Consequences of a food security strategy for economic welfare, income distribution, and land degradation: the Philippine case

Consequences of a food security strategy for economic welfare, income distribution and land degradation: the Philippine case* Ian Coxhead** Department of Agricultural and Applied Economics University of Wisconsin-Madison 412 Taylor Hall, 427 Lorch Street Madison, WI 53706 USA. Tel. 608/262-6390 Fax 608/262-4376 Email: coxhead@facstaff.wisc.edu Abstract Like many developing countries, the Philippines pursues a food security strategy in which self- sufficiency and price stabilization feature prominently. In addition to their widely debated welfare effects, food policies based on price and trade restrictions may also accelerate land degradation by promoting expansion of relatively erosive grain crops. We explore the welfare and environmental implications of food policies first with a simple heuristic model, then with an applied general equilibrium model. Comparing market restrictions with technical progress as alternative food policy strategies, we find that the former increase land degradation and reduce welfare; moreover, anti-poverty and distributional benefits often claimed for such interventions may be illusory. JEL Classification codes: D58, Q18, Q24. Keywords: Food policy, technical progress, self-sufficiency, applied general equilibrium, Philippines * World Development, in press. ** Department of Agricultural and Applied Economics, University of Wisconsin-Madison. This research was supported by funding from US Agency for International Development through the SANREM CRSP. Without implicating them, I am grateful to Jerry Shively, seminar participants at the University of Wisconsin, the University of California-Davis and the University of the Philippines-Los Banos, and two anonymous referees for comments on earlier drafts of this paper.  1 1. INTRODUCTION Staple grains are intensely political goods in most poor countries. In food importers, a combination of import restrictions and consumer price ceilings is frequently pursued with the multiple objectives of attaining self-sufficiency in production, supporting producer incomes, and providing urban consumers with cheap food at stable prices (Anderson). Such programs inevitably involve tradeoffs between economic efficiency on one hand, and their potential redistributive and anti-poverty effects, as well as non-economic impacts such as self-sufficiency, on the other. However, there are other aspects of staple food policies that have not been explored, and which may be important in developing countries. Among these is environmental damage. Although environmental controls are now prominent in the agricultural policies of wealthy countries, they have yet to establish a firm presence in the minds of many developing country planners. In countries where reliance on the natural resource base remains a key feature of the lives of poor people, the value of agricultural development policies that could result in resource degradation must be very carefully assessed. In this paper we identify and quantify some environmental as well as economic implications of current Philippine agricultural policies directed at achieving self-sufficiency and price stability in rice and corn. These crops occupy 50 per cent of Philippine agricultural land. About half of this area is upland or rainfed land, much of which is of generally poor quality and, when used to grow annual crops, is highly erosion-prone due to slope, soil structure, weather, or some combination of the three. For this reason we extend the conventional analysis of the welfare and distributional effects of agricultural policies to include a quantitative assessment of agricultural land degradation. In spite of several decades of relatively rapid economic growth, agriculture in the Philippines retains considerable importance in factor incomes and trade, and basic foods constitute a large share of consumer spending. All Philippine governments in the postwar era have attempted to increase domestic production with the explicit goals of achieving national self-sufficiency and ensuring price stability for producers and consumers alike, as key components of food security and income redistribution strategies (Mangahas; Intal and Power; Pagulayan).1  2 Philippine grain yields are low by Asian standards, and with relatively low spending on agricultural infrastructure and technology, yields have not risen as rapidly as in comparable countries. Consequently, output growth has been due primarily to area expansion, promoted by trade and price policies. The chief agency of food policy has been the National Food Authority (NFA), which from 1950 until the early 1990s held a monopoly over Philippine cereal trade as well as engaging in domestic purchase, storage and release to promote production and to defend producer price floors and consumer price ceilings. Given the political significance of self- sufficiency, NFA grain imports are historically very tightly circumscribed,2 and in spite of liberalization in recent years, nominal protection rates (NPRs) remain high by comparison with other agricultural commodities. The NPR for rice averaged 8 per cent from 1960-86. That for corn has generally been much higher, especially after the mid-1970s when self-sufficiency was made a goal: the NPR averaged 18% in 1970-74, and 42% in 1983-86 (Intal and Power); in 1990 it reached 63% (Pagulayan), and has remained at or above this level since (for comparison, all other crops except sugar have negative average NPRs, while the average manufacturing NPR exceeded 30% until recently). While the rhetoric of agricultural policy also prioritizes technological progress and infrastructure development, the net subsidy received by the NFA for its grain price and marketing programs exceeds the agricultural R&D budget by a factor of 50 (Manasan). Expansion of the permanent agricultural area takes place mainly at the frontier, in upland agronomic zones. It occurs through replacement of perennial and long-fallow agricultural systems (including forests) by short-season crops, among which corn predominates.3 In seasonal cropping the soil is tilled more frequently and exposed for longer periods to the erosive effects of rain and wind, so this land use change has been associated with rapid increases in upland land degradation and soil erosion. Corn in particular is associated with high rates of soil nutrient uptake and erosion in tropical soils, and among upland crops in the Philippines is the chief contributor to land degradation by virtue both of its high relative erosivity and its large share of upland cropped area (David). Agricultural policies that support corn prices thus contribute to increased land degradation.4 Technical progress could inhibit this by raising yields on existing croplands, thus  3 allowing food self-sufficiency targets to be achieved at lower levels of price support. Lower protection for corn producers would in turn diminish incentives to continue expanding corn production in uplands. In other words, investments in technical progress might help the Philippines achieve its food policy goals at a lower environmental cost than the equivalent amount spent in defense of producer prices. Our results in this study suggest that endogenous changes in the rate of land degradation caused by Philippine cereal policies aimed at self-sufficiency and price stability are large in relation to national income, agricultural factor markets, and the budgets of government agencies charged with promoting soil conservation. These environmental costs should be included in assessments of the benefits and costs of agricultural policy. METHODOLOGICAL APPROACHES Natural resource accounting (NRA) studies have quantified losses from agricultural land degradation in several developing countries (Repetto et al.; Barbier and Bishop). However, NRA generates only circumstantial evidence on the causes of land degradation, and thus has limited capacity to inform policy-making. Other methods are required to identify economic causation and thus to indicate specific solutions. In developing countries, where agriculture is frequently large in relation to income, trade and domestic markets, a general equilibrium approach to the measurement of losses from agricultural development is appropriate.5 However, most policy-environment analyses for developing countries are conducted at a high level of generality; in spite of much policy interest there are still few general equilibrium models providing rigorous empirical analyses of specific policies and particular environmental phenomena within such economies.6 The contribution of this paper is first to provide comparative static analysis of a set of food sector policies common in some form to many developing countries, and second to conduct an empirical general equilibrium examination of the environmental, welfare and distributional implications of specific policies in the Philippines. The empirical core is a set of experiments using an applied general equilibrium (AGE)  4 model of the Philippine economy, in which we evaluate the effects of policy changes and technical progress affecting cereal crop sectors under alternative trade and agricultural pricing policy regimes. In order to understand the results from this large and complex model, we first provide a highly simplified sketch of its microeconomic foundations. This exercise provides valuable help in identifying environmental and welfare changes associated with policy changes or technical progress in the presence of trade and price interventions. Subsequently, we quantify these analytical insights by means of experiments with the AGE model. In a final section we discuss the value of predicted economic and environmental changes and evaluate some policy implications. (a) Heuristic general equilibrium model In this section we develop a simple model with which to explore interactions between agricultural price and trade policy, land degradation, and economic welfare. To focus on land degradation, we abstract from other forms of environmental damage--not only non-agricultural pollution, but also the off-site effects of chemical use and soil erosion in upland agriculture. Suppose that upland land is used to produce two crops, and that one, D, is relatively land degrading (in the sense of drawing nutrients from the soil and contributing to erosion and soil structure degradation) while the other, C, is not. Suppose also that D, the land-degrading crop, is a cereal such as rice or corn. Other things equal, reallocating land from C to D production increases land degradation, and in this way land quality is endogenously related to land use. Our assessment of the net social benefits of grain sector policies that induce land use change will then depend on whether we assign a positive or zero value to the land quality change. In the analyses that follow we compare pairs of welfare outcomes following policy or technology shocks. Each pair consists of an outcome in which land degradation costs are ignored, and another in which they are given positive values. In general, ignoring environmental costs gives results in which a particular shock yields an unambiguously positive or negative welfare change. Including environmental costs, however, introduces ambiguity to these results. This ambiguity is the focus of our interest since it draws attention to specific structural features of the  5 economy, policy regimes and technologies as determinants of the final welfare outcome. In the analysis that follows we first define the effective land endowment and develop a measure of welfare change inclusive of the value of changes in this endowment. We then examine the welfare and land degradation effects of agricultural policies and technical progress under alternative assumptions about cereal sector policies, particularly price and trade restrictions that correspond broadly to the Philippine case. (i) The basic model Let T stand for the effective (i.e. quality-adjusted) endowment of upland land available for agricultural production. We define this as the product of the physical endowment of land, T*, and an augmentation parameter, A, representing the effects of technical progress. Technical progress is sector-specific, so in our two-sector representation of the upland economy T = AcTc + AdTd . * * However, production in each sector is also associated with some rate of land degradation, which we denote by j 0. When this is counted, the definition of the effective land endowment adjusted for environmental effects becomes: (1) T = (1 -c)AcTc +(1-d )AdTd * * As indicated above, we distinguish two cases conditional on values of as follows. Free disposal attaches no value to land degradation, so in this case we assign j = 0 for all crops j. In uplands of developing countries, open-access is generally a good approximation of prevailing property rights. This market failure as well as the inherent difficulty of measuring soil quality makes it reasonable to assume that the on-site effects of land degradation may not be fully capitalized into land values.7 The alternative case is weak disposal, in which at least one j is strictly positive. In welfare analysis, accounting for land degradation losses in this way amounts to applying NRA methods for upland land degradation. Under weak disposal, technical progress or price changes that create incentives to increase the use of uplands for relatively land-degrading crops may reduce aggregate welfare in spite of being privately profitable. In order to identify welfare effects, we now locate this agricultural land degradation story in its general equilibrium context.  6 In developing the model we assume competitive markets and constant returns to scale and make use of the following notation and definitions. Aggregate expenditure (by a representative consumer) is denoted by the expenditure function e(p,u) = min{p c | u}, and aggregate income by the revenue function g(p,v) = max{p y | v}, where p, c, and y denote vectors of prices, consumption and production respectively, and u stands for utility. The vector v is the economy's factor endowment, of which the effective land endowment T is one element. Partial derivatives of the expenditure and revenue functions with respect to the j'th price give commodity demands by consumers and supplies by firms. We denote these by ei = e(p,u) pi > 0 and yi =g(p,v) pi > 0 respectively, and note that their difference (ei - yi) is net trade (positive if the good is a net import; negative if it is a net export). The derivative of the revenue function with respect to each factor endowment gives the shadow price of that factor; thus for upland land with price r, r = g(p,v) T . In a competitive economy the optimal factor demands of cost-minimizing producers are exactly those consistent with the vectors of optimal commodity supplies and factor prices obtained from revenue maximization (Dixit and Norman). Thus, using w to represent the vector of factor prices and yicj(w) for the unit cost function in sector j, we can write the demand for land in each sector, Tj, in terms of exogenous price and endowment changes as: cj gj(p,v) cj (2) Tj = yj j=C,D. r = pj r(p,v) This expression demonstrates the general equilibrium nature of the land allocation problem by showing land demand to be a function of economy-wide prices and factor endowments. At the level of the whole economy, firm-level land degradation externalities are internalized in the form of reductions in the effective land endowment. In the absence of price policy or other interventions, aggregate expenditure in each period must be equal to aggregate income minus the current value of land degradation. This implies an economy-wide budget constraint of the form: cj (3) e(p,u) = g(p,v) - jyj . j r We can see from (3) that under weak disposal (i.e., some j > 0), an increase in the price of one  7 agricultural good, with other prices held constant, will influence aggregate welfare through two channels: directly, via changes in production, consumption and net trade, and indirectly, through changes in the natural resource base. As an example, to see the welfare effect of an increase in the price of D, take the total differential of (3) with respect to pd, holding other commodity and factor prices constant,8 to obtain: u yj Tj rj * (4) eu * j = D,C, pd = -(ed - yd)- +yj j= C,DjTj pd r pd Since eu is the inverse of the marginal utility of income, the left-hand side is a money metric of welfare change with respect to pd, holding other prices constant. Under free disposal (all j=0), a rise in pd is just a terms-of-trade shock: welfare will improve if D is a net export, and worsen if it is a net import. Under weak disposal, however, the value of land degradation changes could modify this result. The sign of the term within braces in (4) depends on each upland sector's supply response, on land use responses to factor price changes, and on the effect of the commodity price change on land returns. The second term within the braces can be signed only once we know which upland sector is relatively land-intensive.9 Under weak disposal with d >c, the rise in pd will unambiguously raise welfare only if D is a net export and the sum of the terms within braces is less than ed - yd. The latter is the more likely if the D sector is less price-responsive, and if the technology of production of D is less land-intensive than C (since then r pd < 0). Conversely, welfare will unambiguously decline if D is a net import, is relatively land-degrading, and is more land-intensive than C. Between these polar cases there is clearly a range of intermediate results in which under weak disposal the net welfare impact cannot be predicted without first assigning values to the parameters of land degradation, land-intensity of production, and supply response. The analysis of technical progress runs parallel to that of a price change. For the purpose of developing the model we restrict our attention to the case of factor-neutral technical change (the model is capable of more general forms). Technical progress in sector j can thus be modeled by defining the effective producer price vector p, where the productivity augmentation parameters j take initial values of 1 for all j, and technical progress is represented by d > 0 . Redefining the j  8 revenue function as g(p,v) , and evaluating the total differential of (3) with respect to d gives: u g yj * (5) eu * Tj rj , d = d - Tjj +yj j=C,D d r d which can be rewritten in terms of price change parameters as:10 u * (6) eu * yj Tj rj . d = pdyd - pd + yj j=C,Dj Tj djyj + pj pd r pd wheredj = 1 for j=d, and 0 otherwise. Under free disposal the marginal welfare gain from technical change is the value of an additional unit of D sector output, pdyd. With weak disposal, this gain may be increased or reduced by the valued land degradation effects of changes in sectoral outputs and factor intensities-- the first and second terms respectively within braces in (6).11 Again, the analysis draws our attention to the need to obtain more information on agricultural technology (land-intensity) and supply response if welfare predictions are to be made. Compared with free disposal, in our examples both technical progress and a price increase have additional first-order welfare effects under weak disposal because they induce land use shifts to the more erosive crop. Whether a particular shock causes welfare to increase or decline depends on whether its environmental effects augment or offset its `conventional' economic effects. (ii) Trade restrictions and government purchases Food security policies typically involve interventions in international trade and domestic markets, aiming for self-sufficiency and price stability. If a domestic supply shock occurs in cereals and restrictions on international trade are binding, then either cereals' prices must adjust to clear the domestic market, or the market may clear at a constant price through a quantity mechanism such as government purchases and releases from buffer stocks. We can accommodate either possibility by equating net government grain purchases, DG , to the excess of domestic supply over demand: (7) DG = gd(p,v) - ed (p,u) . In the absence of international trade, when DG is fixed the domestic grain price adjusts to clear the market just as for a pure non-traded good. Alternatively, the government can defend a fixed price  9 by allowing DG to adjust to balance changes in supply with consumer demand. Suppose that the government maintains budget balance by means of a lump-sum tax on households; when DG > 0, aggregate household consumption is reduced by the amount of tax revenue required to fund the grain purchase. We capture this by restating the aggregate income-expenditure constraint (3) as: cj (8) e(p,u) = g(p,v) - pdDG - jyj . j r(p,v) Simultaneous solution of (7) and (8) yields the equilibrium value of aggregate real income and either (a) the domestic grain price pd, with DG fixed, or (b) the level of government purchases DG, with pd fixed.12 These are polar cases of a typical food policy based on both trade restrictions and domestic purchases. Closure (a) is more appropriate when government policy targets a predetermined grain buffer stock; closure (b) more closely resembles the use of food price stabilization as a tool of agricultural growth, anti-poverty and redistributive programs. For brevity, and anticipating the Philippine case study in the next section, we will examine only case (b), setting grain prices exogenous and examining the effects of price and technology changes. (At the end of this section we discuss other closures, including the possibility that some trade occurs but that imports and domestically produced goods are imperfect substitutes). Substituting for DG in (8) from (7) and taking the total differential with respect to pd yields an expression for welfare change analogous to the no-intervention case shown in (4): u yj * (9) * Tj rj , pd = pdZdd - + yj j= C,DjTj pd r pd where = eu - pdedu > 0 in stable models and Zdd = edd - gdd < 0 is the price response of excess demand for D. This expression shows that under free disposal (=0) the welfare cost of raising the grain price is positive when the quantity of grain purchased by government is endogenous. Under weak disposal the value of this loss is increased by the value of environmental damages associated with the movement of land into grain production. The effect of technical progress with a trade ban and domestic price controls can be seen by again substituting from (7) into (8) and taking the total differential, this time with respect to d:  10 u yd yj * (10) 2 * Tj rj , d =- pd pd - + yj j=C,DjTj pd r pd This result may be surprising at first: even under free disposal, technical progress generates an unambiguous welfare loss when government purchases all additional output from sector D. Relative to the initial equilibrium, technical progress in the protected sector simply induces overproduction; part of the D sector supply increase is obtained at the expense of production in other sectors. To preserve budget balance, consumers must pay for government purchases, which they do through lump-sum taxation. At constant prices the value of the aggregate real income gain due to technical progress is outweighed by the cost of financing the grain purchase program.13 Since technical progress also causes more land to be used for the land-degrading crop, relaxing the free disposal assumption merely worsens this prediction. The foregoing analysis serves two purposes. First, it posits a direct linkage between food policies and the environment and permits us to explore the welfare implications of this linkage. Under free disposal, policies that increase grain production and absorb the resulting excess supply through market interventions will impose costs in terms of aggregate real income or in terms of foregone opportunities for growth due to technical progress. Under weak disposal--inclusive of environmental damages--the model shows that food policies can have uncertain welfare results. Second, the analysis then helps us understand the sources of ambiguity in welfare outcomes by identifying characteristics of the economy--agricultural technology, factor mobility, and agricultural supply responsiveness--that condition the welfare impact of a price or technology shock. If a shock causes the grain sector land use to expand, then welfare is more likely to be reduced if that sector makes a relatively large contribution to land degradation, either by virtue of its share of overall land use or the technologies it employs. These effects occur in addition to any other welfare effects associated with price-setting or trade interventions. (b) Analytical versus numerical approaches A heuristic model helps us to think formally and rigorously about land degradation. However, it has clear limitations when we turn to empirical questions. As we have seen, the relative  11 magnitudes and even the signs of the changes in (9) and (10) cannot be identified without additional economic and agronomic information. Moreover, the analysis has thus far been conducted at the simplest possible level in terms of sectors, inputs and technology and with a single consumer; and we have excluded many complicating factors, including policies affecting other sectors. By comparison, a numerical or applied general equilibrium (AGE) approach is less transparent but affords more complexity and thus much greater realism. The AGE approach has the advantage that we can assign magnitudes as well as signs to predicted changes, thus complementing the predictions of the heuristic model with empirical results. One important gain, among many, from the AGE approach is the option of relaxing "polar" assumptions about market structure and price formation. Equations (6) and (10) showed extreme cases of free trade and no trade in grains respectively. However, they shared the property of a fixed grain price-- in the first instance because it is set in world markets and in the second because it is defended by government purchases. The AGE model that follows employs intermediate cases in which producer grain prices are somewhat responsive to market changes. There are several key differences. First, imported and domestically produced cereals are imperfect substitutes, so that even without quantitative trade restrictions a change in the domestic market can alter the grain price. Under this assumption an increase in domestic output, due for example to technical progress, may be reflected in part in a decline in the producer price, thus dampening the welfare and environmental effects shown in (6). Second, the AGE model captures intermediate purchases. Since farmers sell their grain not to consumers but to millers and traders, this is important when price policies effectively target consumer prices rather than prices at the farm gate--as in the Philippine case. Third, the AGE model captures intersectoral linkages operating through factor markets, so that agricultural prices and profitability are subject to many other influences, including policy interventions in non-cereal sectors. These linkages increase the possibility that upland agricultural growth could be welfare enhancing even when land degradation increases. Finally, by increasing the number of factors of production, disaggregating households, and constructing household-specific consumer price indices based on observed expenditure patterns, an  12 AGE model can capture the real distributional implications of price and technology changes. This is important in an empirical assessment of food policies, since reductions in poverty and inequality are major justifications offered for their adoption. THE APEX MODEL OF THE PHILIPPINE ECONOMY (a) An outline of the APEX model We now use an AGE model to examine the economic and environmental implications of technical progress and trade and price interventions in Philippine grain markets, simulating their impacts on factor and product markets, household incomes, government revenues and trade. Our vehicle is APEX, a large, empirically based AGE model of the Philippine economy designed for policy analysis. The following is a brief overview of the model.14 APEX was designed for the analysis of technical progress in agriculture, economic policy and income distribution in the Philippines. It is a comparative-static model in the so-called Johansen tradition, meaning that the model is linear in the proportional changes of its variables from their values in a base year. Several features of APEX set it apart from similar developing- country models. Most notably, all elasticities of production, consumption and trade are econometrically estimated from Philippine time series data using flexible functional forms, a feature which greatly increases the realism of its simulation results. The model is highly disaggregated. It has fifty sectors, of which seven produce agricultural goods. Some of the seven jointly produce several goods; notably, rainfed rice, corn and root crops are jointly produced in the sector called "rainfed crops" (irrigated rice is a separate sector). There are eleven agricultural goods in all; each is produced in three regions corresponding to the main island groupings (Luzon, Visayas, and Mindanao). Agricultural production uses land, labor, fertilizer and other intermediate goods. Primary factors and fertilizer are aggregated into a single primary input in each region with econometrically estimated parameters; this composite input is combined with intermediate inputs in fixed proportions for a given technology (agricultural parameter estimates are reported in Warr (1995)). Given input prices, a representative producer in  13 each region is modeled as buying a production possibility set, upon which the commodity composition of aggregate output is decided by revenue maximization with respect to product prices and technology. Figure 1 summarizes the structure of the agricultural economy. APEX also contains very detailed descriptions of production, input use, and trade in non- agricultural sectors. These sectors employ skilled labor in addition to the primary factors already named; the composite labor input is a constant elasticity of substitution (CES) aggregation of skilled and unskilled labor. Private absorption and factor ownership is modeled for five representative households, each having a unique pattern of factor ownership and consumption expenditures. These are based on Philippine data on the quintile distribution of income and expenditures. Input-output and trade data are obtained from the 1989 Philippine social accounting matrix. Savings and investment, and government revenues and expenditures are also modeled in detail from national accounts data. For importable goods, Armington elasticities of substitution between imports and domestically produced goods are estimated separately for each commodity. For this study we extend the model just described to include measures of agricultural land degradation. We do this by accounting for changes in soil quality associated with changes in upland land use and technology. Our measure of a change in land degradation is a change in the amount of land allocated to more erosive uses, principally corn and upland rice. We assume homogeneous soil conditions and technology, and use the nutrient replacement cost method (see below) to construct estimates of on-site costs associated with upland land use changes. (b) Closure: the policy setting The macroeconomic closure of a general equilibrium model reflects assumptions about economic structure. Alternate closures can thus be used to conduct what might be called 'structural sensitivity analysis', i.e. to examine the robustness of simulation results with respect to the specification of macroeconomic relationships. We can use this to identify the contribution of food policy interventions to observed outcomes. In the following experiments we use two closures: one with and one without grain market interventions of the type discussed earlier. In the first closure,  14 domestic grain prices are endogenous and international trade in rice and corn is unrestricted (although the government levies an import tariff).15 Grain markets thus clear entirely through endogenous adjustments in trade and domestic prices, as in (3) - (6) above. We label this standard representation of market structure the unrestricted closure. In the second closure we capture the effects of interventions by fixing cereal imports exogenously and making government purchases of the output of the rice and corn milling sector endogenous, as in (8) - (10) above. We label this the NFA closure. Without trade, in this closure the government buys the entire excess of domestic cereal supply over demand, with the intent of fixing nominal consumer cereal prices. This in turn supports producer prices of rough rice and corn, since these are strongly influenced by demand from the grain-milling sector from which government is assumed to make its purchases. Both closures share some other characteristics. World prices of imports and exports are exogenously fixed (the small country assumption) as is the nominal exchange rate, providing a numeraire for domestic prices. The current account, budget deficit and real savings of households are also fixed, so the effects of shocks are fully absorbed by current-period changes in real household expenditures. Budget balance is maintained by endogenous adjustments in a lump-sum tax on households.16 In all experiments we hold the physical agricultural land area constant.17 EMPIRICAL EXPERIMENTS In order to compare alternative paths towards food security, we now present the results of three simulation experiments. In the first we adopt the NFA closure and assess the effects of a ten per cent rise in cereal support prices. In the second and third experiments we evaluate the effects of productivity growth in corn, the major crop in environmentally marginal upland and rainfed areas. Experiment 2 does this in a no-intervention setting (the unrestricted closure), while in experiment 3 we adopt the NFA closure, re-evaluating the effects of technical progress in the presence of the price-supporting interventions. In all experiments we maintain the assumption that technical progress and land management practices change exogenously, but that changes in land allocation to  15 different crops are responsive to changes in relative commodity and factor prices. The main results are summarized in Figures 2