Finance & Development, December 2018, Vol. 55, No. 4 PDF version

Targeting the Poor

Developing economies face special challenges in delivering social protection

Rema Hanna, Adnan Khan, and Benjamin Olken

Targeting the Poor

Many people believe that social protection generally involves rich countries aiding those that are poor. Aid is important, particularly for extremely poor countries. Bad shocks can quickly devolve into humanitarian disasters and promote conflict in fragile states, as seen with the current famine in South Sudan, the incipient famine and cholera in Yemen, and the recent Ebola outbreak in Guinea, Liberia, and Sierra Leone.

But for the 108 countries the World Bank classifies as "upper-middle" or "lower-middle" income—for example, India, Morocco, and Peru—overall tax revenue now dwarfs development assistance. Given that growth has been accompanied by increases in global inequality, it is not surprising that redistribution is increasingly taking place within countries. In these environments external support is often important during initial program design and launch, but social protection can be funded primarily through domestic sources in the long run.

As many of these countries increasingly initiate redistribution within their own borders, they are facing challenges different from those in high-income countries. Understanding these differences is critical to grasping how social protection has evolved over time, and how it may change in the future.

In-kind subsidies

The traditional way of providing transfers in many countries, and one that is still common today, is to subsidize particular products. Staple foods are one classic example. Energy is another.

The rationale for this policy tool is simple. In developed economies, the government can use income information from tax documents and other sources to identify who needs assistance. In less developed settings, however, substantial activity occurs in the informal sector, particularly among the poor. There is no paper trail of who is employed and how much they earn—at least none that can be easily verified. Instead, governments aim to subsidize products the poor use disproportionately so that they get a larger portion of the subsidy.

Subsidies tend to be politically popular for several reasons. The first is transparency; for example, with an energy subsidy, consumers see the subsidized price at the pump. A second is that since everyone benefits from subsidies, they may enjoy broader political support than programs that benefit only the poor. Finally, governments can claim that they influence what people consume; for example, subsidizing eggs or milk to ensure that kids get enough protein, rather than a cash subsidy the poor might waste on so-called temptation goods such as alcohol or tobacco.

These arguments have not always been borne out in practice. For the poor to receive more of the subsidy calls for subsidization of what economists call "inferior goods," goods whose demand decreases as people get richer—cassava when everyone prefers rice, low-quality food, and so forth. Subsidizing inferior goods is often unpopular; instead, most subsidies end up being for everyday goods—things people buy more of as they have more income. This undermines redistribution, as these programs end up benefiting mostly the middle class, or even the rich.

Picking up the tab for more popular goods also makes subsidies expensive. Energy subsidies are a classic example. Since everyone can access the subsidy, it must be quite large in order to ensure that the poor get a reasonable share, and in most cases many of the benefits accrue to the middle class, not the poor. In fact, spending is already so high for these subsidies that the amount governments would save by eliminating energy subsidies in emerging market and developing economies exceeds what many of them spend on public health.

Subsidies also have distortionary consequences. Energy subsidies, for example, have serious environmental ramifications. Subsidizing particular foods, such as rice, can backfire in terms of achieving balanced nutrition. And the idea that the poor waste cash by spending it largely on so-called temptation goods has been debunked many times, undermining much of the rationale for trying to influence what people spend the money on in the first place.

Given the cost and distortionary consequences of blanket subsidies, many countries fix the quantity of subsidized goods each household can receive, and potentially restrict them to poor households. This, however, introduces a host of other challenges, such as building a bureaucracy to distribute the goods and monitor who receives them and how much. These systems are generally subject to significant corruption and leakage, given the challenges in managing them.

Reaching the right households

For these reasons, developing economies are moving away from universal (or limited) in-kind subsidies and transfers to the poor and toward targeted cash transfers of various types. Cash is neutral and thus does not distort what people purchase. Moreover, evidence suggests that the labor-supply consequences of cash transfers are small—that is, these programs do not seem to discourage work, often a concern when it comes to transfer programs. Finally, as an additional benefit, cash transfers can be used as a fiscal stimulus to smooth negative macroeconomic shocks by distributing money directly to poor households that have a high marginal propensity to consume.

Targeted cash transfer programs are already widespread in developed economies. For example, the United States provides poor working families with cash through the Earned Income Tax Credit. The US government can do this because the tax system makes it possible to verify which households are poor, and the formal banking system has effective mechanisms to ensure that the poor receive the transfers.

In contrast, many developing economies face challenges in both targeting and distribution. The informality of the workforce means that few people are captured by the tax system, so it is hard to verify income. And many households remain unbanked, so transferring funds to them is logistically difficult.

These challenges are daunting, but developing economies are finding ways to overcome them.

First, there are alternate methods to target transfers to the poor. A common approach is the "proxy means test." Governments predict income using data from periodic, large-scale quasi censuses of the population that collect information on easy-to-verify assets (such as the material of a household’s roof and floor) and demographics. Households with a predicted income below a given threshold receive benefits for a set period of time (for example, until the next census is conducted).

These methods can be quite effective at achieving redistribution. For example, in recent work, we compared what would happen if transfers went to those deemed eligible based on the proxy means test compared with what would happen if the same budget were simply divided equally among everyone (via a universal cash transfer, also known as universal basic income) for two programs in Indonesia and Peru.

Even though targeting through proxy means testing has its flaws, we find that it yields higher overall welfare than the universal transfer, because means testing concentrates benefits among the poor. Put another way, per-beneficiary transfers would have to be much smaller for universal programs than for targeted transfers, usually because of overall budget constraints and competing priorities for government spending (for example, infrastructure development, education, and so forth). In fact, only in the case of significant targeting error would the universal transfer come out ahead.

And although the administrative costs of proxy means testing may seem high given all the data collection, the survey costs are minuscule compared with the amount they save by reducing the number of rich people who receive transfers.

Of course, there are always exceptions. When income inequality is extremely high—there is a large pool of very poor people and a small number of very rich people—a universal transfer may look more attractive. Likewise, if poverty is temporary, proxy means testing, which targets the long-term average of past income, may allow the poor to fall through the cracks. Moreover, errors in targeting that do occur with proxy means tests also generate inequality among equally poor people, and limits on subsidies to the middle class can generate political strife.

But additional methods—from including communities in the process to on-demand application—have been shown to improve targeting along various dimensions, including program satisfaction and more flexibility in the timing of identification. For example, if communities verify the final list of beneficiaries, excluded households can be added. Instead of visiting all households for the proxy means survey, households that believe themselves eligible could enroll at an office; only those close to the eligibility threshold would need home verification. Supplementing a proxy means test with these additional approaches could help resolve some of the challenges of targeting.

How can the cash be distributed? Banking systems and technology are changing rapidly in many developing economies, reducing leakage in cash distribution. Recent studies show that modern banking technology—
biometric smart cards—can help dramatically reduce corruption in cash transfer programs. As mobile money continues to gain prominence, it will likely play an increasing role in delivering transfers.

Increasing capacity

How can developing economies further improve their capacity to deliver social protection?

Strengthen the tax system: Compared with developed economies, lower-income countries collect a smaller fraction of GDP and subsequently face greater budget constraints. But, without sufficient government funds, redistribution is impossible. As more people are covered by the relevant tax authority, taxation can become more progressive.

Develop clear identification systems: In an example we studied, one of the biggest challenges to enrollment within a government health insurance program was a lack of a functioning identification system to track people. A clear and well-functioning identification system is an essential foundation for social protection systems.

Invest in effective targeting: Identifying beneficiaries for targeted transfer programs calls for up-front investment to develop targeting that minimizes both rent seeking and the degree to which poor households are excluded. The sticker price for these systems may seem high, but it pales in comparison with the transfers being paid. Spending 1 or 2 percent of the cost of the transfers on better targeting can make a huge difference in the degree to which the programs improve welfare for the poor.

Improve access to banking: With new banking technology, especially mobile banking, there is an opportunity to improve the distribution of cash transfers. But it will take careful planning—from the types of technology that are feasible to ways to help the poor access and understand this technology.

Recognize that a variety of approaches may be needed: Although the focus of this article is mostly anti-poverty programs, insurance programs—for example, pensions and unemployment benefits—are also an important component of redistributive systems, helping households minimize risk that could distort their behavior.

REMA HANNA is the Jeffrey Cheah Professor of South East Asia Studies at Harvard University’s Kennedy School of Government, ADNAN KHAN is research and policy director of the International Growth Centre at the London School of Economics and Political Science, and BENJAMIN OLKEN is professor of economics at the Massachusetts Institute of Technology.

PHOTO: GETTY IMAGES/SIMONLONG
Opinions expressed in articles and other materials are those of the authors; they do not necessarily reflect IMF policy.