(this is a variant of the principle called “Ramsay pricing” in public finance; see Stiglitz, 1986: 403-406.)
The full impact of a reduction in subsidies in the public sector depends in part on the response of private-sector suppliers and potential suppliers. If private-sector competitors also raise prices when the public sector does, then the overall effect on use of contraceptives of price increases in both sectors may be greater than the impact that would be measured with data only on price and consumption of the publicly provided contraceptives. Molyneaux and Gertler, in simulations based on their models of Indonesian data, showed that the private-sector response to price changes in the public sector could be large.
In Taiwan in the early 1970s, a quasi-experiment was designed to compare distribution of pills free of charge, for a low fee, and for a high fee. The decision was made in favor of the high fee, according to Jeff Tsai, in part because people did not value the service that came too cheaply. Akin's analyses of price effects in Jamaica, Thailand, and the Philippines suggests that there is a point at which contraceptives (condoms, IUDs, pills) become “too cheap,” approximately US$1.35 per year. Above that level, price and consumption are negatively associated, as in normal demand curves; below that level, lower prices are associated with lower, not higher, demand for the products.
Akin summarized the results of his and others' work on price effects: money prices, time prices, and the quality of services all do affect consumption of contraceptives, even if the effects are typically small. This is a nontrivial result, because 10 or 20 years ago many argued that such costs of contraception were inconsequential in the larger balance of couples' decisions on whether or not to have a child.
THE DISTRIBUTION OF BENEFITS OF SUBSIDIES
For most countries, increasing the overall contraceptive prevalence rate is at most an intermediate goal, and reducing population growth rates is only one ultimate goal, not always dominant. Family planning is considered a public good because it promotes the health of women and their children (Committee on Population, 1989) and allows families to control the number and timing of births so that they can invest in the education and socialization of their children. These social arguments pertain most strongly to the poor: it is the poorest women and their offspring whose health is threatened most by large families, and it is they who generally have the least safe and effective alternatives for fertility regulation.
The choice among alternative strategies for reducing overall subsidies (targeting services versus across-the-board user fees) could depend in large part on estimates of their differential impact on poor clients and potential clients. Introduction of, or increases in, user fees is one obvious way to reduce the levels of subsidy. Sliding scales for clients of different income levels are more difficult to
administer, but they may be more acceptable on equity grounds (and more efficient in terms of other social goals, since they allow the program to act like a discriminating monopolist, charging each group of consumers the price it is likely to bear).
Akin presented the results of an analysis of the incidence of the public-sector subsidy across quartiles of the household asset distribution in Cebu, Philippines, separately by contraceptive method. “The total subsidy pattern for this sample of women from a relatively low income Philippine province is certainly provocative,” Akin argued. About 64 percent of the total subsidy accrued to the richest half of the households, and only 17 percent of the total accrued to the poorest quarter of households. Molyneaux and Gertler found almost exactly the same proportion in Indonesia, with 16 percent of the total value of subsidies accruing to the poorest quarter of households. As Table 2 shows, the degree of “income group targeting” in the Cebu study varied considerably by contraceptive method; only 7 percent of the condom subsidy was received by the poorest one-quarter of households, whereas more than half the public-sector subsidy for vasectomies was received by the poorest one-quarter.
Cross-sectional results from one time period may also exaggerate the true long-term differentials in the incidence of subsidies between rich and poor. If diffusion models of contraceptive change are correct, then educated people are typically the early adopters of behavioral innovations like contraception, and the poor catch up, often within a decade or two (see for empirical examples, the
TABLE 2 Shares of Public-Sector Contraceptive Subsidy Received by Quartiles of Households by Asset Ownership, by Method, Cebu, Philippines, 1983-1986
analysis of national survey data from Latin American countries by Rodriguez and Aravena, 1991). The rich would appear to capture a larger proportion of the public subsidy in the early stages of the process than in the later stages.
As Mark Montgomery reminded the participants, the existence of subsidy “leakage” does not by itself mean that more precise targeting to the poor is needed. It will be true of any product or service that is a normal good (in economists' terms) that the rich consume more of it than the poor. If one looks at other subsidized public services, such as hospitals and higher education, in developing countries, the proportion of the subsidies going to upper-income groups is typically far greater than is seen with family planning. More precise targeting of benefits to the poor, for example by means testing, is certainly possible. Whether it would make programs more efficient in meeting their goals depends on how much it would cost to implement such targeting and whether the gains in revenues or lower expenditures outweigh the losses from society's standpoint.
Dean Jamison pointed out that it is not enough to look only at the differences among poor and rich in the consumption of contraceptives to estimate the incidence of the benefit from public subsidies. Citing the example of taxes on tobacco and alcohol, he argued that the behavior of the poor may be more affected by the price mechanism than is the behavior of the rich. The rich who use publicly subsidized contraceptives may well have used contraceptives from other sources in the absence of public subsidies; for the poor this may not be so. The net, as opposed to the gross, impact of the public program in this case could be considered as reaching entirely the poor; the leakage of benefits to the rich is simply a transfer that may be part of the price of reaching the poor who need the program.
Gertler pointed out that surveys producing data intended for use in models of choice of contraceptives and service providers need to devote a great deal of effort to collecting information on what exactly is the “choice set” available to different households. In the data from Indonesia analyzed by Molyneaux and Gertler, there were great differences across households in the availability of methods and providers. For most methods, fewer than half the individuals with some access to the method had access both to public- and to private-sector providers. Of the poorest quartile of households, a fifth were deemed to have no access to pills, a fifth had no access to injectables, and nearly a third had no access to IUDs. The poorest quartile of households was more likely to have access only to public-sector providers. And these are recent data from Indonesia, a country that has made a substantial and sustained effort to make contraceptives universally available and, in recent years, to stimulate private-sector provision of contraception. In other countries, one would have to pay even more careful attention to specifying the choices reasonably available before using household-level data to study how couples choose among methods and providers.
