.8 0.7 0.9 -4.6 Unskilled -5.3 3.7 4.2 -0.7 Urban labor Skilled -4.0 -1.6 -1.5 -5.7 Semi-skilled -7.4 -1.1 -0.9 -8.2 Unskilled -6.8 -0.9 -0.7 -7.3 Capital -5.5 -1.5 -1.5 -6.4 Agricultural land -4.2 11.4 12.4 9.5 Source: Results from the Mozambique CGE model Table 9: Welfare and poverty results for world price shocks Base year Change from base year (%) value Fuel Food scenario Combined scenario Fixed land Flexible land Scenario Equivalent variation National -5.9 -2.1 -2.0 -7.4 Rural households Quintile 1 -3.4 -0.7 -0.9 -3.9 Quintile 2 -3.6 -0.1 -0.1 -3.2 Quintile 3 -3.7 0.3 0.4 -2.7 Quintile 4 -4.2 -0.1 0.2 -3.4 Quintile 5 -5.1 -0.3 0.1 -4.4 Urban households Quintile 1 -5.4 -5.3 -5.8 -11.1 Quintile 2 -6.2 -5.6 -5.8 -11.6 Quintile 3 -6.0 -5.0 -5.3 -10.9 Quintile 4 -7.1 -4.5 -4.5 -11.1 Quintile 5 -7.1 -2.8 -2.7 -9.4 Poverty headcount National 54.1 57.6 55.1 54.9 58.2 Rural households 55.3 58.3 55.4 55.2 57.7 Urban households 51.5 56.2 54.3 54.2 59.5 Source: Results from the Mozambique CGE model 37. The welfare impact for rural households is less strongly negative than for urban households in all simulations. Welfare implications, measured as percent change in equivalent variation, are presented in Table 9. As discussed in Section 3, substantial home consumption amongst rural households provides considerable insulation from both fuel and food prices shocks. In addition, as shown in Table 9, rural wages rise relative to urban wages, particularly in response to the food price shocks. As a result of these consumption and income impacts, rural households are less strongly affected than urban households in all simulations. The stimulating effect of improved agricultural terms-of-trade for the rural economy does not outweigh the negative impacts of the fuel shock, and welfare declines for all households in the `Combined' simulation. The degree of land ownership is the primary factor differentiating outcomes across quintiles in rural areas. The results for the `Fixed land' scenario are consistent with the 22 household-level analysis in Section 3, which showed middle-quintile rural households faring better than others under the food price shock. In urban areas, welfare losses are large in magnitude and relatively constant across the income distribution. 38. Poverty impacts are large, particularly in urban areas. Table 9 and Figure 9 show that the combined shocks result in four percentage point increase in the national poverty headcount rate. The effect is much stronger in urban areas where the poverty rate increases by eight percentage points. In fact, the `Combined' simulation sets the urban poverty rate above the rural rate. Fuel price increases are the principal driver of increased poverty in both rural and urban zones. As would be expected, the capacity to reallocate land reduces poverty with the effect being slightly stronger in rural zones. Figure 9: Impact of food and fuel price shocks on poverty headcount Impact of the shocks on poverty levels 60 59 58 57 ountc 56 head 55 yt 54 ervoP 53 52 51 50 Base year Fuel Food Food Combined value (Fixed land) (Flexibl land) National Rural households Urban households Source: Results from the Mozambique CGE model 5 POLICY RESPONSES 5.1 Principles of a good policy response10 39. Increasing world prices for food and fuel presents policy makers with a range of policy challenges. Allowing the pass through of increases in global prices to domestic prices will reduce real incomes of households, leading to an increase in poverty rates. On the other hand, attempting to prevent passing through higher prices by either subsidizing or reducing taxation can result in a substantial fiscal burden that needs to be financed, and will introduce distortions in consumption patterns, thereby exacerbating adverse balance of payments impacts (or reducing balance of payments benefits). For instance, not passing through the price increases would reduce the incentives to increase agricultural production, and exports, which is a great opportunity for Mozambique. Further, not passing through the international price increases to the domestic markets is often undesirable on equity grounds, since most of the benefits from the 10This section builds on the discussion of policy options presented in the World Bank background paper on the higher food and fuel prices crisis prepared for the G8 Hokkaido-Toyako Summit in July 2008 (World Bank, 2008b). 23 subsidies or lower taxation to shield the domestic prices, would accrue to the richer groups in the population. The best policy is to allow the domestic prices to adjust in line with international prices, while protecting the poorest households by scaling up targeted social safety nets. 40. A few good principles should guide policy makers to identify the best policies to respond to the crisis: (i) short-term policies should ideally support and definitely not undermine, long term priorities; (ii) policies should preserve incentives (e.g. labor, agricultural production); (iii) interventions should be targeted most vulnerable groups; (iv) the costs should be limited, and leakages outside national borders should be minimized; (v) the measures should be easy to implement and introduce; (vi) the measures should entail minimum management and/or governance concerns. 41. This report focuses on interventions to mitigate the impact of the price shocks and does not discuss in detail policies to enhance longer-term food supply. Specific policy interventions in response to the higher food and fuel prices can be divided into: (i) interventions to mitigate the impact of the price shocks; and (ii) interventions to enhance longer-term food supply. In this report we focus exclusively on policies available to mitigate the impact on households of the higher food and fuel prices (see Box 4). They can be subdivided into three groups: (a) interventions to strengthen targeted safety nets; (b) short-run trade policy measures or administrative action affecting food prices; and (c) short-run trade policy measures or administrative action affecting fuel prices. Within these groups, `first best' policies (i.e., preferred options) are those which are more effective and equitable, and introduce fewer distortions, and are best aligned with the principles above. Table 10 summarizes the main policy options and ranks them according to these and other desirable criteria. Table 10: Summary overview of policy options Costs limited, Limited Targeted to Easy to Preserve and management / vulnerable implement / incentives within national governance groups introduce Borders concerns Safety nets programs Cash transfers Food stamps / vouchers Targeted feeding / nutrition programs Public work programs (in cash) School feeding programs Policies to mitigate higher food prices Reducing import tariffs and VAT Targeted consumer subsidies Using buffer stocks to increase supply Generalized consumer subsidies Export bans / restrictions Producer price controls Policies to mitigate higher fuel prices Targeted subsidy to urban transport operators Targeted urban transport subsidies (cards) Subsidized financing for purchase of vehicles Generalized fuel subsidies (or tax reduction) Subsidize/ finance public transport company Reduction in import tariffs for spare parts Source: Adapted from World Bank 2008b. 24 5.1.1 Mitigating the impact of the price shocks by strengthening targeted safety nets 42. The first best option to protect households from the impact of the higher prices is the provision of (means based) targeted cash transfers to vulnerable groups. Protecting vulnerable households by strengthening safety nets is the long run ideal. These measures support the purchasing power of the poor without distorting domestic incentives to produce more food, and without reducing the incomes of poor food sellers. The ideal measure is to provide cash or near-cash transfers that are conditional upon meeting a requirement (such as low income, location or occupation) or engaging in a mandated behavior (such as sending children to school). 43. Cash transfers are complex to administer as they require sufficient institutional capacity to appropriately target and disburse cash to large numbers of people. One drawback of cash transfers, therefore, is that the fixed costs to set up and administer such a program are such that these transfer are not effective if the intention is to have only a short terms program (e.g. to respond to a single weather event). On the other hand, as it makes sense to expect that there will be repeated shocks over time (even if their nature and timing are unknown), an important advantage of cash transfers is that they are flexible so that, once the fixed costs to establish a cash transfers program has been incurred it can be adapted to different shocks. Hence a cash transfers program provides the government with a flexible response mechanism to a range of shocks. Various kinds of cash transfer programs are currently used in Brazil, China, Ethiopia, Egypt, Indonesia, Mexico, Mozambique, South Africa, Sri Lanka, and Tunisia. Several of these countries are adjusting current programs in response to the rise in food prices. For example, in Ethiopia, where food price inflation in February 2008 was 23 percent (year on year), the government has raised the cash wage rate of the largest cash-for-work program by 33 percent. Also, in late 2008 the Ethiopian government set a mechanism to adapt wages in the light of continued price movement. The scale, targeting efficiency and value of such transfer programs tend to be directly related to overall levels of development, given the administrative complexities and fiscal costs entailed. In countries which have a good institutional capacity to target but do not necessarily have the mechanisms to disburse cash it may be best to use `near cash' transfers such as food stamps, or vouchers (e.g. Sri Lanka, Tunisia). 44. Public works programs, preferably with payment in cash, are recommended for low income countries where targeting cash transfers via means or proxy means testing is difficult (e.g. Malawi, Cambodia). Public works programs are generally designed as a self- targeted program by minimizing the incentives the non-poor to take part in the program. This is typically achieved through setting a low level of wages. Public works programs with payment in kind (most commonly `food-for-work') can be useful where local markets for food commodities are not well integrated (e.g. in remote areas); where markets function reasonably, it is best to use cash payments. A number of countries, including Bangladesh, Madagascar, Cambodia, and India, are using self-targeted food-for-work programs, while others, including Afghanistan and Angola, use emergency food aid distribution to ensure food security for vulnerable groups. The food-for-work program in Bangladesh has been expanded recently due to both natural disasters and the rise in food prices. While self-targeting reduces the costs involved in administrative targeting, the physical transfer of food is itself costly and can lead to leakages. Food aid can also have growing disincentive effects on local production if it becomes entrenched beyond the initial emergency or is not tied to a work requirement. 25 45. Targeted nutrition programs to children under-2 and pregnant women is also recommended in parallel with above options as the primary focus is on protecting the most vulnerable (e.g. Burkina, Honduras, Morocco). While food prices are not the main driver of malnutrition, they do affect nutritional outcomes through their impact on real incomes and household purchasing behavior. In compensating for rising food prices, vulnerable households may substitute towards less food, or cheaper, but less nutritious, substitutes for current diets. 46. School feeding programs can also be used under special circumstances to alleviate the higher food and fuel prices. School feeding programs are generally very expensive in terms of the cost per pupil, and are not well targeted to poorer students. Also, school-based programs do not typically address child malnutrition at its most critical point ­ when children are in their infancy. Still many countries, including Burkina Faso, Brazil, China, Kenya, Honduras, Mexico and Mozambique, make effective use of school feeding programs to improve the food intake of school-age children and their families. Cheaper programs are now being experimented (e.g., with crackers or biscuits), and targeting can be improved by limiting the programs to schools in the poorest districts or neighborhoods. Also, take home rations can work as a conditional transfer, since the children have to attend school to collect the rations. 5.1.2 Interventions to mitigate the domestic food prices through short-run trade policy measures or administrative action 47. First best options to lower domestic prices include reducing tariffs and other taxes on key staples. Reduction in import tariffs, VAT and other taxes to grains and milled products are recommended for countries with large significant taxes and tariffs on food grains. Many countries impose tariffs on food imports, both to encourage domestic production and boost domestic revenue. In times of sharply increasing prices, reductions in tariffs and taxes can provide some relief to consumers, albeit at a fiscal cost. The revenue loss from reducing tariffs can be significant and the fiscal implications of combining this with additional social protection expenditures may well require cutbacks in lower priority areas. In April 2008 a World Bank survey found that some twenty-four of fifty-eight countries sampled have recently reduced import duties and VAT in the wake of rising food inflation (Zaman 2008b). 48. Selective grain/bread subsidies targeted to poor consumers are a second best option, which may be appropriate in countries where targeted safety net programs cannot be scaled up sufficiently during crises (e.g., Bangladesh). The main reason for avoiding these subsidies is that the extent of possible targeting is modest and that once set up they are hard to change. Also, selective subsidy programs commonly suffer from the pitfalls of two-tier pricing, notably the risk of `back door' sales (Alderman, 1988). Several countries, mainly in the Middle East-North Africa region (e.g. Egypt, Morocco) have a long history of using bread or grain subsidies specifically targeted to the poor to cope with household food insecurity. Others have introduced consumer subsidies for staples following the recent rise in food prices. For example, the Government of Yemen is supplying wheat in select markets at subsidized rates following a sharp rise in food prices. In early 2008 the Government of Pakistan announced that it was reviving a ration card system to distribute subsidized wheat. 49. Policy options which should be avoided include generalized consumer subsidies, price controls for basic staples and grain export bans or taxes on exports. Generalized consumer subsidies entail very high fiscal costs, and tend to be regressive. Further they act as a dis-incentive to food producers. Such subsidies are particularly bad when they subsidize foods 26 which are consumed mainly by the well off. Also, a risk with such measures is that they can become entrenched, incurring high fiscal costs. Price controls for "strategic" staples or on trader margins (e.g. several countries in Eastern Europe and Middle East) are not targeted to the most vulnerable and lead to the creation of parallel markets, with related governance issues, and also generally result in substantial leakages outside the country. Further, measures to keep producer prices low can create disincentives for domestic food producers, and end up being counterproductive. The one exception is when price controls are explicitly introduced as a temporary measure and are widely felt to be justifiable in terms of a higher social goal. In such cases, the risks of entrenchment will be minimized, as observed in recent interventions to limit price increases for staples during Ramadan in Morocco. For countries that are grain exporters, there may be political pressures to ban or tax grain exports in high price years. Grain export bans or taxes (e.g. India, Argentina, Croatia, Pakistan, Vietnam, Russia) are not targeted, and lead to leakages and governance issues, and result in a dis-incentives for food producers. Further, they bring negative externalities for other countries (international `beggar-your- neighbor' competition). Unfortunately, several countries have now implemented these types of measures. These policies tend to have a limited impact on domestic price levels and a significant negative effect on earnings for domestic producers and exporters. They can also lead to sharp price fluctuations in countries that depend on imports, proving harmful to the global system. 5.1.3 Interventions to mitigate the domestic fuel prices through short-run trade policy measures or administrative action 50. Policies to mitigate the impact of higher fuel prices on poor households include targeted urban transport subsidies (vouchers / cards to specific vulnerable groups) and/or targeted subsidies to urban transport operators. Both these options represent a second best policy for countries which do not have a well functioning safety nets policy. 51. Policy options which should be avoided include generalized fuel subsidies (or reduction of taxes), subsidizing a public transport company, providing subsidized credit for private purchase of public transport vehicles, and reducing import tariffs for spare parts for public transport vehicles. Generalized fuel subsidies (or reduction of taxes) entail high fiscal cost, and are highly regressive. They also introduce distortions and entail negative externalities on pollution. Subsidizing a public transport company also entails high fiscal cost, and is not well targeted, and difficult to manage. The provision of subsidized credit for private purchase of public transport vehicles is not targeted, and raises management and governance issues. Finally, a reduction in import tariffs for spare parts for public transport vehicles introduces distortions, and leads to a parallel market in spare parts (and related governance concerns), as well as leakage to other countries. In fact, as a result of the recent oil price hikes many countries have introduced measures to lower domestic fuel prices. Most developing countries control the domestic pricing and distribution of petroleum products. The IMF Fiscal Department carried out a brief survey in mid-2008 and found that from 48 countries in Africa only 15 had fully liberalized systems (including Kenya, Tanzania, Uganda), 8 had functioning automatic pricing formulae (and a further 8 suspended the use of their formula as a result of the crisis), and a further 21 had ad hoc pricing (Zambia; Cameroon, Chad, Gabon, DR Congo, Rep. of Congo). 27 Box 4: Measures to stimulate a medium-term food grain supply response While higher grain prices are clearly a burden to poor net purchasers of food, they also present an opportunity to stimulate foodgrain production and enhance the contribution of agriculture to medium-run growth. For example, higher prices weaken the rationale for costly floor prices or import tariffs for grain, and may facilitate the implementation of politically difficult trade reforms. Higher grain prices can also help to reverse a generally declining trend in government, private sector and donor investment in the agricultural sector. Agricultural producers such as Brazil, Malaysia and Thailand have made significant progress in agricultural commercialization in recent years, and have increasingly undertaken investments in research and extension necessary to promote increased agricultural productivity and reduced agricultural risk. However, some of the short-run policy options discussed above may limit the scope for longer-term solutions. For example, policy responses that seek to control markets through mandated grain prices, export restrictions, forcible procurement, or direct government involvement in marketing activities are likely to lower the food supply response over the medium term. Similarly, the introduction of anti-hoarding measures, such as restrictions on local storage and trade, can have lasting negative effects on the development of domestic food markets. In contrast, alternative measures such as the piloting of market-based risk management tools in Malawi, and the improvement of publicly accessible market information systems in India and Mali, are all likely to mobilize significant new resources in the private sector to cut marketing costs and improve efficiency of grain markets over the medium term. For many low-income countries, transport and logistics costs are a key component of food prices and are generally far higher than OECD benchmarks of around 9 percent. While countries can do little to reduce ocean shipping costs (which for high volume, relatively low value goods such as grains and edible oils represent a significant part of the final price), they can act to lower the overall cost of domestic distribution. The importance of strengthening inland transport links in mitigating price spikes was recently underscored in Congo Republic. Improvements in transport capacity stemmed the rise in food price inflation that was experienced in 2006, and further investments in transport links with Brazzaville are expected to be an important part of controlling price spikes. Investments in basic transport infrastructure have a proven record in reducing prices, particularly in remote locations in countries such as Nepal. Moreover, improvements in customs facilitation, logistics performance, and efficient grain storage can also have significant benefits for consumers, while generating a favorable supply response. 5.2 Possible policy responses in Mozambique: An analysis of the fiscal cost and benefit incidence of alternative policy measures using first-order partial equilibrium analysis 5.2.1 Accelerate safety nets expansion 52. In response to the increase in food and fuel prices, the government could scale up the existing social protection programs. Rising food and fuel prices are expected to push vulnerable households towards coping strategies that can have irreversible impoverishing impacts on families and children, such as asset depletion, removing children from school or reducing their daily caloric and nutrient intake. Social protection interventions in response to the crisis should therefore offer relief from this likely increased deprivation. In as far as possible, the government can expand and scale up existing government programs, in order to ensure that the proposed interventions can come into effect in the very short term and also are in line with government policies and priorities. The response to the crisis also needs to take into account the existing capacity constraints. 53. A basic system of safety nets has been established in Mozambique to support the extreme poor and the most vulnerable. Poverty is widespread with 54 percent of Mozambicans (or about 11 million people) living below the poverty line in 2002/03 (the year of 28 the most recent national household survey). Social protection programs benefited about 150,000 people directly in 2008, and approximately 900,000 including their dependants, equivalent to about 4 percent of the population. Hence the reach of the social protection programs is very small when considering the extent of poverty. Unfortunately the available household survey data do not allow us to assess the incidence of the safety net programs, in terms of whether the safety net programs successfully benefit the poorest and most vulnerable households. The main programs are run by National Institute of Social Action (INAS), under the Ministry of Women and Social Action (MMAS). (see Box 5). The Programa Subsídio de Alimentos (PSA) is a targeted cash transfers reaching about 128,000 direct beneficiaries in 2008, and largely targeted to the elderly. The Programa Apoio Social Directo (PASD) provides in-kind support to the chronically ill, and orphan and vulnerable children, but is limited to about 7,000 beneficiaries in 2008. Other small programs managed by INAS include the Social Benefit for Work (PBST), the Income generation Program (PGR), and the Community Development Program (PDC). Additional social protection programs to address malnutrition are managed by the Ministry of Box 5. Main Social Protection Programs in Mozambique The main Social Protection programs are run by National Institute of Social Action (INAS), under the Ministry of Women and Social Action (MMAS): The Food Subsidy Program (Programa Subsídio de Alimentos, PSA) is a cash transfers program targeted to extremely poor individuals with no capacity to work or support themselves. The name reflects its origins but it is not specifically linked to food subsidy. While the PSA is the largest existing social protection program, its coverage is very limited, and the amount of the cash transfers is small. The PSA makes monthly cash disbursements range between US$4 (MT100) for a single-person household to US$12 (MT300) monthly for a household with 4 or more dependants. The PSA reached about 128,000 direct beneficiaries in 2008 (and approximately 880,000 including their dependants, equivalent to about 4 percent of the population). The PSA is mainly targeted to poor elderly people, which account for 92 percent of beneficiaries. More than 80 percent are women, many responsible for Orphans and Vulnerable Children (OVC). The program also targets disabled, chronically ill and malnourished pregnant women. The Direct Social Support Program (Programa Apoio Social Directo, PASD) provides a one-off in-kind support to the absolutely poor who are unable to work and in need of immediate assistance. It is mostly targeted to chronically ill, and orphan and vulnerable children and malnourished children and malnourished pregnant women, family victims of fires or other disasters. It has very limited coverage, reaching only about 7,000 beneficiaries in 2008. The number of beneficiaries, however, is much larger when considering the food assistance program managed by WFP, in coordination with MMAS, in the areas affected by natural disasters, which benefited approximately 150,000 vulnerable households in 2008. The Income Generation Program (PGR) provides capital input to households or communities. The program provides support by identifying potential income generating activities and providing loans to beneficiaries. The loans are made available to groups of individuals who fulfill the eligibly criteria of the program. Repayment is soft and with amounts depending on the scale and type of the project. There are about 5,300 reported beneficiaries (not projects), of which 62 percent are women. The graduation criterion is the full repayment of the loan. In practice scarcely any group repays their loan at any time. INAS staff lacks capacity in financial management as identified in the INAS 2007 Annual Report. The Social Benefit for Work Program (PBST) provides monthly payments conditional on work. Total number of beneficiaries is about 3,200 of which more than 80 percent are women. This scheme is meant to give training for unskilled labor through jobs in other public sector bodies against receiving a monthly subsidy from INAS. The PBST is intended as a transitory program with beneficiaries graduating to employment where they were trained or self employment in income generating activities. The program has not proved successful as beneficiaries are not likely to secure future jobs in their training posts or move to self employment. 29 Health. These include supplementary feeding programs for malnourished children and the provision of general food rations to AIDS patients. There are also programs focused on retaining learners in schools which are managed by the Ministry of Education, and provide school feeding to children in selected primary, secondary, and boarding institutions. 54. An assessment of the possible safety nets response to the crisis has highlighted the need for an integrated approach. Taking into account the specific nature of the crisis and the status of social protection programs in Mozambique we recommend two parallel sets of measures:11 (i) Targeted interventions to reach the most vulnerable and socially excluded through the expansion of cash transfers schemes and the scaling up of food assistance; and, (ii) Targeted interventions to address specific vulnerabilities (chronic and acute malnutrition) and ensure that access to education is maintained. In this report we focus exclusively the possible scaling up of the PSA cash transfers program. Nevertheless the other interventions targeted at specific vulnerabilities should also be implemented in parallel. Such complementary measures include increased food assistance12 and labor based social assistance mechanisms, including urban employment schemes13, and are of particular importance to ensure that highly vulnerable groups are effectively protected from the worst impact of price increases in the very short term. 55. The main social protection policy response could be to strengthen, expand and scale up the existing PSA targeted cash transfer program implemented by the Ministry of Women and Social Action (MMAS). The PSA is the only governmental program (non- contributory transfers) with legally established implementation guidelines and is presently undergoing a process of consolidation with the technical support of bi-lateral donors and the United Nations as well as an expansion in line with targets set out in the Government's 5-year Poverty Reduction Strategy (PARPA II). The PARPA-II already envisages the gradual scaling 11A detailed set of recommendations on social protection options to respond to the crisis was prepared by the United Nations in May 2008 ("Outline for a National Response to the Rising Food Prices in Mozambique". UN System in Mozambique. Draft Discussion Note. May 2008), and subsequently refined by the World Bank, IMF, UNICEF, WFP, DFID, and the Netherlands. Here we focus on the main points suggested in that note. 12The need to reach vulnerable groups that are not directly targeted by the PSA requires a reformulation of the governmental in kind transfer program, Programa Apoio Social Directo, also implemented by MMAS, to ensure an increased focus on child-headed households. This recommendation would imply the standardizing of the basic package of materials provided to the different categories of beneficiaries covered by the program and of the periodicity of transfers. These changes can, however, be introduced within the current budget for the program. In addition, the expansion of the food assistance program towards orphaned and vulnerable children as well as the chronically ill is of critical importance. The existing program, active since 2003, supports 43,000 OVC and 12,500 AIDS affected households in the Provinces of Maputo, Gaza, Inhambane, Manica, Sofala and Tete with monthly food baskets. This program provides food commodities at a value of US$16 million per annum. It is proposed to expand the program by 50 percent (with a further cost of US$8 million per year). 13Social Protection also targets individuals that are able to engage in productive activities. Using different kind of mechanisms, many countries have introduced programs providing direct social assistance to the active poorest whilst enhancing their potential to take up the future income generating activities through paid or self-employment. Nevertheless, the experience has shown that to make these programs an effective tool in poverty reduction, a special attention has to be given to their design and implementation. As such, supporting the government in the design and implementation of programs or policies that could provide access to recurrent/seasonal income and increase the employability of the most vulnerable groups may be critical. In the immediate, WFP has proposed the possibility to support the Government response in urban areas through an urban employment scheme for young, unemployed city dwellers in the cities of Maputo and Beira. The program would target young people between the ages of 16 and 30 years old in the poorest neighborhoods of Mozambique's two largest cities for two years (January 2009 through December 2010). This program is not currently part of WFP's program portfolio in Mozambique and will be introduced specifically to assist in addressing the impact of high food prices in urban communities. 30 up of the PSA program over a number of years (i.e., independently of the higher food and fuel prices). In the context of the recent price shocks, however, the scaling up of the PSA could be accelerated to protect the poorest households from the impact of the crisis. The PSA program presents the advantage of being geared towards a relatively easy group to target (the elderly), and for which social assistance enjoys higher levels of social acceptance. Further, the PSA's focus on the elderly as well as orphaned and vulnerable children appears particularly appropriate in the context of the crisis, as these groups are extremely vulnerable to price increases and, unlike small landholders, not be able to benefit from the sale of their agricultural produce at increased prices. Based on preliminary simulations carried out by UNICEF in Mozambique, it would be possible to immediately scale-up targeted cash transfers (PSA) by increasing the amount of the transfer by 25 percent (from 100MT ­ 300MT to 125MT ­ 375MT), and expanding the number of direct beneficiaries to 290,000 by 2009 and 370,000 by 2010, concentrating in most affected urban areas. This increase in the PSA amount will mitigate the increased vulnerability experienced by households in a situation of extreme poverty. Considering the greater vulnerability in urban areas it is suggested that priority be given to Provincial and District capitals. 56. Scaling up the cash transfers program would be fiscally manageable and can be expected to be reasonably well targeted. The simulation of an immediate scaling-up scenario of the Targeted Cash Transfers (PSA) from 4 percent in 2008 to 14 percent of population by 2010, suggests that the fiscal cost would increase from the current US$14 million, to about US$33 million in 2010, which would almost entirely reach the beneficiaries given the relatively low administration costs of the PSA. No data exist on the benefit incidence of the PSA program, but based on the experience with reasonably well targeted cash transfers schemes in other countries we can expect a leakage of less than 20 percent of funds to the non-poor, with the richest quintile getting less than 5 percent of the transfers and the poorest quintile receiving about 40 percent of the transfers, corresponding to an increase in incomes of about 17 percent for the poorest quintile (Figure 10). It is worth noting that the distribution of benefits in poorly targeted cash transfer programs is roughly equally distributed across the population (i.e. approximately 20 percent of benefits per quintile of population). Figure 10: Assumed distribution of benefits from cash transfers safety nets program Assumed distribution of benefits from cash transfers 70 60 Assumed Shares 50 of benefits egatnecr 40 Subsidy as % of total consumption 30 Pe 20 10 0 1 2 3 4 5 Quintiles of pc expenditures Source: Assumed distribution based on experience from other countries. 31 57. There is a need to clarify the objectives of the PSA and refocus its targeting towards the poorest and most vulnerable households. While the short term response is focused on existing programs, more analysis is required to strengthen the PSA program. At present the PSA lacks a clear policy focus, as it is not clear whether its primary goal is to promote equity or to prevent poverty traps. The stated goal is to reduce poverty by focusing (mostly) on the elderly. However, no analysis has been done to assess whether targeting the elderly is the best possible targeting strategy / group. The PSA's targeting criteria draws on South Africa's experience, which suggests that children also benefit from a social pension scheme (see Box 6). However, it should be pointed out that evidence that children also benefit from a transfer to the elderly is not evidence that this is the best way to reach that younger group were it the main objective of the transfer policy. Further, while this criteria/focus may address poverty, it is Box 6: Does the PSA contribute to economic not clear that there is no better targeting growth? possible to reach the extremely poor. While the elderly are not at risk of work Some studies have shown that the elderly often use cash disincentives, there is no evidence that transfers for productive activities as well as to support household members (in particular children). A study on they are poorer than other groups. In fact, South Africa, for example, has demonstrated that a the 2003 household survey data suggests combination of cash transfers, including child benefits and that the elderly are relatively less likely to social pensions has resulted in a 23% reduction in poverty be poor than other groups, such as levels (Samson, et al., 2002). According to an children, or female headed households. econometric study carried out in South Africa employment rates for household members of `active age' were 8% to Hence, while in the short term it may 15% higher for those households receiving a transfer than make sense to scale up the PSA program for those where no transfer was received (Samson, 2002). in response to the higher food and fuel Additional cash increases the risk taking ability of younger prices, it is also important to start an in- household members in the search for better paid depth evaluation of the effectiveness of the employment. On the other hand, including cash transfers as part of the solutions to this issue enables an increase in PSA program in reaching the most the purchasing power of the poorest without creating a vulnerable households, and to start work distortion in domestic incentives to the increase of to develop a more cohesive Social agricultural production and without reducing the income Protection Strategy over the medium-term. of the poorest sellers of agricultural produce. Finally the The work should include a vulnerability cash amounts are often used to reinforce the productive capacity of the household and can stimulate local and mapping exercise, and assessment of the community level economic activity. Data from South impact of current programs with a view to Africa establishes a correlation between cash transfers and their reformulation, and a program of the expansion of micro-enterprises as well as other types experiments with different targeting of family-based economic activity (Barrientos and Holmes, 2002). approaches to improve effectiveness of social protection programs. 58. Accompanying interventions would be required to strengthen the capacity of INAS to administer a scaling up of the program. Recent studies have highlighted the administrative constraints in PSA, and have outlines a series of capacity building measures. Scaling up of the size of social protection programs will also require an upfront investment in institutions and strengthening capacity, including strengthening the financial and human resources management system, the information management systems, existing delivery mechanisms, the institutional capacity to appropriately target, and the institutional capacity to disburse cash to large numbers of people. Institutional capacity building is currently underway at the National Institute of Social 32 Action (INAS), covering a series of activities to strengthen management and information systems, supported by DFID, the Netherlands Embassy, and UN Agencies. The capacity building efforts are required to strengthen INAS' information and management systems. Capacity to link vulnerability analysis with the expansion of the PSA and to establish a mechanism for indexing disbursement amounts to inflation will be crucial for the medium to long- term expansion of the program. 5.2.2 Reduce import tariffs on milled grains to 2.5 percent 59. Almost all tariffs and taxes on foods have already been (close to) eliminated in Mozambique, except for the import tariff on milled grains. All basic foods are VAT exempt in Mozambique, and import tariffs on basic foods are also negligible (maximum 2.5 percent), except for milled grain products which face 20 percent import duty. The rationale for the import dury on milled products has been to protect the milling industry in Mozambique. However, the higher food prices have highlighted that it would be best to separate the industrial policies from trade policies. If support to the milling industry is warranted, it would be best done directly through a lump sum transfer, perhaps on a declining scale. This would constitute a less distortive industrial policy. The removal of import duty on milled wheat and maize would be equivalent to a transfer from the milling industry (and the Government) to consumers. Figure 11: Distribution of benefits from removal of import tariffs on milled grains Distribution of benefits from the tariff removal 50 45 Share of benefits 40 35 Subsidy as % of e total consumption 30 agt 25 cen er 20 P 15 10 5 0 1 2 3 4 5 Quintiles of pc expenditures Source: Own calculations 60. Lowering the import tariff on milled grains would be fiscally inexpensive and greatly benefit consumers, but there may be need for transition measures to avoid a collapse of the milling industry. The results of a back-of-the-envelope simulation using 2006 values and prices of the impact of the reduction in import duty on milled wheat and maize to 2.5 percent, highlights that the fiscal cost to the government from the loss of tariff revenues would be minimal, at approximately US$0.3 million. On the other hand the loss of protection by the milling industry would correspond to approximately US$29 million, which would be transferred from the milling industry to consumers if the import duty were to be reduced to 2.5 percent. Of course, the fiscal costs for the government could be higher if it decides to compensate the milling 33 industry for the loss of protection by providing a direct lump-sum transfer of resources from the budget. In any case, the total benefit to consumers would be around US$29.3 million, of which the richest quintile would receive about 35 percent, and the poorest quintile would receive approximately 10 percent, corresponding to a very small increases in their income of around 1 percent (Figure 11). 5.2.3 Food price subsidies: maize, rice and wheat 61. Food subsidies are notoriously difficult to administer and target, and Mozambique does not have any food staple products with the characteristics required for a successful food subsidy. Substantial international experience has highlighted that food subsidies generally do not work well (Per Pinstrup-Andersen 1988; Alderman and Lindert, 1998; Farrar 2000). Food subsidies work best when the subsidized product is (a) self-targeted (i.e. consumed mainly by the poor), (b) occupies large share in budget of the poor; and (c) is traded across the country and enters wholesale (or import) markets for central subsidization (so that prices can be affected). As discussed in Section 3, however, Mozambique does not have any staple food products that have those characteristics. Maize occupies a significant share of the poor's budget in the urban areas in the north and the center, but overall it is only slightly self-targeted. Rice has significant share of the poor's budget in Maputo and urban areas in the south and the center, and in rural south. Hence, it is well matched to the areas that are worst affected by crisis, as highlighted in Section 3. However, rice is only slightly self-targeted. Wheat products are consumed in significant quantity in Maputo city and in the south, but they are not mainly consumed by the poor. In fact wheat is a luxury for household in the north and the center regions, even in urban areas. Those households mainly consume cassava and maize. Cassava is well targeted to the poor in the north, but the lack of an integrated national market for cassava makes it impossible to subsidize its price.14 Yellow maize had been used in the past with some success, as it is not the preferred type of maize, which facilitates self-targeting (Dorosh, del Ninno, and Sahn, 1994; Tshirley and Santos, 1995). However, there are some political costs associated with yellow maize in Africa since advocates for the poor regularly claim that the government is giving the poor 'animal feed'. In sum there are no obvious products for a food subsidy in Mozambique, and as a result any subsidy would not be well targeted and would be expensive because the benefits from the subsidy would have to be spread across the entire population. Table 11: Simulation of introducing a 20 percent price subsidy on wheat, maize, and rice Fiscal Value of Share of Increase % in Share of benefits costs benefit to benefits income of (US$ consumers poorest quintile Poorest 20% Richest 20% Poorest 20% Richest 20% million) (US$ million) (national) (national) (urban) (urban) Wheat 10.4 10.6 0.2 6.1 42.4 6.8 38.8 Maize 1.7 6.6 0.3 14.3 25.9 14.6 27.0 Rice 18.3 21.0 0.8 9.8 38.4 13.2 25.5 Note: Preliminary estimates based on 2006 quantities and prices, for indicative purposes only on relative costs and distribution of benefits. Only first order effects. No substitution effects. Import demand assumed inelastic. 14Notably there are no wholesale markets for cassava. 34 62. A simple simulation highlights that a food subsidy would be expensive to the budget and would mostly benefit the richest households. In order to get a better idea of the likely costs and distribution of benefits, we carried out a back-of-the envelope simulation of introducing a 20 percent price subsidy on wheat, maize, and rice. The basic idea is that it would be sufficient to subsidize the imports entering the border to affect the domestic market price. Our calculations are based on data from the 2006 national accounts. The results are summarized in Table 11. The fiscal costs of the subsidy could be relatively low for maize, because most of the consumptions is produced domestically. On the contrary the fiscal costs would be substantial for wheat and rice since most of the domestic consumption is imported (and hence most of the cost of the subsidy would have to be borne by the government). The distribution of the benefits is heavily skewed in favor of the richer households, since they tend to consume more of the products. The share of benefits accruing to the richest quintile of the population ranges from as much as 42 percent for a wheat subsidy, 38 percent for a rice subsidy, and 26 percent for a maize subsidy. The share reaching the poorest quintile ranges from only 6 percent for a wheat subsidy, 10 percent for a rice subsidy, and 14 percent for a maize subsidy. In sum, food subsidies in Mozambique would be expensive (except perhaps on maize) and poorly targeted. In addition they would introduce a dis-incentive to local production, thus undermining the long term response to the crisis. 5.2.4 Fuel price subsidies 63. Measures to respond to the crisis could also focus on reducing the increase in fuel prices, through a subsidy or a reduction in taxes. Mozambique uses a formula to link retail fuel prices at the pump to world prices. The formula has been in place for the past few years and has worked fairly well. It would be possible, however, to introduce a subsidy as part of the formula, or to remove some of the taxes on fuel. Alternatively, failure to pass through the price increases according to the formula would in fact be a price subsidy (which would eventually have to be reimbursed to fuel importers).15 64. International experience suggests that subsidizing fuel prices is expensive. A fuel subsidy is expected to be very expensive since all domestic consumption is imported. Experience from other countries shows that fuel subsidies can easily reach enormous fiscal costs. In Cameroon fuel subsidies accounted for 0.3 percent of GDP in 2005, and 0.6 percent of GDP in the Central African Republic in 2007, and as much as 4.2 percent of non-oil GDP in Gabon in 2006, and 2.2 percent of Ghana's GDP in 2004, and 5.6 percent of non-oil GDP in the Republic of Congo in 2007. 65. A simulation suggests that a generalized fuel subsidy would be extremely expensive and highly regressive. In order to examine the likely impact of a fuel subsidy in Mozambique, we have carried out a back-of-the-envelope simulation of introducing a 20 percent price subsidy on gasoline and 10 percent on diesel. The simulation used data from the 2006 national accounts. The fiscal costs of the subsidies would be extremely high, at around US$100 million per year (equivalent to 1.25 percent of GDP in 2007). Further, as discussed in Section 3, fuel is used most by the richer part of the population, and as a result a blanket subsidy to fuel is expected to be highly regressive. As expected, most of the benefits would accrue to the richest quintile, who 15In June 2008 the government temporarily removed import duties and VAT on diesel and import duties on kerosene, thus partially offsetting the increase in world prices since January 2008. The taxes are scheduled to be reinstated from January 2009. 35 would receive around 71 percent of the value of the subsidy. The poorest quintile of the population would only receive 3 percent of the benefits. In both cases, the subsidy would correspond to just a small increase in incomes, at 1 percent for poorest quintile and 3 percent for richest (Figure 12). Figure 12: Distribution of benefits from a generalized fuel subsidy Distribution of benefits from the fuel subsidy 80 70 Share of benefits (National) 60 Share of benefits egatnecr 50 (Maputo) Subsidy as % of total 40 consumption (National) 30 Subsidy as % of total Pe consumption (Maputo) 20 10 0 1 2 3 4 5 Quintiles of pc expenditure Source: Own calculations 5.2.5 Support to the semi-collective urban transport system 66. An alternative the short-term instrument to mitigate impact of the higher fuel prices on the poor is to focus the subsidy to the urban transport system. For instance in Mozambique the Government introduced a monthly lump-sum compensation to registered-and- tax-compliant urban transport private operators (minivans, buses) from February 2008, in exchange for keeping fares constant in spite of higher fuel prices.16 The subsidy has been designed as the difference between market retail price of diesel fuel at the pump and the price per liter in early February 2008, corresponding to subsidy of around 15 percent of pump prices for most of the period between February to September 2008. The subsidy to semi-collective urban transport operators started in February 2008, and was expected to cost of 0.15 percent of GDP in 2008. Approximately 85 percent of the vehicles are licensed to operate in Maputo city, so the subsidy is mostly to Maputo urban transport operators/users. The subsidy, which is really a lump-sum transfer to the urban transport industry, is only targeted in as far as the users of urban transport (minivans and buses) are mainly the poorest 80 percent of the population. However, only 21 percent of the subsidy accrues to the poorest 40 percent of the population. 16It should be noted that there is a need to develop longer run solutions for urban transportation, in order to improve the quality of services and keep prices competitive. 36 Figure 13: Distribution of benefits from a targeted subsidy to semi-collective transport Distribution of benefits from the transport subsidy 60 Share of benefits (National) 50 Share of benefits (Maputo) egatnecr 40 Subsidy as % of total consumption (National) 30 Subsidy as % of total consumption (Maputo) Pe 20 10 0 1 2 3 4 5 Quintiles of pc expenditure Source: Own calculation 67. A simulation of a subsidy to the transport industry suggests this is a better alternative compared to a generalized fuel subsidy, but is not targeted to the poorest. In order to examine the likely impact of a subsidy to urban transport operators in Mozambique, we carried out a back-of-the-envelope simulation of introducing a 20 percent price subsidy on diesel consumed by the industry (based on based on estimates regarding the level of fuel consumption, in line with the existing program). The fiscal cost would be expected at about US$17 million, of which about 32 percent of benefits would accrue to the richest quintile of the population in Maputo, and only about 7 percent to the poorest quintile. Since the benefits would accrue largely to the residents of Maputo, the subsidy could represent a significant increase in incomes in Maputo, at around 3.5 percent for the poorest quintile and 3 percent for richest quintile (see Figure 13). 5.2.6 Summary of the impact of alternative policy measures using first-order partial equilibrium analysis 68. The policy simulations highlight that the first best short-term measure would be to scale up the direct cash-transfers PSA program. Reducing tariffs on imported milled grain (wheat and maize flour) is also a good policy response. However, in order to avoid a collapse of the milling industry, it would be best to accompany the removal of the tariffs with a temporary direct transfer to the milling industry (possible decreasing over time). The targeted lump-sum compensation to the semi-collective transport industry is reasonably efficient in the short term. 69. The results highlight that the least preferable policies are the provision of blanket fuel subsidies (or fuel tax removals). Food subsidies are also not recommended, as they introduce a dis-incentive to domestic production. In terms of commodities, subsidies to wheat and rice are expected to be relatively more expensive compared to maize (the actual level depends, of course, on the target price). However, they are better targeted to the poor compared to maize subsidies, and particularly rice is well targeted to the areas most negatively affected by the higher food prices. 37 Table 12: Summary table of simulated impact of policies responses using first-order partial equilibrium analysis1 Fiscal costs Value of % increase in Share of benefits Share of benefits (USD consumer income of (national) (Maputo) millions) benefit (USD poorest quintile Poorest Richest Poorest Richest millions) (national) quintile quintile quintile quintile Cash transfers (expansion) 33.0 27.0 17.0 40.0 5.0 Price subsidy (Wheat) 10.4 10.6 0.2 6.1 42.4 6.8 38.8 Price subsidy (Maize) 1.7 6.6 0.3 14.3 25.9 14.6 27.0 Price subsidy (Rice) 18.3 21.0 0.8 9.8 38.4 13.2 25.5 Fuel price subsidy 99.0 99.0 1.0 3.0 71.0 3.5 60.0 Transport subsidy to 17.0 17.0 3.5 (Maputo) 5.0 51.0 7.0 32.0 industry Remove import tariff 02 29.3 1.0 10.0 34.0 on milled grains Notes: 1. Preliminary estimates based on 2006-2007 quantities and prices for indicative purposes only on relative costs and distribution of benefits. Only first order effects. No substitution effects. Import demand assumed inelastic. Reasonable assumptions made on administrative costs and targeting for cash transfers. 2. Fiscal cost of removing import tariffs on milled grains is zero. However, it would be appropriate to provide the milling companies a lump sum compensation decreasing over time to assist them to adjust to the loss of protection. 5.3 Possible policy responses in Mozambique: An analysis of the impact of alternative policy responses using the CGE model 70. A number of policy responses to the rising food and fuel prices are simulated in this section using the Mozambique CGE model.17 First, in the `Subsidies' scenario, we examine the impact of applying fuel and food subsidies at the border. These subsidies are designed to eliminate 25 percent of the international price increases for all of the commodities shown in Table 5. Secondly, in the `Liberalization' scenario, we eliminate import tariffs on agricultural products and processed foods. However, fuel taxes, which generate significant revenues, are maintained. Finally, in the `Agricultural technology' scenario, we model investments in the agricultural sector that are presumed to lead to 10 percent improvement in total factor productivity across all agricultural sectors. A primary difficulty in analyzing such a policy revolves around uncertainty as to the costs and institutional arrangements required to achieve the productivity gains. These issues are not addressed here. However, in order to emphasize that costs will be incurred, a 10 percent increase in recurrent government spending is imposed on the model alongside the agricultural productivity gains. Consistent closure rules are applied implying that the incremental spending is deficit financed. Tables 13-16 present results of the three policy response simulations, and are of the same format as Tables 6-9, which presented the price impact 17Some of the specific policy options discussed in Section 5.2 have not examined with the CGE. This is because the CGE model is not well suited for such simulations. For instance, while CGE models can be used to examine large scale social protection programs, generally they are not well suited to model poverty traps and the benefits from social protection programs. Notably, it is difficult to model the degree to which a price shock leads to reduced investments in health and schooling (as well as to other investments), and therefore the long term benefits of social protection. In addition, the current program in Mozambique is too small to generate substantial general equilibrium impacts; hence, there is limited value to analyzing the program in a CGE framework. 38 results. All three policy simulations are compared to the `Combined' simulation, which depicts the price shocks under the assumption of a constant policy environment. Table 13: Macroeconomic results for policy responses Change from base year value (%) Combined Subsides Liberalization Agricultural scenario (food & fuel) (food only) technology Quantities GDP -1.2 -1.1 -1.2 1.2 Absorption (C+I+G) -5.1 -5.3 -5.2 -3.1 Consumption (C) -7.3 -5.3 -7.1 -3.9 Investment (I) -1.2 -8.2 -2.0 -7.8 Recurrent government (G) 0.0 0.0 0.0 10.0 Exports (E) 5.9 7.6 6.7 7.1 Imports (M) -9.6 -9.5 -9.4 -8.8 Prices Nominal exchange rate -1.5 0.5 -0.7 13.5 Real exchange rate 15.2 17.4 16.2 -2.9 Terms-of-trade -16.2 -16.2 -16.2 3.5 Source: Results from the Mozambique CGE model Table 14: Sectoral production results for policy responses Base value- Change from base year value-added (%) added share Combined Subsides Liberalization Agricultural (%) scenario (food & fuel) (food only) technology Agriculture 25.9 0.7 1.0 0.7 8.7 Cereal crops -0.9 2.9 3.0 2.7 11.6 Roots crops 1.1 -1.8 -1.6 -1.9 6.8 Pulses and groundnuts -1.2 4.2 4.4 4.5 14.5 Horticulture 9.4 -1.7 -1.4 -1.7 7.4 Export crops -0.4 21.2 20.3 22.1 37.1 Livestock -0.3 4.2 4.4 3.8 10.7 Forestry 3.8 -1.9 -1.4 -1.8 -2.6 Fishery -0.2 -5.9 -4.0 -5.2 3.0 Non-Agriculture 1.0 -0.5 -0.6 -0.5 0.4 Industry -0.1 1.4 -0.1 1.2 0.4 Mining 1.0 -0.5 -0.1 -0.4 -1.5 Manufacturing 1.3 3.2 3.6 3.2 4.8 Primary product processing 1.1 5.7 6.5 5.6 8.7 Other industry 1.7 -1.2 -5.8 -1.6 -6.0 Electricity -2.2 0.2 0.6 0.4 -1.3 Water 1.0 -2.1 -1.7 -2.2 -1.5 Construction -0.7 -1.5 -7.6 -2.2 -7.4 Services 0.0 -1.4 -0.9 -1.4 0.4 Source: Results from the Mozambique CGE model 39 Table 15: Factor price results for policy responses Change from base year factor prices (%) Combined Subsides Liberalization Agricultural scenario (food & fuel) (food only) technology Rural labor Skilled -1.6 1.5 -1.5 0.4 Semi-skilled -4.6 -2.9 -4.7 -0.5 Unskilled -0.7 2.4 -0.6 -1.4 Urban labor Skilled -5.7 -3.8 -5.6 0.7 Semi-skilled -8.2 -6.0 -8.2 -1.8 Unskilled -7.3 -5.6 -7.4 -3.2 Capital -6.4 -5.0 -6.4 -7.0 Agricultural land 9.5 13.3 9.8 5.5 Source: Results from the Mozambique CGE model Table 16: Welfare and poverty results for policy responses Base year Change from base year (%) value Combined Subsides Liberalization Agricultural scenario (food & fuel) (food only) technology Equivalent variation National -7.4 -5.4 -7.2 -4.1 Rural households Quintile 1 -3.9 -1.8 -3.6 0.0 Quintile 2 -3.2 -0.8 -2.8 -0.2 Quintile 3 -2.7 -0.3 -2.3 0.3 Quintile 4 -3.4 -0.8 -3.0 -0.9 Quintile 5 -4.4 -1.9 -4.1 -2.9 Urban households Quintile 1 -11.1 -9.8 -10.9 -4.5 Quintile 2 -11.6 -9.9 -11.4 -6.2 Quintile 3 -10.9 -9.4 -10.7 -5.3 Quintile 4 -11.1 -9.2 -10.9 -6.9 Quintile 5 -9.4 -7.7 -9.3 -6.0 Poverty headcount National 54.1 58.2 56.8 58.0 55.7 Rural households 10.7 57.7 55.9 57.3 56.1 Urban households 9.6 59.5 58.6 59.5 54.9 Source: Results from the Mozambique CGE model 5.3.1 Food and fuel subsidies 71. Food and fuel subsidies provide benefits to consumers, but these short-term gains come at the expense of investment for the future. Table 16 indicates that household consumption declines by two percentage points less than in the `Combined' simulation. However, investment declines by a further six percentage points due to the finance required for subsidization. Consistent with the household consumption aggregate, household welfare analysis shown in Table 16 illustrates a two percentage point gain relative to the `Combined' scenario. Interestingly, rural households gain relatively more than urban households across all quintiles even though urban households are, by far, the principal consumers of imported foods. This is also reflected in the poverty rates, which show a larger mitigation effect for rural than urban households. Within each zone, the incidence of the subsidy is relatively even across household expenditure quintiles, with a slight tendency for higher-income households to benefit more. 40 72. The relatively larger gains registered by rural households in the `Subsidy' scenario stem from second order macroeconomic impacts. By subsidizing expensive fuel and food imports, the subsidy effectively increases the macroeconomic burden of adjustment, particularly with respect to the balance of payments. Greater imports of fuel and food imply increased foreign currency needs, which in turn require an even more dramatic export response and/or greater import compression in non-subsidized sectors. From Table 13, one sees that the subsidy forces the largest cut in absorption (the best economywide indicator of welfare) of any scenario. This is because it requires the largest increase in exports and reduction in imports. These adjustments are achieved via a substantial depreciation of the real exchange rate, which provides even greater stimulus to agricultural exports. 73. Overall, while large-scale subsidies enhance household welfare in the short-run, they are expensive, are not particularly well targeted, and exacerbate the burden of macroeconomic adjustment. If financing the subsidies reduces the investment budget, future growth is likely to be sacrificed. Though not modeled, subsidy policies are often difficult to administer and subject to fraud. Finally, international experience indicates that, once enacted, general subsidies can prove to be exceedingly difficult to remove, thus generating a long-term drain on government finances. 5.3.2 Trade liberalization 74. Trade liberalization is the second policy response considered. In principle, trade liberalization is equivalent to subsidization at the border if the subsidy simply offsets the tariff. However, because tariffs are relatively low in Mozambique, the subsidy analyzed above brought domestic prices below world price equivalents. Hence, trade liberalization implies a much smaller loss in revenue. In addition, a zero tariff is much easier to administer than an overlapping tax and subsidy policy. Though not modeled here, reduced border tariffs also tend to reduce evasion, thus providing a further cushion to the revenue effect of reduced border tariffs through greater collection of VAT at the border.18 Finally, reduced or eliminated tariffs are consistent with Mozambique's fundamental open economy policy stance. 75. Since tariff rates are already relatively low (though effective protection rates for some processing sectors, such as grain milling, are high), the economywide impacts of reducing tariffs are relatively small. Household welfare, shown in Table 16, increases marginally with the gains fairly evenly distributed across rural/urban areas and across expenditure quintiles. A shift in the components of absorption (i.e., a reduction in investment) contributes to these gains. Trade liberalization also opens the economy to the world engendering an increase in exports and a decrease in imports. These adjustments may be unwelcome during a period when similar adjustments are required to confront the commodity price shocks. 76. In summary, the world price shocks may provide an opportunity to reduce high effective protection rates afforded to some food processing sectors. Given the regional differentiation of the Mozambican economy, these tariff reductions may provide some relief to consumers in urban centers, particularly Maputo where the import-intensity of demand is highest. However, at the same time, the moment is likely inopportune for a policy induced shut- 18See Arndt and Tarp (2008) and van Dunem and Arndt (2006) for a discussion on the relationship between tax rates and tax evasion in Mozambique. 41 down of some food processing factories. The prudent way forward is likely to involve incremental liberalization and thus commensurately smaller gains for consumers. 5.3.3 Agricultural technology 77. The final policy scenario indicates that improved agricultural technology is the preferable policy response to higher world prices. Agricultural technology improvements represent a powerful impetus to the economy (see Table 13). As a result of the productivity gain and associated increase in agricultural production, the economy achieves substantial gains in exports and a reduction in imports. Unlike in previous scenario, the real exchange rate actually appreciates due to better export performance. The reduction in absorption is about 40 percent smaller than in the `Combined' scenario. In addition, by increasing marketed surplus, agricultural technology gains reduce agricultural commodity prices. Thus, the gains from agricultural technology accrue primarily to urban households (see Table 16). This usefully offsets the impacts of the world price shocks, whose adverse affects are concentrated on urban households. Despite the domestic commodity price declines, rural households also experience significant gains in welfare. Within rural and urban zones, the registered gains are strongly progressive across income quintiles. 78. While the benefits of improved agricultural productivity are pronounced, tapping into the known sources of these gains remains a challenge in the short and longer term. Enhancing agricultural productivity in Mozambique has been on the policy agenda since the end of the civil war in the early 1990s but unfortunately little has been achieved to date (Uaiene, 2008). Whatever gains have taken place are difficult to ascribe to actions undertaken by the government (Arndt et al., 2007), beyond the fact that the government followed an overall policy framework which did not hinder smallholder initiatives which have been associated with some productivity gains. At present, the use of agricultural inputs and improved technology (seeds, irrigation, mechanization, etc.), and consequently agricultural yields for major food crops in Mozambique remains among the lowest in Africa (World Bank, 2008). Analysis of improved agricultural technology adoption indicates that households with access to credit and extension advisory services as well as members of agricultural associations are more likely to adopt new agricultural technologies (Uaiene, 2008). Households with higher levels of education are also more likely to adopt. Finally, results suggest that outgrower schemes by providing credit to farms can help stimulate agricultural technology adoption. Following over a decade of expansion-based agricultural growth, there is an urgent need in today's context, to intensify agriculture to achieve higher production levels and sustain the growth rates the sector has achieved until recently. 79. In this context, the government has embarked on a `Green Revolution' adopted in 2007, whose implementing agenda is the 2008-2011 `Food Production Action Plan'. The Action Plan was designed as the government's supply response to the higher food prices crisis, and its main objective is to reduce the country's dependency from imports in all major food crops. The Action Plan aims to increase agricultural productivity by promoting the adoption of improved technologies, facilitating credit and building storage facilities. The government commitment to the Green Revolution and the Action Plan has translated into a substantial increase in the budget allocations for the public institutions involved in the agricultural sector, as reflected in the 2009 budget and the country's 2009-2011 Medium Term Expenditure Framework. 42 80. While the model simulations underline the importance of expanding agricultural production, experience suggests that this will be difficult to achieve in the short term and is therefore unlikely to address the immediate impacts of the current food and energy crisis. Experience in Mozambique and other countries shows that the broad-based availability of key determinants of agricultural productivity (e.g. new varieties and improved seeds, functioning irrigation systems, etc.) requires a longer period of time than the 3-years envisaged in the Action Plan. Notably, as mentioned above, a sustained increase in agricultural productivity requires efficient systems to deliver basic services to smallholders, such as extension and rural finance, which are critical for the adoption by the smallholders of the new technologies promoted, yet they take a long time to be developed. Hence, the considerable gains in agricultural productivity are likely to require a longer time frame than the 3-years envisaged in the Action Plan. 6 CONCLUSIONS 81. The analysis carried out (using a number of different approaches) indicates that the macroeconomic and poverty impacts of world price increases in food and fuel will be negative and substantial, particularly for urban households and the South region. The impacts of increase in world food and fuel prices experienced in 2007 and 2008 have been analyzed using three approaches, aimed at capturing the short and longer-term effects as well as differential impacts across rural/urban households in different regions of the country. The findings indicate that the world price increases constituted a substantial negative terms-of-trade shock for Mozambique. Moreover, significant policies to insulate domestic fuel and food markets from the international price increases have not been put into place. Evidence from domestic price series indicates that the world price rises are being transmitted to the domestic economy. A poor persons' consumer price index (PCPI), developed for this analysis, indicates that the increase in the cost of the basket of commodities consumed by lower income households is similar to the increases registered for the average economywide basket. However, regional differences were observed. Net benefit ratio analysis indicates that urban households and households located in the South are generally more vulnerable to food price increases, while rural households often benefit from their net seller position, particularly those in the middle of the income distribution. Analysis using a CGE model of Mozambique indicates that the fuel price shock is more important from both macroeconomic and poverty perspectives. The CGE model simulations also highlight the importance of agricultural production response in general and export response in particular. The findings from all approaches conclude that the macroeconomic and poverty impacts of the world price increase will be negative and substantial, particularly for urban households. 82. The first best short term measure to alleviate the impact of the shock on households would be to scale up the direct cash-transfers PSA program, while the least preferable policies would be the provision of blanket fuel subsidies (or fuel tax removals). The analysis of possible policy responses highlights that a blanket fuel subsidies (or fuel tax removal) is expensive and highly regressive. Similarly, food subsidies are not recommended, as they are expensive and introduce a dis-incentive to domestic production. Social protection programs (preferably cash transfers) offer the best policy responses to mitigate the impact of the shocks for poor households in the short term. Other recommended policy measures to alleviate the impact of the price shocks in the short term include reducing tariffs on imported milled grains (accompanied with a temporary assistance to the milling industry), and scaling-up of the targeted lump-sum compensation to the semi-collective transport industry. The government already has 43 initiatives in all three of these most preferred options, since it has already decided to scale up the PSA cash transfers program in 2008, it is pursuing an agenda of trade liberalization within SADC, and it introduced a lump-sum transfer to the semi-collective urban transport operators in February 2008. 83. The analysis of policy responses using the CGE model also points to difficult trade- offs between short-run mitigation measures and long-run growth. In the long term, improved agricultural technology is the preferable policy response to higher world prices. Agricultural technology improvements represent a powerful impetus to the economy (see Table 13). As a result of the productivity gain and associated increase in agricultural production, the economy achieves substantial gains in exports and a reduction in imports. Moreover, while improving agricultural productivity is most effective in addressing the adverse effects of higher food (and fuel) prices, expanding agricultural production will remain difficult despite improved agricultural terms-of-trade, and is unlikely to address the immediate impacts of the current food and energy crisis. 84. The volatility of commodity prices highlights the need to develop analytical tools to quickly assess the impact of different shocks, such as the ones presented in this study. In recent months, commodity prices have decreased substantially from their peak levels of mid- 2008. Notably, fuel prices are currently below their level in early 2007 and food prices have decreased to an intermediate level. Already, therefore, the permanent impact of the shocks is expected to be smaller than simulated in this study. Furthermore, if prices were to go back all the way below their level in early 2007, the projections of the model are not going to be realized, and the impact of the shocks would be limited to a temporary delay in the pace of poverty reduction. Of course, the model could be used to run a new set of projections using the new level of prices. More generally, this study illustrates the type of analysis that can be used to assess the impact of a range of terms-of-trade shock, and what policy responses might be best applied. 85. The recent crisis emphasizes the value to develop flexible policy programs which can be adapted to respond rapidly to different shocks. The recent volatility in commodity prices is not unusual and underlines the need for policies designed to allow the government to implement a rapid and flexible response, which can be adapted to the level of the shocks. The recent global financial crisis is soon expected to have a negative impact on Mozambique, and again enhances the importance of the role of social protection programs as recommended in this policy note. Protecting vulnerable households from the impact of the external shocks by strengthening safety nets is the long run ideal. These measures support the purchasing power of the poor without distorting domestic incentives to produce more food, and without reducing the incomes of poor food sellers. The ideal measure is to provide cash or near-cash transfers that are conditional upon meeting a requirement (such as low income, location or occupation) or engaging in a mandated behavior (such as sending children to school). One drawback of cash transfer programs is that the fixed costs to set up and administer such a program are significant, such that these programs are not cost-effective if the intention is to have only a short term intervention (e.g. to respond to a single climatic event). On the other hand, as it makes sense to expect that there will be repeated shocks over time (even if there nature and timing are unknown), an important advantage of cash transfers is that they are flexible so that, once the fixed costs to establish a cash transfers program has been incurred it can be adapted to different shocks. Hence building up a sizeable and cost-effective cash transfers program provides the government with a flexible response mechanism to a range of shocks. 44 REFERENCES Aksoy, A., Isik-Dikmelik, A., 2007. Are Low Food Prices Pro-Poor? Net Food Buyers and Sellers in Low-Income Countries. Policy Research Working Paper N. 4642. The World Bank, Washington DC. Alderman H., and Lindert K., 1998. The Potential and Limitations of Self-Targeted Food Subsidies. The World Bank Research Observer, vol. 13, no. 2 (August 1998), pp. 213-29 Alderman, H., 1988. The twilight of flour rationing in Pakistan. Food Policy, 1988, 13, (3), 245- 256. Arndt, C. and Tarp.F., 2008. "Trade Policy and the Missing Revenue." Journal of African Economies. 17(2008): 131 - 160. Arndt, C. and Tarp, F., eds. Taxation in a Low Income Economy: The Case of Mozambique. Routledge. Forthcoming. Arndt, C., Benfica, R., Tarp, F., Thurlow, J., and Uaiene, R., 2008. "Biofuels, Poverty, and Growth: A Computable General Equilibrium Analysis of Mozambique." Eleventh Annual Conference on Global Economic Analysis. Helsinki, Finland. June 2008. https://www.gtap.agecon.purdue.edu/events/Conferences/default.asp#Past Arndt, C., Jones, E.S., and Tarp, F. 2007. Aid and Development: The Mozambican Case. In: Lahiri, Sajal ed. Frontiers of Economics and Globalization: Theory and Practice of Foreign Aid. Amsterdam: Elsevier, 2007, pp. 235-288. Barrett, C. B., Dorosh, P. A., 1996. Farmers' Welfare and Changing Food Prices: Nonparametric Evidence from Rice in Madagascar. American Journal of Agricultural Economics 78 (August): 656-669. Barrientos, A. and Holmes, R., 2002. Non contributory Pensions and Social Protection. Issues in Social Protection, International Labour Organization. Boughton, D., Mather, D., Barrett, C.B., Benfica, R., Abdula, D., Tschirley, D. and Cunguara, B. "Market Participation by Rural Households in a Low-Income Country: An Asset-Based Approach Applied to Mozambique". Faith and Economics, 50: 64-10. 2007. Budd, J. W., 1993. Changing Food Prices and Rural Welfare: A Nonparametric Examination of Côte d'Ivoire. Economic Development and Cultural Change 41 (3): 587-603. Campa, J., Goldberg, L., 2005. Exchange Rate Pass Through into Import Prices, Review of Economics and Statistics, November, 87 (4): 679-690. Campa, J., Goldberg, L., 2006. Distribution Margins, Imported Inputs, and the Sensitivity of the CPI to Exchange Rates. NBER Working Paper 12121. Chen, S., Ravallion, M., 2000. How did the world's poorest fare in the 1990s? Policy Research Working Paper Series 2409, The World Bank, Washington DC. Christiaensen, L., Demery, L., 2007. Down to Earth: Agriculture and Poverty Reduction in Africa. Directions in Development. Washington DC: World Bank. Christiaensen, L., Demery. L. 2006. Revisiting The Role of Agriculture in Poverty Reduction in Sub Saharan Africa. Mimeo, World Bank. Cirera, X., Nhate, V., 2008. An Empirical Estimation of the Degree of Price Transmission from Border to Consumer Prices in Mozambique. Discussion paper No. 39E Chapter 11, April 2007, National Directorate of Studies and Policy Analysis, Ministry of Planning and Development, Republic of Mozambique. http://www.mpd.gov.mz/gest/documents/chapterbook/chapter11_An%20Empirical%20Esti mation%20of%20the%20Degree.pdf. 45 Curtis Farrar, 2000. A Review of Food Subsidy Research at IFPRI. Impact Assessment Discussion Paper No. 12. January 2000. International Food Policy Research Institute. Washington, D.C. Deaton, A., 1989. Rice Prices and Income Distribution in Thailand: A Non-Parametric Analysis. Economic Journal 99 (395 Supplement Conference Papers): 1-37. Dimaranan, B. V., Editor. Global Trade, Assistance, and Production: The GTAP 6 Data Base, Center for Global Trade Analysis, Purdue University. 2006. Dorosh P., del Ninno C., and Sahn D., 1994. Food Aid and Poverty Alleviation in Mozambique: The Potential for Self-Targeting with Yellow Maize, Cornell Food and Nutrition Policy Program Working Paper 50. Ithaca, NY: Cornell University, 1994. Frankel, J., Parsley, D., Wei, S. 2005. Slow Pass-through Around the World: A New Import for Developing Countries? NBER Working Paper 11199. Goldberg, P., Knetter, M., 1997. Goods Prices and Exchange Rates: What Have We Learned? Journal of Economic Literature, vol. 35, 1243-92. INE (National Institute of Statistics). "Consumer price index electronic data." Maputo, 2008. INE (National Institute of Statistics). "Household consumption survey 2002-03 electronic data." Maputo, 2003. Klytchnikova, I., Diop, N., 2006. Trade Reforms, Farm Productivity and Poverty in Bangladesh. World Bank Working Paper No. 3980. The World Bank: Washington, D.C. McCool, K., Thurlow, J.T., Arndt, C. "Documentation of Social Accounting Matrix (SAM) Development" in Arndt, C. and Tarp, F. eds. Taxation in a Low Income Economy: The Case of Mozambique. Routledge. Forthcoming. Ministry of Planning and Development, and UNICEF, 2007. Expandir o programa de Subsidios Financeiros em Moçambique Simulação de custos e benefícios de cenários alternativos. Maputo , Mozambique. Nicholson K., O'Laughlin B., Francisco A., Nhate V. (2003). Poverty and Social Impact Analysis: Fuel tax in Mozambique. Oxford Policy Management. Mimeo. March 2003. Obstfeld, M., Taylor, A.M., 1997. Non-Linear Aspects of Goods-Market Arbitrage and Adjustment: Heckscher's Commodity Points Revisited. Journal of the Japanese and International Economies, Vol. 11, No. 4. Per Pinstrup-Andersen (Editor), 1988. Food Subsidies in Developing Countries: Costs, Benefits, and Policy Options. Published for the International Food Policy Research Institute. The Johns Hopkins University Press, Baltimore and London. Pollard, P. S., Coughlin, C. C., 2004. Size Matters: Asymmetric Exchange Rate Pass-Through at the Industry Level. University of Nottingham Research Paper No. 2004/13. Rashid, S., 2002. Dynamics of agricultural wage and rice price in Bangladesh. MTID Discussion Paper 44. International Food Policy Research Institute (IFPRI). Washington DC. http://www.ifpri.org/divs/mtid/dp/papers/mssdp44.pdf Ravallion, M., 1990. Rural Welfare Effects of Food Price Changes under Induced Wage Responses: Theory and Evidence for Bangladesh. Oxford Economic Papers 42 (3): 574- 585. Ravallion, M., 1990. Welfare changes of food price changes under induced wage responses: theory and evidence for Bangladesh. Oxford Economic Papers, 42, 574­85. Ravallion, M., 2000. Prices, wages and poverty in rural India: What lessons do the time series data hold for policy? Food Policy, vol. 25, issue 3, pages 351-364. 46 Samson, M., 2002. The Social and Economic Impact of South Africa's Social Security System. Economic Policy Research Institute. Samson, M., Babson O., Haarmann C., Haarmann D., Khathi G., MacQuene K., van Niekerk I., 2002. Research Review on Social Security Reform and the Basic Income Grant for South Africa, M. EPRI Policy Report 31, Economic Policy Research Institute and International Labour Organisation. Tarp, F., Arndt, C., Jensen, H.T., Robinson, S. and Heltberg, R. "Facing the Development Challenge in Mozambique: A General Equilibrium Perspective." International Food Policy Research Institute, Research Report 126. 2002. Thurlow, J. "Options for Agricultural Growth and Poverty Reduction in Mozambique." Mimeo. International Food Policy Research Institute, Washington D.C. 2008. Tschirley, D. and Abdula, D. "Toward Improved Marketing and Trade Policies to Promote Household Food Security in Central and Southern Mozambique: 2007 Update", Ministry of Agriculture/Mozambique. RP62E. 2007. Tshirley D. and Santos, A.P., 1995. Who Eats Yellow Maize? Preliminary Results of a Survey of Consumer Maize Preferences in Maputo, Mozambique. No 53, International Development Working Papers from Department of Agricultural Economics, Michigan State University. Uaiene, R., 2008. "Determinants of Agricultural Technical Efficiency and Technology Adoption in Mozambique." PhD Dissertation. Purdue University, August 2008. United Nations 2008. Outline for a National Response to the Rising Food Prices in Mozambique. UN System in Mozambique. Draft Discussion Note. May 2008. Van Dunem, J. and Arndt, C. "Confronting the Issue of the Elasticity of Customs Evasion in Mozambique: An Empirical Study." Ninth Annual Conference on Global Economic Analysis. Addis Ababa, Ethiopia. June 2006. https://www.gtap.agecon.purdue.edu/events/Conferences/default.asp#Past. Weber, M.T., Staatz, J. M., Holtzman, J. S., Crawford, E. W., Bernsten, R.H., 1988. Informing Food Security Decisions in Africa: Empirical Analysis and Policy Dialogue. American Journal of Agricultural Economics 70: 1044-1052, December World Bank, 2008. World Development Report 2008: Agriculture for Development. The World Bank, Washington DC. World Bank, 2008b. Double Jeopardy: Responding to High Food and Fuel Prices. Background Paper for the G8 summit in Hokkaido-Toyako, Japan, July 2008. Zaman, H., 2008. The distributional implications of rising food prices: A partial literature review and a proposal for further work. Mimeo. The World Bank, Washington DC. 47