Health-sector reform initiatives around the world are grappling with the problem of how to allocate scarce resources—how to balance public health priorities with consumer demand, and how to maximize equity as well as public health impact in settings where public resources for health are sadly inadequate. In this context, equity is generally understood as equal access to health services, based on need, with payment based on ability to pay. There is little empirical evidence about how equity concerns can be reconciled with goals of improved efficiency, cost-effectiveness and sustainability, however.¹

Among the many possible strategies for reforming health systems, attention most often focuses on recovering costs through user charges.² Many observers argue that it is both feasible and necessary to collect user fees for basic health and family planning services in most settings, but they also acknowledge that tiered pricing systems may be needed to avoid excluding the poorest clients. However, when user fees are introduced or increased, the effects on equity appear to be mixed. A recent review³ cites examples in which rates of health-care utilization are not adversely affected by small increases in user fees, especially when improvements are made in quality of care.⁴ The same review cites evidence from other cases in which user fees appear to have forced poor women and children to forgo health care.⁵ It has also been argued that means testing is difficult to do and that the potential revenue to be gained by collecting charges for essential services in resource-poor settings is often too small to justify the negative impact on access for the most disadvantaged clients.⁶

In 1997, the Bangladeshi government adopted a health and population sector strategy that outlines measures for improving the impact of and equitable access to primary health services. Reproductive and child health care, communicable disease control, limited curative care and behavioral change communication are to be provided as part of an Essential Services Package.⁷ Nongovernmental organizations (NGOs) provide reproductive health and other services in areas allocated to them by the government. Initially, such services were often set up where government programs were lacking or in areas considered hard to reach. Beginning in late 1997, a consortium of NGOs supported by the U.S. Agency for International Development (USAID) began to implement the government's Essential Services Package approach. The NGOs have adopted a variety of strategies to improve the quality of services and to increase their sustainability and impact on health. These strategies include expansion of the range of services through clinics, satellite clinics and village depots.* Village depots have replaced household contraceptive distribution in rural areas.

Users of the reconfigured NGO services are being charged modest fees for consultations with service providers, for contraceptives and for drugs. Prices vary from place to place, but at all sites they are set well below what would typically be charged in the private sector. Even so, from the client's perspective, costs for most services are higher than they used to be at NGO clinics and higher than fees at government health facilities. In many NGO facilities, consultations were formerly free or provided for a nominal charge of a few taka (50 taka equal about U.S. $1.00). At the time of the study, clients in the urban research sites were being charged 20-25 taka to consult a doctor and 10-20 taka to see a midwife. The typical cost of a two- or three-month contraceptive injection (previously available free) was 10 or 20 taka. Prices in rural clinics were generally lower. Exemptions were available at all sites, but operational guidelines for granting exemptions were lacking.

As the program has evolved, the NGOs have been assessing their experiences and modifying their strategies in an effort to balance the goal of cost recovery with that of expanding access to the Essential Services Package for clients in greatest need of services. This article analyzes some of the challenges they face, one of which is the coexistence of less expensive or free government services in jurisdictions not served by the NGOs. Thus, some people are being asked to pay or to pay more for similar (albeit better-quality) services because they happen to live in areas served by NGOs.

In the following section, we describe strategies the poor have developed to obtain affordable health care from the NGOs. We then discuss how the policy changes have both affected and been affected by these health-seeking norms and behaviors. The article identifies access barriers imposed by poverty, gender and other factors, and highlights the interplay between institutional policies and practices and social norms and relationships.

DATA

The data come from a multimethod, qualitative study designed to examine the effects of program reorientation on patterns of service use, client-provider relations, client satisfaction, and household-level decision-making about family planning and other reproductive health services. Our goal is to document how both communities and programs are responding to the transition from a long-established service delivery model to the new program, and to understand the changes from the perspectives of the intended clients and the service providers. The research was conducted at two rural and two urban sites selected in late 1997 in consultation with USAID and the NGO consortium. We also included a third rural site where we had been working since the early 1990s. All were considered average in terms of performance. We collected data in 2-3 subareas of each site, one within a one-mile radius of the clinic, and one or two 2-4 miles away, where satellite clinics were regularly held.

The field research team consisted of two male and five female interviewers. All had bachelor's degrees and a few had master's degrees. All were experienced and had extensive training in qualitative research methods. Data were collected during several visits of 2-3 weeks each between January 1998 and July 2000. The data come primarily from in-depth, semistructured individual interviews with a broad range of people—for example, women who had received contraceptive services from community distribution workers, NGO clinics or depot holders; the husbands of some of these women; various types of providers; and other community members. The individual interviews were supplemented with a few group interviews and observations in clinics and satellite clinics. In addition, the researchers' unstructured ethnographic field notes were collected and analyzed. Respondents were asked about their past and present experiences and perceptions of service providers available in and around their communities, including the NGOs implementing the new service delivery model, other NGOs, government services and private practitioners. Respondents were also asked about their experiences with health and family planning services prior to the study period. Changes over time were explored through follow-up interviews.

Respondents were identified through a variety of methods: snowball sampling, key informants, service records and recruitment of clients at service sites. Although we did not use random sampling procedures, we attempted to ensure representation of key population subgroups, including women working outside the home, abandoned women, slum inhabitants, and women considered economically solvent. Families not using NGO services were also identified and interviewed at each site.

This article is based on approximately 420 transcripts from interviews and observations at the five sites. We identified key themes while reviewing and discussing the transcripts with the field research team. Using an ethnographic database package (SPData), we created thematic categories, and then reread and coded the transcripts. The organization of coded material into categories made it possible to systematically consider all relevant material on a particular theme. After specific interpretations were formulated, we went back to the data to look for counterevidence, and we rejected or reformulated some interpretations accordingly. In addition, we asked the field research team to review the findings and interpretations, and to provide counterevidence for us to consider. The quotes presented in the paper were selected to illustrate themes that had been articulated in various ways by the study respondents. The variable contexts and sources of the information make systematic comparisons between sites difficult, but we note any urban versus rural differences that emerged.

HEALTH-SEEKING STRATEGIES OF THE POOR

The implications of the new NGO service delivery model for access to services, and for the NGOs' ability to recover costs through service charges, should be understood in the context of health-seeking strategies used by the poor. These strategies, which predate the program changes and reflect social and cultural norms, developed in response to the government's long-standing health policies as well as to informal institutional practices.

Reliance on Free Government Services

Low-income respondents were clearly grateful to have access to some free or nominally priced health care through the government health system. However, this care was not consistently available, and obtaining it was sometimes difficult and unpleasant. Many poor respondents said they expended considerable time and effort to obtain health services free or at the lowest possible cost; they traveled long distances (often on foot), waited for long periods in crowded government facilities and used different places (including but not limited to government facilities) for different health needs. In some cases, they simply let a disease run its course because they lacked money for treatment.

Reliance on Personal Connections

Bangladeshi society is extremely hierarchical, and many poor people depend on patron-client relationships to obtain health care.⁸ In an emergency, a poor person may appeal to a better-off person in the community or to an employer to arrange or pay for hospital care. For less serious health problems, less influential intermediaries—for example, people who have a friend or relative working in the facility or who have used the facility frequently and become acquainted with the staff—are often used to obtain access to care.⁹ The assumption is that one will not be ill-treated or overcharged when there is a personal connection, and that special favors such as credit or fee waivers may be granted.

Interviews with clients and former community-level family planning workers about the household distribution system indicated that workers typically established personal relationships with their clients and were often seen as intermediaries, because they knew the system and the clinic staff and could use their influence to help clients get what they needed without unnecessary payments. There were also informal intermediaries, mainly in the rural communities, who had become familiar with the health facilities and enjoyed accompanying other women. One such woman explained that she liked this role because it gave her a legitimate reason to go out in public and mix with other women. Some informal intermediaries took pride in being known as knowledgeable, public-spirited people who had connections, whereas some simply liked helping others.

Relaxation of Restrictions on Women's Mobility

Although the opportunity cost of going to health facilities is generally seen as lower for women than for men (women's work and time are undervalued because of gender inequality), social restrictions on women's mobility have historically limited their access to services outside the home—a situation that formed the rationale for the previous doorstep program. In this study, however, many people commented that women's use of health facilities had increased dramatically in recent years. The study findings suggest that this change is a function of several factors, including wider acceptance of family planning and other public health interventions, the partial erosion of norms limiting women's presence in the public sphere and greater availability of free and low-cost health services.

Our data from these sites and from other villages where we have been working since the early 1990s suggest that physical seclusion of women (purdah) is less important now than it was in the recent past, when men who could afford it might have purchased medicines or brought local practitioners to their homes rather than let their wives go out to obtain health services. At present, both men and women speak about observing purdah in varying degrees, but they seem unwilling to sacrifice much to maintain the custom. One of the rural NGO staff, for example, offered to provide a contraceptive injection at home for a woman who usually wore a burqua to cover her head and body when she left the house and whose husband was known for his religiosity. The woman declined because she saw no reason to pay five taka (approximately 10 cents) extra for this service. Similarly, men sometimes encouraged their wives to go to clinics where they could get free treatment or medications. (In some cases, the women collected drugs for their husbands.) One rural woman said her husband used to purchase medicines for her and her children, and had brought doctors to their home because he did not like her to leave the house. When he heard that free drugs were available at the NGO clinic, he insisted she go there (without him), despite the necessity of a one-hour journey on foot. She made the journey, but discovered that medicine was not available for free. The husband subsequently forbade her to return to the clinic and said he would resume getting drugs from the market. Thus, it appears that gender norms are being redefined to accommodate a rising demand for modern health care in an environment where access to a variety of sources of care is also increasing. Not surprisingly, families are more inclined to follow new norms when they see an economic advantage in doing so.

Restricting Use of Family Money for Women's Health

Economic pressures may also work against women's access to clinic-based services. As one poor rural woman explained, "I am a woman. If I stay in bed for two days (because of illness), it will not be a problem, but if he (my husband) remains in that condition, who will support the family?" Similarly, although the majority of men in the study acknowledged the importance of women's health and did not object to their wives' using clinics, many seemed unwilling to allocate family resources to health care for their wives. As another rural woman said:

"They tell their wives to let it be (ignore the illness) because it is already there and the family's work is still getting done. Finally, when she is unable to work or to serve her husband's [sexual] needs, then maybe the wife is treated. Or else she is sent to her parents' home."

Another woman commented that some women were unable to attend clinics because of transportation costs. The mix of statements about economic constraints to women's use of health facilities seems to reflect a blend of poverty and gender inequity.¹⁰

Many respondents in this study, as in our previous research,¹¹ indicated that when men oppose women's contraceptive use, it is often because they are afraid the methods will have side effects and that they will have to pay for treatment.¹² Even in families that were not under severe economic stress, women often worried because they knew their husband and his family would be reluctant to make expenditures to treat problems believed to be caused by contraceptives.

Making Contraceptive Choices to Minimize Risk

The NGOs offer a cafeteria plan of options, but women's choices of contraceptive methods are often constrained more by the reluctance of families to bear the costs and risks of family planning than by the availability of methods.^†

"I know that if my health fails my husband will not arrange treatment for me. I have to manage it myself, maybe by secretly taking something from the house and selling it to pay for the treatment. That's why I am afraid of the injection. I have to choose something [a contraceptive method] after considering all the pros and cons."—urban woman

Some women in the study saw the government's practice of providing compensation payments to those who adopt long-acting contraceptive methods (sterilization, the implant and the IUD) as a way of helping them overcome any problems they might encounter as a result of using these methods.¹³

THE IMPACT OF THE POLICY CHANGES

The common strategies used by the poor to obtain health care often did not function as expected in the new context. Poor people will certainly find new strategies, but doing so will take time, especially insofar as the new policies are still not fully developed and are not well communicated.

Some features of the new system make previous strategies unnecessary. There was some evidence, for example, that clients were beginning to see that personal connections were no longer needed to obtain adequate services at fair prices (as long as they could afford the prices). In addition, many women who had previously relied on intermediaries to guide them through the service delivery system and to talk to health providers on their behalf no longer felt this was necessary.¹⁴ Some strategies were dysfunctional in the context of the new NGO model. Some problems were due to poor information. For example, some clients failed to differentiate between NGO and government services, or did not realize that fees could be waived or reduced. Some problems reflected inconsistencies between government and NGO policies, gaps in the new NGO service delivery model and a lack of resolution within the NGOs regarding how policies should be implemented.

Credit: A Gap in the New Model

Even in families with economic resources, women often have limited access to cash. Clients in areas served by the NGOs had to pay for contraceptives when the door-to-door system was operating, but many established informal credit with the community-based family planning workers, with whom they had personal relationships. When this option was withdrawn, it created hardship for many women, particularly in rural areas. "I had a really good relationship with her [the family planning worker]," said one woman, "She used to ask me to pay on time, but once in awhile I would tell her that I didn't have any money, so I could not take [the supply] of pills. She would give me the pills, and tell me to pay later." Some women turned to private (and, therefore, more expensive) practitioners because they offered credit.¹⁵

There was no formal system for providing credit in the reconfigured NGO services, but an informal system seemed to be emerging. Statements by clients and clinic staff revealed a shared view that credit should be provided only when a personal relationship exists: "The apas do not know us, so how can we make demands on them?" explained one client shortly after a clinic had been established in her community. Later in the study, another woman said, "I have a card and I know people. Do they have any other choice besides giving me the [family planning] methods on credit?"¹⁶

Because this emerging credit system was informal, there were no rules and procedures to guide staff as they attempted to reconcile client needs and their own impulses to be helpful with the cost-recovery mandate that they had been given. Outside the clinics, the rural women who ran village supply depots responded to the increased emphasis on cost recovery by becoming more reluctant to extend credit, although they continued to do it. Many of these depot holders were either former door-to-door workers whose jobs had become lower-paid and less demanding, or former depot holders who had received a stipend and therefore had not needed to concern themselves much with collecting payment for contraceptives. In either case, they disliked having to go to women's houses to collect debts, and they sympathized with their clients. One complained:

"Now I have decided that I will no longer permit them [to take contraceptives] on credit. Because those who took credit last time caused me a lot of trouble. I had to go to them three times to recover the money so I could submit it to the office. But what can I do, Apa—they are poor!"

Some of the women running village supply depots responded to this dilemma by dropping clients who had great difficulty paying.¹⁷

The program's dual mandate has therefore put service providers who deal directly with clients in a difficult position. Client demand for credit persists—in fact, the need for credit may be even greater now than in the past because of higher charges—but now there is more emphasis on cost recovery. Service providers complained that they were under pressure to provide credit or to accept payment in kind (for example, eggs or vegetables)—another traditional practice that helped women overcome the barrier imposed by their irregular access to cash.

Fee Waivers: Policies Unclear

Under the new system, need-based fee waivers are intended to mitigate gender- and class-based constraints to access. The NGOs have an official policy that no client should leave a clinic without receiving services, but clients often did not seem to know this. In accordance with this policy, many clients received discounts and exemptions. In 1999, average revenues in the urban clinics were only about half what they would have been if all clients had been charged the posted prices.¹⁸‡ However, the clinic staff found it difficult to provide subsidies openly and systematically. They feared that openly waiving or adjusting fees for some would raise concerns about fairness and make it difficult to enforce the posted prices for the majority of clients.

Some service providers improvised ways to implement the flexible charges policy by telling clients to pay the balance at their next visit; they made a show of writing the amount owed on clients' receipts, even though both sides often assumed payment would never be made. This process was carried out both to avoid embarrassing the client and to avoid advertising the availability of discounts and fee waivers. In most cases, the staff discussed payment in very delicate ways and were kind to clients who were unable to pay. For example, a paramedic was heard saying to a woman who did not realize she would be charged for an antenatal check-up, "You can pay the money later. Today you should have the check-up. You need this." The practice of bargaining over prices in clinics, and the granting of personal favors to the needy may be so familiar to both clients and providers that they engage in this behavior without much reflection.

Staff also clearly tried to direct subsidies to clients who really needed them. A well-dressed woman whose husband worked abroad, for example, was told, "Apa, this [free] medicine [contraceptive injection] is for those who do not have the ability to pay for it. Your husband lives abroad [implying that he earns a good income] so we should not give it to you, but we have." She probably hoped that the woman would understand and reimburse the clinic later.

Despite their efforts to identify clients in need and to provide them with services, clinic staff also understood that cost recovery is a priority under the new program, and they attempted to reconcile these two mandates. A paramedic in a satellite clinic seemed genuinely distressed because she could tell that the very poor woman to whom she had given a prescription for a vaginal infection would not be able to purchase the medication. Medications subsidized by the government and donors can be provided free, but those purchased through the NGOs' revolving drug funds cannot be, according to policy. The paramedic gave the woman a couple of free medications for other problems and showed her a lot of sympathy, but she did not provide the medication the woman needed for her vaginitis.

The clinic staff were virtually unanimous in saying it would be a mistake to make the availability of discounts and fee exemptions widely known. A paramedic in one of the clinics explained that although they had been instructed that no one should leave the clinic without services for lack of money, they always told people to pay later. "If we provide free service to someone, then others will also demand it. It may create a disturbance," she said. Asked whether information about the availability of credit, discounts and free services was mentioned on the posted list of charges, she said it was not, and that if it was, no one would want to pay. Asked whether she thought people in the surrounding communities knew, or should know, that discounted or free services were available, she said that people generally did not and should not know. Because virtually everyone in these villages was poor, she said, it would be very difficult to justify waiving charges for some but not others.

The failure to advertise the availability of fee waivers may be creating hidden problems. First, the NGOs may be perceived by some as overly concerned with cost recovery and unconcerned with the welfare of the poor.¹⁹ Second, clients may not take advantage of the full range of services available because they cannot afford everything they need, or they may not return to the clinic for fear that they will be asked to pay the remaining balance from the previous visit. Third, some may not go to the NGO clinics at all because they think they will not be able to afford the service charges.

Volunteers who run supply depots in rural areas are supposed to refer clients to the clinics. Their incentive for doing so is a proportion of the fee paid, an arrangement that may have the unintended effect of discouraging referrals of women likely to need fee waivers. Similarly, because part of their income consists of commissions from the sale of oral contraceptives, condoms and basic medicines, they are unlikely to seek out clients who need free commodities.

One woman had been encouraged to go to a nearby NGO. When her children fell ill, however, she failed to seek medical care for them because she had heard that the NGO's system had changed and that clients now had to pay:

"The apas of the family planning office^§ sometimes come [to my area] to give children oral vaccinations. They told me to come to their clinic. They said if I went there they would provide free treatment, but I know that they will not even speak to me if I do not have money. No matter what they say, they will not provide treatment without money. This is why I do not go."

Concerned that fees might be discouraging the poorest from coming to their clinics, NGO staff in the rural sites modified their promotional messages. Near the end of our study, they began telling people that the clinics charge for services but that people should not avoid coming to the clinic just because of money. How people are to understand this ambiguous message is unclear. If the NGOs hope that only those genuinely in need will understand that fee waivers are available, they are probably being unrealistic. If they want to make known the availability of need-based exemptions, a direct approach would surely be more effective.

A client's need for subsidy is not always apparent. In one village, a clinic promoter had announced that people who could afford fees would have to pay, whereas those who could not would get services for free. This led to misunderstandings and resentment in several cases. For example, when three women who considered themselves poor were asked to pay, they felt that they had been misled by the promoter. One left the clinic without getting the services she was seeking, and two were given credit (and did not plan to return to the clinic).

One respondent was a sociable woman who said she enjoyed going to the clinic and would stay there and gossip with the staff when they weren't busy. She said she liked to listen and watch them interacting with other clients. She herself didn't have any problem paying for services, she told us, but she questioned whether the poor should be charged. She believed there were many people who did not use the clinic because they could not afford the fees. When she recognized a beggar woman who was being asked to pay for medicine, she remonstrated with the clinic staff:

"Apa, this woman begs from door to door. She has no husband, she has no one. She can't even support herself. She doesn't even get 10 paisa when she goes to someone's house (to beg)—where will she get 10 taka to pay you?"

The paramedic then wrote the woman a prescription without charging her.**

Extent of Need for Fee Waivers and Credit

Although fee subsidies appear essential for the poorest clients, the percentage of clients who require them may not be excessively high. Among our sample of 154 recent clients from the five NGO sites between mid-1998 and mid-2000, only about one in five needed fee reductions or waivers. The researchers reviewed transcripts from interviews with these 154 clients and sorted them into three groups: those with no apparent problem paying; those who could afford the fee but needed credit; and those who appeared to need a partial or total subsidy.†† Of the urban respondents, 24 women were able to pay, eight needed credit and five needed a subsidy. Among rural women who responded, 52 were able to pay, 37 needed credit, and 28 needed a subsidy.

Thus, about half (49%) of the 154 clients had no problem paying for their most recent clinic visit. Somewhat fewer than a third (29%) needed credit, and about a fifth (20%) needed a discount or fee waiver. Waivers and credit were often given for services and contraceptives, but rarely, if ever, for medications.

These findings imply that current prices were within reach of the majority of this small sample of clients, but that many clients needed credit and a substantial minority needed a fee reduction or waiver. Among the clients who needed either fee waivers or credit, very few left a clinic without receiving any service whatsoever, but some were unable to receive some of the services they needed—usually medications or laboratory tests. A few women were sent home to get money. Of these, some returned and received services, but others became discouraged and either delayed going back or did not return. The risk of this happening is significant, especially in satellite clinics, which usually take place once every 2-4 weeks.

In at least a few cases, women who needed credit or fee waivers failed to make their situation known to the service providers. Because this question was not always asked in the interview, however, we cannot specify the number. The data suggest that some types of clients are more likely than others to make their needs known. Unfortunately, it may be the most disadvantaged who hesitate to assert their need for waivers. These clients probably underrepresent the number to whom cost is a barrier to access; potential clients may fail to seek services because they expect them to be unaffordable. As a promotional strategy, urban clinics have begun to offer family health cards, which entitle the purchaser to unlimited services for all family members for one year. Very poor families receive the cards for a nominal charge. This strategy has been popular with clients because it mitigates the problem of women's inconsistent access to cash and makes it unnecessary for poor clients to request fee exemptions each time they visit the clinic.

CONCLUSIONS

Reviews of evidence from health programs in other developing countries yield few, if any examples of successful large-scale strategies that have simultaneously expanded access for the poor, improved service quality and increased cost recovery and sustainability.²⁰ The NGOs in Bangladesh recognize the complexity of pursuing cost recovery while making their services accessible to the poor. They are still experimenting with pricing policies and with strategies for communicating information about their services and prices to the communities they serve. The results of this study highlight the difficulties they face.

Findings from surveys commissioned as part of the new NGO initiative suggest that a substantial minority of women needed fee exemptions. In one such study, interviews were conducted with a sample of 3,000 households in urban areas served by 10 different NGOs to determine what they had recently paid for various types of services from various sources, including the NGO clinics. The proportion able to pay current prices was measured by the percentage who actually paid (assuming lower-priced sources of care were available). A second estimate was derived by calculating what a typical family would pay for basic health services as a proportion of total expenditures; households with cost-to-expenditure ratios exceeding 1.5% were considered unable to pay.²¹ The authors concluded that about 14-17% of households in the areas served by the NGOs would require a price subsidy.²²

Our analysis shows a somewhat higher percentage of women requiring fee reductions or waivers (about one-fifth overall at our five NGO sites), but the majority of recent clients in our study were rural. Our data and findings from other studies suggest that user charges present greater barriers for the rural poor than for the urban poor.²³ In addition, approximately 29% of clients in our sample needed credit, a proportion that might not be reflected in a survey using hypothetical questions, or in calculations based on household expenditure levels. For this group, the price levels were not in themselves a problem—the problem was the women's inconsistent access to cash.

It might be argued that the NGOs' current configuration of prices and exceptions already provides for clients who need credit or exemptions. Both credit and fee waivers are available in the clinics and supply depots; fee waivers are provided as a matter of official policy, and credit is given when individual staff take the initiative to help clients. However, neither credit nor fee waivers are openly advertised or mentioned in the price lists posted at the clinics, and the manner in which they are discussed and provided is personalized and opaque.

Although these practices help many disadvantaged clients, they both disempower clients and undermine the NGOs' goal of improving equitable access. Some people—particularly those who assume rules are rules—are discouraged from coming to the clinics. Opaque practices also reinforce the idea that access to credit and fee exemptions depends on the whim of the service providers and the client's personal relationships with them, as it often does outside the NGO system. Thus, those who have established personal relationships with staff are likely to feel personally offended when denied credit or discounts, staff are under pressure to provide this assistance to clients with whom they are friendly, and clients may be put in the position of having to beg for something to which they are entitled.

Experience from around the world confirms that it is no simple matter to design need-based subsidy systems that are at once transparent and effective in reaching the poor.²⁴ There is increasing recognition that when user fee policies for basic health services are implemented in poor countries, systems are needed to exempt those unable to pay; most countries that have introduced user fees have some sort of exemption policy.^‡‡ The effectiveness of systems for means testing is not well documented, however, and the literature points out numerous weaknesses in existing systems for targeting need-based subsidies.²⁵ It appears that targeting and means testing systems are often informal, as was the case among the NGOs in our study. The investigators in another study concluded, for example, that "highly informal means testing is not adequate, is inaccurate, and is not as effective as it might be in protecting the poor and improving equity under cost recovery."²⁶

We argue that the Bangladeshi NGOs need a clearly defined, transparent system for providing credit and fee exemptions. We propose that communities be enlisted to help define such a system and to modify it to fit conditions in the various areas where NGOs work. Expansion of the prepaid family card system could reduce the volume of fee waivers required at the time services are provided. Involving communities in identifying families and individuals in need of exemptions could enhance the perceived fairness of the policies and increase the likelihood that they will be widely communicated and enforced. If such a system cannot be made to work, then perhaps user charges should be reconsidered.