FOOD SECURITY Concepts, Basic Facts, and Measurement Issues June 26 to July 7, 2006 Dhaka, Bangladesh Rao 15: The Scope and Limits of Targeted Policies Learning: Trainees will learn to appreciate the basic and subtle issues that arise in targeting, both normative and measurement issues that must be faced up to in the choice between universalism and various degrees of targeting. Brief Contents • universalism versus targeting in social welfare approaches • type I (false negative /exclusion) & type II (false positive /inclusion) errors • administrative costs in identifying targeted beneficiaries • tradeoffs between type I error and administrative costs versus type II error • application to food subsidy, food for work and other food security policies Constraints on Policy • In standard economic theory, any desired distribution can be attained in a market economy provided only that “lump-sum redistributions of initial wealth” can be arranged. Given this assumption, states do not have to confront any budgetary constraint or any conflict over alternative instruments of policy. This is called a 1st best policy context. • In reality, policymakers lack lump-sum instruments, and so confront both budget constraints & painful choices among alternative policies. This is called a 2nd best policy context • In achieving welfare gains (poverty & FIS alleviation being prime examples), the most general forms of constrained policy choice is between pro-growth versus pro-poor growth strategies and between “universal” versus “targeted” policies. Growth versus Pro-Poor Growth • The “great tradeoff” thesis is that welfare state policies necessarily sacrifice growth to get immediate gains in welfare i.e., sacrifice future gains in welfare that only growth can bring about for the sake of current gains. • This ignores much evidence showing that faster economic growth does not always translate into more equitable growth e.g., faster growth may improve total food supply but with unequal entitlements distribution of growth gains, access to food may increase little or even decline. • Policies are not always formulated optimally under given structural economic constraints. Ideological bias and governance failures often underlie the fitful and fad-ridden approach to poverty and food insecurity. Universalism VS Targeting • `Universalism’ in policy means making a publicly provided benefit available to all sections of society e.g., a universal policy of subsidized food distribution makes food grains available at reasonable prices to anyone who wants to avail of that benefit. • By contrast, a targeted policy makes the publicly provided benefit available only to defined segments of society e.g., a targeted policy of subsidized food makes the food available only to those identified as being below a poverty line. • In practice, universalism does not mean everyone will actually avail of the benefit e.g., self-selection by the poor (and self-exclusion by the non-poor) in a universal foodfor-work program Universalism and Rights • The discourse of rights (e.g., human or citizenship rights) is necessarily universal. A right is intended to be available to all, without discrimination or `targeting’. • Thus, the idea of a fundamental right to freedom from hunger is itself meant to be universal. It does not follow, of course, that this right necessarily requires direct state interventions designed to make food available to all. In some contexts, targeted policies to enhance food entitlements of the hungry/FIS may suffice. • But in other contexts, targeted policies may fail to redress lack of access to food for many. Judging each context to determine what policies are required to achieve the goals becomes a crucial element of strategic policy choice. The Scope of Targeting • We may distinguish between two types of targeting: • Positive Targeting: aims to support the vulnerable, a group larger than the actually poor or food-deficient: here, the idea is to support those who cannot achieve assured access via the market. No explicit budgetary concern motivates the targeting. The motive rather is to protect the unprotected, so the budget is adjusted to achieve the goal. • Negative Targeting: is determined by a real or self-imposed budgetary constraint: here, the idea is to avoid subsidies & admin. expenditure for the benefit of those who are otherwise food secure while focusing outlays for the benefit of those who are vulnerable. Implicitly via the choice of target groups, the goal is adjusted to conform to the budget. The Scope of … (contd.) • Clearly, the choice between universalism and negative targeting is a choice of policy goals often disguised by political rhetoric as merely a choice of policy instruments. • Equally clearly, positive targeting is compatible with if not actually identical to universalism. So when we use the term targeting without qualification, negative targeting is intended. • With negative targeting, “fiscal prudence“ and "balanced budget” become the watchwords that constrain policies in general, food security policies in particular. The motive here is clearly budgetary in the sense that where the goal of food security conflicts with the budget, the budget wins. • With positive targeting (and universalism), any such conflict is resolved in favor of the goal. Targeted Elements, Groups, Channels • We have seen various elements in targeting: Targeted asset distribution e.g., land reform Public works and Food-for-work (FFW) Targeted food subsidies e.g. fair price shops Direct food transfers e.g. relief assistance • Typical targeted groups might be: subsistence farmers, small farmers, tenants, urban poor, poor urban dwellers, rural landless, rural & urban un- and under-employed • Specific vulnerable groups, e.g. displaced, disasteraffected, female headed HH, underweight children, etc. • Targeted policies raise entitlements via varied channels: increased AG income, HH food supplies from subsistence production, cash income, real income due to lower food prices, and through direct food transfers Information and Identification • Targeting requires information, identification, channeling and monitoring to fulfill the targets within the budget. These can be very costly and must be themselves covered by the budget • To identify target groups, relevant information must be available about the wider population, allowing it to be separated into the target and non-target groups. Alternatively, their behavioral responses must be known so programs can be designed to elicit self-targeting (by eligible beneficiaries) and self-exclusion (of the ineligible groups). • In turn, administrative arrangements must be made to channel the benefit (say, subsidized food) to the targeted and to monitor the inclusion only of the eligible/targeted and the exclusion of all the ineligible/non-targeted population Type I (exclusion) & Type II (inclusion) errors • Identification is said to incur a Type I or inclusion error (also termed false positive) when ineligible people are included • Identification incurs a Type II or exclusion error (also termed false negative) when eligible people are excluded • Apart from admin/budget costs of identification, inclusion errors entail financial costs (budgetary/admin resources) of including the ineligible and exclusion errors entail social costs (malnutrition/health impacts) of excluding the eligible. • But these costs are incommensurable. It is meaningless to compare either the error probabilities or the costs involved without attaching explicit social weights to them. Considerations in the Choice • To sum up, then, the choice between universalist versus targeted policies involves – Clarity of what the policy goal (benefits) is – Consideration of various implementation costs • Budgetary and administrative costs in regard to information and identification of target groups • Social weights of Type I and Type II errors of targeting • Social costs of the errors of targeting • Budgetary and administrative costs in regard to channeling and monitoring General Limits to Targeted Food Subsidies • Food subsidies can only benefit those population groups who have, although minimal, cash income. Subsistence farmers, jobless, old and sick persons, children, refugees, etc.) cannot be reached by food subsidies. Yet, these groups may be the most vulnerable to FIS. • Food subsidies are relatively ineffective in alleviating food deficiencies resulting from unequal intra-HH distribution. • Food subsidies are relatively ineffective in alleviating specific nutritional deficiencies. In all these cases, direct food transfers appear as the only feasible alternative of targeted food interventions. Specific Limits to Targeted Food Subsidies • Target groups are typically identified by means of HH income/expen. surveys (also used in identifying the poor). Targeting errors arise from having a HH classification based on a survey in one year, and then following it for many years, the reason being HH income fluctuations. • The use of a poverty line in identifying the food insecure creates its own problem since poor ≠ FIS. • Often, budgetary pressures lead to greater implicit weight for inclusion rather than exclusion errors. The upshot: a large part of the nutritionally vulnerable may be excluded. • Universalism avoids socially costly exclusion errors and can also serve to stabilize market prices via state procurement and distribution e.g., through fair price or ration shops. Leakages and Governance • A standard criticism of universalism is that it leads to black-marketing of subsidized grains, and leakages in transfer from procurement to distribution points. • But universalism also obviates the identification problems that targeting gives rise to. And targeting by itself does not avoid the aforementioned leakages. • Universalism also provides a powerful check on abuses since it is not only the voiceless poor and FIS who benefit but also the better-off i.e., it makes for better governance. • By providing the operational basis for making food a basic right and the food system a collectively shared good, universalism can raise the social priority given to food and foster food sovereignty. Direct Cash Transfers • Some argue that direct cash transfers are a more efficient instrument of subsidizing the poor, since non-cash subsidies for food, fertilizer & fuel involve massive leakages. • 1st where they have succeeded (e.g., Brazil’s bolsa familia), direct cash transfers have complemented rather than supplanted existing schemes of providing welfare services. • 2nd with cash transfer to be confined to the poor and FIS, the financial costs of identification and social costs of exclusion errors will be perpetuated, not eliminated. • 3rd cash transfers cannot ensure that they will be spent on socially preferred welfare-enhancing or `merit’ goods e.g., in patriarchal contexts, the money may well be spent on individual (male) priorities such as liquor, tobacco, etc. Cash Transfer with Changing Prices • Under targeted subsidies reliant on identification exercises for a number of years in common, exclusion errors arise due to fluctuating incomes over time. Cash transfers compound this problem in conditions of price instability. • This is because cash transfers are usually initiated along with deregulation of market prices (of food, fuel, fertilizer). When prices rise, the “real” income (or vector of goods) represented by the cash transfer may be rapidly eroded. • True, if the relevant prices trend down, the cash transferred will convey more than the intended "real" transfer. But even if the probabilities of the price trend are symmetrically distributed, the poor themselves can be rationally expected to prefer a real transfer to an equivalent cash transfer since Cash Transfer … (contd.) • the gain in low-price years cannot compensate for the loss in high-price years. Put differently, the system of real transfers (e.g., subsidized food distribution) includes an insurance value that the cash transfer system lacks. • Since food-fuel-fertiliz. prices are expected to rise over time, type I errors will, on average, accumulate over time and so the insurance value of real transfers will rise over time. So the growing preference for cash transfer among economists, donors and lending agencies adds a dynamic exclusion error to the (static) exclusion error of targeting. • Arguing for cash transfers is consistent with arguing for targeting. But in no way does it address the social costs they impose by replacing real transfers and universalism.