Estimating Service Provision Costs from Survey Data: Family Planning Service Provision Structure and Cost in the Philippines

Transcription

1 Estimating Service Provision Costs from Survey Data: Family Planning Service Provision Structure and Cost in the Philippines John F. Stewart University of North Carolina at Chapel Hill Carolina Population Center Alejandro N. Herrin Rachel H. Racelis University of the Philippines Presented at the Souther Economic Association Meetings November 2002 Revised October 2002 The funding for this research was provided by the Rockefeller Foundation and the EVALUATION Project. The survey used in this study could not have been completed without the cooperation and assistance of the community of international donors, NGOs, local government officials, and the Philippines Department of Health. Their cooperation and generosity with their expertise are much appreciated. We also wish to acknowledge the hard work of interviewers from the Upecon Foundation of the University of the Philippines, the Office of Population Studies of the San Carlos University and the Regional Institute for Mindanao Culture of Xavier University who administered the surveys in Luzon, Cebu, and Mindanao respectively. The paper has benefitted from the comments of Brad Schwartz, David Guilkey and three anonymous referees. Corresponding author: John F. Stewart Department of Economics, CB 3305 University of North Carolina at Chapel Hill Chapel Hill, NC

2 Abstract Estimating Service Provision Costs from Survey Data: Family Planning Service Provision Structure and Cost in the Philippines John F. Stewart, Alejandro N. Herrin, and Rachel H. Racelis February 2002 This paper uses survey data to construct estimates of the per CYP cost of family planning service provisions at 226 public, NGO, and private health care facilities in the Philippines. Family planning service provision costs show substantial variation across facilities. Regression analysis is used to explore the relationship between per CYO cost, scale, method mix and facility types provides is performed. While the technology would suggest that certain methods (such as sterilization) should result in low cost per CYP, in practice we find that this is not the case. We do find that the output scale has a significant effect on costs with higher output facilities having significantly lower costs. Facilities that specialize heavily in family planning service appear to have higher cost than less specialized facilities, and some types of facilities have higher costs than others.

3 Estimating Service Provision Costs from Survey Data: Family Planning Service Provision Structure and Cost in the Philippines February 2002 Abstract: This paper uses survey data to construct estimates of the per CYP cost of family planning service provisions at 226 public, NGO, and private health care facilities in the Philippines. Family planning service provision costs show substantial variation across facilities. Regression analysis is used to explore the relationship between per CYO cost, scale, method mix and facility types provides is performed. While the technology would suggest that certain methods (such as sterilization) should result in low cost per CYP, in practice we find that this is not the case. We do find that the output scale has a significant effect on costs with higher output facilities having significantly lower costs. Facilities that specialize heavily in family planning service appear to have higher cost than less specialized facilities, and some types of facilities have higher costs than others.

4 I. Introduction The cost of providing maternal and child, reproductive health, nutrition, and family planning health care services in developing countries has become an increasingly pressing issue over the last decade. A number of factors drive the concern over cost. These include the staggering estimated total cost of providing these services, 1 the scarcity of funds to meet policy objectives, and the resultant pressure from donor countries and agencies for programs to be more accountable for results and cost effectiveness, and for programs to demonstrate the potential for long run sustainability. Many of the research and evaluation questions that must be confronted in this cost-conscious environment require more detailed and disaggregated information on the costs of service provision than has been generated in the past. The cost of providing a health service is not a single parameter but can vary depending on a wide variety of factors, many of which are a result of policy decisions. Examples are numerous. Among the current issues requiring detailed cross sectional analyzes of service provision costs is the relative efficiency of private versus public delivery of services, of integrated versus vertical service delivery programs, and of decentralized versus centralized service structures. 2 The cost of providing a health care service may also depend on the characteristics of the population to be served. For example the costs of serving an urban population may be different from the cost of serving a rural population. To assess the costs of serving different segments of the population one would need a sample of facilities serving populations of varying characteristics and have cost estimates of each. Finally as donors and countries are faced with the problem of allocating scare funds among 1 The United Nations Fund for Population Activities (UNFPA) has estimated that worldwide spending on family planning (FP) and reproductive health (RH) will reach 17 billion dollars annually by the year 2000 (UNFPA, 1996). 2 In many countries, including the Philippines, donors are supporting quasi-market service delivery systems as an alternative to government provision. In such cases, one must be able to generate estimates of service provision costs in both government and alternative systems of service delivery to evaluate the relative cost effectiveness of alternative service delivery models. Though policy-makers now view maternal and child health, reproductive health, and family planning as an integrated package of services, service provision is often less than fully integrated. For example family planning services are often provided by specialized vertical programs. In some areas, for example parts of west Africa, these vertical programs are major, if not dominant, supplies of the services. See Mancini, Stecklov and Stewart (1999). Even in countries where services are nominally fully integrated, such as the Philippines, one finds that certain organization and service delivery pathways are heavily specialized in the provision of certain services such as family planning. A large number of countries have been experimenting with devolving control of the public heath care system to lower levels of government in the hopes of increasing the responsiveness of the health care system and increasing its efficiency. -1-

5 various types of health interventions such as immunizations, prenatal care, family planning and so forth, knowledge of the costs of providing various types of services must be estimated. 3 Many of the important issues in providing health care in less developed countries will require estimates of cost that can be disaggregated by the type of service provide, the structure of the systems providing those services, and the characteristics of the populations being served. Because there are many factors that will be determinants of observed variations in service delivery costs, fairly large samples of facilities will be required. Such estimates require data that are not readily available from existing sources. Existing data obtained at the facility level such as the DHS Service Availability Module and Situation Analysis, while providing much information about the facility s operation, provided essentially none of the information required to make estimates of service provision costs. Those cost studies that have been made are usually done on a case by case basis for a limited number of facilities. This paper reports the result of a detailed analysis of family planning service provision cost in the Philippines. The cost estimates are based on a survey administered at the facility level for a large sample of family planning service providers in the Philippines. These cost estimates allow us to compare the cost of providing service under a variety of different structures of service provision including government service provision, NGO service providers, and a quasi-market franchise system of service providers unique to the Philippines. Regression analysis is then employed to examine the relationship between cost, method mix, scale, degree of specialization in family planning, and facility type. The paper will proceed as follows. The next section will discuss the conceptual issues in measuring cost. Section III describes the structure of the family planning service provision system in the Philippines, the sample, and the survey used in this study. Section IV describes the methods used to produce estimates of facility level per CYP service provision cost. The cost estimates and regression results are presented in Section V. Section VI concludes the paper. II. Conceptual Issues At the most basic level, the concept of cost is relatively simple. The production of a good or service 3 A extremely important and related issue here are the benefits or outcomes of the different types of interventions which will also be required for informed allocation of funds across various activities. -2-

6 requires the use of various scarce resources (e.g., labor time, materials, supplies, etc.). Each of the resources has a cost or price associated with its use (e.g., a wage rate or purchase or rental price). The calculation of the cost of producing a given quantity of goods or services merely involves counting up the quantity of each type of resource used in production, multiplying the quantity of each resource by its unit price and adding up the results to a total. The conceptual issues surrounding the calculation of cost boil down to which input quantities and prices to use. Unfortunately, there is no fixed correct answer to which input quantities and prices to use, but rather the answer depends on which questions are to be asked of the cost estimates generated and the data that is available. Economic Cost or Expenditures The first conceptual issue is one of semantics. When an economist uses the term cost, she is referring to the opportunity cost of the resources required to produce a given level of output assuming cost minimizing input selection and technical efficiency in production. When a layperson refers to cost, he typically means the total amount of money that was paid for something. That is, the amounts actually used and the prices actually paid. The apparatus of economic cost theory, with it the assumptions of technical efficiency and cost minimizing input selection, derives from an assumption of profit maximizing behavior by the producer. While this may be a reasonable assumption for business firms in private market activity, it may not be appropriate for government agencies and NGOs providing health care services. Where an economics approach to cost represents the minimum value of resources that would be required to produce a given level of output, an expenditure approach to cost approximates what is actually being spent given existing institutions and behavior to provide observed levels of service. For most of the types of questions described earlier, expenditures come closer to providing the desired information than does economic cost. 4 We wish to make it very clear that our estimates will be made from the expenditure rather than from the economic cost point of view. However, after numerous attempts to come up with a terminology that would reflect this difference, we have decided that it is better to risk offending the economists rather than totally confuse 4 The presumption here is, that from a practical point of view, one is working from a perspective of the institutions and practices as they exist not from what they could be if they operated in a world of technical and allocative efficiency. It seems more relevant, in most contexts, to ask the question what expenditures are required to provide services under the current set of institutions, practices, and circumstance to provide service rather than what should it cost under ideal circumstances, though there is certainly a role for economics cost notions as a basis for comparison with current practices. -3-

7 our other readers and will continue to refer to our estimates as cost estimates. Level at which to measure cost Two issues are relevant to the level at which cost is measured. The first issue is who actually pays for the inputs. Many of the resources used in the production of family planning or other health services may be without cost to the facility produce the services. An example would be donated family planning commodities received by the facility. While the facility bore no cost for these commodities, the donating agency did. In this paper we will estimate cost based on the total expenditures made on the resources used to produce the output without regard to who makes the expenditure and will thus include the cost of donated inputs. 5 Related to the issue of who actually pays for the resources used to provide the service is the issue of how far up the system above the facility to trace costs. The typical facility will not be a standalone organization but will likely have administrative oversight, logistics, training, IEC and other functions provided by levels of the organization above the facility directly providing services. These costs are part of the costs of producing the service provided by the facility and ideally should be included. Doing so presents at least two practical problems. First, it increases the burden of data collection in that facilities are very unlikely to have data on the overhead costs and costs of ancillary activities (such as training or IEC) that are born by the parent organization so these data must be obtained from parent organizations rather than facilities. Second, because these activities often are not undertaken at the facility level, their cost must be allocated in some way across a number of facilities, which is a difficult process. Our approach in this study is to measure the quantity of resources used at the point where the services are actually delivered, at the facility level. This is what we call a bottom-up approach to cost estimation. The bottom-up approach has the advantage of making it easier to associate costs with particular localities and particular services. The disadvantage of the bottom-up approach is that it is difficult to observe cost of administration and other support activities that occur at a high level in the health care delivery system, In the 5 An issue related to donor supplies inputs is user supplied inputs. It is important to keep in mind that the client is a key input into to the production of health services such as family planning. All the staff, equipment, and commodities in the world will not produce a family planning outcome unless there are clients using these services. Resource supplied by the client including the value of their time, transportation expenses born by the client and so forth should be included in the total cost of providing the service. However, inclusion of these costs is beyond the scope of what we report here. -4-

8 current study we will concentrate only on the direct costs of service provision at the facility level. The alternative is a top-down approach in which all expenditures made by any party supporting the provision of health care service are accumulated starting with the central government and external donors and moving down toward the points of service delivery. Herrin, et. al (1997) made a detailed top-down estimate of family planning expenditures in the Philippines for The study covered expenditures of all external donors, all level of government, and private individuals. 6 Of the estimated $59 million spent on family planning services, Herrin estimates roughly 77 percent of the expenditure was for the direct cost of service provision with 5 percent for training and research and evaluation, 9 percent for IEC and 9 percent for administration. 7 The Problem of Joint Costs and the Allocation of Cost to Individual Services Joint costs are costs for resources that produce multiple services, for example a clinic administrator whose activities cover family planning and other services at a clinic. There are two levels of issues with respect to joint cost. The first is conceptual, whether or not there is any real basis by which one can allocate the cost of a resource across the multiple outputs it produces. For a true joint input, there is no way to allocate its cost across outputs. The second issue is practical. For many inputs, for example staff time, one could reasonably allocate the cost of staff time between activities based on the proportion of time spent in each. However for most such inputs, available data does not allocate inputs across the various outputs they produce. Total staff hours may be fairly easy to obtain but how those staff hours are divided between family planning service production and the production of other services will not typically be available. Two approaches are possible. The first is to allow the data to allocate total costs across activities by using models of multi-output production processes. 8 The second, and the approach that will be 6 Estimates of spending on FP by private individual was bases on consumer expenditure data from the 1993 National Demographic Survey. Spending by government and external donors was based on a detailed examination of budget records. 7 A substantial portion of the total expenditure (27 percent) could not be allocated to the functional categories of service provision, training & research and evaluation, IEC, or support. The numbers presented assume that the unallocated expenditure was distributed by functional categories in the same way as the allocable expenditure. If none of the unallocated expenditure was used for service provision, the share of direct service provision in total expenditure would drop to 56 percent. 8 In a multi-product production model a vector of outputs is assume to be a function of a vector of inputs (Q 1,..., Q n ) = f(x 1,..., x m ), where the Q i are output quantities and the x i are input quantities. A multi-output cost function can the be derived, TC (Q 1,.., Q n ) = g(q 1,..., Q n, p 1,..., p m ), where the p i are the prices of the inputs. This approach requires a very large number of facilities if the number of outputs and inputs is large and requires a great deal of assumed structure on the nature of the -5-

9 followed here, is to collect the data that allows a prior allocation of resources across activities. Choice of Output Measure The final choice is how to measure output. The provision of family planning services will clearly require differing amounts and types of resources depending on the method of family planning that is used. A visit of an IUD insertion may provide years for protection of unwanted pregnancy but will consume a substantial amount of resources. A visit for pills may consume substantially fewer resources but provide birth control for only a few months. For the purpose of this study we choose to use a standard Couple Years of Protection (CYP) measure. The advantages of using CYPs as a measure of output are twofold. The first is a practical data consideration. Facilities, in general, had more complete and accurate information on the distribution of contraceptives than they did on visits. Contraceptive distribution can easily be converted to CYP output but not to visits. Because the a priori expectation is that method mix should have a large impact on per CYP cost, method mix was included as an explanatory variable. As a further check, cost estimates were also constructed for individual methods. The second is conceptual. The ultimate service being provided is a time period of protection from unintended pregnancy. While clearly methods differ significantly in characteristics such as side effects and reversibility, the CYP comes closer to being a consistent measure of the service being provided. The choice of this measure, however, requires the careful consideration of method mix in comparing costs across facilities. III. The Sample and the Survey A. Overview of Approach This study estimates the direct cost of service delivery at the facility level. This was accomplished by using a variety of data collection strategies to identify and measure the quantities and cost of the various resources used to produce family planning services at each facility in the sample. The production of health care service requires inputs from four general categories resources that can be observed at the facility: 1) staff time, 2) supplies and medicines, 3) equipment, and 4) physical space. The approach of this study was to design a facility level survey production and cost functions. -6-

10 to measure the amounts of the various resources that are being used at each facility in the provision of family planning services and cost these resources at prices that reflect the amount of funds that are being expended in support of service provision. 9 Service output measures were constructed from the facility s service statistics and contraceptive distribution records and then converted to CYPs using the conversion factors shown in Table 1. In the remainder of this section we will describe the sample of facilities surveyed and the methods used for calculating the various components of cost. ---Table 1, about here--- B. Structure of Family Planning Service Provision in the Philippines To understand the sample selection procedure that was used, it is first necessary to provide a description of the structure of the FP provision structure in the Philippines. The Philippines have a particularly diverse set of structures providing family planning services in a clinical environment. 10 Family planning services are provided by three different levels of government (National, Provincial, and Local Government Unit 11 (LGU)), by a variety of traditional NGO s, by industrial (employer operated) clinics, and by private physicians and clinics. In addition, there is a mixed NGO-Private provision structure of family planning service provision in which private clinics are franchised by The Institute of Maternal and Child Health (IMCH) or Integrated Maternal Child C Services Development, Inc. (SDI). Each of these systems is described more fully below. Public Sector Service Delivery: Prior to devolution 12, there were direct and straight administrative and policy links from the top to the 9 As noted elsewhere, the issue of who is making these expenditures is an important one. The approach we will employ is based on the total expenditures regardless of who is making them. Thus donated commodities and supplies are costed based on the donors expenditures, staff costs are based on wages and salaries paid to staff regardless of who actually makes the payment. Under a pure economic concept of cost, resources would be valued at their opportunity costs; our approach is to value resources at their financial cost. 10 Family planning services may also be provided in non-clinical environments including pharmacies and CBD programs. These are not included in the current study. 11 For the purpose of our study, local government units included cities and municipalities. 12 Under devolution, which began in 1992, the direct responsibility for many of the social programs in the Philippines including health, welfare, and education was moved from the national government to local governments. Financial support for these programs was provided in part by block grants by the national government to the local governments. The local -7-

11 bottom of the public health care provision system. At the bottom of the chain was the barangay health station (BHS) 13 providing basic primary care to the immediate population with a midwife or nurse being the sole staff. The BHS s were (usually) linked at the municipal level to a rural health unit (RHU), or in urban areas, to a Health Center, which provided more resources and highly trained personnel to support more complicated medical procedures for the group of barangay health stations. Overseeing a group of RHU s would be a Provincial Health Office (PHO) and overseeing a group of PHOs would be a Regional Health Office (RHO), providing logistic and administrative support. These were linked back to central administration and policy formation at the national Department of Health (DOH). The public clinics were supported by DOH and Provincial hospitals for the provision of more complicated inpatient care. The devolution essentially broke the chain at the RHO level. Rural Health Units, Health Centers and the barangay health stations that they support now operate under the authority of local government units with finances being supplied by these units from block grants provided by the national government. Though the DOH still has a role in the overall formation of health policy, the link between policy formation and implementation is considerably more tenuous and less direct. 14 Thus the public sector health system in the Philippines currently has three, largely independent, segments serving different functions within the health care delivery system. At the national level are DOH-retained hospitals where a range of family planning, reproductive health and other services are offered at the facility. The primary health care facilities operated under Provincial government authority are provincial hospitals, though some clinics and some family planning clinics are reported in this group. Provincial hospitals tend to be smaller and less compartmentalized than DOH hospitals. At the local level of government FP services are provided through Health Centers, rural health units and the associated barangay health stations. governments decided how the funds would be divided among the various social services programs and how those programs were to be operated. Health services employees and employees in the other functional areas were no longer employees of the national government but became employees of the local governments. 13 The barangay is the basic administrative unit in the Philippines. In a rural setting it would be a small village and surrounding area. In an urban context it is a neighbor or ward. 14 The DOH still influences the operations of the RHU s and BHS s in a number of ways. With respect to family planning, the Regional Health Offices may transfer resources usually in the form of supplies and logistics to the public health providers of the LGU. Local governments also receive assistance from the Family Planning Service of the DOH through the Comprehensive Health Care Agreement (CHCA). The CHCA is an assistance package offered by the DOH to LGUs to encourage them to provide some basic health services including family planning and other public health programs. -8-

12 NGO and Private Sector Service Delivery: There are large numbers of NGOs that are involved in the provision of family planning and RH-MCH services in the Philippines. Some of these are local affiliates of international NGOs such as Family Planning Organization of the Philippines (FPOP) which is the local affiliate of International Planned Parenthood. Others were are national and some are purely local. One unique FP service provision structure in the Philippines is the franchised clinic. John Snow, Inc./Research and Training Institute (JSI/R&T) has pioneered a concept of franchising in health care/fp service delivery through its NGO Strengthening Project. Two large nationwide FP NGOs are engaged in franchising arrangements; SDI and IMCH. The parent NGO (SDI or IMCH) acts as the franchisor while its affiliate clinics function as the franchisees. In 1994, IMCH had 65 affiliate clinics while SDI had 49 clinics as franchisees. Under the franchising arrangement, the franchisor offers benefit packages to franchisees: equipment and testing benefits, service protocols, management development resources and financial assistance. In return for these benefits, each franchisee pays the franchisor a lump-sum joining fee which is paid in advance, a fixed percentage of its gross revenues as royalty and a monthly amortization if the clinic made use of a loan. Franchisees are obligated by their contracts to provide a single and fixed menu of family planning/maternal and child health services, performed in accordance with a set of established protocols. 15 Franchisors conduct regular monitoring and evaluation visits to all clinic franchisees and provide each franchisee with a list of equipment and instruments, medical and other supplies, contraceptive supplies, and cleaning materials. The contraceptives dispensed by franchisees are provided by DOH through provincial or city health offices. 16 Finally family planning and RH-MCH services are also provided by private clinics and hospitals and a number of large private employers provide health care facilities for their employees. 15 The menu of services includes (1) family planning for new and continuing acceptors (IUD insertion and reinsertion, condom supply, pill dispensing, injectable, spermicide and natural family planning methods); (2) laboratory services (pap smear, pregnancy and urine/albumin test); (3) maternal and child health services (pre- and post-natal care, immunization, well-baby care, other pediatric consultations, minor gynecological consultations); and (4) other health services (ear piercing, physical examination and cautery). 16 Contraceptive supplies are predominantly USAID donated, with Family Planning Logistics Management (FPLM) handling the logistics of the distribution of FP supplies. -9-

13 C. The Sample of Facilities A total of 253 facilities was included in the survey. 17 First a non-random selection of six Provinces was made. 18 Within each selected Province, random municipalities were selected where each of the selected municipalities contained a cluster used in the 1993 National Demographic Survey (NDS). This selection mechanism was used to allow us to link facilities in our survey with information relating to the characteristics of the population in the area they served. The number of facilities selected in each area was then determined partly using the regional distribution of public and NGO facilities providing FP services in the country. In 1995, roughly 45 percent were located in Luzon, 25 percent in Visayas and 30 percent in Mindanao. The final sample includes 153 facilities from the NCR and other provinces on Luzon, 60 from Cebu, and 40 from Misamis Oriental. The selection of facilities within each municipality or city barangay was then made to reflect the range of facilities available to the NDS household. In general, the following rules were applied: (a) All RHUs in selected municipalities were included (b) At least one BHS within the municipality was included. The first to be chosen were those within or closest to the NDS barangay. 19 (c) At least one Health Center either within or close to the selected NDS barangay was included. (d) Private clinics and hospitals within the selected areas who would agree to be interviewed were included. 20 (e) At least one public hospital in each province either within or close to the selected NDS municipality was 17 A complete description of the sampling procedure can be found in Stewart, J.F., D.K. Guilkey, A.N. Herrin, and R.H. Racelis, Selection of Provinces was based in part on the availability of trained survey personnel, budget and logistic constraints. These provinces do provide broad coverage of the country and include the National Capital Region (a highly urbanized area of Luzon including Metro Manila) three less urbanized provinces also in the island group of Luzon (Batangas, Laguna, and Cavite), Cebu in the Visayas island group (which includes the second largest metropolitan area in the Philippines,but also rural areas), and the Province of Misamis Oriental on the island of Mindanao. 19 In general, more than one BHS were interviewed for each municipality. 20 Since no listings of private clinics and hospitals were available, these facilities were identified only when the interviewer got to the field. To keep the total sample size intact, BHSs were replaced when private facilities allowed interviews. -10-

14 included. (f) NGO clinics either within or close to the selected municipality were included. (g) Industry-based clinics within the selected NDS municipalities were included. ---Table 2, about here Of the 253 facilities surveyed, there were 226 generated with sufficiently complete data to calculate average cost per CYP estimates. The distribution of facilities by type and operating authority is shown in Table 2. It is interesting to note the extreme diversity in the facilities in the sample. There is substantial variation in the sample with respect to the degree of service integration, the scale of operation, method mix, and internal organization. Nominally, family planning has been completely integrated with maternal and child health and reproductive health services (MCH-RH) in the Philippines. Of the 226 facilities in the final sample, only nineteen of the facilities showed indications of providing family planning services using what might be called a single service (vertical) structure. Eleven of these facilities were hospitals that had a separate family planning departments, five were LGU operated health centers that had separate family planning departments and three were standalone family planning clinics, two operated by provincial governments and one operated by an NGO. Though the vast majority of the facilities offer family planning services in an integrated service environment, the degree to which they specialize in family planning varies substantially. Figure 1 shows the distribution of family planning and MCH-RH visits for different types of facilities. The provincial clinics, FPOP and the other NGO clinics all had well over half their patient visits being for family planning services. At FPOP clinics it was 90%. Almost half the visits at IMCH-SDI clinics were for family planning. The other types of facilities average about 20% family planning visits. There was also substantial variation in the scale of operation. Figure 2 shows the total monthly number of FP visits and CYPs produced for the different types of clinics in the sample. The output of family planning services is very unequally distributed among the suppliers of these services with a very few of the facilities providing the majority of the service but these facilities are operated by a variety of organizations. The fifteen largest producers in the sample generated 65% of the total CYP production. This group includes seven hospitals (four National, one Provincial, one Private and one other NGO), two other NGO clinics, four FPOP clinics, and two local government -11-

15 Health Center. There is also substantial variation in method mix across facility types. Figure 3 show the distribution of method mix for the various categories of facilities. Female sterilization is a very important method in the Philippines and the provision of this method is dominated by hospitals and a few NGO facilities. The output of IMCH-SDI clinics is heavily weighted toward IUDs and the provincial clinics are heavily weighted toward natural family planning. 21 Output distribution is still heavily skewed to a few large facilities even if hospital and the large NGOs with high proportions of sterilizations are excluded. The 207 clinics in the sample produced 63% of the total CYPs in our sample. Over half of the CYPs produced by clinics was produced by just the ten largest clinics. These clinics tended to have client loads heavily weighted toward family planning. Six of these eight had more than 85% of their patient visits for family planning. The unequal distribution of production across clinics is a result of both the scale at which the clinics operate and the method mix they provide. Table 3 shows the average number of CYPs per visit produced at the different types of facilities. It is highest in hospitals and NGO clinics where sterilizations and natural family planning are most prevalent and lowest at BHS and industry clinics where barrier methods predominate. D. The Facility Survey Figures 1, 2, 3 and Table 3, about here The facility survey was conducted in the summer and early fall of As noted above, the basic data requirements for estimating service cost include the quantities of the various inputs used in the production of the services, the prices of these inputs, and the quantity of the service produced. Because virtually all of the facilities in our sample produced other services as well as family planning and typically inputs were shared across these multiple services, data to serve as the basis for allocating resources across activities were also required. The first step in the survey process was to identify whether or not there was a sub-unit or department within the facility where the resources used to provide family planning services were located. In most small clinics there were no relevant subdivisions, however in some larger facilities the survey could be limited to the outpatient department, the maternal 21 Some of the provincial clinics can be viewed as a political response to the religious debate on the morality of artificial methods of contraception. These clinics provide only natural family planning. The Provincial government s position has also had effects on the provision of services in other public health facilities. -12-

16 and child health department, or, in rare instances, the family planning department. Once the organizational home of family planning was located, the individual most responsible for managing family planning was identified. In most cases, this was the facility manager; but in facilities that were departmentalized, it was the department manager. Resources were then inventoried. For staff resources, the manager was queried about the number of staff members in various job categories (doctors, nurses, midwives, and auxiliary personnel). For each type of personnel, the total hours per month they were employed at the facility, and the typical wage was also supplied by the manager. The staff inventory was divided into to groups on the questionnaires. The first group was staff members and hours for staff that provided exclusively family planning services. 22 The second group covered staff members who provided family planning services but also provided other services for patients. For this group the manager was asked to estimate the percentage of their time these staff members spent on family planning service provision. Other inputs used in the production of family planning services were inventoried by the survey teams through direct observation and examination of facility records. In virtually all but the private facilities, contraceptives were supplied through Family Planning Logistics Management (FPLM) working out of the DOH with donated supplies. Each facility had records showing the month s starting count of each type of contraceptive, the number received during the month, the number distributed during the month, and the final balance. 23 The interview teams recorded the data from these records which served both as the measure of the quantity of inputs used, and as the basis for calculating the CYP production at the facility. Facility patient logs were abstracted to collect data on patient visits for family planning and MCH-RH services. 24 The survey teams inventoried equipment at the facility. Equipment was divided into two groups based on the managers responses: to queries about whether specific family planning. 22 Only about six percent of the facilities reported having any staff members whose sole job responsibility was to 23 The survey team independently inventoried FP commodities at the facility (also checking for date expired commodities) and found an extremely close correspondence between the actual counts and the records. 24 The survey questionnaires were designed based on the DOH programs for MCH and other basic health programs. The services statistics for most government facilities were based on the DOH Field Health Service Information System (FHSIS). Where the responses could not be obtained directly from the survey, the FHSIS reports were used if available. Hospitals often kept aggregate patient statistics in a different form and for some hospitals the Bureau of Licensing and Regulation (BLR) hospital reports were used as the primary source of patient statistics. A complete description of the construction of the service statistics is available from the authors. -13-

17 equipment was used exclusively for providing family planning services or whether it was used jointly to provide FP and other services. The survey team measured the dimensions of the facility s physical space and recorded information about the building s conditions, materials used in construction, number of rooms and so forth. Again, physical space was categorized into space used exclusively for family planning and space used for FP but also used to provide other services. 25 The managers were also asked detailed set of questions in which, for each type of family planning service provided, the typical staff time used for the procedures and the equipment, supplies and facility space that would be used was ascertained. This scenario analysis followed a new and continuing user of each method through a complete visit at the facility and, as will be described below, was used to estimate cost on a method by method basis. Finally, the manager was asked questions on the facility s budget and expenditures covering all cost categories including building rent, general medical and other supplies, utilities, and so forth. E. Discussion of Survey Several general comments on the survey method should be made at this point. First, the data requirements for estimating the cost of service provision are large. A very large number of individual inputs are used in the production of any health service and choices have to be made with respect to the level of detail that will be used in enumerating the inputs for which data will be collected. Second, the complexity of the data collection can result in missing values for some data elements in the survey. Because, as will be described below, each cost calculation requires the use of a large number of variables, missing one small piece data can make it impossible to complete the cost calculation for a facility unless various techniques are used to backfill missing values. 26 Third, the quality of the data generated through the survey will depend crucially on the respondents knowledge. In the course of the survey we found that there were many things that could be known and measured with considerable accuracy at the facility level and others that could not. At one extreme we found that very complete and up to date records on 25 Many of the facilities had equipment or physical space that was reported in the survey as being used exclusively for FP. About 30 percent of the facilities reported building space exclusively used for FP, and 60 percent reported equipment exclusively used for FP. 26 In the interest of brevity, much of this tedious detail has been omitted from this paper. A version with a much more complete, and tedious, description of the calculations and data is available from the author. -14-

18 conceptive distribution were available at facilities that received contraceptive supplies from FPLM. At the other extreme, direct inquiry as to the total amount that was spent by the facility on various types of inputs, supplies, and services yield very few complete responses. While the quantity of the various resources employed at the facility could be observed, the prices or cost of the resources typically could not. As will be described in more detail below, it was necessary to go to sources outside of the facility to obtain prices of contraceptives, equipment, and other resources. Finally, the most troublesome issue is the allocation of cost to specific services. Our interest here is the estimation of cost of family planning service provision. Virtually all facilities provided other services as well and virtually all facilities used shared resources. It is fairly easy to get numbers that reflect the total of staff cost at a facility, and the total number of FP visits and visits for other services that were performed at that facility, however it is generally not easy to get information on what proportion of staff time (and thus staff cost) was devoted to FP and other services. Such information is however crucial if one wishes to cost family planning services. A number of different approaches have been tried including direct observation of practice, staff kept logs, and survey of staff or clinic supervisors. Considerable debate exists of the relative costs and benefits of the alternative approaches. IV. Estimation of Cost Family planning service cost estimates are presented on a per CYP basis. Four components of cost were considered and estimated. These included per CYP direct labor cost, per CYP contraceptive cost, equipment cost per CYP and facility space cost per CYP. Two components of cost are not explicitly estimated because of data and conceptual problems. These are the per CYP cost of other supplies and consumables and the per CYP cost of general overhead and administration. The general approach to estimating cost in this study is to use the survey data to quantify the physical amounts of resources (staff time, quantities of contraceptives and equipment, and the amount of physical space) used in producing family planning visits of various types. The quantities of input resources are then costed using price and wage data to produce a cost estimate for each type of visit which can be aggregated using the facility s composition of visit types to produce a cost per CYP estimate of cost. The details of the actual calculations are tedious, so what follows is a summary of how each cost component was calculated from the data. A complete -15-

19 description of the calculations can be found in Stewart, J.F., D.K. Guilkey, A.N. Herrin, and R.H. Racelis, Labor costs: Average Labor Cost: The average labor cost per CYP of family planning output was calculated by taking the monthly hours of each type of staff time used in the provision of FP service 27, multiplying by the wage rate, summing across staff types and dividing the total by the facility s monthly CYP output. The staff hours devoted to family planning is the sum of staff hours for staff that exclusively provided FP and the allocated portion of hours of staff members who provided FP service and other services. Staff wage rates were obtained from the survey data. Because there were a fairly large number of missing values for wage rates, missing wages were predicted using regression analysis. Wage rates were allowed to differ based on the operating authority employing the staff, the type of facility where the staff was employed (hospital, BHS, or other clinic), whether the clinic was in an urban or rural locations, and the Province in which the facility was located. 28 The average labor cost per CYP is best interpreted as what is actually spent by the facility on staff to produce a CYP given the current composition of staff employed, the division of their time between family planning and other activities, wage rates paid by the facility, and the quantity and method composition of family planning output actually produced by the facility. Contraceptive Costs: The calculation of the contraceptive cost per CYP was made based on the distribution of contraceptives, the method mix of the clinic, the CYP conversion factors shown in Table 1, and the price of the contraceptive. The total expenditure on contraceptive distributed in the month was then divided by the output of CYPs for the facility for the month. All contraceptives were priced using the value of USAID contraceptives. 29 Because common 27 Staff time was accumulated separately for nine different categories of staff (doctors, nurses, midwives, institutional workers, community volunteers, other medical staff, traditional birth attendants, administrative staff, and auxiliary staff). 28 We do not report the results of all nine wage prediction regressions here. As a general description, the regression produced a good fit of the data and would support several generalizations about salary structure. Namely wages were generally higher for the National Capital Regions, wages for employees at FPOP and IMCH-SDI facilities are lower for most staff categories than at government facilities, and wages at industrial sponsored clinics on average tended to be higher. 29 The USAID price list did not have spermicides so an IPPF price list was used to price these commodities. While the vast majority of the contraceptives used by government facilities and NGO facilities were donated by USAID, the sources for private clinics varied and the USAID prices may not accurately reflect the cost of contraceptives for these clinics. Government clinics typically reported that 95% or more of their contraceptive supplies came through USAID channels. IMCH- -16-

20 contraceptive prices and common conversion to CYP factors are used for all clinics in the sample, cross facility differences in contraceptive cost only reflect differences in method mix. 30 Equipment, Building, and other Cost Components: Capital Inputs: The production of services requires, in addition to labor and consumable supplies, physical space (building space) and equipment. For some types of inputs, the use value of the input in producing the service is fully consumed at the time the service is provided. For example, when a packet of pills is distributed to a client, it can no longer be used to provide additional services at the facility. The same would be true for staff time used in a visit or other consumable supplies. For other types of resources, the input s services are provided over an extended period of time. That is, the input provides a flow of service over time. An example would be the flow of services provided by building space or a piece of equipment. The appropriate costing concept for such inputs would be a rental cost of capital, the allocation of the cost of the resource over time. The appropriate calculation of a rental cost would include the actual cost of the input, the service life of the input (the length of time over which the input would provide service), depreciation (an allowance for the loss of use value of the input over time because of its age, obsolescence or other factors), and the cost of money capital (basically an interest rate reflecting the cost of using funds for a capital purchase). Most of these dimensions are conceptually and practically difficult to measure. For this study we will take a very simplified view of these capital costs. For the cost of equipment we will value the input at replacement cost. This is the cost of purchasing the equipment at current prices. There are two rationales for this choice. The first is practical. It is fairly easy to obtain current price lists for the various types of equipment used in service provision from the various international agencies and donor groups. It would be virtually impossible to obtain information on the prices that were actually paid for the equipment and facilities currently in use (original SDI clinics reported roughly 90% from USAID, 5% from IPPF and 5% from the private market (mostly IUD clients who purchase their own IUDs to be inserted at an IMCH-SDI facility. Other NGOs reported USAID and IPPF were the source of 85% of their contraceptive supplies with the remainder being purchased on the open market. Industry clinics had a similar source pattern but with a slightly larger percentage coming from market sources. Two thirds of the contraceptives used in private clinics came from market sources. 30 Given the USAID contraceptives prices, the CYP conversion rates that appear in Table 3, and an assumed exchange rate of 26 pesos to the dollar, the contraceptive costs are P70 for pill, P151 for condoms, P8.40 for IUD, P87.88 for injection, and P31 for foaming tablets. -17-