Policy, Research, and External Affairs WORKING PAPERS Population, Health, and Nutrition Population and Human Resources Department The World Bank March 1991 WPS 628 The Indonesian Family Planning Program An Economic Perspective Dov Chernichovsky Henry Pardoko David De Leeuw Pudjo Rahardjo and Charles Lerman The intrauterine device (IUD) and (less so) the injectable are relatively cost-effective methods of contraception that could probably improve contraceptive prevalence. They both require capital investment and trained medical manpower- which are beyond the means and jurisdiction of Indonesia's family plan- ning agency but would probably pay off, especially in improved health care. Thc Policy, Resarch. and tExtenial Affairs `'mplen distrhuite.s PREW 1 orking Papers to dicseinnate tic fiidlings ofwi trk in progrcss anid to encourage tie exchange of idcs among liarik sialf and all othcrs increslcd in development issuc 'I hese papwrs carry the names of the authors, reflcct onl) their sicws, dnd shiuld hc used and (tied accordingly 'Ihc findings, niterprtaciaons. and coniclusions are the authors'ow ti. I hev shnul. not he a:tribu:udil viciie Wcrld liank. us Board of D)irectors, L's managanent, or an of its memrier countries Policy, Research, and External Affairs Population, Health, and Nutrition WPS 628 'rhis paper is a product of thie Population, I lcaldi, and Nutrition Division, Population and 1 luman Resources Departmcnto lh'e World Bank, incooperation wil ilihc Indonesian National Family Planning Coordinating Board, anid tlic Miistry of Forcign Affairsisotthe RoyalGovernincintol the Nctiicrlanids. It s part ol a lalrcr efIlort iin PREtocxamilnie thceconomicsol familyplanning. Copicsare avavilablef ree I Vom thte World BaBllk, 1818 11 Strece NW, Washington DC 20)433. Plcase contact Otilia Nadora, room S6-065. exteisioil 31')1 (185 pages). A comparative analysis of three provinces in so the most cost-efiective metlhods are also tihc Indonesia indicat's that Ihc IUD and, to less most efficicnt - probably also in terms of' extent, the injectable, arc .icthods that, if demographic impact. Difkfrcinces in thc mean available, would probably be usWd and would age of users foi the IUD (32.5), pill (30), and contribute to high contraceptive prevalence. injectable (29) are slighlt -- so reproductive potential and risk of pregniancy are about equal Moreover, the IlD appears to be relatively among different user groups. cost-ei'fective. Clearly, altering the delivery system --- But the IUD (and to less extent thc inijcct- particularly in favor of miethods that recluirc able) requires capitai investment and trained mcdical facilities and staff- requires invest- medical manpower (which are beyond the mcans mcnt in facility, stafl, anld the cost of initiating a and control of Indoniesiai's National Family new method. This merits a detailed cost-benefit Planning Coordinating Board, BKKBN). analysis, as the data strongly suggest that suci investments might pay olf, especially because Thc relative delivery cost of differenlt hcy would also improve medic;al care. methods are inversel) relatcd to their efficacy - Thc lIRE Working Paper Serics disseminates the Findings of work under way in the Bank's Policy, Research. and External Affairs Conmplex. An ohjectiveofthc series is to gctthcse findings out quitckly\. even if presentatiols are lcsc tian fully lpolshcd. Thel findin, I, mterilet.olown and colclthlolln iI Olie- pa1etr Jo n;l lt al j l ropresenit official Rank polic\ ltr(iticO(l( 1\ the iR RE Di sse in in Ition (Ce nter PREFACE ...... . . . . . . . .. . . ..... . . . . . . * . . 1. IIIODUCrION . . . . . . . . .... . . . . . . .... . . . . . . .... . . .. 4 2. POPULATION AND FGLY PLAIMNNNG IN IDNESIA - AN OVERVTE . I. . . . . . . 9 2.1. PopulationSize, Growth, and Distribution 2.2. The Family Plannin Program 2.3. Program Cbjectives and Strateies 2.4. New TrencL 2.5. summary 3. IHE SY .................................. . 19 3.1. Introduction 3.2. Political Economy of Family Planning Delivery 3.3. Analytic Franework 3.4. Study Areas 3.5. Data 3.7. Summary 4. PROGRAM Li ..... . . . . . . . . ........ . ...... . . . . . 42 4.1. Introduction 4.2. Program Output Data 4.3. Program Statistics and Survey Data 4.4. Levels of Contraceptive Use 4.5. Contraceptive Use by Method 4.6. New Users 4.6. Summary 5. PGRAM DESIGN AND DELIVERY SYSTM . . . . . . . . . . .3......... . 55 5.1. Introduction 5.2. Modes of Delivery 5.3. Significance of Modes of Delivery 5.4. Income-Generating Schemes 5.5. Summary 6. FIELD PERSONNEL AND OPERATIONS . . . . . ... ... .......... . 66 6.1. Introduction 6.2. Field Personnel 6.3. Field Worker Allocation 6.4. Worker Characteristics 6.5. Training and Experience 6.6. Workers' Activities 6.7. Summary 7. PROGRAM COST . . . . . . . . . . . . . . . . . . . . .. .. . . . . e . 83 7. 1. Introduction 7.2. Lator cost 7.3. Capit 1 Cost: Types and Source of Funding 7.4. CDs& of Contraceptives and Other Supplies and Utilities 7.5. Ibtal Cost of Resources and Sources of Funding 7.6. Sumnary 8. PR)AM COSTaETENESS . . . . . . . . . . . . . . . . . . . . . . . 108 8. 1. Introduction 8.2. Program Cost Per User 8.3. Adjustment of User Cost for Altenative Prevalence Estimtes 8.4. Summary 9. WORKER PRODUCIIVIIY AND EFFICINCY OF FIELD OPERATIONS . . . . . . . . . 122 9.1. Introduction 9.2. Field Worker and Supervisor Productivity: A Model and Hypotheses 9.3. Field Worker Productivity Estimates 9.4. Field Workers' Time Allocation 9.5. Worker Effeativeness 9.6. Supervisor Productivity 9.7. Worker Pay and Efficiency 9.8. Summnary 10. PRCGRM EFFICIENCY . . . . . . . . . . . . . . . . . . . . . ..... . 153 10.1. Introduction 10.2. The Consumer's Perspective 10.3. Efficiency of Method and Mode of Delivery 10.4. Conclusions 11. SUiMARY AND CONCLUSIONS . . . . . . . . . . . . . . ......... . 166 11.1. Introduction 11. 2. Summary 11.3. Implications 11.4. Future Research BIBLIOGRAPIY . . . . . . . . . . . . . . . . . . . ... . 183 TABLES Table 2.1: Estimates of Indonesian Fertility Rates . . . . . . . . . . . . 11 Table 3.1: Iand and Population Sizes . . . . ... 33 Table 3.2: Demographic Features. . . . . . . . . . . . . . . . . . . . . . . 35 Table 3.3: Urban Residence . . . . . . . . . . . . . . . . . ...... . 36 Table 4.1: Mean Number of users Per Subdistrict and First Order Correlation Coefficients Between Different Sources of Program Output,bySourceofData . . . . . . . . . . . . 45 Table 4.2: comparison of Cnttraceptive Prevalence and Method Mix Rates for the Three Provincial Study Areas Using 1985 BKKBN Monthly Service Statistics, the 1985 Annual Report ),U the 1985 Intercensal Survey, and the 1987 Dmxegraphic andHeal.thSurvey . . . . . . . . . . . . . . .. 47 Table 4.3: Eligible Couples, Their Status and Distribution of Family Planning Methods, by Method . . . . . . . . . . . . . . . . . 49 Table 4.4: nX-tribution of New Users .iy Method . . . . . . . . . . . . . . 53 Table 5.1: lacility Distribution . . . . . . . . . . . . . 58 Table 5.2: Contraceptives Distrikuted and Distri.lAtion Channel. . . . . . . 61 Table 5.3: Availability of Inccme-Generating Sch30es . . . . . . 64 Table 6.1: ELtXs, Villages, Area, and Med. Staff per Field Worker . . . . . 71 Table 6.2: Determinants of Number of Field Workers, Linear Regression Results; No. of Field Workers in Sub-district as Dependent Variable . . . . . . . 71 Tlakle 6.3: Worker Characteristics a, by Type . . . . . . . . . . 73 Table 7.1: Labor Costs by Type of Staff in Rupiah per ELMX per mnnth . . . 85 Table 7.2: Earnings by Worker Type and Type of Inomne . . . . . . . . . 88 Table 7.3: Distrikution of Earnings by Type of Staff and Source of Furdirig . . . . . . . . . . . . . . 91 Table 7,4: Labor Cost, by Source of Funding in Rupiah per ELOO per month . 92 Table 7.5: Zero-Order Correlation Coefficients of Sub-district Level Labor Costs per EXM by Source of F iding . . . . . . 93 Table 7.6: Capital Costs by Capital Type in Rupiah per ELM per month . . . 97 Table 7.7: Capital Costs by Facility Type in Rupiah per EI m per month. . . 97 Table 7.8: Capital Costs by Source of Funding in Rupiah per ELrJ per month. 99 Table 7.9: Disposable Costs by Type of Contraceptive in Rupiah per ELLfl per month ... ....... . . ... . *. 101 Table 7.10: Cost of Disposables, by Mode of Delivery in Rupiah per EL.O per month ... . . . . .. o . . . .. . . 101 Table 7.11: Month-ly Total Program Costs in Rupiah per ELC0 per month . . o . 104 Table 7.12: Monthly Total Costs, by Source of Funding in Rupiah per ELOO per month . .. . . . . . . . 105 Table 7.13: Ratio of Total Cost to BKN's Cost . . . . . . . . . .106 Table 8.1: Program Cost per User by Type of Cost, in Rupiah per month . . . 110 Table 8.2: Unit User Cost by Region for Similar Output Levels . . . . . . . 113 Table 8.3: Marginal Costs per User (all cost included) . . . . . . . . . . 115 Table 8.4: Ratios of Prevalence Rates from Surveys to BKKBN Annual Reports 118 Tlable 8.5: Adjustment of User Costs per Alternative Prevalence Rate Estimate (Ratio: Tangerang = 1.00) 119 Table 8.6: Adjustment of Marginal Cost per Alternative Prevalence Estimate (Ratio: Tangerang = 1.00) 119 Table 9.1: List of Variables and Reference to Analytic Framework . . 134-135 Table 9.2: Field Worker Productivity Estinates, No. of Users or Log of Number of Users as Dependent Variable . . . . . . . . . . . . . 137 Table 9.3: Log of Number of Users as Dependent Variable, Regression Coefficients ................ . 140 Table 9.4: Field Workers' Allocation of Time to Delivery as Dependent Variable, Regression Coefficients . . . . . . . . . 142 Table 9.5: Logarithm of User Rate as Dependent Variable, Regression Coefficients . ..... . .* ..... . 144 Table 9.6: Determinants of Number of Field Workers by Supervisor, Regression Coefficients . . . . . . . . . . . . . . . . . . . . 147 Table 9.7: Worker Salary Determinants for BKMN Field Workers Log of Salary Measures as Dependent Variables, Regression Ooefficients ................ . 150 Table 10.1: 1UD, Pill, Injectable and Total Prevalence Rates Regression Coefficients ..... . 158 Table 10.2: Estimated Unit Cost of Method, by Type of Cost . . . . . . . . . (in Rps. per Month per User) 160 Table 10.3.: Cost of Method per Month by Region . . . . . . . . . . . . . . . 163 FIGURES Figure 3.1: location of Study Areas . . . . . . . . . ..... . . . . . . . . 31 Figure 3.2: Study Areas . . . . . . ....... .... . 32 Figure 4.1: Use and Method Mix in Tangerang . . . . . . . . . . . . . . . . 50 Figure 4.2: Use and Method Mix in Kulon Progo and Bantul . . . . . . . . . . 50 Figure 4.3: Use and Method Mix in South Kalimantan . . . . . . . . . 51 Figure 5.1: ELCOs Served by Facility (Logarithmic Scale) . . . . . . . . . . 57 Figure 6.1: Family Planning Personnel .... . . . . . . 69 Figure 6.2: Activities of Field Workers and SuLpervisors . . . . . . . . . . 79 Figure 6.3: Activities of Physicians . . . . . . . . . . . . . . . . . . . 79 Figure 6.4: Activities of Non-Physician Medical Staff . . . . . . . . . . . 80 Figure 7.1: labor Costs by Type of Staff .... . . . . . . . . . . . . . . 86 Figure 7.2: Capital Costs per ELCO, by Capital Type . . . . . . . . . . . . 96 Figure 7.3: Capital Costs by Facility . . . . . . . . . . . . . . . . . . 98 Figure 7.4: Capital Costs per FJIX, by Source of Furnding . . . . . . . . . . 99 Figure 7.5: Total Program Cost ..................... . 104 Figure 7.6: Monthly Total Costs, by Source of Funding . . . . . . . . . . 105 Figure 8.1: Predicted Average Costs per User . . .. .... . 117 Figure 9.1: Field Worker Productivity: Interaction between Supply and Denand . . . . . . . . . . .. . 130 Figure 10.1:Percent Quarterly Contraceptive Method Mix, DI Yogyakarta, 1981-1987 157 Figure 10.2:Percent Quarterly Contraceptive Method Mix, West Java, 1981-1987 157 Figure 10.3:Percent Quarterly Contraceptive Method Mix, South Kalimantan, 1981-1987 - 157 The views expressed herein are those of the authors and should not be attributed to the World Bank, the Dutch or Indonesian Governments or their affiliated organizations. EDUtJWIVE SUMMARY Indonesia has achieved one of the most impressive reoords in fertility reduction over :he past two decades. The country's total fertility rate has declined from an estimated 5.5 in 1967 - 1970 to 3.4 in 1987. Population growth has been estimated at 2.1 percent during the eighties. Many observers credit Indonesia's National Family Planning Coordinating Board (known by its acronym, BKKBN) as being instrumental in this fertility reductior. and slowdown of populatior. growth. BKKBN is a public sector organization responsible for planning and coordination of almost all family planning activities in Indonesia. The study objective is to provide UBON and the Government of Indonesia with data that can help improve the cost-effectiveness of family planming delivery in Indonesia. The study examines resource allocation, cost, funding institutions, and output of the program at grassroots level in selected regencies in three provinces: West Java, the Special District of Yogyakarta, and South Kal mrantan. It is based on data about the prcgram's field operations collected during November 1986 - March 1987, and routine service statistics of BKKBN. The six regencies participating in the study are case studies; they are not meant to represent the entire National Family Planming Program. Tangerang, in West Java is a densely-populated in-migration area next to the capital Jakarta. It has a high ratio of population to health facilities and other resources for family planming delivery. The Yogyakarta study areas. Kulon Progo and Bantul, are more traditional agricultural areas, close to a slow-grcring urban center. They have a relatively strong social infrastructure and medical resources for family planming. These densely populated areas contrast with the study areas of South Kalimantan, Banjar, Baritu Kuala, and 'rapin, which are sparsely populated and have more resources per capita for gamily planning, but which are spread over large areas and are not easily accessible. BEKBN operates in this diverse and quickly evolving social environment. Its activ.' ty is conditioned by four constraints: a) national administrative regulations, b) availability of medical and community resources, c) consumer demand, and d) BKKBN's budget. Regardless of area size, population, or other features which might influence resource allocation, BKKBN only has one family planning supervisor per administrative subdistrict. Medical and commumity resource constraints wti.ch influence the nature of the program are to a substantial degree external to BKK3N, which uses medical infrastructure controlled by the Ministry of Health. While BKKBN can advise about the supply, distribution, and administration of these resources, it exercises minimal control over them. These resources and consumer demand shape the nature of the program, exhibited in contraceptive method mix. Availability of nedical infrastructure per eligible couple (ELM) varies ii greatly between regions. One health center is available for anout 9,500 eligible couples in Tangerang. This is twelve times the ratio of population to health facilities in the Kulon Progo regency of DI Yogyakarta, where there is one health center for each ELOO and fifty-five times that of sparsely- populated Tapin province of South Kalimantan where there is one health center for eadh ELM. Medical personnel are also unevenly distributed. Different regions possess very different medical resources w. ich are available for famiiy planning. BKKBN must adopt modes of deliveyy and method mix accordingly. Considerable variations are also observed across regions in the allocation of BKKBN staff. Tangerang's field workers are assigned to above 1,400 eligible couples, compared with about 1,200 in Yogyakarta, and 900 in South Kalimantan. Field workers in Tangerang handle far larger populations with fewer resources per ELCO than their counterparts in the other areas. Tangerang and Yogyakarta workers benefit, however, from a relatively high population density. At the same time, within each of these three provinces, allocation of field workers varies systenatically with number of ELCOs, villages and size of catchment area. This is less She case in DI Yogyakarta and more so in Tangerang, where resources are relatively stretched. BKKBN field workers oversee village family planning volunteer activities in the community. However, the volunteers who play a major role in outreach activities, primarily via distribution of pills and condoms, are not entirely answerable to BKKBN, in part because of other responsibilities they share in the community. BKKBN staff spend most of their time searching for new acceptors and promoting family planning. The staff report working "by the book", which gives rise to the hypothesis that on the whole they may be underutilized. they do not seem to be under obvious pressure to respond to local variations in need. There is a positive correlation across study areas between levels of all types of resources provision per eligible couple. The inplication is that availability of medical infrastructure may be a key -ctor in the developoent and nature of the program.. Resource availability, and to an extent age of program, are reflected in prevalence rates and method mix. According to BKKBN's serAice statistics, both Tangerang and the South Kalimantan regencies have contraceptive prevalence rates of approximately 60 percent. Injectables, however, predominate in Tangerang and pills in South Kalimantan. The Yogyakarta regencies have the highest prevalence rates, around 80 percent, and the IUD is the mcst common method. Data on prevalence rates used in this study (from BKKBN sources) are higher than rates available from survey data; however, both types of data lead to the same implications because data on method mix is highly consistent amongst different sources. The value of all resources allocated to family planning delivery ranges from about 270 Rps. (about $0.18) per month per EO in densely-populated Tangerang to 630 Rps. in sparsely-populated South Kalimantan. BKKBN mnarkges to nobilize an additional one rupiah from other government and ccmmunity agencies, for each rupiah it invests in field operations. That is, BKKIN bears about 50 .iii percent of family planning delivery costs, the Mini--try of Health about 40 percent, and the conmiLnity the remairng 10 percerc. BKKBN bears less of the cost burden in areas where there is a medical infrastructure, becauise it does not pay full labor value for medical personnel and can rely on-longer-lasting and less costly methods, notably the IUD. It is noteworthy that physicians and other medical staff, who report spending roughly 20 percent of their time on family planning activities, receive only 8 percent of their government income from BKKBN. It is costliest, 900 Rps. per month, to maintain an average user in South Kalimantan. It is hali that cost in Tangerang, the least costly area. The costs are higher and regional discrepancies wider when lower prevalence data from surveys are used. In all regions, especially those of South Kalimantan and Yogyakarta, scale of operations is a crucial variable influencing user cost. User cost is lower where operational units (field personnel) cover larger population and higher numbers of users. When scale of operation is controlled for and variable and marginal user costs are examined, D.I. Yogyakarta has the least costly type of program. The IUD is the relatively more cost-effective method in the long run, followed by the injectable. These methods are probably also the most efficient when their efficacy, compliance and potential demographic impact are considered. The mean age of users of theIIUD, pill, and injectable for Java and Bali is 32, 30, and 29 respectively, differences meaning almost identical risk of pregnancy. Moreover, comparative analysis of the three provinces indicates that the IUD and the injectable, are methods which, if made available, might be popular and contribute to high prevalence. Major gains in cost-effectiveness can therefore be brought about primarily through altering contraceptive method mix in favor of the mare permanent methods - the IUD and the injectable. While IW delivery incurs low recurrent cost in labor and supplies, this method demands a relatively high investment in infrastructure, personnel, and start-up costs which, of course, are subject to substantial economies of scale. EKKMN, even if it manages supplies out of its own budget, cannot induce high IU rates in any substantial manner without the required medical infrastructure and personnel. BKEK has no direct control over these resources whatever consumer preferences may dictate. The structure of user cost underscores the political economy of the family planning program. The major user cost component of the program, borne by BKXBN, is supplies, about two-thirds of EKKBN's cost in its field operations. This cost element is almost entirely influenced by method mix which, in turn, is to a substantial degree set by availability of medical infrastructure and personnel, and consumer demand. This means that BKKBN, the responsible agency has rather limited scope or latitude in terms of its own cwst control, let alone the cost of the entire program. A long-term cost-effectiveness strategy should therefore be cc'nsidered within the context of the political econmy of the Indonesian health and family planning system. It must combine investmnt in infrastructure, adoption of new family planning technology and mcdes of delivery, and influencirg consumer demand. This can be done by the Goverrmient. Even if coysumers pay the full iv cost of IUD use, including recurrent capital cost, the private sector cannot be expected to finance at the outset the investment required to support IrU. The proposition that the government supports clinically-based methods, at least in financing the investment, would be less acceptable on economic grounds in sparsely-populated areas such as South Kalimantan. Ihlere the injectable would be appropriate in the long-run as a relatively efficient nethod. This would require government support for training personnel. These argumentsr do not suggest that the government slould rpfrain from supporting non-clinical methods, at least in the short run, acoording to conditions. For its part, BKKBN might improve the cost-effectiveness of its operations and incur savings through modification of its target-setting policy, allocation of field workers, and improvement of workers' time use. 2KB1N does control its labor cost, about one-fifth of total recurrent cost of the program's field operations, or one-third of BMNW's own cost. The effects of apparent target- setting are observed throughout the analysis; there is a very high correlation between numbers of users and numbers of ELCOs across observations. This confounds sound evaluation of the field operations that can lead to better management of BKKBN's field resources. Taken at face value, the data strongly suggest that the population and area assigned to the average field worker, colld be extended since workers seem to operate undez concitions of excess capacity; their catchent areas can be raised with no loss in production. This finding implies that present prevalence levels can be maintained, on the average, with fewer field staff, or more feasibly, that new field workers need not be added even though populations of eligible couples grow. In areas where the program is well-established and not too dependent on outreach activity, as in Yogyakarta, savings may be gained by reducing BKKBN's field staff and limiting their activity to information, education, and communication (ImC.) Inccme-generating schemes serve areas with high prevalence rates, like Yogyakarta. Such a policy is consistent with the idea of rewards rather than inducement. There may be scope to transfer these funds as incentives in low prevalence areas, especially if there is excess production capacity and field worker productivity could be highest. The study findings can help deal with the issue of shifting some of the finzncial borden of the program tcr -:he community, as implied by the KB Mandir or privatization idea adopted by C-KBN. The ccmmunity bears about one-tenth of total delivery costs. mTis share increases in outreach activity. Ihe pill and condoms are delivered almost exclusively through volunteer outlets. The share of the burden on the community is somewhat less in the case of the injectable because of the involvement of medical personnel. The contribution of the community is least in the case of delivery through medical facilities, such as in the case of the IUD or even the pill in South Kalimantan. The estimated monthly cost to service a pill user ranges from 450 - 600 Rps. Most labor and capital cost are borne by the community. The lion's share in cost of pill delivery, however, lies in its supply. Supplies' cost account for some 60'to 80 percent of total cost. V Even if cl-ients pay under a private system the full cost of the pill, which is the major method to be affected under the IK Mandi-ri scheme, the net cost recovery fram the cotmnunity would be less tqan 100 percent in view of the community's current contribution. Any subsidy, most likely of supplies, would even further reduce net cost recovery from the commumity. A subsidy in the range of 10-30 percent, by conservative estimates, might not amount to a ret cost recoverv because roughly this cost is currently borne by the cammumity. As the program now stands, at lease, would be less penalized by a shift toward payment for pills than South Kalimanitan, with its high pill use. Since DI Yogyakarta alread&y enjoys advantages (particularly with regard to health infrastructure), assessing fees for pills, rather than IUDs, would increase its privileges even further. Charging for pills and not IUDs could aggravate regional disparities. The cost recovery potential of IUD is higher than that of the pill. First, the community's contribution in IUD delivery is miiidmal: any charge for IUD delivery would therefore entail a net shift of burden from the government to the community. Second, the IUD is used by presumably better-off people who live close to health centers, and therefore may be able and willing to pay more than others. There is an additional reason: IUD fees are collected in clinics and may be applied towards financing oommunity health facilities, whereas the pill is largely sold by non-medical outlets. No single universal cost recovery strategy may be appropriate across Inidonesia. The study clearly opens a broad research agenda. While the study is based on its own data collection efforts, it is clear that the analysis oould have been based on a slightly rpdified data collection system other than BKKBN's current system, and an appropriate analytic infrastructure. The issues and inplied programmatic changes all merit more focused research. Operations research into the allocation and activities of field workers should be high on the agenda. A crucial element clearly missing in the data is the consumer's perspective. No strategy and prograrmatic change can be adequately assessed without regard to consumer response, especially in the diverse and fast changing demographic, economic and cultural environment of Indonesia. PREFACE Cost-effectiveness in family planning delivery has become especially important in recent years in view of the slowed growth of public resources for family planning, at a time when the need for it remains pressing. This study is part of an operational research work program in family planning initiated by the World Bank. Urder this program, a quantitative ecoric.mic appit-ach with a cost-effectiveness orientation has been developed to assist policy-makers and program managers to learn from their own experience about resource allocation, cost, and Linanc , and how these relate to program objectives. The present work program inclided development of guidelines and studies of two distinct family planning programs: the Indonesian National Family Planning Prograni arxd the Colcobian Profamilia Program. This stuldy is of the Indonesian program. The Indonesian Family Planning Program is well-established. The National Fami'Ly Coordinating Board (BKKeN) is a public sector organization responsible for planning and coordination of almost all family planning activities in Indonesia. The study cbjective is to provide BKMW and the Govi-naent of Indonesia with techniques and data which can help inprove the efficiency of family planning delivery in Indonesia. This study represents a collaborative effort of BKFN, the World Bank, -2- and the Royal Goverrment of the Netherlands, all of whom funded the study. We thank BKKMN's management, which ensured full support for.the study; Dr. Haryono Suyono, le Chairman of BKKBN, who took personal interest in the study, and his staff, Dr. Srihartati P. Pandi, Dr. soetedjo M'beljodihardjo, Dr. Sageng Waluyo, Dr. Sahala Pandjaitan, Mr. Gary Lewis, and especially Dr. J. Malynoux for initial cost analyses. We also thank Dr. Budi Soeradji of BAPPENAS. We would also like to thank the administrative coordinator, Sen,usia Bantas. We thank the staff of the Ministry of Foreign Affairs of the Payal Goverrment of the Netherlands: Mrs. Bergsma, Messrs. Doll, Vehmeyer, Van Rinsum, Mollema; and World Bank staff: N. Birdsall, and Messrs. I. Hhssain, A. Hamilton, M. Chocksi, K. Nordlander, B. Herz, A. Williams, B. CarlsGn, D. de Ferranti, and especially Ms. S. Cochrane, who did not spare time and trouble to review legthy drafts, making most instructive cauemtus. Special thanks go also to Professor G. Jones who reviewed the monograph and provided additional useful conments. Special thanks are due to the field coordinator of the study, Mr. Lalu Soedermadji, the staff at BKKBN Head Quarters and the many people participating in the study at the field level; provincial and regency heads of EKKBN, regional coordinators, field workexs, supervisors, medical staff, village heads, and volunteers in the Indonesian Family Planning Program. This mmnograph is a tribute to Dr. Henry Pardoko, an honorable aid - 3 - dedicated Indoneian, who passed away while it was under revision. -4- 1. rINaO=N Indonesia has achieved one of the nest impressive records in fertility reduction over the past two decades. qhe country's total fertility rate has declined frmm an estimated 5.5 in 1967 - 1970 to between 3.3 - 3.7 in 1985 (Hull and Dasvarma 1988; Suyono and Shutt 1988). Population grcroth has been estimated at 2.1 percent during the eighties (Prescatt £ al. 1986). Many observers credit Indonesia's National Family Plawning Coordinating Board (known by its acronym, BKKEN) as instrumental in this fertility reduction and slowdown of population growth (Hull ek al. 1977; Sinquefield and Sungkono 1979; Chernichovsky and Meesook 1981). Key reasons for the success of CKBNM's program include its commumity-based distribution system, involvEmnt of local cmmunity leaders to prcmote family plannig, administrative decentralization, and an effective reportinr, recording, and monitoring system. About twenty years since its initiation as a national program, BRe other operation would still reduce overall unit caost. Seoond, when it operates at too large a scale; introducing a new operation, e.g. a new clinic, wuld reduce such unit cost. In the second case, this my require new (long-term) investments and reorganization. Identification of cost, especially in relation to labor productivity is therefore critical elements to a cost-effectiveness analysis of family planming services. It helps identify efficient patterns to allocate resources, or delivery strategies. In an environment where various instituLtions control different resources and pay for them, individual institutions may find it hard to recognize an overall or social cost-effective strategy, because the different institutions respond to different aspects of cost and productivity, and therefore overall social efficiency in delivery may be hard to attain. EKXBN need not, for example, consider in full to cost borne by the Ministry of Health (MnH). Ihe government, including M1H, BKKBN, and other ministries may not respond to the cost shouldered by the coamunity. All instituticns combined may be insensitive to cost of inpits provided by international donors. From a policy perspective, it is therefore crucial, in addition to evaluating unit cxst, to assess the extent of funding control various institutions have on different resources participating in the family planning effort, especially when those institutions have different objectives canpeting for the same resource. Given the institutional framewrk discussed above, thiee sets of resouroe - 24 - constraints are conidered: BEJ's, the Government's, and the society's. As is shown later in the discussion (chapter 7), the marginal cost of family planning delivery in Indonesia is borne mainly by BKK and the ocimmumity, while the fixed cost is borne by the Ministry of Health (MOH). Cost-effectiveness can be seen as having t corKeptual aspects: allocative efficiency and internal efficiency. Allocative efficiency issues involve the question of how additional resources should be allocated across operations and, within each operation, anong the different services or methods offered. The allocation question in the case of Indonesia stems primarily frcm the different levels of infrastructure that can benefit EKXBN's operations in the different provinces. Intarnal efficiency issues here concern the allocation of resources in any given operation or service. These issues concern the question of what resources or productive attributes should be expanded at the margin, possibly at the expense of others, in order to increase the cost- effectiveness of an operation. This seregation of efficiency questions guides the study. Each of the distinct study areas is first examined separately as an individual program, then ccnpared with others. In this manner, we study ways in which BKKMW and the Indonesian government might improve overall efficiency by pramting particular operations and types of programs.3 3 Any implied prograLUwatic changes are assumed to be marginal within the realms of current contraceptive technology, consumer preferences, and EKXBN's current structure and patterns of delivery. The analyses stemiing from the tWo efficiency questions may suggest prcmotion, at the margin, of one family planning method at the expense of another, and the serving of different populations. It must be borne in mind, however, that no data are available on demand for alternative methods. Suggestions about a "preferred" method (cum - 25 - The quest for cost-effectiveness calls for imed design in terms of targeting progran resources, choioe of delivery modes, method mix, and personrel traits. The task is even nore cozmplicated for BEW3N, as it mist also allocate same of its resources to resource nmbilizaticn from other instituticns and the oammunity, and then make the best use of all resources within its dbmain.4 In this discussion, we focus on the allocation of resources to delivery given resource availability for that purpose. While the optimum may never be achieved, marginal changes and fine-tunmng of program design and operations according to some basic econcmic principles can lead in the right direction. To establish the oost levels of family planmnig delivery - the first necessary step for a cost-effectiveness analysis - the study exmaines: a) Levels and nature of resources for family planning efforts coordinated by EKOMB; b) ownership or funding sources of these resources; c) Allocation of these resources and its determinL. '4, and d) Cost of resources. A demand and supply framework, outlined here in general terms, is employed mode of delivery) and program frmm a cost-effectiveness perspective, must ultimately consider consumer preferences. A program may be cost-effective in terrs of unit costs of protection offered, bit socially inefficient when oonsUmer preferences are not considered. 4 mhat is, part of BKMIO's budget must be allocated to resource mobilization. The optimal program indicating how a nonr-profit organization should allocate its initial budget between resource mobilization or recovery and delivery in order to maximize delivery is under development (Chernichovsky 1990). - 26 - to determirne labor productivity and efficiency of operations. Acoordingly, demand (QD) for contraceptives in a given operaticn is depicted by: Ql = d (population size, population characteristics, protion activities, the "full price" [FP) of servioe to clients) That is, the quantity of fertility control dernaned in a given catchment area is a function of: a) pol-ilation size, which influences potential demand for contraception and scale of operations; b) population characteristics, which determine the desand for children, fecurxity, and attitudes toward family planning (Easterlin and Crimmins 1983), all of which determine demand for contraception; c) prLortion activities such as Information, Education, and ommmnication (IBC) activities and incentives to households; and d) the full price of service to the client, which is determined by fees (when applicable) and ease of aocess to outlets and support services.5 Effective demand levels vis-a-vis potential size of operation are important to cost-effectiveness because scale economies, as suggested above, are a major means of reducing user cost or increasin oost-effectiveness. Therefore, consumer demand is a major concern to program planners and managers; in addition to identifying the size and nature of a catchment area, they may need to 5 "Access" can relate to many aspects of the client's perceptions of cost and comfort of service. Therefore, the nature of inputs, e.g. female vs. male workers, might be oonsidered in order to assess their impact on the full cost of contraception to the oonsumr, especially in the absence of fees for service. See Ciernichovsky and Mk±aughlin (1988). - 27 - influence client behavior. Program managers have two options: praomtion, largely through IEC activity, and manipulation of the full price of service. Both require resources and therefore present managers with the challenge of an optimal allocation between the options. The supply of contraceptives in the cnmmunity concerns a program's capability to influence and accommodate potential clients by affecting the full price of contraception or the client's perceived access to servioe. This price can be lowered when the program has more and better resources for delivery. Generally, for a given external program budget6, the following relationship obtains: "Full price" to the client (FP) = s (QD, money resour-ces or budgets allocated to delivery, input prices, infrastructure) (3.2) That is, the price is a function of: a) quantity demanded (QD), which determines the actual scale of an operation. When it is "too high" for a particular operation in the short run, it may involve a high prioe to potential consumers and a lower the program's ability to acoc%mnodate all of them;7 b) money resources allocated to delivery, which determine how well oonsumers can be aomnrdated, when input prices are given; 6 The term "external" program budget concerns those resouroas available to the program at its outset, before resource msbilization in the comwnity. 7 Under-demand will not affect the consumer price. A fuller discussion of this issue follows in chapter 8. - 28 - c) input prices, which determine the level of real resouroes available for delivery. The higher the prioes the lower the real resources; d) (comunity and other) infrastructure, which enhances a program's potential by augronting productivity of other resources; A third relationship concerns mobilization of omm=nity resourcs, which are xncluded under infrastructure: I = i (program inputs for resource mobilization, ccr -"ity infrastructure) (3.3) This quantity is determined by: a) the inputs the program allocates to imbilize resources. The higher the input levels, the higher the level of resources mobilized; b) the infrastructure, health and other ocomunity facilities, which determine the potential for resoute rowbilization. The discussion thus coawans allocation of resources between promotion, delivery, and commnity rescurce mobilization, so that delivery unit cost is minimized. When all oosts and institutions participating in delivery are considered, the third relationship can be integrated in relationship 3.2.8 The nature of the institutional oonstraints disoussed above is such that from the delivery perspective, vethod mix may be greatly influenced by 8 This may be also warranted for subsequent statistical analysis because it may be imposrAble to separate the impact of the contrtzition of the comunity to infrastructure frcm the impact of its characteristics on demand. - 29 - availability of MOH infrastructure and personnel.9 This also means that fixed c06ts are largely borne by MCH, and that BKKBN may have little latitude in controlling method mix and hence the variable cost of supplies. BKKM controls only its labor cost. A basic working hypothesis here is that BKKEN may indeed have a relatively limited scope in the determination of cost-effectiveness. Thmrough allocative decisions, it can affect the internal efficiency of its operations only by enhancing the productivity of its labor. In sum, labor productivity is thus envisioned as resulting from an interaction between demand and supply factors that charactexrize a particular operation. Labor is believed to be the major instrumnets by which P ti my influence the cost-effectiveness of family planning delivery. The study areas and data have been selected with a view towards identifying the resources participating in family planning delivery, their cost, and what might affect unit cost through alternative delivery strategies. 3.4. Study Areas The Indonesian Family Planning Program covers all 301 regencies and municipalities in the country's twenty-seven provinces. Six regencies were selected in three provinces as study sites: a) Tangerang in the Province of West Java; 9 This affects not only potential supply patterns bit also consumer demand. - 30 - b) Kulon Progo and Bantul in the Special District of Yogyakarta; and c) Banjar, Barito Kuala, and Tapin in the Province of South Kalimantan. Figure 3.1 shows the study provinces in relation to the entire country. The study ccvers family plang operations in 83 subdxistricts and 1016 villages, where 406 field workers and 83 supervisors operate (figure 3.2). Villages in Java are divided into hamlets, with an average of three hamlets per village. These areas do not constitute a representative sample. Indonesian provinces vary greatly in terrain, population size, ethnicity, and religion. Short of conducting a large-scale national survey, the country's heterogeneity precludes representativeness. The areas were selected, however, so that the populations and conditions are fairly hintgeneous internally, yet with distinct variations between them. The study areas do reflect part of the wide spectrum of socioeconomic enviroments aind progranmatic conditions which shape family planning operations through both the nature of the population and the availability of infrastructure. - 31 - Figure 3.1: Location of Study Area- IMD privOlbe Fractime Bf ~ private pY>ctice - 62 - In all regencies surveyed, more than three-quarters of pills are delivered through the VCDCs, followed by clinics and outside-clinics. The amount of pills delivered through private sector channels is negligible, but is highest ir; Tangerang, where supply of program facilities per eligible couple is lowest. The two DI Yogyakarta regencies each deliver over 92 percent of all pills through the VCDC. It is noteworthy that in the three South Kalimantan regencies, where the pill is most common, a greater proportion of pills is delivered through the clinics (8.1 percent to 18.3 percent) and outside-clinic activities (6.0 percent to 7.3 percent) than in other regencies. The South Kalimantan data are consistent with the hypothesis that medical infrastructure is inportant for new programs even when no clinical intervention is required. In Indonesia, a medical check-up is required for new pill acceptors. condom delivery is quite varied among the six districts surveyed. In Tangerang, the only regency where private sector channels play more than a minor role, virtually all condoms are delivered through private pharmacies. mTe number delivered, however, is the mallest of all the surveyed regencies. Condcm in DI Yogyakarta are almost entirely delivered through the VCDCs (98.6 percent and 91.2 percent). In the Barito Kuala and Tapin regencies of South Kalimantan, clinics deliver over half the cond=, VCDCS one-third, and the remainder by outside-clinic staff. Outside-clinic activities are responsible for 57 percent of injectable distribution in Tangerang, being the regency with the heaviest injectable use. Clinics deliver most of the remainder. In all other regencies surveyed, injectables are delivered predcminartly through the clinics, followed by - 63 - outside-clinic activities and private sector ctannels. The relationship between availability of medical facilities and extent of outreach activities is not clearly a priori. On the one hand, outreach activities may substitute for a lack of medical facilities. On the other hand, as outreach activities may depend on medical infrastructure, they may be complementary. Complementarity between modes of delivery may be important in new areas where even for non-clinical methods, clients need ard want medical attention. Table 5.1 and figure 5.1 show that where there is a relative scarcity of medical facilities per EIMn, there is also a relative scarcity of outreach activity per ELOO. The data suggest that all delivery mechanisms appear to complement rather than substitute for each other. This will be further explored in the next chapter. 5.4. Incaoe-Generating Schemes It is difficult to quantify thLis non-delivery aspect of the program. Data were collected just on availability in the areas operated by field workers in such schemes, and on the number of participating T. No data were available on the resources available for these sch .12 Income-generating schemes appear most commonly in DI Yogyakarta with the coconut seedling scheme being the most ommon, as in other areas (table 5.3). This evidence, namely that these relatively new schemes are most omman in the 12 Education benefit schemes are not reported because of the neqligible number reporting such schemes. - 64 - region where the program is oldest and most developed, suggests that these schemes are a reward to areas with high prevalence, and hence their inpact as an incentive to recruit new users rather than retain old users remains to be investigated. Table 5.3: Availability of Incone-Generating Schemes % field workers % ELOsr No. reporting partici- report- availability pating ing Coconut seedling 50.7% 9.8% 134 Credit 5.9% 1.3% 134 Public utilities 8.2% 2.4% 134 DI Ycgaat Coconut seedling 89.7% 18.5% 78 Credit 67.9% 8.3% 78 Public utilities 39.7 24.3% 78 South Kalimantan Coconut seedlings 14.1% 1.0% 64 Credit 10.9% 3.9% 64 Public utilities 7.8% 10.1% 64 5.5. Sumyary There are dramatic differences in availability of medical infrastructure per ELT) among the study regions. These differences oorrelate with modes of delivery and method mix. The IUD is ccmTan where medical facilities are - 65 - relatively common and where programs are well-established (DI Yogykarta). Outreach activity and related methods, pills and injectables, are more common where health facilities are relatively scarce. In the peri-urban areas of Tangerang, the presence of the private sector in delivery of ckndxzs and pills is noticeable. This strict observation does rot neaot sa ily imply causality between method prevalence and infrastructure, especially snmce perceptions about IUDs change in new programs. At the same time, sut percepticns in favor of IUDs may be facilitated by the availability of infrastructure. Still, as availability of medical infrastructure is by-and-large an exogenous factor to the family planning program, it is mxst likely to influenne prevalence and method mix, rather than the other way around. - 66- 6. FnD PERSONNEL AND OPERATIONS 6.1. Introduction Locally-recruited family planning field workers (PLKB) and their supervisors (PPLI3) are key members of the National Family Planinj Program. They are employed directly by BKKBN, and therefore are unmer its direct influence. Field workers receive a salary and have civil servioe status. Each field worker is responsible for two or three villages within a subdistrict. This worker's main tasks include coordinating and training village family planning volunteers (VCDC volunteers), commuricating with village chiefs and leaders about family planning matters, maintaining non-clinical contraceptive supplies, meeting with and motivating acceptor group members, and writing logistic and user reports. In the past, an important field worker activity was house-to-house canvassing for new acceptors. As community volunteers became more familiar with their duties, and as people became more knowledgeable about family planning, this task became considerably less important. Hence, this activity is relatively ommon in thie Outer Islands where the program is comparatively new. One family planning field worker supervisor (PPLKB) is assigned to each of Indonesia's 3,539 subdistricts. These supervisors oversee the activities of the four to six field workers, on the average, who work under their lurisdiction. The supervisors' responsibilities include family planning liaison with the health centers, on-the-job training of PLKB and community volunteers, coordinating activities with other agencies and departments, and - 67 - data collection, recording, and processing. The Field Control Activities Reports, which they fill out ronthly, provide information about IEC wDrkers, PLKB, VCDC and sub-VCDC volunteers, acceptor groups, and mobile medical teams. Understanding how field workers are allocated and the factors that may affect worker productivity is an important objective of this study. BKKBN has direct administrative control only over field workers and their supervisors, and of these, has latitude only over the allocation of field workers. In order to deal insightfully with efficiency issues later in thiL study, it is necessary to know on what basis BKKBN distributes these workers across subdistricts, who they are and what they do, in tandem with their catchment areas, populations, and other personnel they coordinate in the ommunity. These issues are examined in this chapter through a study of the allocation of overall field personnel participating in family planning activities, the "quality" of the personnel in terms of perscnal characteristics, education and training, and BKKBN's allocation criteria. 6.2. Field Personnel Tangerang has the highest number of ELms per workers of any type, followed by Yogyakarta (figure 6.1). Variations between subdistricts within areas are small. one explanation for differences across provinces may lie in urbanization ard migration patterns. Tangerang is an in-migration area (see chapter 3) and these high ELr staff ratios may sinply reflect slow response of infrastructure and labor allocation to influxes of potential clients. Bantul and Banjar also have relatively large urban populations and may be areas of net migration -68- surplus; the former is adjacent to Yogyakarta Municipality and the latter is a transmigration destinaticn region. Allocation strategies based on past trends may be relatively insensitive to changirn client conditions. One clear implication of this picture is that BKKBN personnel, supervisors and especially field workers need to deal with larger target populations in Tangerang, which is a relatively growing urban area. They also need to coordinate resources which are relatively scarce in this area. From the data presented in table 6.1, howevrer, it is evident that although Tangerang field workers handle a much larger number of eligible couples than their colleagues in the other provinces, they do so in a vastly smaller average areas and numbers of villages. South Kalimantan field workers cover the smallest number of eligible couples, but this client pool is widely scattered over a wide area and relatively many villages. Tangerang's field workers also have acoess to fewer non-physician medical staff. These staff are the major resource available for family planning delivery in the community. Their relative scarcity in Tangerang may imply, more pressure on field workers who need to deal with resources which are more scarce (per ELCO) compared with other regions. On the other hand, this may suggest a need for lesser coordination efforts because of fewer medical personnel in a more oonoentrated area. - 69 - Figure 6.1: Family Planning Field Personnel in the Population Thousands capita per worker 14 12- 10 4- 2 0 Tangerang K.Progo Bantul Banjar B.Kuala Tapin MD 7.493 3.203 4.543 4.717 3.436 2.377 Other Med 1.886 0.736 0.794 0.349 0.573 0.37 Supervieor 12.38 3.737 4.81 4.354 2.291 2.08 Fieldworker 1.515 1.043 1.168 1.231 0.833 0.64 MIMD M Other Med =:I Supervisor v tleldworker - 70 - 6.3. Field Worker Allocation Is there any rule guiding the allocation of field workers, given the observed variance in their allocation to target populations, areas, and clinical personnel they need to coordinate? Regression analysis (table 6.2) indicates the impact that selected predisposing variables have in determining the allocation of field workers by subdistricts. In all three areas, field worker-s are allocated by numbers of eligible couples; for Tangerang and South Kalimantan, number of villages also appears to act as an allocative criterion. It is interesting to note that the coefficients for eligible couples are virtually identical across regencies. This implies a rather strict allocation rule. Th:.s rule does not correct, however, for past and other regional differences. mTe "constant" is both positive and significant for Yogyakarta, and regression R2 is by far the lowest. This suggests that the distribution of field workers is relatively uniform across sub-districts in Yogyakarta, and "responds" less to some concept of need. In the other two provinces however, it responds to target population size and population distribution. The R2 in Tangerang is the highest of all regions, which indicates that it responds most to some concept of need because the predisposing variables explain more than in other regencies variations in field worker allocation.13 13 All statistics in this study refer to the entire population in a study area. Statistical inference must be done with this fact in mind. Statistics should be used mainly for their predictive value. - 71 - Table 6.1: ELC)s, Villages, Area, and Med. Staff per Field Worker DI South Tangerang Yogyakarta Kalimantan ELIXs 1437 1199 882 Villages 1.84 1.23 5.71 Area (Sq. hec.) 658.85 3477.20 7098.81 Non-physician .81 1.44 2.50 medical staff Table 6.2: Determinants of Nuiber of Field Workers, Linear Regression Results; No. of Field Workers in Sub-district as Dependent Variable DI South Tangerang Yogyakarta Kalimantan Constant 2.441 3.260 -3.77 (1.72) (2.61) (-0.38) Eligible couples 3.5E-4 3.4E-4 3.9E-4 (4.05) (2.68) (3.11) Children per -2.300 -1.656 2.699 eligible couple (-1.34) (-0.77) (1.92) Villages .313 -.052 .052 (5.68) (-0.80) (2.50) Area -2.7E-4 1.OE-4 -1.7E-5 (-1.42) (0.81) (-0.21) Non-physician .154 .064 .050 medical stuff (1.52) (0.95) (-1.36) F 23.96 1.99 12.49 Adjusted R2 .86 .16 .66 - 72 - The estimated coefficients suggest that nore field workers appear to be found in South Kalimantan villages with more children below five per eligible oouple. It appears that in Tangerang worker allocation "responds" scarnhat to medical Mnfrastructure measured by non-medical staff. This indicates that the program there may inde3d take advantage of availability of meeical staff who support the injectable. On the whole, infrastructure does not appear to influence BKKBN's field staff allocation. 6.4. Worker Characteristics Worker characteristics could play an important role in labor productivity, especially when the number of workers constrains output irprovement. In BKKBN, assessment of labor need and labor recruitment occur at the regency level. With the recent decree grantinj civil servioe status to family plawuiig field workers, BKKBN promulgated regulatiors stipulating that workers have at least high school education. -73 - qlh 6.3: Wxr tj,, a I, TA q W. D .- FP sevam 28 11 14 12 11 8 84 FP Fil W3ox 8s 43 63 43 33 26 396 Madimi Stf 189 74 122 174 53 52 664 FP Voube 3DS 654 687 362 268 196 3497 FP pvio 35 42 39 32 33 32 36 FP Fied S3Q3 32 34 36 29 27 293 32 MPf Mral Staff 35 36 38 39F 31 34 35 FP vt&urte3o 37 43 4D0 * 36 39 39 EP upervisot9 86 73 71 75 lCO 75 En FP Field W3Jhess 61 46 41 58 52 62 55. MpbfCTi SFERff 35 3e 27 41 58 48 38 FP Volurtnbas 29 19 2n) 55 52 32 30 FP SDpwAsior 89 91 100 92 100 88 93 FP Eie-ld kWboks 81 70 84 65 52 65 75 H3dical FSbff 85 89 89 84 66 85 84 FP')blurteer; 91 94 94 90 86 97 91 MESN U13RCF CIRE EP apsrvLsXrs 2.78 3.55 2.36 1.38 1.82 1.12 2.39 EP rield Wtrhers 1.91 1.38 2.0B 0.74 0.45 0.77 1.61 MF^icnl Staff 2.22 2.23 2.25 2.32 1.5B 1.90 2.18 FP voitSImmr 3.42 3.68 3.19q 3.42 2.81 2.77 3.34 PFECEN MUMI H 5R ABCU EP 54petvECso 87 91 lC0 100 100 100 91 EUP Fie3d MJoes -1 E3 85 99 lOD IOD 87 Mebdcal StafF 86 86 81 62 8n 81 78 E!P vohtiltEr 27 45 58 44 30 175 40 a) IrrJutes all wztkgrs IarIiciLpting in the stbJy. FEEsnse rabe vas cuer 95 % ftr all cWmgxrisc. b) Includ ohysxmans, nurs=s, midhievm sua iFry nmddvwi arnd prarEEis. - 74 - Table 6.3 reveals differences in worker characteristics between personnel types within one region and also differences in the characteristics of one personiel type across regions. The mean age, in the mid-thirties, of the four types of workers is relatively similar. This indicates a program decision to recruit mature and educated personnel. Workers are slightly younger in Tangerang and South Kalimantan, which may reflect the age structure of the local labor pool as well as the age of the program. Field workers are the youngest group and volunteers the oldest. The majority of family planning supervisors are male, as are a slight majority of field workers. Most medical staff members are female, although within this general category, physicians are primarily vale. Family planning volunteers tend to be wives of village heads or other cammunity elites. Compared with other areas, DI Yogyakarta regencies have nigher proportions of female workers, which may reflect the relatively high status accorded to Javanese women. IUDs are the favored method in DI Yogyakarta (see chapter 4), not older than Tangerang's, and this may partly be a function of the higher female canponent of the family planning and medical personnel. Marital status patterns follow age patterns, with older w-rkers the most likely to be married. This holds true both for regions and for different worker types within one i-egion. The same pattern pertains to mean number of children; older workers tend to have more children. DI Yogyakarta workers thus are mDst likely to be married and have high mean numbeners of children compared with their counterparts in the other study areas. - 75 - Educational patterns, however, are mixed. Field workers and their supervisors are uniformly well-educated, especially in the South Kalimantan regencies. The higher proportions of well-educated staff in South Kalimantan may be a product of their younger average ase and also their lower seniority. As mentioned, regulations now mandate secondary education credentials for recently-hired family planning workers. Medical staff education, largely influenced by the education cf para- medical staff, is low in Banjar. This is in part a result of the medical personnel mix in the area. One possible reason could be the difficulty in recruiting qualified medical personnel for Outer Island areas. The wide range of educational levels within the medical staff category reflects both the university degrees of physicians, and the elementary school edac .ion of many auxiliary nurses and paramedics. Family planning volunteers have the lowest education of all persomrel categories, although levels are highest in DI Yogyakarta. TIhis may reflect their age and mirror their population of origin. 6.5. Training and Experience Experience and training are considered important variables influencing worker effectiveness. There are few seniority differences between the variouss personnel types (table 6.3). Generally, personnel in Kulon Progo ani Bantul are more experienced than their counterparts elsewhere, this of course being - 76 - related to their older average age and age of the program. Comparing mean work experience with rean age reveals that I!st personnel were hired in their mid-twenties, and that they have been working steadily in the health and family planning arena, evidently without much turnover. Their civil service status offers them a secure income and position, and thus there is little incentive to seek employment el¢ewhere in a comntry where employment, especially of people with high education, is an issue. Supervisors received the most family planning training, followed by field workers and medical personnel. DI Yogyakarta supervisors and field workers received substantially more training than those in the other regions, although this was not the case for medical personnel. Most personnel, no matter what type, receive basic training in family planning dnd IC. Strikiing interregional differences do not appear on this dimension, except that South Kalimantan medical personnel receive less family planning and IEC tradiing than their counterparts elsewhere. Again, this may reflect the relatively short history of the program in this area and distaroe from Java. This finding ma- reveal a pro.gram sihrtc=uing, because, as shan in the previous chapter, clinics in South Kalinat*an are a p-raninent delivery mode for recurrent ontraceptives. Medical personel receive more training about long-term methods (IUDs, implants, and female sterilizations) than otker personnel types. This finding particularly pertains to Twngerang and DI Yogyakarta medical personnel. Given its high IUD prevalence rate, the high proportions of workers receiving training on this method in DI Yogyakarta is predictable. Fbr reasons not fully clear, Tapin workers also receive subetantial training on IUDs. Both Bantul and Tapin personnel get noticeably more training in female sterilization than do those elsewhere. More training is given on IUDs than on the other methods, an entirely predictable finding given its higher prevalence rate. Nevertheless, considerable training is also offered on inplant and female sterilization in Bantul, suggesting that these methods may be slated for greater prominenoe in the mature program regions. Most field workers and their supervsors receive training in reporting and administration, although the proportions are comparatively low for South Kalimantan field workers. The majority of medical personnel also receive reporting and administration training; again, the figures are lower for South Kalimantan than elsewhere. TraMinrg appears to mirror local delivery conditions and the history of the program in any area. In South Kalimantan, distances may also adversely affect training, as it may be costlier to assemble workers. TtLis wide variation in training curricular both within and between provinces suggests that much of the decision-making on this dimension originates at the local level, perhaps in resporne to method allocaticns and client demands. On the whole, the data may be confounded by a tendency of workers to confuse formal tradinii with experience. - 78 - 6.6. Workers' Activities Field workers and their supervisors spend about 40-50% of tieir tine in promotion activities ("search for new acceptors and mapping", IEC, and "1eyond family planning"); about a quarter in supply-related activities ("organization of delivery", "coordination and organization in the ccmmunity" and "supervision"); and an equal share of work time in administration (Fig. 6.2). That is, BKKBN's staff spends most of its time on search and recruitment of new users rather than on delivery. In most areas (data not shown), especially those of Yogyakarta, field workers spent more time on administration than on delivery. Only the field workers of Tangerang and Tapin spent more time on delivery-related activities than on administration. Differences in the allocation of time of BKKBN's field personnel across regions are generally similar. Moreover, across regions and especially within regions, field workers and supervisors have fairly similar distributions of time allocation. The latter group spend somewhat less time on recruitment and promotion. Hence, supervisors are "first among equals"; they do not spend more time than their subordinates on supervision and administration. These data indicate that the variance prevalence rates and method mix across provinces is not mirrored in what field workers and their supervisors do; their activities are relatively uniformly distributed. CC - , _6 S I p O 'S~~~~-, _1~~~~~, L , , . _,_ ., , _ , _ S , ~~-~ t -o0- Figure 6.4: Activities of Non-Physician Medical Staff Overhead v e 3 ,~12 ,,,..Workers R%Th. Exram 17% '~'"~~26% 24% IE Ref erral Medical 4% 64% ther 9% identify -ollow up 13% 3 Health center physicians spend about cne-fifth of their work time in farRily planning activities (figure 6.3). One-half of this time is devoted to IEC and supervising and coordinating workers. Physicians allocate approximately two-fifths of their time to activities involving face-to-face contacts with family planning clients, namely, cbtaining information about new acceptors, dealing with referrals, and performing follow-up procedures. Non-physician medical staff spend about one-quarter of their time con family planning (figure 6.4). Approximately one-third of this time is devoted to supervision and coordination, and the remainder largely to face-to-face contacts with clients. Initial clinic exaninations are almost exclusively - 81 - performed by these staff members. 6.7. Summary Considerable variations are observed across regions in the allocation of labor in relation to their jurisdictions. Tngeramng has the highest ratio of eligible couples to all personnel types, which may suggest the inability of the program to respond quickly enough to rapidly changing demographic conditions. Tangerang's field workers need to deal with far larger populatiozs using femer other resources per ELC4 than their c in the other areas. They Lenefit, however, from a re]ct ively high population density. At the same tize, within each of the areas, labor allocatioms appear to be based more evenly and there is same rational decision-making in all three provinces: allocation of field workers is on the basis of ELCos, villages and area of cove-rage. It is less so in DI Yogyakarta and more so in T¶ngerang, where resources are relatively stretched. This may suggest that local managexs are more responsive to scame criteria of need in allocating scarce resources than central ranagement, especially in newly emrging program areas. Wmere the need is more pressing, as in Tangerang, responsiveness is greater than whkre it is not, as in Yogyakarta. Hcxwever, differenoas in terms of method mix and other aspects beteen the areas may be justified in some degree y variations in labor inputs across regions. - 82 - Workers basically mirror their target population and the age of the program. DI Yogyakarta has a larger share of workers who are older, female, more experienced, and better-trained. This may suggest few changes in personnel in view of need as the program matures in ssme areas, and allocation across regions. Time allocation for both field workers and their supervisors is mainly divided between demand generation, supply, and administrative activities. This is similar across regions. Both groups of workers spend most of their time on search and promotion activities, with rather uniform time allocation patterns across areas. This may suggest that workers report working "by the book", and gives rise to the hypothesis that on the whole they may be underutilized; there is no obvious pressure to respond to local variations in need. If there is such pressure, it may exist in Tangerang. Medical staff spent about one-quarter of their time in family planning activities. Family planning worker characteristics and time allocations will be used in subsequent analysis in order to determine what effect they have on output. - 83 - 7. PROGRAM COST 7.1. Introduction Although it is essential in order to assess the social efficiency of the program and the cost-effectiveness of alternative operations, the value of the resources used in proamting and delivering family planning discussed in the previous chapter has never been fully assessed in indonesia. Cost elements, such as capital costs, the value of volunteers' time, and even the value of staff time, are not cxmpletely accounted for in recurrent budgets. Moreover, in the aksence of detailed accounts, the allocation of oost to specific operations is often oamplicated or outright impossible. file issue is particularly acute in Indonesia, where many institutions cmntribute to family planning operations. However, knowledge of the level and comprsition of the cost of contraceptive delivery, and of who shares its burden, is essential to program evaluation, policy fonumlation, and prograMding. In this chapter we seek to identify: a) the value or cost of the resources used in family planning delivery; b) the cost structure by types of inputs - fixed vs. variable costs; c) the contribution of EKXBN, MDH (Ministry of Health), and the community to the cost; and, d) the unit cost per ELn and user of alternative family - planning operations - prcmotion, delivery and adnird stration. - 84- Resources are groupoed by major types: labor, capital (buildings and equipmnt), and supplies (contraceptive supplies and purchased servioes). costs of irpits per ELCO are evaluated in order to measure the intensity of program effort in each of the study areas. Miae cost data are related to program output or number of users in order to assess the relative cost- effectiveness of 8KKBN's operations and alternative program strategies. 7.2. Labor Cost14 Labor costs are allocated by the following personel types: - BEULKB, or family planning field worker supervisors; - PUKB, or family planning field workers; - Mg, or clinic physicians; - Other medical staff, or clinic nurse-midwives, auxiliary midwives, and paramedics; other staff, or non-medical personnel such as clinic administrative staff; and - PEKBD, or village family planning volunteers.15 14 Labor cost is ccmputed acoording to the actual staff size in each region. This is marginally different from the number of workers participating in the study [See note in Table 6.3]. Average waqes were conputed based on the information received from participating workers. 15 We include only the family planning portion of "NDs" and "other medical staff" cost, basing it both on the time these personnel reported they worked in family planning activities and on their average government wages. We inputed "wages" of village family planning volunteers based on an equation adopted from ine function estimates of 3,000 workers from East Java in 1982. Characteristics of these workers matched characteristics of the volunteers in this study. We estimated their work time as follows: for each gosyandu we designated three volunteers for two days per mornth, and for each VCDC we designated one volunteer for four days per month. These are considered -85- Ta 7.1: T TA C Stf in Ria per E pr m W. b D I .YwJL M& U-ti a&aVEr%is 1% 15% 13% 1% 23 16: 30.7 15.01% Fiel^dwa8ars 53% 3CP.3 38% 27% 35% 30%: 67.2 32.9% MYs 7% 5% 4% 3% 6% 9%: 12.8 6.3% Othw mid. 1ff 1% 19% 36% 43% 2: 53.0 2.9% odw ff 3% 1% 11.% 6% 6% 5M: 14.3 7.0% VobxtflES 7% 21% 17%6 32 18 9% 11%: 26.3 12.9% S 100% 100% 100% 100% 10% 100*: 2)D4.3 1 M.0 (in FES.) (73) (203) (160) (L98) (231) (361): a SiRie MSE rer miniml time inputs. It is assumed throughout the discussion that all recurces have oPPortunity ccst; MDs and other medical staff could treat !atients when they handle family planning users, and volunteers could use their time producti4ely and leisurely when not engaged in family planning activity. -86- Figure 7.1: Laxr Costs by Type of Staff Rupiah/Month 120 __ 100 190 I At rangerang K. Progo Bantul Banjar B.Kuala Tapin PPLKB 8 30 21 20 46 59 PLKB 39 60 61 54 80 109 MOs 6 10 7 6 15 34 Other Med 14 38 26 83 54 103 Other Staff 2 22 18 11 16 18 Volunteers t; 43 27 24 21 _ 38 D PPLKB 1 ! PLKB EJ MD$ Other Med Other Stat? Volunteers - 87 - There are substantial differences in worker costs per EL between Tangerang, which has the lwest mnxthly labor cost, and South Kalimantan, which has the highest (table 7.1, figure 7.1). In Tapin, labor costs are between four and five times higher than those in Tangerang. The DI Yogyakarta regencies, Kulon Progo and Bantul, lie between these extremes. These variations reflect in part population densities, because, as shown in chapter 6, eligible couples by personnel type ratios are highest in Tangerang, followed by fairly uniform levels in the other areas. BKK9N staff, field workers, and supervisors account for more than 48 percent of labor costs, while medical persomnel account for about 32 percent. That is, BKKB personnel account for only about one-half of the value of labor resources employed in family planning delivery. Administrative and support staff account for the remaining seven percent. The imputed labor cost of cxmmunity family planning volunteers constitutes roughly 12 percent of the total labor cost. This cost is lowest in Tangerang and highest in Kulon Progo. 7.2.1. Types of Labor Income Salaries and honoraria constitute 87 percent of average total incomes of PLKBs (field workers) and 79 percent of average total income of PPLKBs (field worker supervisors) (table 7.2). Compared with PLKB labor-related incomes, PPLKB salaries are more varied and on tthe average, PPLKBs report incomes 58 -88 - Al7.2: Mmily Elmitlp b. Wx'w Ti arid, Te cE Im ,sq Rxw tu jEl WRq T (in RQ.) hid EED W3KS(P ) W.;M 606 77% (a5b 67% 82Rc 706 5;3.0 70.Wo4 Prm 01 N N 7% O% m k 1% 1.2 1.54% 'IyL 3io C% G% 1% 1% 2to 1.0 1 .325' Hormcari 226 17% la& 3A 13 20% 13.2 17.40% CX1X3r Eqy 3-IV 6% 31% Lai 7io a. 7.3 9.69P* QtItoa 100% 10DO 1D% 10D0S 100W OC 75.7 100.00% (73)a (69) (88) (72) (71) (81) gSEm (PEMR) iE(g3 64% 78% 74%' 83% 83% 65% 84.0 74.12A- Per- Dian 17% 128 11% 3%. 6t 69 10.5 9.wt TWl!I3 0% 1% 1% C% 2% 1% 0.8 0.74%o Hxt-codra lD0o 2% 2% 8% 3% 6% 5.3 4.71% CXSl3 PRY 9% M% 13t 6% 7io 23io 12.7 1 1.18M Totanl 10D% lDO% lDO 100% 100% 100% 113.3 I100%00 (111) (lSM) (114) (90) (18) Uv2) BL(^3B 41% 64% 41% 46%- 55% 79% 165.8 53.09% Per rufn 4% Ot 1% 2% 2% 2% 6.0 1.92% S>2wul 5io 0% 2% 2% 3% 7% 1D.5 3.36% H2axrwara 36% 3D% 49P.- 31% 28% 10% 99.3 3 1. 8C%* Cd,fir pa 14% 6% 8% 18% 311% 1% 30.7 9.82% TI-til 100% 100% 100% lDO lOQ% lDO% 312.3 I M. M% (364) (205) (402) (289) (303) (311) Sm 67% 80io 66io 82& 71% 81% 93.0 74.5C%o Fer Dfimy 2% 0% 1% 1% 2% ift 3.5 2.8D% q]wru3 2% 3% 2% 3i0 3% 1% 3.0 2.40% HEmrsaria M5 12% is% 311% 1D% 5% 15.0 12.02% CXSwr PR 14% 4% 13% 3% in6 2541 10.3 8.28% qwesa 100% 100* 0DC% lOQ% lOQ% l0DO% 124.8 1M.DO0% (123) (123) (136) (107) (129) (331) a TcEIe;dg (in 100 I;p pE3r rt) in p -89- permet '.igher than those of PI(s. Most of the difference is accounted for by salaries and honoraria, which are centrally determined by N Tangerang reports the lest base salary (60 peroent of Rp.73,000/mcnth, or Rp.44,000Atnth) caqraed with the other regencies. Htrice, BaM labor costs in Tangerang are loest not only because of BKU's lower allocation of resources per EMLM to the regncy, bIt also beause of the relatively lcow salaries it allocates to Tangerang's persnnel. Given the different characteristics and activities of field personnel as outlined in the previous chapter, and the relatively small variation in salaries, the data suggest a fairly uniform pay scale with little consideration to variaticns in characteristics and types of operation. Ms stand out as above average both in levels of reported incomes and the shares they report as honoraria. 7.2.2. Labor Funding EKKBN funds about 4.5 percent of the physicians' government salaries and a smewhat higher percentage of other medical staff salaries (table 7.3). Notably, even ERKSN staff report incomes from other sources, including MoH, for travel and related expenses. In Tbngerang, where the pay from BKXB fdr its workers is low and alternative employment opportunities are possibly better, field personnel report greater earnings from other sources. BKBeN contributes a greater share of physicians' inc=es in South Kalimantan where, because of the apparent dearth of private practice opportunities, MD salaries - 9o - are lower and work in family planning is higher. BEKKN is responsible for aboult 49 percent of labor costs, covering its own personnel and part of the cost of medical personnel (table 7.4). M)H handles some 38 percent of labor costs, and the comunmity 13 percent.16 From BKKBN's viewpoint, it is basically its own share that matters. From the viewpoint of the government, it is the cost of HKKEN and MLH. From the viewpoint of [ndonesian so-ciety, it is all contributions, including those of the ccmuAnity. It is noteworthy that the s of cost to BMOBN is highest in Twngerang where the public sector medical infrnstructure is lagging. In .absolute terms, labor cost is still higher in the other areas than in Tangerang because of low population densities ,nnd contributions to medical staff in those areas in comparison with Tangerang. 16 These percentages are based on the data presented in table 7.3. -91 - 'Tshle 7.3: Disb iekn aE Em3nixgb ty TA4 of Stff ard 93rz aE Rndirig Mm B-- B- rtD R i 9, DsAi- mkS. bEt tu jr E6a Tii (inlZ.) b FILD W3FES (FPI3) IHE 3% 1% 1% 4% 1% 9% 2.5 3.30% EKWEN 79% 90% 80% 88% 93% 79% 64.0 84.40% Cthwr esrmErit r1% 1 1% 1% 1% 3% 8% 2.0 2.64% CtbEr 16% 7% 18% 7% 3% 5% 7.3 9.67% 'Ita. 100% 100% 100% 100% 1OC% 100% 75.8 100.00% (73)a (69) (88) (72) (71) (81) samwimis (PR) 4% 0% 0% O0% 0% 5% 1.7 1.47% EEN g 9%93% 9 90% 98% 89% 94% IM.3 92.94% OCk amomwt 0% 2% 0% 1% 3% 2% 1.3 1.18% Cdrw 4% 3% 10% 1% 8% 0% 5.0 4.41% Total 100% 100% 100% 100% 100% 100% 113.3 100.0% (11) (120) (114) (90) (118) (27) NLEML DO- HEI 45% 73% 47% 53% 55% 8% 180.8 57.90% KME 3% 2% 1% 7% 5% 9% 13.7 4.38% cd-fir OW,umut 2% 3% 4°. 4% 6% 0% 9.8 3.1% Orei 51% 22 48% 36% 34% 6% 108.0 34.5B% 100% o100% 100% 0 10 100% 312.3 1O.M00 (364) (205) (402) (28) (3M3) (311) Ca ER r1D@ S-F HE1! 72% 76% 66% 84% 76% 91% 96.3 77.17%. E1 3% 8% 4% 6% 7% 8% 7.5 6.1% Cta ut 0% 2% 0% 1% 2% 1% 1.0 0.8a 0dIsr 25%0o, 1% 30% 8% 1 1% 20.0 16.02% Total 100% 100 1000% 100% 0 1 124.8 100.00% (i12) (M3) (136) (107) (129) (131) a Ual mmng (in MCO FP. pw rath) in p3, - 92 - Tale. 7.4: T 03t, bl 3 CiE RFU in Rih pEr EL pErl a M. bM tiL jw Ra T.Ti (inEFD) htic EBEN 64% 46t 52?6 392 58% 46% 10.0 48.96 ;6 4 33% 31% 49% 32% 44% 78.0 38.18% Cmnity 7% ZL% 17% 12% 9% 11% X.3 12.8D% iai loo. lOA 1m6 mm% 1m 10mm AO4.3 100.00% (73) (201) (162) (1.2) (11) (361) T-El T lIt 1.6 2.2 1.9 2.6 1.7 2.2 BEN 03tibti -93 - cl1e 7.5: ZMO-Qt O=Vlatin clSfiCiat aE Stx.tist . Il r at FBr ELM bk c e Rr irg 1.00 ME1 0.82 1.00 ainity 0.57 0.61 1.00 The data in Table 7.5 denonstrate that BKE, the MiH and the ccxmmity support health and family plarnning workers in a complementary manner.. Higher- than-average expenijtures per PT by the MQH are typically associated with higher-than-average expenditures by BE1K. Higher-than-average co=mnity expenditures are also associated with higher-than-average EKKHN and MaH expenditures, though the association. is not as great as that between EKKBN and 14H. It is important to keep in mind for the forthcoming discussion that in respect to labor costs, the cost of medical staff borne by MDH is mainly a fixed cost, as it does not vary much with the level of family planning activity. HKKBN's labor costs are quasi-fixed (for individual operations) with respect to its c.± staff, and are variable with respect to medical staff because the latter are paid by level of activity. Ciuiunty costs are variable. 7.3. Capital Cost: Types and Source of Funding Although usually not acocunted for in the recurrent kudgets of family planning, the relationship between this cost and variable labor and supply - 94 - ccsts is essential to the identification of efficient method mix and delivery modes. Disregard o..r this cost distorts the real cost of delivering different methods of family planning, as is often the case with clinically-based methods such as IUD. Capital cost comprises five categories: a) buildings and land; b) general equipment, including most furniture, and non-medical and non-family planming equipment; c) family planing equipment, inco uding all equipment used exclusively for family planning, such as IUD and sterilization kits; d) medical equipment, including all equipment which has medical uses, regardless of whether it has family planing uses; and, e) transportation equipment, including cars, vans, motorc~ 'les, etc. operated by the different facilities. Capital cost relates to the five ditferent types of facilities which are the basic modes of contraceptive delivery of the program. They include: a) health centers or p (HC); b) health sub-centers (SubHC); c) mother and child health facilities (MCH); d) integrated health posts ( e) village contraceptive distribution centers (VCDC). The discussion here concerns only that portion of total capital cost that is allocated to family planning. The capital cost of medical facilities is allocated to this activity according to the proportion of time medical staff report in family planning activities. The equivalent of rental values was put - 95 - on buildings, and a depreciation rate, straight line on the basis of assumed life of 10 years, plus a 5 percent real annual interest rate, was used to calculate the coct of other capital. Total capital cost per ELCO varies dramatically, from only Rp. 18 per ECrn in'Tangerang to over Rp. 140 in Kulon Progo (table 7.6, figure 7.2). This reflects the low public sector medical infrastructure per Ernn in Tangerang and the high infrastructure in Kulon Progo (see chapter 5). Medical equipment is the largest item of capita-, accounting for about half of the total capital cost allocated to family planning. Yquipmient makes up for the lion's share, between 40 and 50 percent, of capital costs (table 7.6). Interestingly, in Tangerang building costs, belonging to the community and servicing VCDCs and Posyandu, contribute highly to cost of capital services in this region. Health centers generally have the highest capital cost, -nqing from 41 percent in Bantul to 91 percent in Barito Kuala, with an overail average of 67 percent (table 7.7, figure 7.3). Distribution of the remaining costs show great ".ariability. VCDCs, which account for 55 percent of capital cost in Tangerang and 43 percent of capital cost in South Kalimartan, are responsible only for 10 percent and 12 percent of capital cost in Kulon Progo and Bantul, respectively. While the remaining facilities oonstitute only 5 percent of Tanerangl's total costs, they constitute 42 percent of Kulon Progo's costs. Health sub-centers -96 - consume between one-fifth to one-quarter of Kulon Progo's and Banjar's total costs, but elsewhere no mcre than 10 percent. Mother and child health facilities, which appear only in DI Yogyakarta, acaxunt for almost 40 percent of Bantul 's costs and 15 percent of Kulon Progo' s. Finally, the ec6zar are not a major contribitor to family plarning capital cost: at their highest, s mmndue-mke up only 8 percent of Kulon Progo's total family plannirg capital cost. Notably, capital cost is high also in Soutn Kalimlntan, although the pill ib the uwt ccmn method there. Figure 7.2: Capital Cost per EW, by Capital Type in Rupialh/Elco/Montl 60 40 30 20 - 10L Tangerang K.Progo Bantul Banjar B.Kuala Tapin Building 4 17 7 6 7 26 FP 3 39 18 8 13 V Medical 5 51 22 25 21 37 General Eq. 2 13 6 5 6 a Transport 3 23 II 19 17 21 Building r FP e Mdical General Eq. 1'ransport Fi4re 7.3: ital asf bj Rdality Rupiah/Elco/Month 80 80- 40 Tangerang K.Progo Bantul Saniar B.Kuala Topin HEALTH CTR 15 59 42 62 58 76 SUB HC 0 20 3 6 4 3 MCH 0 10 11 0 0 0 POSYANDU 0 51 5 1 0 0 VCDC 3 4 3 3 2 22 - HEALTH CTR S UB HC C] MCH on POSYANDU .. f vcoc Capital cost is clearly associated with health facilities. Consequently, the MOH shoulders most of this cost (67 percent) through the provision of facilities and equipment (table 7.8, figure 7.4). As in the case of labor, BKFBN's contribution to capital cost (largely in relation to IUD kits) is closely correlated with total capital cost. - 99 - UK7.8: apital a3f by dmmc nJ In R4piah pE ELD pw mUth T3. Rm KLi jE 1ala TA Mm hiH 'ri~ 'S \\67% 66% 68% 82% 77% 69% 53.2 69.OD% '.17X 27% 96 13% 23 9% 15.0 19.47% (CmrLnity 17% 7% 3% S7 3% 2a. 7.0 9.09 ItYzall 'lcx lOt lOnD iC0? l0n lOnG 75.2 i7.56% (18) (144) (63) (62) (64) (100) Figure 7.4: Capital Costs per ELC), by Source of Fwxling Ruplah/Month 100 40- 20 Tangerang K.Progo Bantul Ban jar B.Kuala Tapin BKKBN ~~3 39 18 8 13 9 MOH 12 95 43 51 49 89 Community 3 10 2 3 2 22 2 BKKBN 3 MOH Community - 100 - 7*.4. Costs of Contraoeptives and Other .Azlies ard Utilities Cntraceptive supplies ard purchased servioes, such as utilities, etc., are most closely associated with levels and copmositicn of contraceptive delivery.17 Hence, they are the most significant variable and marginal cost cuponent. They are also the direct responsibi lity of EWN.18 VCDCs, which distribata pills and condmrs, account for the bilk of the overall nran supply costs (tables 7.9, 7.10), reflecting the relatively recurrent nature of the supply costs with these w methods. The clear exception is Tangrerad, wher injectables are the primary method.39 Injectables cannot be distributed through VCDCs, and cnsewguently are distributed instead throuch the VgM%Ddm and health clinics. Costs of supplies in medical facilities are rielatively low in YoLylakarta, where they deliver mostly ILCD, and relatively high in Tangerany and South Kalimantan, where they deliver p'lls and injectables. 17 Utility expenses for health and family plarming facilities (including electricity, water, and otaher expenses) are negligible compared with other costs in the analysis. These expenses are not presented here. 18 Contraceptive supplies are a major form of foreign donor assistance to the Indonesian program. This notwithstardiny, supply costs are ac-.onted for here. Tmis is important for assessing strategies which may need to be considered, from an efficiency perspective, as future supplies may not be free. 19 In Tangerang 5% of the injectables are distribated by private MDs and midwives, and condcms are also sold by pharmacies. These data are included in Table 7.9 but not in 7.10. Tdble 7.9: 1sE bi TA3f CE , pri in E1a PEr NM EW H I Ulit E tl Ban Ban B8t m Disr- prim mq. Em tLi jw Xa Upi hi*ai ILD 225 o6 2% it 06 oi6 at 0.9 0.51% PL1 3315 3D%6 306 5;R6 896 93% 9A LD.6 67.65W omdm 762 1% 63 42 3% 1% 2% 27.7 15.52 l eoJu 92D 65% St 5%6s 6 5% 29.1 16.32 Mbal dis 100% 10D 100% lN 100N6W I 178.3 U)0.oa0 (177) (ns) (183) (196) (229) (167) qgl7.10: amt ciE r ,bl M33e czE DliELy inRta pEr ELMpr nuX tt 'A, mi- B2- Brito istD - ilmty 7mg. EM tliL Nm3 Mqd b*i IC 3OD 6ic 626 23% 1% I2 27.1 15.47% -K: a2 0 3% 1% 1% 13 1% 2.8 i.5Q KR 0?S 1% 396 0% at otc 0.9 O.539- Rsprdi 43% it 3% 696 'A St 19.9 11.37% V= 2EN% 90% 87% 70% 75c 79% Z24.7 71.069 R 100 10 100% 100% M c D 175.5 100.aa% (167) (n18) (1]3) (LS2) (229) (166) - 102 - 7.5. Total Cost of Resources and Souroes of Funding Tangerang's resource cost is lowest, and Tapin's highest (table 7.11, figure 7.5). Supplies constitute a large proportion of expenditures in Tangerang, while labor costs are relatively high in the other areas, notably in Tapin because of the involvement of medical personnel. The data presented in table 7.12 and figure 7.6. separate these same oosts by the providers of the different resources: BKKBN, the Ministry of health, and the conuity. Cammunity cost figures reflect computed salaries for volunteers, and the space and equipTmnt used by pyau and VCDC facilities. M xst remarkable is the uneven distribution of Ministry of Health -esource , from Rp. 32 per eligible couple per month in Tangerang to Rp. 226 in Tapin. This reflects medical service distribution patterns per capita. In view of these variations, it is most interesting that the contribution of BKKBN is relative even, ranging from Rp. 226 in Tangerang to Rp. 376 in B. Kuala. It follows that BKKBN pursues a fairly uniform rule of resource allocation, in spite of different regional leverages. This in part reflects the relatively even allocation of its PLKBs (see chapte 6), and the evidence that where supply costs are high, as in Tangerang, its labor and capital costs are low because of BERIN lower reliance on medical staff and facilities. The ratio between the total spending on family planning and each rupiah spent by EKXBN, can indicate the agency's relative effectiveness in mobilizing and coordinating resources (table 7.13). The ratio is lowest in Tangerang and highest in Kulon Progo and Tapin, where medical infrastructure is highest per - 103 - ELmO and a are there is less need for supplies. - 104 - Tlde 7.11: Mcly Ital Progm ;l .n RP per ELV per mtihli Tam. .MP tA jar 14m]a Tn b 27% 44% 39t 44% 44% 57% 234.3 44.67% capLita 7% 31% 16% 13% 1S 16% 85.0 16.38% N,qMSd:flfZ 66% 256 45% 43% 44% 27% 178.2 38.95% Toa 100 6 100c 6 6 106 457.4 100.0 (267) (464) (406) (45) (52A) (629) Fig 7.5: T-a PLum Ct Ruplah I Month X ELCO 400 300- 200 - 100 Tangerang K. Progo Bantul Ban jar B.Kuala Tapin Labor 73 203 180 198 231 361 Capital 17 143 64 63 64 101 Disposables 177 118 183 195 229 167 ED Labor z Capital M Disposables ']hl 7.12: Ibnl1y To1t1 (XiS, k t cf Ridin In R#Bh pEr Elr =ith q2m. RM tu jw 1ma T*dn MaE Inim HW Mt85 54% 76 62 72%6 54% Z3.2 64.20% 12% 35% 23% 33% 24% 36% 130.7 28.61% Q2un.nity 3% U% 7% 6% 4% 10% 32.8 7.19 Totail 1OC1 100% 100% 100% 10CA 100% 456.7 100.00% (267) (463) (405) (454) (523) (628) Figure 7.6: Monthly Total Csts, by Saurce of Funding Rupiah/Month 400 _ 300- 200 100 Tangerang K. Progo Bantul Banjar B.Kuala -Tapin BKKBN 227 250 284 280 377 341 MOH 32 161 92 148 124 227 Community 8 52 29 26 22 60 2 BKKBN MOH t Community - 106 - Table 7.13: Ratio of Total Cost to BKKBN's Cost by Regency Tang. K.Progo Bantul Banjar B.Kuala Tapin 1.18 1.85 1.43 1.62 1.39 1.84 7.6. Summary The value of resources allocated to family planing delivery ranges from about 270 Rps. per month per ELO in densely populated Tangerang to 630 Rps. in sparsely populated South Kalimantan. BKB manages to mDbil.ize from other govenmsent agencies and the community, an additional 1 Rupiah for each Rupiah it invests in its field operations. The ratio would fall, of course, if BKKBN's administrative overhead costs were included. These figures reflect the finding that, on the whole, BKKBN bears about 50% of family planning delivery costs, the Ministry of Health 40%, and the commnity the remaining 10%. BKKBN does especially better in areas where there is a medical infrastructure because it does not pay full labor value for medical personnel and can rely on longer lasting and less costly methods, notably the IUD, especially relatively lower resupply costs. It is noteworthy, that physicians and other medical staff receive only 8 percent of their government income from BKKBN, even though they report spending roughly 20 percent of their time on family planning activities (chapter 6). The marginal cost to EKKBN of the time that these staff allocate to family - 107 - planning is thus lower than the total value of said time.20 Moreover, in performing family planning activities, the medical staff use buildings and equiprent which are part of the medical infrastructure. Due to the relative scarcity of 'xained medical personnel and facilities in Indonesia, these inputs should not be viewed as free. mTe opportunity or social cost of this activity by medical staff to Indonesian society, therefore, has been treated here as the sunr of both MOH and BKKBN salary and capital cost, a cost clearly higher than what BKKBN itself pays.21 The data on labor cost and its funding, combined with the data about BKKBN staff allocations discussed in the previous chapter, are indicative of the allocation issue confronting the agency. On the one hand, it must respond to some perception of need, as manifested by the size of the target population and a lack of the resources that could provide family planning. On the other hand, it must respond to productivity considerations which are high where resources to coordinate are high - that is, where there is medical and ccmmunity infrastructure. 20 The private incentives for time allocation in this setting do not automatically induce medical personnel to allocate their time in a 'socially' optimal manner; the private opportunity cost of reallocating time between family planning and medical activities is essentially zero for medical personnel whose regular wrking hours are fixed by the government. Therefore, small honoraria which EKKBN pays for medical personnel activities during their regular working hours may have large impacts on time allocation decisions. 21 It is beyond the scope of this discussion to assess the shadow price or opportunity cost of medical personnel engaging in family plannIng. This would entail assessing the "sacrifice" in medical services because of family planning activity. It should also be emphasized, however, that family planning has a clear impact on health. - 108 - 8. PRoGRAM COST-'EFFECTIVENESS 8.1. Introduction The cost to serve a user (C/U) is a measure of program cost effectiveness. It is the ratio between prevalence rate (U/E), discussed in chapter 4, and the value of resources allocated per ELOO (C/E), summarized in the previous chapter. That is, unit user cost reflects both resource productivity as measured by prevalence rate, and resource allocation in a target population as measured by cost per ELC0. The objective in this chapter is to examine the level and structure of user cost, and to start exploring how it correlates with scale of operation, types of inputs, method mix, productivity of inputs, and scome aspects of consumer demand as outlined in chapter 3. Thereby we should identify the delivery strategies and means that can increase the cost-effectiveness of the program, and the institutions who control various aspects of such strategies and means. It is important to stress that long term cost is discussed here: the cost of providing protection to a group of users in a given population in a steady state marked by uniform flows of new users and dropouts vis a vis the size of the grolup and method mix.22 Cne would ideally wish to compare different 22 For an elaboration on this and other conceptual issues, see Chernichovsky, (1990). - 109 - programs but of che same age and where method switching does not take place. New and growing programs tend to have (especially when delivering IUD and sterilization) high initiation cost. And, for the same reason user cost tends to be high where method switching to more permanent methods takes place. That is, programs which have more new users of any kind, tend to be more expensive in the short term than the same programs when stable. This issue will be considered throughout the discussion. 8.2. Program Cost Per User It is costliest - about Rps. 900 per month - to serve the average user in South Kalimantan (table 8.1).23 23 The reader is reminded that the user cost is the cost of serving a user in the population. Ihis definition is not as mXch of an issue in the case of pill users as it might be in the case of ILD users. For the pill, the entire group of users is serviced unremi.ingly, and there is almost an identity between the group of users and the group serviced during any particular period. For the IUD, just new acceptors and those who need replacement are actually treated by the program during, say, the year. Thie cost of an IUD insertion wuld clearly be higher than the cost of the average IED user. The discussion here concerns the long term recurrent cost of protecting a group by two alternative methods. This *a,qarison is particularly valid here because the IUD-based program, in Yogyakarta, is the oldest, about 25 years and has no new users (see Table 4.5). Consequently, the program does not incur initiation cost. If we assume that on the average every user replaces her IUD very 3 years, then a 1/3 of the wcmen are treated by the program every year, and treatment cost is three tunes as much as user cost. - 110 - Table 8.1: Program Cost per User by Type of Cost, in Rupiah per month W. JavaD I YS. Kalim. Mean Distri- Tangerang Kulon Progo Bantul Banjar Barito Tapin (1000 Rp) bution Labor 27% 44% 39% 44% 44% 57% 314.0 44.58% Capital 7% 31% 16% 13% 12% 16% 109.9 15.61% Supplies 66% 25% 45% 43% 44% 27% 280.4 35.81% Tbtal 100% 100% 100% 100% 100% 100% 704.3 100.00% (452) (559) (539) (819) (890) (970) This region has the youngest. program, relatively low population density and ELCO's per resources, and a method mix almost whojlly composed of the pill, which is delivered largely by medical facilities. It costs about half the above figure to maintain a user in Tangerarxg, where population density is highest and resources per ELM is lowest, and where the rnst popular method delivered to the czmm.nit, is the injectable. The cost per user in Yogyakarta is in between these two extremes. The breakdown of user cost by type of cost shows that the share of capital cost is lowest in Tangerang where the target populEtion is largest in relation to medical infrastructure. Capital cost shares are highest in Yogyakarta, which has a heavier medical infrastructure per ErE than the other areas (table 8.1).24 24 In the discussion which follows we combine Kulon Progo and Bantul to "Yogyakarta" and the three South Kalimantan areas to "S. Kalimantan"- - 111 - Labor cost follcws a similar pattern. This appears to reflect the allocation of labor per ELCO and the quasi-fixed nature of labor cost. Cost of supplies ranges from 37 percent of user cost in Yogyakarta to 66 percent in Tangerang (table 8.1). They are highest in absolute terms in South Kalimantan where the pill is most common, and lowest in Yogyakarta, where the IUD is most prevalent. Cost of supplies are of special significance because of their variable and recurrent nature, and the way they figure in public budgets. From a substantive viewpoint, programmatic changes that alter cost of supplies, for given groups of users and level of protection, determine the economic rates of return o such changes, and thereby tlin overall cost effectiveness of a delivery strategy as characterized by method mix. That is, a strategy based on pills, for example, requires a different investment in the delivery setup and the initiation of the group than a strategy based on IUD. The recurrent cost of the two strategies will be different too. Strategies are considered, from the providers perspective, alternatives, when there is a trade-off between level of investment and subsequent recurrent cost. This raises the need to deal with the rate of return to a required investment, say for replacing the pill by ITJD. This return is determined by the recurrent cost savings (of servicing the population with IUD) vis a vis an ongoing alternative (servicing with the pill). From an institutional and budgetary viewpoint, unlike fixed cQst, - 112 variable cost show in recurrent budgets, and of all variable cost coclponents, supply cost are less subject to political and institutional considerations such as the case may be for labor cost.25 Hence, there may be scope to alter them, if the required investment can te made. 8.3 The Effect of Scale of Operations The user cost data just presented, reflect many programmatic aspects related to program efficiency: scale of operations, resource productivity, method mix cum delivery system, and consumer demand, all of which determine the cost-effectiveness of one operation when compared with the other. Indeed, these data conceal basic information about the underlying causes for the user of the different types of prograws. We start with the study of the effect of scale of operation. TIb this end, unit user costs of different programs are ccmpared for similar levels of output. It has been already established above that larger operation per worker tend to be less expensive, but without adjustment for other factors. The point estimates (table 8.2) indicate that to the extent that there are coTmmn levels of output or users, in the range of 2,000-5,000 users per subdistrict, South Kalimantan is still the costliest area and Yogyakarta the least costly. 25 In Indonesia, maintaining employment is a clear policy objective of the government. At the same time the government wishes to ocntain the growth of civil servioe. Such policies may overide efficiency considerations with regard to personnel. - 113 - Table 8.2: Unit User Cost by Region for Similar Output Levels Region Output Mean no. No. of Average range of users subdistricts user cost Tagerang 3,703-5,125 4,457 4 647 Rps. Yogyakarta 2,086-5,158 3,380 38 526 " S. Kalim. 2,252-5,156 2,898 11 837 " These findings are corroborated by the data illustrated in figure 8.1, based on predicted values from linear regression estimtes within the similar cutput ranges. Yogyakarta is still the least expensive within the entire range. In the output range above 4,000 users, South Kalimentan tends to become less costly than Tangerang. These data also support the hypothesis that larger operations of fiald workers in any program type are associated with lower user cost. Clearly, the age of program or its growth rate are of relevance here. The program in S. Kalimantan is the youngest and the fastest growing of the programs and is therefore expected a priori to be the most expensive, ceteris paribus. mis is, however, not as much of an issue here primarily because the older program, in Yogyakarta, is based on the ID which wculd incur the highest initial cost. The youngest program in S. Kalimantan, is based on the pill which incurs couparatively low initiation cost. Still, Yogyakarta has no new users while S. Kulimantan has most (table 4.5), and under more - 114 - long term and stable conditions it might be less expensive. In addition, the discussion of scale of operations is adjusting somewhat for potential initiation cost. As field workers are allocated on the basis of ELCOs, workers with larger numbers of users (given number of ELCOs), are likely to have fewer new users. Hence, one of the reasons larger operations, as defined here, are less expensive is because they have fewer new users. 8.4. Marginal and Average Variable User Cost Both marginal cost - the cost of servicing an additional user - and related average variable cost, comprise cost of supplies (SUP) and those labor costs which vary with output.26 Consequentl.y, this type of cost is less influenced by level of operations (across field workers) than average user cost discussed thus far. As cost of supplies is constant per user, marginal cost would rise (and influence average variable cost to rise as well), only if the marginal productivity of labor is falling when workers need to service a larger population.27 Data were available on cost of supplies, but rct on the variable elements of labor cost.28 Tb establish the marginal cost associated with labor, we estimated this relationship: 26 "Cost of Supplies " is per the discussion in Chapter 7. 27 The high initiation cost in S. Kalimantan would be measured here in labor cost, mainly MDs time required to subscribe pills to new users. 28 These labor cost elements wuld include honoraria, travel, etc. that tend to vary with output. Although this information was available in principle, its association with number of users was not clear. -15 - total labor cost = a + (b x users) where "a" is the fixed element of labor cost and "b" its variable element. Consequently, the marginal cost per user in each region are (SUP + b). Accordingly, marginal oosts per user were obtained (table 8.3).29 Table 8.3: Marginal Costs per User (all cost included) Supplies Estinmted Estirated per variable Marginal user labor costs costs Tangerang 298 44 342 Yogyakarta 204 92 296 South Kalimantan 346 166 512 29 Actual Estimates are (y = no. of users): Tangerang 'btal Labor Cost = 649 + 0.044y; Adj. R2 = .22 (in Rps. 1000) (4.36) (2.58) N = 21 Yooakrt Total Labor Cost = 406 + 0.917y; Adj. R2= .23 (3.83) (3.04) N = 29 Kalimntan Total Labor Cost = 431 + 1.66y; Adj. R2 = .41 (6.06) (4.80) N = 33 - 16 - Here again, servicing an additional user is least costly in the Yogyakarta areas primarily because of their relative dependence on the IUD which has the loest supply costs per average user. It is nost costly in South Kalimantan because of high dependenc on thie pill. supplied through medical infrastnx±ure and personel, in part because of the young age of the prgam. A general comparison of nmrginal cost with average unit cost estimtes suggests tbat rarginal cxst terds to be lowr than average cost in South Kalimantan, and in Yogyakarta in particular. This is usually an indication of '"tco" small operations where economies of scale - vis a vis labor inputs - - are not fully exploited with regard to the quasi-fixed labor costs. Namely, the same wrkers could do more, or the same work could be done, on the average, with fewer workers. - 117 - Figure 8.1: Predicted Average CbstB per User I .1 X1 I { No~~~~~~~~~l. of Users " - 118 - 8.5. Adjustment of User Cost for Alternative Prevalence Estimates It was noted in chapter 4 that the MOWBN program statistics used in this study deviate from the Intercensal Survey (ICS) and the Demographic and Health Survey (CMS). The differences between the data sources are basically in scale and not in method mix. Table 8.4: Ratios of Prevalence Rates from Surveys to BKKBN Annual Reports Intercensal DH Survey Survey West Java (Tangerang) 0.77 0.76 DI Yogyakarta (Yogyakarta) 0.74 0.92 South Xalimantan 0.61 N/A Source: table 4.2 As the prevalence rate affects user cost, an adjustment is made in user cost estimates according to the scale coefficients shown in table 8.4. Since user rates from the surveys are lower than those from BKKBN statistics, user cost is inflated (table 8.5) accordingly. This adjustment shows that the ordering of average user costs by region, does not change, except for an equalization of the cost of Tangerang and Yogyakarta. Adjustment of marginal cost estimates are shown in table 8.6. This - 119 - adjustment is more significant than the previous one because cost is less affected by scale of operation than user cost. Here again, while the ratios between the original estimates change, their ordering does not. If anything, the adjustment amplifies difference between the areas, suggesting that the Yogyakarta marginal oost per user relative to the other areas is even lower thin estimated originally.30 Table 8.*5: Adjustment of User Costs per Alternative Prevalence Rate Estimate (Ratio: Tangerang = 1.00) Study Original Adjustment Adjustment area cost per ICS per DHS Tangerang 455 591 598 (1.00) (1.00) (1.00) Yogyakarta 547 739 592 (1.20) (1.25) (0.99) South Kalimantan 866 1,419 N/A (1.90) (2.40) Table 8.6: Adjustment of Marginal Cost per Alternative Prevalence (Ratio: Tarigerang = 1.00) Study Original area estimate ICS DHS Tangerang 342 444 450 (1.00) (1.00) (1.00) Yogyakarta 296 400 321 (0.87) (0.90) (0.71) South Kalimantan 512 839 N/A (1.50) (1.90) 30 It should be noted that these differences might be even larger as wastage is more likely where the pill is most common, in South Kalimantan, ard least likely where the IUD is most common, in Yogyakarta. - 120 - 8.7. SMMary In this chapter, data on prevalence and resource allocation were combined to estimate the long term cost of servicing a user in each of the study areas. It is costliest (900 Rps. per mnmth) to maintain a user in South Kalimantan. It is half of that cost in Tangerang, the least costly area. In all regions, especially South Kalimantan and Yogyakarta, scale of operations is an important variable influencing user cost: larger (field worker) operations are less costly. mmen scale of operation is controlled for and variable and marginal user costs are examined, vogyakarta has the least costly type of program from a long term perspective. there is a clear indication in the data of a negative association between cost of supplies per user and the permanence of the most common method in use. It is also clear that the IUD and to a lesser extent the injectable require medical attention. Hence, user cost to BKKBN is almost entirely influenced by method mix that, in turn, is to a sutetantial degree set by availability of medical infrastructure, personnel, and consumer demand. Over those, which are rather intimately interrelated, BKKBN has no control. BKKN controls its labor cost, which represents one-fifth of total recurrent cost of field operations, and about one-third of its am cost in those operations. Ihis means that the agency in charge of the program - 121. - appears to have rather limited latitude in terms of controlling its aAm costs, let alone the cost of the entire program, urder current family planninx technology. The next chapter corcerns the question of how BiKBM might improve the oost-effectiveness of its an operations. - 122 - 9. WORKER PRODJCTIVTY AND EFFICIENCY OF FIEID OPERATIONS 9.1. Introduction The prime resources BKKEN can control are its staff of field workers (PLKB) and supervisors (PPLKB). How BKKBN recruits and allocates them determines the effectiveness and the cost-effectiveness of its individual operations. It has been established (chapter 6) that BKKBN allocates its field workers largely on the basis of number of ELJXs, and to a lesser extent on the basis of number of villages. Supervisors are assigned on an administrative basis. The objective in this chapter is to examine these allocation strategies for their cost-effectiveness, and whether EKXBN could improve the cost- effectiveness of its field operations through better staff recruitmeent and allocation strategies. 9.2. Field Worker and Supervisor Productivity: A Model and Hypotheses The resources available to field workers and supervisors are their own time and the community resources they can mobilize: volunteers, distribution points operated by volunteers (VCDCs), and medical resources. Let us assume that the average field worker's potential production in the ccmmunity (Us) is a function of time spent in delivery and coordination (td) and of infrastructure (I) available to him: - 123 - US = f(td, I).31 (9.1) This is a technological relationship representing a worker's potential in the community: the number of users he or she can service with different amounts of time allocated to delivery, given a particular method mix and community resources.32 This function is depicted by curve f() in the upper right quadrant of figure 9.1. It is further assumed that all workers share the same production function or technology, that each strives to maximize output, and that the maximum can be reached by exhausting all working time.33 Clearly, the greater the output, the lower the unit cost per user. A worker is considered a "quasi-fixed" input and the marginal cost of his or her operation entails primarily costs of supplies and possibly some honoraria and travel costs (chapter 7).34 optimal output and minimal unit cost would be 31 For clarity of discussion, the number of variables in this and the functions which follow in this section, is kept to a minimum, without loss of generality. 32 As resource mobilization enhances the worker's productivity by providing more help in the canmmunity, the function f() may be viewed as the function expressing optimal solutions for allocation of time between delivery and resource mobilization. It is further assumed that workers wish to deliver the most efficacious nwetod mix. They are constrained, however, by availability of medical infrastructure and consumr demand levels. 33 These assumptions, the last in particular, are kept to maintain clarity of discussion. 34 The worker's wage is a fixed cost because he or she neither works nor is paid on a part-time basis. It is noteworthy that the marginal costs to society are larger than th, marginal cost to BKKM if other community resources are indeed diverted to family planning. - 124 - achieved at output level USM-35 Let us further assume that demand, delineated by number of would-be users (Ud) in an individual worker's catchment area, is a function of number of ELCOs (E), their socioeconomic characteristics (SE) and promotion efforts, - - field worker time allocated to promotion (tp), or IEC activity, and consumer incentives (CI): Ud - g(E, SE, tp, CI) (9.2) This function is delineated in the lower left quadrant of figure 9.1. Clearly, the nurber of would-be users cannot exceed the numter of ELCrs in any catchnEnt area (Ud = Usm. Staff works to capacity and unit costs are minimal, but demand is not fully exploited. This situation shculd be indicative of a program which does not have enough resources to procure sufficient workers and supplies to service the deand in its target population. The second regime is depicted by g(E ) or g(E2), situations where the worker's maximom capacity to serve exceeds the population's latent demand, - 126 - Udt < U.. In this case the worker has more time than needed to satisfy latent demand for services, and therefore can allocate time to demand promution. This situation should be indicative of a program which can afford to hire more workers than needed to satisfy latent demand. Therefore, the program should promote demand. This can be done through consumer incentives to clients, or by reducing the target populations assigned to each worker so that there is time for delivery, as well as promation. The optimal situation under the second regime would be the singular supply and demand situation, g(EO), leading to point 00, where a worker is assigned to a population in which he or she car, allocate all working time in a way that leads to Udc = U.0, and workers allocate their entire time productively, or are fully employed, t°+tP9=T. Two non-equilibrium situations may prevail. The first is portrayed by point 0', associated with function g(EI) and excessively low demand. A worker produces more than is demanded. This is wasteful: more resourcoe are allocated to delivery than used, and user cost can be reduced. The worker should increase promotion efforts at the expense of delivery efforts, until reaching equilibrium or close to it. The second situation is delineated by point 0" , associated with function g(E2), and excm-;sively low supply. This is not wasteful in terms of program resources; it is socially wasteful in terms of the client's queuing time. The workzr may fine-tune time allocation, more time allocated to delivery and less to promotion, keeping fully employed and minimizing the social cost entailed in queuing. - 127 - Given the worker's productivity potential, delineated by f(.), there may be only one demand function g(EO,.) which makes possible a full equilibrium such as depicted by 0°. The issue is that the worker's marginal promotion efforts may not 'match" their marginal deiivery efforts. For example. in a given range of operations there is excess supply. A worker may decide to reduce delivery efforts by one hour and increase promotion efforts by the same amount of time. The two outcomes may not match; he or she may generate less or more demand than the reduction in supply. If workers are assigned relatively low numbers of ELCOs, their marginal productivity in prcmotion may begin declining ?Atively low levels of time allocation to prcmotion, and it may fall fasL likelihood of excess supply is greater in such situations. Consumer incentives may aggravate the situation in this case. This might manifest itself in staff underemployment and waste of other resources.37 When pills are required to satisfy consumer demand, worker poxxluction potential might be high, as he or she are not constrained, in a relative sense, by infrastructure. Consequently, redundancy of supply (efforts and supplies) is more likely than in the case of, say IUDs, where supply may be constrained by medical facilities and staff. This hypothesis is supported by the evidence suggesting (see chapter 4) that when ccmparing service statistics and survey data, discrepancies are most likely to exist in the case of pills than of other methods. 37 If a orker does not actually sit idle, the social waste of commzn- -y resources which the worker mcbilizes may be higher than the waste of just BKKBN resources. - 128 - It should be further noted that circumstances of excess supply mean a waste for BKKBN, which is paying the worker and for stocking of supplies in any case. There may be savings to the government nonetheless, or society-at- large, if the efficient worker saves on health and other community resources.38 Only under demand situations g(E>E,) indicating likely quRuing, a more productive worker would produce mire and benefit the entire system at the expense of queuing time of the population. Two basic allocation problems are presented here. The first is the allocation of population cum environment to an individual worker. This allocation decision must be taken by program management, at least at the supervisor's level. The second is the worker's time allocation. This is mnst likely to be the worker's own decision which might be influenced by guidelines and his or her supervisor. It must be recognized that in any of the situations discussed above, workers may risallocate their time: too much in delivery and too little in prcomtion, or vice versa. 'The same may be said about promotion inceintives.39 They would be useful when a lack of effective demand is nct an obstacle to higher worker productivity and when the number of ELCCs per worker is not 38 This exemplifies a situation where BMW's incentive to save is lower than what the government's or society's wculd be, as the agency wuld not benefit directly fram such savings. 39 This is part of the optimization process not dealt with explicitly here. For a formal discussion, see Chernichovsky (1990). - 129 - low enough that even with incentives the worker remains eremployed, and incentives just reduce his or her marginal productivity in prcuction.40 40 In the long run, there may be a trade-off between incentives and time allocation to prmotLion, or IBC and the search for new acceptors. Workers are traded with incentives so that there is an optiual allocation betwee hiring workers and incentives. - 130 - Figure 9.1: Field Worker Productivity: Interaction between Supply and Devard in a Catchment Area Us 9 I ~ ~ * I I I X IUs 'I~~~~~~U l l I . l I t *tI Un~ ~ I. I Ud Uto u - I t tdr g(E3) g(EG) g(E5) g(E1) tp - 131 - According to the model, tiia logical sequence of efficient resource allocation should be as follows: a) satisfy latent demand for as long as possible with available resources by hiring workers. Then, if resources are still available - b) hire more workers, reducing the catchment area per worker, and at the same time raise demand through incentives so that workers work to capacity in delivery. If resources are still available, - c) continue hiring workers, reducing further the catchmert area per worker, and start prtaoting demand through IEC and related activity, while balancing time allocated to prawtion and time allocated to delivery.41 The empirical analysis aims to test the hypotheses derived from this framework. These are: a) BKMW's field workers are underutilized (situation 0'), implying that: worker productivity is influenced by demand conditions: only more demand through a larger target population to the average worker, would generate higher productivity; and, productive workers would utilize fewer resources, including their own time (without cost savings to BKKBN). 41 There is an optimization problem ccrcerning the optimal allocation of resources to prnmotive incentives to clients and to staff working in prcmotion or IEC. The principle guiding this discussion is that the situation is such that supply efforts lag behind demand to assure maximum productivity and least unit costs. - 132 - b) B{N's field staff work to capacity (situation O" and above), implying that: denmrd cmrditions do not affect wormker producivity; and, relatively produtive wrkers wuld prodce uKre, and produce at lower unit costs. Each has different irmplications on how Mom might increase the efficiency of its operaticns. 9.3. Field Worker Productivity Estimates The first task of the empirical analysis is to test under which regime EKKBN's field workers operate: one consistent with the first hypothesis, or another, consistent with the second. Uxnder the first, any changes in demand would be met by the workers. In this case the upaard slope of the production function f() would be "traced" (across observations) by the data, without reaching iminq potential producticn within the relevant range. Under a regime consistent with the seocnd hypothesis, upper levels of demand would not be met by the individual (average) worker, and the number of users would not increase beyond a cerLain of number of ELCOs within the range of ELE observed in the data. Hence, the effect of the variables influencing effective demand are the clues to the answer of which regime BK1MM's average field worker operates under in the - 133 - study areas. Ihe statistical analysis therefore concerns the estimation of structural relationship 9.2 under the assumption that workers attempt to satisfy denand arxi that demard conditions determine productivity. It employs the variables listed in table 9.1 with reference to cnoeptual franeork and operational category. - 134 - Table 9.1: List of Variables and Reference to Analytic Framework Conceptial Variablec framework reference Operational reference U -No. of current users Ud and Us Output Scale of Operation and nature of Catchment are E -No. of ELCOs Size of target population Program design, - relationship (9.2) targeting, and scale of operations AE-Area in square Size of catcbment area hectares - relationship (9.2) VL-No. of villages Nature of catchment area in jurisdiction - relationship (9.2) Population Characteristics CH-No. of children under age 5 Socioeconomic Targeting characteristics - relationship (9.2) Worker Characteristics and Pay AGE Qualitative Recruitment SEX - (Male = 1, Female = 0) Attributes of and retention MS - Married (yes = 1, No = 0) td and tp in of workers ED - Years of schooling relationship (9.1) EX - Years of experience and (9.2) Trg - Weeks of training Demand Promoti-on CO-Availability of a CI in relationship (9.2) Proitive coonut seedling scheme incentives to (yes=l, no=0) consumers CR-Availability of a credit scheme (yes=i, no=O) PU-Availability of public facilities (yes=l, no=0) IEC-Proportion of field tp in relationship (9.2) Worker's time worker's time allocated allocation to IDC - 135 - Table 9.1 (continued) Worker Delivery and Coordination Activitie ORG-Proportion of field td in relationship (9.1) Worker's time workers' time spent on allocation such activities Md-No. of MDs in subdistricts I in Program design, VC-No. of VCDCs in area relationship (9.2) and allocation VO-No. of family planning of time to volunteers resource PO-No. of integrated FP and mobilization health posts (yad)42 Several functional relationships are consistent with falling marginal productivity and with an upper level of productivity as depicted by function f () in figure 9.1. Of several such functions, the two providing the best estimates are reported here: the linear-quadratic, and the double and semi- logarithmic combination. Of these, the double-logarithmic (between users and ELmDs) does not subsume the number of ELCOs (per field worker) as the upper limit to worker productivity within the observation range. The results are presented in table 9.2. The data show a close association between the number of ELCOs each field worker needs to serve and the number of users he or she "produces". In all study areas the number of users increases nearly proportional to the number of ELms.43 Accor-ding to the estimates (eq. 1) a 10 peroent increase in the number of ELmos is associated on average with about a 9.4 percent increase in the study areas 42 Trhis operation can be listed under medical resources and modes of delivery 43 This association yields the high "Adjustod R2" values. - 136 - of Tangerang and Yogyakarta, and 7.8 percent in the correspo,ding areas in South Kaiiantan. The non-linear association between number of users and T across observations is also suppoted by the estimated ooefficients (on the numer of ms) in the quadratic equation (equation 2:). This implies some peak, except for Tangerang, in the number of ELCOs that can be served by a field worker. Interestingly enough, in Tangerang, where the numbers of EUE.s served by field workers, are relatively high, no such peak is suggested by the data, the linear-quadratic estimates are insignificant. still, worker productivity, measured by the number of users, increases when each worker, on the average, is assigned to larger numbers of EL(Xs. - 137 - Table 9.2: Field Worker Productivity Estimates, No. of Users or Log of Number of Users as Dependent Variable Indlependient TANtANG YOGYAKARTA S. KALIMAN Variable Eq. 1* Eq. 2+ ECq. 1 EqI. 2 Ecl. 1 Eq. 2 Ln(No. of EIMs) 0.9396 - 0.9358 - 0.7781 - (66.53) (62.83) (9.24) No. of EL s 0 0.6465 - 1.0421 - 0.4097 (24.26) (8.41) (5.78) (No. of ELOOs)2 - -5E-05 - -0.0002 - - (-12.21) -6.35) No. of children 9.70E-06 - -3E-05 0.1565 0.1391 0.1052 (0.28) (-1.44) (1.53) (1.57) (1.17) (No. of children)2 - SE-05 - -2E-05 - (3.42) (-1.68) Availability of 0.0832 49.6565 -0.0425 47.8554 0.0866 47.2504 oomxnut scheme (2.05) (1.79) (-0.55) -0.74) (0.92) (0.83) Availability of - - 0.0331 -46.3409 0.1221 58.1500 credit schem (0.67) (-1.08) (1.26) (1.00) Availability of 0.0445 44.5013 -0.1183 -27.083 public works schme - - (0.96) (1.11) (-.096) (-.37) % time in IBC -0.0048 -2.7302 0.0058 -0.2661 0.0001 1.1827 -2.52) (-3.70) (2.56) (-0.14) (0.06) (1.32) Adj. R2 0.99 0.97 0.99 0.97 0.99 0.89 F 27188 868 15023 325 5628 124 N 134 134 78 78 88 88 Note: t-statistics in parentheses. + No. of users * Log of no. of users - 138 - These findings, suggesting relatively slight falls in user rates as the number of ELCOs per field worker rises, are consistent with the first hypothesis. Productivity is determined by the levels of effective demand in each worker's catchment area.44 Field workers appear underutilized: each worker could service larger populations than they service now. Findings for Tangerang suggest that more can be gained if workers spend less time in promotion, and more resources are invested in consumer incentives. That is, controlling for average number of ELX)s, workers who spend more on delivery and whose populations benefit from incentives, do better on average.45 This finding is reasonable in view of the relative scarcity of resources in Tangerang (chapters 5 and 6). Workers can do better by coordinating medical personnel resources, mainly because they service the larger population with the injectables, which require medical attention. The findings for Yogyakarta suggest that IEC activity might help. But for the study areas in this region and those in South Kalimantan, the results, beyond the strong effect of number of ELOOs on worker productivity, are less 44 It must be stressed that these findings may be strongly influenced by the implicit or explicit target-setting system whereby workers may be assigned relatively fixed user rates, and report accordingly. Data about these rates (see chapter 4) suggest that absolute user rates could be lower. This might seriously hinder any evaluation of the data. 45 As far as incentives are concerned, they may reward areas already achieving high rates. Hence, causality may be the opposite of that suggested by the estimates. - 139 - conclusive.46 on the whole, even if somewhat obscured by target-setting, according to these estimates, field-worker productivity appears to be basically influenced by the size of the target population, and to a much lesser degree by other factors affecting demand, IEC and promotion incentives. In this situation, consistent with the first hypothesis, additional field workers in the population would have little or no impact on output. The model suggests that under these conditions, more workers would imply smller catchment areas to each worker, and no consequent rise in output. To test this proposition, the effect of the number of field workers across subdistricts was estimated for each of the three provinces. Total number of users at the subdistrict level was correlated with the number of EL0Cs and field workers. The relevant regression estinates are shown in table 9.3. There is no effect of the number of workers on number of users when the number of EL:rs is statistically controlled. Even at this subdistrict level of obeervation, there is a close relationship between number of ELCOs and the number of users. These findings reaffirm the previous estimates at the individual field-worker level, indicating that there is underemployment of these workers. Additional workers do not increase output. In Tangerang, they may in fact 46 Because of the simultaneity bias, stemming from the effect of number of users on allocation of more time to delivery efforts, the estimated coefficients on time in IEC are downward biased. The effect of IBC may be stronger than implied by the estimates. - 140 - be associated with lower output.47 Table 9.3: Log of Number of Users as Dependent Variable, Regression Coefficients Ind. Variable Tangerang Yogyakarta S. Kalimantan No. of ELCXs* 1.1988 0.8447 0.9527 (10.95) (11.92) (19.17) No. of Field -0.1808 0.0136 -0.0159 .Workers* (-1.40) ( 0.17) (-0.21) Constant -1.9414 1.0405 -0.1363 (-2.18) (1.040) (-0.47) Adj. R2 .90 .87 .95 F-Stat. 91 99 383 N 21 29 33 Note: t-statistics in parentheses. A Log of 47 This may be an attempt to boost output in low prevalence areas. - 141 - 9.4. Field Viorkers' Time Allocation Another way to examine field worker productivity is to study their time allocation to delivery and coordination. If, as suggested by the firdings just presented, demand driven by number of ELCs is the basic determinanrt of each worker's scale of operations, the number of users should affect the amount of time spent in delivery (td) when workers are fully employed. If this hypothesis is not borne out by the data, workers are underemployed. A simple linear estimate has be tried in order to correlate the potential determinants of the worker's time allocation to delivery. As seen in table 9.4, no variance in this allocation can explain meaningfully. These findings too are consistent with the hypothesis that field workers aze underemployed. This may also explain the obeerved high proportions of their time - 60-70 percent (chapter 6) - workers spend in non-delivery activities. - 142 - Table 9.4: Field Workers' Allocation of Time to Delivery as Dependent Variable, Regression Coefficientsl Tangerarg Yogyakarta Kalimantan No. users 0.00002 0.00002 0.00007 (0.64) (0.69) (0.89) Area (in Sq.) 0.00001 -0.00000 7.28723 (1.18) (-0.69) (0.14) No. villages -0.02621 0.01316 -0.00432 (-1.20) (0.82' (-0.42) Worker seniority (yrs.) 0.00057 0.00058 0.00013 (0.46) (0.98) (0.08) Worker education (yrs.) -0.00644 -0.00680 -0.00151 (-1.17) (-1.25) (-0.11) Constant 0.47731 0.31504 0.40613 (6.20) (4.24) (2.33) Adjusted R2 0.00 0.00 0.00 Note: t-statistics in parentheses. I The dependent variable is log[p/(1-p) where p = percent of time allocated to delivery. - 143 - 9.5. Worker Effectiveness Under the conditions established in the previous sections, there is little or no scop.. for productivity gains in delivery by the individual worker. The average worker is not working to capacity because of a lack of effective demand. Clearly anything, includinr particular traits of workers, that can promote denand, would benefit the system. There is little theoretical or empirical knowledge of what makes family planning workers more effective in given populations. Is it their ability to prcmote demand? Or, is it their ability to manage delivery efforts? Alternatively, there is no knowledge of what worker attributes affect demard and what attributes affect supply, or both. To test worker effectiveness in line with the nodel and the evidence presented thus far, we explore the effect of worker characteristics on their effectiveness, measured by the prevalence rate: the ratio of number of users to number of EL(Xs. Thsxse worker characteristics that make a difference in this rate across observations a priori affect demand. The estimated equation is an extension to the previous estimates, and is based on relati-Lship (9.2): lwig(U/E)j =hariXi s + vj where Xi is the ith characteristic of the jth wcrker, andf vj is a rardam - 144 - disturbance term. The results are reported in table 9.5. Table 9.5: Logarithm of User Rate by Field Worker as Dependent Variable, Regression Coefficients Independent Tangerang Yogyakarta S. Kalimantan Variable Age -0.003 -0.002 0.000 (-1.35) (-0.55) (0.07) Sex (male 1) 0016 -0.124 0.007 (0.35) (-2.57) (0.10) Marital status (married=l) -0.127 -0.029 -0.001 (-1.77) (-0.29) (-0.13) Number of children -0.014 0.011 -0.007 (-1.26) (0.50) (-0.13) Years of schooling -0.027 -0.008 -0.042 (-3.74) (-1.18) (-3.37) Using family planninq (Yes=l) 0.180 0.056 0.082 (2.64) (0.066) (0.077) Adj. R2 .84 .56 .74 F 111 17 40 Note: t-statistics in parentheses. - 145 - The estimated coefficients indicate that educated workers are less effective; worker effectiveness declines with level of schooling. Following the model, this suggests that relatively well-educated workers are poor prcm*ters of fandly planning. There is also some indication in the table 9.4 findings that they spend less time than other workers in delivery. Hence, the hypothesized "efficiency" of the workers may manifest in their ability to realize the situation. Their effectiveness my in fact manifest in that they do not waste oommunity resources by idle operations. Not surprisingly, in the study areas of Yogyakarta where the IUD is common, female field workers do better than their male counterparts. There is some indication that workers in Tangerang who themselves use faimily planning, perform better. Other worker characteristics do nat appear to make any difference to worker performance. 9.6. Supervisor Productivity Supervisors are in imiediate control of field operations. It would be expected therefore that they use their prerogative to allocate field workers in a raticnal fashion on the basis of locl potential and field worker productivity. It was established in chapter 6 that field workers are allocated largely on the basis of ELCms and to a lesser extent on the basis of number of villages and area the field workers cover. It is expected therefore that - 146 - under general circunstances of apparent underemployment, more efficient supervisors will allocate fewer workers than their less efficient counterparts, all else being equal. To test this hypothesis, the equations on field workei- allocation (table 6.2) were re-estimated with selected supervisor characteristics.48 The data show (table 9.6) that experienced and more educated supervisors in Tangerang tend to "employ" more field workers than their less experienced and educated counterparts. A similar pattern, although with less predictive power statistically, is observed in South Kalimantan. In Ycgyakarta there is ome evidence that more experienced supervisors employ fewer field workers. Tums, except for Yogyakarta, the evidence does not support the hypothesis that a priori more efficient supervisors may save on field workers in a situation where there is scope to economize on this resource. 48 "Area" and "Med. Staff" were omitted because of statistical insignificance, and to economize on statistical degrees of freedom. - 147 - Table 9.6: Determinants of Number of Field Workers, by Supervisor Regression Ooefficients DI South Tarigerarg Yogyakarta Kalimatan Constant -4.281 4.287 -2.982 (0.281) (0.009) (0.111) Education 0.299 0.068 0.115 (0.288) (0.364) (0.256) Worker experience 0.258 -0.095 0.160 (0.023) (0.082) (0.138) Eligible couples 0.0003 0.0002 0.0003 (0.000) (0.036) (0.001) children per eligible couple -2.034 -0.636 3.730 (0.118) (0.653) (0.006) Villages 0.285 -0.102 - (0.000) (0.071) N 18 21 27 Adj. R sq. .89 .39 .57 F 29 4 10 Note: t-statistics in parentheses. - 148 - 9.7. orker Pay and Efficierx Are there anry guiding principles in setting salaries that can be eventually related to workerts productivity? To answer this questiorn, two salary equaticns were estimnted: one with two alternative dependent variables: a) the base salary, corsidered a salary policy variable set by MKKBN management; and, b) all labor incomes, including all ooponents (salaries, per diem, travel, honoraria, other pay) which reflect work activity. t These equations conform to commn wage equations which take the form: Log Wj = a +Zbi Xij where Wi is the munthly salary of the jth worker and Xjj is the ith characteristic which is hypothesized to determine his or her salary.49 "a" is some base salary, and "b" is a vector of coefficients relating "Xi" to salary. The estimated coefficients for field workers are reported in table 49 This function also turned out to have the best fit of several tested. - 149 - 9.7.50 The most important explanatory variables are education and experience.51 In the base salary equation, we see that workers receive, on average, salary increases of 3.6 percent annually. We can also see that junior high school graduates earn 21 percent more than elementary school graduates; senior high school graduates earn 38 percet more; and university and academy graduates earn 30 percent more. Allowing for all sources of labor inccme also increases the earnin3s of South Kalimantan workers, who earn higher total incomes. In both equations, Tangerang workers earn marginally less than Yogyakarta workers, though the differences are not statistically significant. Mien we allow for additional sources of labor inoome, the wage equation estimtes change. The predictive power of the equation falls; the education variables becane less significant in explaining wage variations. 50 Similar estimates for supervisors, did not yield any results of statistical significance. 5- The zero order correlation coefficient between age and experience is 0.18. This rather low correlation suggests that BKKBN hires workers of all ages and that worker trnover may be high. - 150 - Table 9.7: Worker Salary Determinants for BKKBN Field Workers Log of Salary Measures as Dependent Variables, Regression Coefficients Dependent Base All labor Variable salary costs paid by govt.a Constarnt 3.263 3.254 (16.22) (17.22) Sex O.BE-4 0.073 (male - 1) (-0.62) (1.47) Age 0.002 0.002 (0.46) (0.35) Work experience 0.036 0.046 (4.89) (5.89) DIEtic Junior high* 0.215 0.137 (2.49) (1.41) Senior high 0.379 0.169 (4.23) (1.72) Higher education* 0.302 -0.020 (1.51) (-0.32) Tangerang -0.014 -0.076 (-0.24) (-1.20) South Kalimantan 0.147 0.243 (2.22) (3.34) adj R2 0.157 0.157 F 6.83 6.83 N 250 251 Note: t-statistics in parentheses. * Dummy variables - "primary education" excluded. ** Dummy variables - "DI Yogyakarta" excludel. a Includes travel expenses, honoraria, etc. - 151 - BKKBN's pay to its workers cannot be justified on productivity principles. But, then in this environannt where workers appear to be assigned to too small populations, the effect of worker traits on productivity cannot manifest Other labor-related incones tend thus to equalize earnings amnmg field workers, except for the effect of experience or seniority. This suggests that more educated workers may be investing less effort; they travel and participate less in seminars, etc. This may be further evidence that educated staff work less - an indication of efficiency under excess production capacity - and indeed be less wasteful in terns of utilizing resources, beyond their base salary. 9.8. Summary It was argued in chapter 8 that BKXEN has relatively little latitude to affect the cost effectiveness of the family planning program. The major cost component it can affect directly is the organization's own labor cost, which constitutes about one-fifth (1/5) of the total unit user cost, and about one- third (1/3) of the user cost immediately under BKKE1N's control. Within these limits, the data about labor productivity presented here indicate that there is scope for savings by assigning larger catchment areas, mxre ELOsc, to the average field worker. This would mean savings in labor - 152 - cost and would bring to bear the full potential of the field staff. More educated and experienced workers might do better than their less experienced and educated counterparts when conditions of extra production capacity are eliminated. This would also provide supervisors with mDre leeway to exercise some managerial control, which appears to be lacking. Supervisors engage in the same activities as field workers, and more experiencxd supervisors tend. to miwbilize mcre workers, contrary to what might be warranted on cost- effectiveness grounds. - 153 - 10. PRGRAM EFFICIECY 10.1. Introduction In the previous chapter we explored ways in which BKKB could improve the allocation and management of a resource under its direct control: family planning field workers. This, however, has been shown to be a rather narrow option to improve the effectiveness of family planning delivery because BKKBN has almost no control over capital expenditures, and by implication over infrastructure which determines to a substantial extent method mix and cost of supplies, the major recurrent cost element to the system and to BKKBM. An overall strategy that aim to increase program efficiency must consider all resources, consumer preferences, and possibly changing technology. The question becomes: how can the Government of Indonesia increase the long-run cost-effectiveness of family planning delivery? A comparative analysis of the three study areas - which may represent three models of delivery - can help to answer this question. It was shown in chapter 8 that for identical levels of operation, the study areas of Yogyakarta are the mnst cost-effective in terms of unit cost. It was argued in chapters 8 and 9 that raising scales of operations and improving allocation of workers can improve cost-effectiveness in all regions, in Yogyakarta in particular because the potantial for labor savings there is highest. That is, Yogyakarta may potentially become more cost-effective, through savings on labor cost, than current data indicate. - 154 - The major issues that can explain the regional differences leading to the observed cost-effectiveness differentials must relate to the ccmbination of three intimately related factors: consumer demand, method mix, and the delivery system. 'The objective in this chapter is to study the potential contribution of these factors to program efficiency through a comparative analysis of the three regions. 10.2.. The Consumer's Perspective Observed output (contraceptive prevalence) levels indicate an interaction between consumer preference or demand and contraceptive availability or supply (chapter 3). No data on consumer preference was collected as part of this study. Consumer preference, however, is a major variable affecting program success, since a program is mest productive when it can satisfy clients' unmet demand. TIo gain sce understanding about the effects of this variable, we analyze how method mix relates to levels of prevalence. We assume that where consumers have access to IUDs, they also have access to all other methods. If they choose this method, they must prefer it. Alternatively, if in areas where the use of ID is relatively high, prevalence rates are high, this would suggest that the 'cafeteria approach' may be productive. People would respond favorably to the IUD, if available. - 155 - To examine how prevalence correlates with the ratio of users of a particular method among ELCOs, the following relationship is escimated: UMi/E = a(U/E)b where UMi represents the users of method i, U represents total contraceptive users, E is the number of ELCOs, "all is a constant indicating the average ratio of users of method i, and "b" represents the rate of change. The estimated coefficients are reported in table 10.1. Tn each region, the coefficient and the common method is approximately unitary (=1), suggesting that prevalence rates and the user rates of the ocrman methods change proportionately. The estimates show that IUD is the method mcst associated with high prevalence rates. In all regions, but especially in Tangerang and South Kalimantan where IUDs are not commKon, a 10 percent rise in prevalence is associated with a higher than 10 percent rise in the ratio of IUD users. The equivalent figures are 15 percent in Tangerang and 26 percent in South Kalimantan. In South Kalimantan the same holds true for injectables: a 10 percent rise in prevalence is associated with a 13 percent increase in the ratio of ELOs using injectables. This tendency is Corr-o,rated by the data presented in chapter 4 suggesting a rise of use of injectables among new acceptcrs. These data suggest that, at the margin, higher prevalence rates are positively associated with the m,re permenent methods in all tbree regions, - 156 - especially where these methods are not oomuon. Moreover, even in Yogyakarta where the IUD is most oommnn, the other methods are not associated with high prevalence rates. These findings support the hypothesis that, when permanent methods are available, users tend to employ them. Moreover, the demand for permanent methods may increase with time because clients who start with pills usually will continue with a permanent method, namely, inje-tables, IUD, and others such as implants and sterilization, should they be available. Additional support to the preference of a permanent method can be found in the times series shown in figures 10.1 - 10.3 relating to method mix in the three study areas. It is evident that user rates of permanent methods (IUD, injectables and sterilization) rise once they are made available. The reader =m.st bear in mind that the study areas differ not only in medical infrastructure, but in cultural make-up as well. There are a variety of hypotheses concerning the effects of cultural factors which are far too involved to undertake in the present study. Briefly, Trangrang, which has a relatively high non-MUslim population, way be amenable to injectables, unlike the Mhslims who tend to sanction its use. By the same token, the Christian population of DI Yogyakarta, who follow a more syncretic form of Islam, may be agreeable to the IUD, contrary to the Muslim population in this area, who tend to avoid its use. Education and tradition, as well as the history of the program, also have far-reaching effects. The influence of these factors merits additional data and research. - 157 - rig. 1'. I , 1`'6gntl ouni lclly Co,il,rneplive Meltiod Mlx, Di '(ogynkatin, IlB0-WI'd 40 40 *.-----* 7-- - -- SO0~ - ,_.__ , ,____ 20t -- ----- -4- r: s to 1 t4 1 2 1 4 1 12 3 41 t 2 Ft 4 t 2 1 4 t 2 1 4 t I n~tet 1 L902 I t1Z. LT WeI 19s4 I t.-. - leo11 * nnt 4- pill * Condom l" ngaoIebIC _"- SI.fl.llontIrIloth -tS lninlan rtIttr-9l Monthlr Ste, rlee ftlhallelle ri9. 10.2 : 'otoolut Co,uitfcepllvo Melltod Mlx, West Java, 19ot-of *0g '° I - -~'N --i;;., ".., -, I 41141142.41341311 , ' 4 1 2 't 4 t Jt J 4 t 2 't 4 t 2 3 4 1 2 st 4 1 2 I 19t i1902 I 1903t 1 9eo lecet Ileot 1 luo ~~~-t- rml Ct ondeom -0. tnolsllona _ra- g9gq9g9ggZ9qj,,eQhe -t tmelint SoOroet frKfllIl Uenfl,y Saortis SiSflleite Fig. 10.3 . Ieroetil Oumilerly otillrooeWlilve Metliod Mix, Boull Kolmontlan, lg 1f -I9Or loo so. 40t rto X e 4 1 t1 1t 41 11 te 411 *t o 4 1 t O1 4 ;T * 4 t a -w*| x - en-er I'll"411 Ilost 11141114 1114118ties I- rnlealab1 - *lln*llantOlha Sourfee, IIKIII nthi Sailin *elo SIat9s|teS - 158 - Table 10.1: ULD, Pill, Injectable and Total Prevalence Rates, by Region: Regression Coefficients South Tangerang Yogyakarta Kalimantan IUDs Constant -2.705 -0.618 -2.992 (-12.78) (-12.89) (-8.61) U/E 1.487 1.059 2.645 (4.24) (7.03) (4.59) Adj. R2 0.11 0.39 0.21 F 18.01 49.37 21.09 Pill Constant -5.519 -2.038 -0.218 (-14.80) (-13.00) (-7.07) U/E 0.972 0.544 0.902 (6.02) (1.10) (19.26) Adj. R2 0.20 0.00 0.81 F 36.27 1.22 370.99 Ifniectables Constant -0.467 -3.030 -2.650 (-9.75) (-22.91) (-14.13) U/E 0.975 0.423 1.282 (12.92) (1.02) (4.50) Adj. R2 0.55 0.00 0.18 F 167.04 1.03 20.21 Note: t-statistics in parentheses. - 159 - 10.3. Efficiency of Method and Mode of Delivery Because of the differences in infrastructure, and probably because of the age of program, the three study areas are markedly different in modes of delivery (see chapters 4 and 5). In Tangerang, the pill and injectable - the most common methods there - are largely delivered through c and VCDCs. In Yogyakarta, the IUD, the most common method, is delivered through medical facilities. In South Kalimantan, the most ccmmon method, the pill, is still largely distributed by VCDCs and medical facilities (table 5.1). The scale of operation for each program is different. These differences are clearly reflected in user cost of methods in all the study areas as shown in chapters 7 and 8. It is important to establish how much it costs to deliver similar methods through alternative delivery modes in order to identify efficient delivery mechanisns. There is, however, the problem of allocating cost to different methods when delivered by the same mode, especially when no detailed accounting data are available. To overocme this problem to a certain degree, hedonic price equations (Chernichovsky and Zmora, 1986) are estimated for each method in order to establish method cost by delivery (table 10.2). These estimates, which are combined with accounting data, are crude and should be used as general indicators. - 160 - Table 10.2: Estimated Unit Cost of Method, by Type of Cost (in Rps. per Month per User) Method Fixed Labor Supply Total cost cost cost cost (Statistical Estimates) (Acocntir) Tame= IUD 121 179 6 306 (4.39) (4.78) Pill 22 103 341 466 (1.24) (4.30) Injectable 19 22 306 347 (2.04) (1.87) Adj. R2 .44 .56 F 15 23 N 53 53 IUD 91 56 6 153 (3.97) (0.84) Pill 22 226 341 589 (0.40) (1.36) Injectable 460 924 306 1690 (2.47) (1.71) Adj. R2 .31 .06 F 15 3 N 90 90 IUD 294 82 6 382 (3.11) (0.57) Pill 10 90 341 441 (0.85) (5.01) Injectable 541 591 306 1438 (5.76) (4.18) Adj. R2 .42 .36 F 27 21 N 108 108 Note: t-statistics in parentheses. - 161 - The estimates shown in table 10.2 indicate that even when delivered in srall quantities and initiation costs are comparatively substantial, the IUD in Tangerang and S. Kalimantan is the least costly method per user, largely due to low supply cost. It is least expensive in Yogyakarta, where it is delivered in large quantities.52 Unlike the other methods, however, the IUD involves relatively high capital cost. The pill, by comparison, is more expensive than even the injectable in Tangerang, where the injectable is commn, because of the labor and supply costs associated with pill delivery. The injectable is expensive, (as is any other method), if delivered in small quantities which is the case in Yogyakarta and S. Kalimantan. These estimates and their implications are not affected by the differences in reported prevalence rates between alternative data sources reported in chapter 4. In table 10.3 the cost estimates for the different methods were adjusted per the method specific prevalence rates reported in table 4.2. In the first instance (center column) all costs, including those of supplies are adjusted. That is, wastage is assumed in supplies; each actual user is consuming on the average more than is alloted to him through the distribution system.53 In the second instance (right col.) only labor and capital cost are adjusted. That is, each user is "assigned" supply costs as if she or he use 52 The reader is reminded that the programs in Tangerang and especially in S. Kalimantan, are younger than in Yogyakarta and therefore would be mare expensive because they grow faster. The reader is further reminded that the user cost for the IUD, is not the cost of an actual insertion which would be about three times as much. 53 The reader is reninded that BKKBNs statistics are based on the distribution system. That is, if there are, for example, 80 actual pill users for every 100 users implied by the cycles of pills distributed, it is assumed that each actual user, is "consuming" 1.25 cycles; 0.25 are wastage. - 162 - the appropriate supply. The index numbers suggest that the above conclusions hold almcst regardless of data source on prevalence. The 1UD is the least costly method in the long run. The pill and the injectable appear sensitive to scale of operation. Each of the methods is less costly where it is relatively common. In Tangerang, where the pill and injectable share comparative prevalence rates, the pill in the range of 42-47 percent and the injectable 30 percent of users (table 4.2), the injectable is by all accounts less costly to deliver than the pill. From a financial perspective, the IuD requires government investment in appropriate medical infrastzuture and staffing. This problem is somewhat lass acute for the injectable which, as dsmnstrated in Tangerang, can be delivered in the coamunity (at least in an urban area) at relatively low cost. Because of its relatively high supply .oost, the pill will remain expensive to the goverrment, even when it is assumed that a substantial portion of its labor and capital costs to deliver it are borne by the community through volunteers and the Village Contraceptive Distribution Centers (VCDC) network. - 163 - Table 10. 3: C0st of Method per Month by Region Original All cost adjusted capital & labor costs cost icS DM ics iis Rps. Ird- Rps. Irxd- Rps. Irxd- Rps. Irmd- s. Ind- ex ex ex ex ex MN. IUD 306 100 244 100 312 100 245 100 312 100 Pill 466 152 566 233 524 168 493 201 482 154 Inject 347 113 345 141 341 109 347 142 346 111 YOG. IUD 153 100 121 100 139 100 122 100 140 100 Pill 589 385 718 593. 1218 877 643 526 854 612 Inject 1690 1105 1037 856 1082 779 1155 944 1192 854 S.. .L. IUD 382 100 520 100 NA 518 100 NA Pill 441 115 456 88 NA 444 86 NA Inject 1438 376 1208 232 NA 1257 243 NA Source: tables 4.2, 10.2. - 164 - 10. 4. Conclusions A comparative analysis of the three provinces indicates that the IUD, and to a lesser extent the injectable, are methods which, if available, may be used and contribute to high prevalence. Moreover, the IUD appears, from a cost perspective, to be a relatively cost-effective method. It requires, however, capital investment which is beyond the means and control of EKKBN. This holds true to a lesser extent for the injectable, which requires trained medical manpcwer, again outside the current jurisdiction of BKKBN. The relative delivery cost of the different methods are inversely related to their efficacy. Hence, the cost effedtive methods are also the =K,st efficient. These methods are probably also the most efficient when their potential demographic ispact is considered. The mean age of user of the IUD, pill, and injectable for Java and Bali is 32.5, 30, and 29 respectively. That is, the relatively small differences in ages between the groups, indicate that the reproductive potential and risk of pregnancy are about equal amcng the different user groups.54 This discussion is partial. Clearly, altering the delivery system, especially in favor of methods which require medical facilities and staff, requires two types of investment: in facility and staff, and in the cost of 54 We are indebted to S. Cochrane for suggesting this particular and essential adjustment. More data on the age distributions is needed for a mxre corLclusive evaluation. - 165 - initiating a new method. mese costs are not considered here. The data merit a detailed cost-benefit analysis, as they strongly sugzgest that such investments may pay off, especially because they would cotrihbte to medical care as well. - 166 - 11. SUMMARY AND CONCLUSIONS 11.1. Introduction The study objective is to explore potential ways to improve the cost- effectiveness and efficiency of the Indonesian family planning program. The study examines resource allocation, cost, funding institutions, and output of the program at grass root level in selected regencies in three distinc provinces: West Java, Central Java, and South Kalimantan. It .s based on data about the program's field operations collected during November 1986 to March 1987, and routine service statistics of the National Family Planning coordinating Board (BKKEN). The six regencies participating in the study are case studies; they are not meant to represent the entire National Family Planning Program. Beyond describing and measuring all program field resources and relating them to cost and output, the study has program policy implications that fall into three cate-gories. The first deals with potential improvements in cost- effectiveness of BKKBN's field operations through allocation and management of the basic resource under the immediate control of BKKBN: its field staff. The second category concerns potential improvements in program cost-effectiveness through a broader program strategy related to all resources and agencies participating in the program. The final category deals with the issues of resource mobilization and cost recovery. - 167 - In this chapter the major study findings are summarized and their implications discussed. The chaptor concludes with a note about future research. 11.2. Summary The study areas reflect Indonesia's varied and rapidly changing cultural, demcgrapic and socioeconomic environwznt. Tangerang, in West Java, an area next to the capital Jakarta, is a densely-populated in-migratiom area. It has a high ratio of population to health facilities and other resources for family plamning delivery. This area may be indicative of things to ocm: rapid urbanization with a lagging social infrastructure. The Yogyakarta study ar3as, Kulon Progo and Bantul, are more traditional agricultural areas, close to a slow-growing urban center. They have a relatively strong social infrastructure and resources for family planning. Tangerang has a subetantial Chiinese Buddhist population, while Yogyakarta has a Christian population. These two densely populated areas ccntrast with the study areas of South Kalimantan, which are sparsely populated and have more resources per capita for family planning, but which are spread over large areas and are not easily accessible. All residents are prcminently Muslims. Understanding the political econamy of the program is crucial to identifying ways to improve its cost-effectiveness. MKKw is a coordinating agency which cperates in a diverse and quickly evolving social environment. - 168 - BKKBN's options towards i%proving the cost-effectiveness of the program, are conditioned by four constraints: a) national adminstrative reulations, b) availability of medical and community resources, c) consumer demand and d) KKBNM's own budget. Regardless of area size, population, or other features which might influence resource allocation, BKKB only has one family planing supervisor per administrative subdistrict. TIhe medical and cmmunity constraints concern the control of those resources ulsed in family planning which are to a substantial degree external to BKKBN. These resources and consumer demand shape the nature of the program, exhibited in contraceptive method mix. While EKKBN can provide some advice about the supply, distribution, and administration of these resources, it exercises minimal control over them. BKKBN is usinq medical infrastructure controlled by the Ministry of Health. In the community, BKKBN field workers do oversee village family planning volunteer activities, but the volunteers who play a major role in outr.ach activities, primarily via distribution of pills and ooxnoms, are closely associated with the local village civil admninstrators and leaders and thus are not entirely answearable to BKKBN. Clearly, BMON operates within its own budget which covers labor and supplies. Availability of medical infrastructure per eligible couple (ELTa) varies greatly between regions. One health center is available for about 9,500 eligible couples in Tangerang. This is twelve times the ratio of population to health facilities in the Kulon Progo regency of DI Yogyakarta, and fifty- five times that in the sparsely-populated Tapin province of South Kalimantan. - 169 - Medical personnel are also unevenly distributed. Distances to facilities are short in Tangerang and long in South Kalimantan. These differences correlate with modes of delivery and mix of contraceptive methods. Hence, different regions possess very different medical resources available for family planning. BKKN adapted to this differential allocation in spite of its potential influence of modes on delivery and method mix. Considerable variations are observed across regions in the allocation of EXKKN staff in relation to their jurisdictions. Tangerang has the highest ratio of eligible couples to all personnel types. There are above 1400 ELOOs per field worker in Tangerng, compared with 1200 in Yogyakarta, and about 900 in South Kalimantan. This suggests the program's inability to respond quickly enough to rapidly changing demographic conditions. At the same time, within each of these provinces, labor BKKBN's staff allocations appear to be based on rational decision-making: allocation of field workers is on the basis of ELrOs, villages and size of catchment area. This is less the case in DI Yogyakarta and more so in Tangerang where resources are relatively stretched. Such a situation suggests that local managers are more responsive to objective criteria of need in allocating scarce resources than central management. Where need is more pressing, as in Tarwerang, responsiveness is greater than where it is not, as in Yogyakarta. There is also evidence that senior and educated supervisors tend to oversee more field workers than do their junior end less-educated peers. - 170 - Workers basically mirror their target population and the age of the rrogremi. DI Yogyakarta has a larger share of workers who are older, female, more experienced, and better-trained. Time allocation for bcth field workers and their supervisors is mainly divided between demand generation, supply, and admLnistrative activities. This is similar across regions. EKKBN staff spend most of their tim on demand generation: search for new acceptors and promation activities. The staff report working "by the book", which gives rise to the hypothesis that on the whole they may be underrutilized. they do not seem to be under obvious pressure to respond to local variations in need. If there is such pressure, it may exist in Tangerang. Medical staff spend about one-quarter of their time in family planning activities. There is a positive correlation across study areas between levels of all types of resources provision per eligible couple. The implication is that availability of medical infrastrLcture, chiCh is not under the control of BKKBN, may be a key factor in the developlment and nature of the program: BKKBN allocates more of its own resources where medical infrastructure is more available. The different environments and relative availability of medical resources, as well as program age, are reflected in prevalence rates and method mix. According to BKMWN's service statistics, both Tangerang and the South Kalimantan regencies have contraceptive prevalence rates of approximately 60 percent. - 171 - Injectables, however, predominate in Tangerang and pills in South Kalimantan. The Yogyakarta regencies have the highest prevalence rates, around 80 percent, and TUDs are the most common method. Levels of prevalence data used in this study (from BKKBN sources) are higher than those from available survey data. Both types of data, however, lead to the same implications because method mix information is consistent between all data sources. Modes of delivery are closely related to contraceptive methods. In all regencies surveyed, more than three-quPrters of pills are delivered through the Village Contraceptive Distribution Centers (VCDC), followed by clinics and outside-clinics. The amount of pills delivered through private sector channels is negligible, but is highest in Tangerang, where availability of program facilities per eligible couple is lowest. In the three South Kalimantan regencies, whare the pill is most common, a greater proportion of pills is delivered through the clinics and outside-clinic activities than in other regencies. The two DI Yogyakarta regencies each deliver over 92 percent of all pills through the VCDC. This phencmenon is oonsistent with the hypothesis that m9dical infrastructure is important for new programs even when no clinical intervention is required. In Indonesia, a medical check-up is required for new pill acceptors. Condom delivery is quite varied among the locales surveyed. In Tangerang, the only regency where private sector channels play more than a minor role, virtually all condoms are delivered through private pharmacies. The number of condoms delivered, however, is the smallest of all the surveyed regencies. In DI Yogyakarta condom are delivered almost exclusively through the VaCCs. In - 172 - the Barito Kuala and Tapin regencies of South Kalimantan, clinics deliver over half the condcms, VCDCs one-third, and the remainder are provided by outside- clinic staff. Outside-clinic activities are respcnsible for 57 percent of injectable distribution in Tangerang, the regency with the heaviest injectable use. Clinics deliver most of the remainder. In all other regencies surveyed, injectables are delivered predrni*r.ntly through the clinics, followed by outside-clinic activities and private sector chanrnels. The value of.all resouroes allocated to family planning delivery ranges from about 270 Rps. per moth per ELMO in densely-populated Tangerang to 630 Rps. in sparsely-populated South Kalimantan. MKXBN manages to mobilize, on tLLe average, an additional one rupiah fram other government and comnunity agencies, for each rupiah it invests in its field operations. The ratio would fall if EKKBPN's administrative overhead costs were inicluded. These figures reflect the finding that on the whole, BK[OEN bears about 50 peroent of family planing delivery costs, the Ministry of Health about 40 percent, arxi the community the remaining 10 percent. EKKXN bears less of the cost bkrden in areas where there is medical infrastructure, because it does not pay full labor value for medical xersonnel arnd can rely on-longer lasting and less costly methods, notably the rJD. It is noteworthy that physicians and other medical staff, who report spending roughly 20 percent of their time on family planning activities, receive only 8 percent of their goverrwrent incaue from EGXN. - 173 - Data on prevalence and data on resource allocation were combined to estimate the cost to service a user in each of the study areas. It is costliest (900 Rps. per month) to maintain a user in South Kalimantan. It is half that cost in Tangerang, the least costly area. The figures are higher and the discrepancies wider when survey data are used instead of BKKBN data. Ir all regions, especially those of South Kalimantan and Yogyakarta, scale of operations is a crucial variable influencing user cost. Mmen scale of operation is controlled for and variable anr marginal user costs are examined, Yogyakarta is the least costly type of program. The structure of user cost underscores the complex nature of the politioal economy of the family planning program, and the rather limited soope available to BKKBN to improve the program's cost-effectiveness. The major uiser cost component borne by the agency is supplies, about two-thirds of BKKBN's cost in its field operations. This cost element is alirst entirely influenced by method mix which, in turn, is to a substantial degree set by availability of medical infrastructure and personnel, and cansumer demand. BKK2N has control cver its labor cost, about one-fifth of total recurrent cost of field operations, or one-third of its own cost in these operations. This means that the agency responsible for family planning activity appears to have rather limited latitude in terms of its own cost control, let alone the cost of the entire program. The productivity of BKKMW staff is therefore the basic cost-effectiveness - 174 - element under the agency's immediate control. BKKBN appears to raise staff productivity by having the average field worker run a larger operation. Since workers seem to operate under conditions of excess capacity, their catchment areas can be raised with no loss in troduction. More field workers in the average jurisdiction does not correlate with more output. A ccmparison of Tangerang and Yogyakarta is fruitful. Yogy-akarta appears to be relatively "saturated" with field workers, so that their time allocation does not appear associated with productivity: the allocation of field workers' time to coordination, IBF and administrative tasks cannot be explained in a systematic way. There is same indication that wurkers may be more productive in areas where there are more children below age five per EFT . Taken at face value, the data suggest that labor productivity is determined by number of ELCOs. Of course, this may be a statistical artifact: in an administrative framework where targets are set, workers may report according to these targets. Omparative analysis of the study areas irdicates that the IUD and the injectable are methods which, if made available, would be popular and contribute to high prevalence, assuming that cul.ural and religious conditions shaping consumer demand do not restrict its use. Data about new acceptors suggests that the injectable is increasing in popularity. The IUD is also relatively cost-eftectiv- cause of relative low loig run supply and maintenance cost. It is followed by the injectable. These - 175 - methods are probably also the most efficient when their efficacy and potential demographic impact is considered. The mean age of users of the IUD, pill, and injectable for Javra and Bali is 32.5, 30.0, and 29.0 respectively. That is, the relatively small age differences between the groups, indicate that their reproductive potential and risk of pregnancy are about equal. 11.3. Implications BKKBN has limited scope to alter the nature of the program in any particular area under current family planning technology and the program's political economy. Major gains in cost-effectiveness can be brought about in the long run primarily through altering cortraceptive method mix in favor in the more permanent methods on which this study is focused, the IUD and the injectable. Method mnx affects the program's major cost compconert - cost of supplies - which even if managed by BKKBN, is not determined by the agency because method mix is influenced strongly by medical infrastructure. CNost-effective gains can be achieved by a strategy that combines investmnt ir infrast.-.-ture, adoption of new family planning technology and modes of delivery, and influencing consumer demand. Within this general constraint, the study suggests 3everal ways NE4XN might improve the cost-effectiveness of its operations and incur savings through modification of its target-setting policy, allocation of field workers, - 176 - and improvement of workers' time use. First, the effects of target-setting are observed throughout the analysis. Numbers of users closely parallel numbers of eligible couples, a relationship unlikely to be so exact in a program lacking targets. Empirically, field worker characteristics fail to correlate with output levelc. This is exceptional, because one would likely predict that staff characteristics, such as education, age, and years of experience, would relate with contraceptive output. The lack of such association may be a function of target-setting, where field workers habitually report their targets. Those failing to achieve their targets may misrepresent their performance by reporting higher figures, whereas those who could potentially achieve higher levels than what is targeted may only achieve expected norm. Second, taken at face value, the data strongly suggest tha. field workers can be assigned larger target populations, even in areas where their presence Patters on the margin (as in Tangerang and South Kalimantan). That is, present prevalence levels can be maintained with fewer field staff, or more feasibly, that new field workers need not be added eren as populations of eligible cuuples grow. It appears that BKKBN allocates its main resource, field workers, in a relatively uniform fashion across regions, without particular attention to productivity. In areas where the program is well-6stablished and not too dependent on outreach activity, as in Yogyakarta, savings may incur by reducing the MKKBN's field staff, and limiting their activity to IBC. In these areas, - 177 - the cost of outreach activity is high because it serves a rather limited number of users, and the cost of pills and injectables - typical outreach methods - is relatively high. Income-generating schemes serve areas with high prevalence rates, like Yogyakarta. Such a policy is consistent with the idea of rewards rather than inducement. There may be scope to transfer these funds as incentives in low prevalence areas, especially if there is excess production capacity and the potential for worker productivity is higher. A long-term cost-effectiveness strategy must be considered within the context of the political economy of the Indonesian health and family planning system. The health infrastructure required to deliver efficient methods is not under the control of BKKBN, but under the Ministry of Health and other government agencies. A cost-effective strategy would thus require that both BKKBN and Ministry of Health inputs be well coordinated. The relevant strategy and policy changes would necessitate that fundamental decisions be made outside BXKBN as part of a wider welfare decision-making process. While IUD delivery incurs low recurrent cost in labor and supplies, this method demands relatively high investment in infrastructure, personnel and start-up costs. Even if consumers pay the full cost of IUD use, includirg recurrent capital cost, the private sector cannot be epected to finance at the outset the investment required to support IUD. This argument is even strcnger considering that private health facilities would be placed in the position of servaing ccmmnity health needs. Although the IUD represents the - 178 - most cost-effective and efficient method, the public sector will still need to provide many of the inputs required to support it. The proposition that the government supports clinically-based methiods, at least in financing the investment, would be less acceptable on economic grounds in sparsely-populated areas such as South Kalimantan, although on moral and political grounds of equity there is a need to do so. There the injectable would be appropriate in the long-run as a relatively efficient method. This would require government support for training personnel. EKKBN's recent moves to hire medically trained field personnel who can deliver the injectable and possibly the implant, is certainly a cost-effective mNove. It provides for an outreach program not based on the pill, on the one hand, and not on medical infrastructure, on the other hand. These argtments do not suggest that the government should refrain from supporting non-clinical methods, at least in the short run, given favorable oonditions. The study findings can help deal with some issue of shifting some of the financial burden of the program to the con .nity, as implied by the KBManiri or privatization idea adopted by BKKB. tIese cmnsidera.tions include the cost recovery potential of the fee, and its impact on equity and prevalence or program efficiency. The community bears about one-tenth of tota delivery costs, a share that increases in outreach activity. The pill and the condom are delivered for - 179 - BKKBN almost exclusively through volunteer outlets. The burden on the community is sonewhat less in the case of the injectable because of the involvement of medical personnel. -he contribution of the community is least in the case of delivery through medical facilities, such as in the case of the ID or even the pill in South Kalimantan. The estimated nonthly cost to service a pill user ranges from about 450 - 600 Rps. Most labor and capital cost are borne by the community. The larger share in cost of pill delivery, however, lies in its supply. Supplies' costs accouht for some 60 to 80 percent of total cost.55 Under program privatization, entrepreneurs would have to bear labor and space costs currently borne by the volunteers or the ocomnity. The price they charge for c=ntraceptives would need to embody these costs. That is, even if clients pay the full cost of the pill (and condom), the net cost recovery from the conmunity would be less than 100 percent in view of the community's current contribution. Any subsidy, mnst likely of supplies, would even further reduce net cost recovery from the community. A subsidy in the range of 10-30 percent, by conservative estimates, might not amount to a net cost recovery because this cost is currently borne by the community in the case of the pill. It would merely constitute a shift of income from government to entrepreneurs. 55 Another social efficiency consideration not discussed here is foreign exchange. While Indonesia currently receives pills and condoms as donations to encourage provision of these methods, this may not be the case in the future. These methods would then also become a foreign exchange burden. One should note, however, that only minimal numbers of pills and condoms are bought from foreign assistance. - 180 - Tihere are additional consideraticns. As the program nrw stands, at least in the study areas, each cammunity shold possess a VCDC regardless of its physical, social, or ecarhic enviroment. Under a private system, entrepreneurs may be able to make a profit in somr comanities bit not in others. In this case, =cmunities would receive differential advantages. ihose living in poor communities would either have to bear the cost of travel ir to wealthy camanities to receive their pills and ondcmns (provided they were available within a reascnable distance), bear the time cost of waiting for the mnothly Vgxu= visit, or opt to do without. This lack of availability, however, could spur clients of poorer coumunities to use more permanent methods. Privatization may differentially affect entire regions if the pill requires a fee and the IUD does not. DI Yogyakarta, a province with high IUD use, would be less penalized by a shift I , payment for pills and condoms than South Kalimantan, with its hiqh pill use. Since DI Yogyakarta already enjoys advantages (particularly with regard to health infrastructure), assessing fees for pills and condcms, rather than IUDs, wculd increase its privileges even further. Charging for pills and not IUDs could aggravate regional disparities. Assessing ai fee for non-clinical methods, while continuing to subsidize clinical rethods, should have the effect of motivatirg clients to use the latter. Since perismnent methods are cost-effective, this shift could clearly help improve program cost effectiveness. It my, however, not ptvvide much by way of cost recovery for the goverrnent, since it would have to furnish health facilities and personnel to meet increased demand. As there is a shortage of medical resources in rural and remote areas in Indonesia, an expansion of - 181 - demand for clinical methods may run up against those shortages. Cxnsequently, in spite the government's intention to continue to subsidize costs for those unable to pay, privatizing pills and condome could negatively affect the decision of those on the economic margin and young users to oontinue to use contraception. The consequences of a general policy to recover cost from non-clinical methods might have adverse consequences both in terms of equity and prevalence. The cost recovery potential of IUD is higher than the pill's. First, the cmmunity's contribution in IUD delivery is minimal: any charge for IUD delivery would therefore entail a net shift of burden from the government to the ccmlunuity. Second, the IUD is used by older and presumably better-off people who may have a low price elasticity of demand, and who live close to h3alth centers. There is an additional reason: IUD fees are collected in clinics and may be applied toward financing community healtlh facilities, whereas the pill is largely sold by non-medical outlets. No single universal oost recovery strategy may be appropriate across Indonesia. Fee-setting may be appropriate for Yogyakarta for all methods, with different reasons for each. Privatizing pill delivery may be most effective in areas like Yogyakarta, where it is expensive because a few are served, and those few have other more efficient and effective options if they wish and can use them. Loyal users of a particular method may, in fact, have low price elasticities (and high cost recovery potential). The reasons for a fee for the IUD were mentioned abcve. There may also be scope for charging fees - 182 - for conrdms ard pills sold in medical facilities, as is relatively common in South Kalimantan. This would discourage expensive use and delivery of methods that do not require these facilities. It would simultaneously encourage use of IUD, and to a lesser extent injez-tables, that require such facilities. 11.4. Future Research The study clearly opens a broad research agenda. While the study is based on its cwn data collection efforts, it is clear that the analysis could have been based on a lightly modified data collection system than BKKBN's current system, and an appropriate analytic infrastructure. The issues and implied prograimatic changes, all merit more focused research. Operations research into the allocation and activities of field workers would be high on the agenda. A crucial elemeait clearly missing in the data is the consumer's perspective. No strategy and progranatic change, especially concerning cost recovery, can be adequately assessed without regard to consumer response, especially in the diverse and fast changing demographic, economic and cultural environment of Indonesia. - 183 - BTRT.OGRAPHY Ananta, A. and Molynoux J., "Population Dynamics in Jakarta: Its Impact on Food Needs", Unpublished paper. 1987. BKKBN, Master List: Bulan November 1986, Biro Pencatatan dan Pelaporan. Jakarta: Badan Koordinasi Keluarga Berencana Nasional, Jakarta 1986A. BKKBN, Hasil Pendataan PUS dan Peserta KB Bulan Oktober-Noember 1986, Biro Pencatatan dan Pelaporan. Jakarta: Badan Koordinasi Keluarga Berencana Nasional, Jakarta, 1986B. Biro Pusat Statistik, Survei Prevalensi Indonesia 1987: Laporan Sementara, Jakarta: Central Bureau of Statistics, 1988. Chernichovsky, D. and Meesook, O., Regional Acpects of Family Plarning and Fertility Behavior in Indonesia, World Bank Staff Working Paper Nc'. 62, Washington, D.C., 1981. Cher.ichovsky, D. and Zmora I. "A Hedonic Prices Approach to Hospitalization Costs", J. of Health BEcnomiC, 5: 179-191, 1986. Chernichovsky, D., Lernmn C., and Rahardjo P., "Choice of Family Planning Method among Field Workers and its Impact in the Population". Mimeo. Jakarta. 1988. Chernichovsky, D. and McLaughlin R., "Cost Recovery, Efficiency and Equity in Family Planning Delivery: A Proposal for a Work Program". Mimeo. IPPF. New York, 1988. Chernichovsky, D., "Cost and Cost Effectiveness in Family Planning; Theoretical considerations and Practical Guidelines" Processed, 1990. Easterlin A.R. and Crimmins M.E. The Fertility Revclution: A Supoly-Demand Analysis. University of Chicago Press, Chicago & London, 1985. Freedman, R., Siew-Ean K., and Bondan S., Modern Contracecgtive Use in Indonesia: A Challenge to Conventional Wisdom, Scientific Reports, International Statistical Institute, No. 20 March, 1981. Haryono, S., and Shutt, M., "Strategic Planning and Management of Population ProgramTes: An Indonesian Case Study," Monograph Prepared for the Eleventh International Conference of the International Council on Management of Population Programmes, Beijing, 1988. Hugo, G., "New Conceptual Approaches to Migration in the Context of Urbanization: A Discussion Based on Indonesian Experience," Revised draft of a paper prepared for a seminar on "New Conceptual Approaches to Migration in the context of Urbanization" organized by the International Union for the Scientific - 184 - Study of Population's Comuittee on Urbanization and Population Redistribution, Bellagio, 1978. Hugo, G., "Indonesia: Patterns of Migration to 1971," in Migration arnd Development in South-East Asia: A Demographic Perspective, edited by Robin J. Pryor, OVLP, pp. 173-221, 1979. Hugo, G., Hull, T., Hull V., and Jones G., The Dema sic Dimension in Indonesian Develcment, Oxford University Press, Singapore. 1987. Hull, T., "Changing Patterns of Marriage," Research Note 74, The Australian National University International Population Dynamics Program, May 1987. Hull, T., and Dasvarma,G.L. "Fertility Trends in Indonesia 1967-1985," Bulletin of Indonesian Econnmic Studies, Vol.XXIV, No. 1, April 1988. Hull, T., and Bhakta G., "Multivariate Analysis of Infant and Child Mortality in Java and Bali," Research Note 27, The Australian National University International Population Dynamics Program, September 1984. Hull, T., and Hull, V.,"Health Care and Birth Control in Indonesia: Links Through Time," Research Note 53, The Australian National University International Population Dynamics Pr-A.ram, March 1986. Hull, T., and Hull, V., "Population Change in Indonesia: Findings of the 1980 Census," Bulletin of Indonesian Eoonomic Studies, Vol. XX, No. 3, December 1984. Hull, T., Hull, V. and Singarimbun, M., In.nesia's Fam_ily Planning Stc-En Suocess and Challenoe, Population Bulletirk, Vol 32, No. 6, Novembar 1977. Judd, M., Kadeis in Indonesiia, USAID. Jakarta, 1987. Tawrence International Ltd., Long Term Family Planning Strategy Final Report, Jakarta, 1988. Lerrman, C., Moeljodihardjo, S., Pandjaitan, S., and Molyroux, J. ,"Assessing the Net Correlation of Family Planning Program Inputs With Indbnesian contraceptive Prevalence and Method-Specific Use Rates", Studies in Family Plannii , 20:1. 1989. Population Reference Bureau, Wbrld Population Data Sheet, 1988. Prescott, N., Carlson, B., Bongaarts, J., and M"Nicoll, J., Indoresia: Trend in Fertility and ContMgMaceive Prvl en World Bank Report, WItshington, D.C. 1986. Rietveld, P., "Urban Develownent Patterns in Indonesia", Bulletin of Indonesian EHonomic Studies, Vol. 24, No. 1, April 1988, pp. 73-96. Sinquefield, J., and Sungkono, B., "Fertility and Family Planning trends in - 185 - Jdva and Bali." International Family Planning Perspectives, Vol. 5, No. 2, June, 1979, pp. 43-58. Streatfield, K., "A Comparison of Census and Family Plamzing Program Data on Contraceptive Prevalence, Indonesia."Studies in Family Plannin., Vol. 16, Number 6/Part I, Nov/Dec 1985, pp. 342-349. Streatfield, K., Reliability of BKKBN Prevalence Statistics: A Comparison of BKKBN and Census Figures. Gadjah Mada University, Population Studies Center, Yogyakarta, 1984. Streatfield, K., and Larson, A.,"The 1985 Intercensal Survey: Infant and Child Mortality Levels." Research Note 14CS, The Australian National University International Population Dynamics Program, Canberra, 1987. Suharso, A. Jr., Redman, H. and Husain, I., Rural and Urban Migration in Indonesia, (National Institute of Economic and Social Research), Jakarta, 1976. PRE Working Paper Series Contact BIQ ~~~~~~~~~Author Die for p2aper WPS607 Abolishing Green Rates: The Effects Donald F. Larson March 1991 D. Gustafson on Cereals, Sugar, and Oilseeds Simon Glance 33714 in West Germany Brent Borrell Merlinda Ingco Jonathan Coleman WPS608 Cross-Ccuntry Studies of Growth Ross Levine March 1991 CECMG and Poiicy: Methodological, David Renelt 39175 Conceptual, and Statistical Problems WPS609 A Sensitivity Analysis of Cross- Ross Levine March 1991 CECMG Country Growth Regressions David Renelt 39175 WPS610 Can Preshipment Inspection Offset Alexander J. Yeats March 1991 J. Jacobson Noncompetitive Pricing of Developing 33710 Countries' Imports? The Evidence from Madagascar WPS611 Tariff-based Coanimodity Price Jonathan R. Coleman March 1991 S. Lipscomb Stabilization Schemes in Venezuela Donald F. Larson 33718 WPS612 Education and Productivity in Lawrence J. Lau March 1991 WDR Office Developing Countries: An Aggregate Dean T. Jamison 31393 Production Function Approach Frederic F. Louat WPS613 Price-Wage Dynamics and the Simon Commander March 1991 0. Del Cid Transmission of Inflation in Socialist Fabrizio Coricelli 39050 Economies: Empirical Models for Hungary and Poland WPS614 Accountability in Public Services: Samuel Paul March 1991 E. Madrona Exit, Voice, and Capture 37496 WPS615 Socialist Economic Growth and Heng-fu Zou March 1991 A. Bhalla Political Investment Cycles 37699 WPS616 Optimal Nonlinear Income Taxation Ravi Kanbur March 1991 J. Sweeney for the Alleviation of Poverty Michael Keen 31021 Matti Tuomala WPS617 International Poverty Projections Sudhir Anand March 1991 J. Sv'eeney Ravi Kanbur 31021 WPS618 Poverty and Development: The Ravi Kanbur March 1991 J. Sweeney Human Development RP.port and the 31021 World Development Report, 1990 WPS619 Foreign Direct Investment in Laurence Cockcroft March 1991 S. King-Watson Sub-Saharan Africa Roger C. Riddell 33730 PRE Working Paper Series Contact Tnle Author XL for paqer WPS620 Have Commercial Banks Ignored Sule Ozler March 1991 S. King-Watson History? 33730 WPS621 Sensible Debt Buybacks for Highly Enrica Detragiache March 1991 S. King-Watson Indebted Countries 33730 WPS622 How Factors in Creditor Countries Sule Ozler March 1991 S. King-Watson Affect Secondary Market Prices Harry Huizinga 33730 for Developing Country Debt WPS623 World Bank-Supported Adjustment Vittorio Corbo March 1991 A. Oropesa Programs: Country Performance Patricio Rojas 39075 and Effectiveness WPS624 Choosing Policy Instruments for Gunnar S. Eskeland March 1991 A. Bhalla Pollut,vn Control: A Review Emmanuel Jinienez 37699 WPS625 How Trade and Macroeconomic Ramon Lopez March 1991 K. Cabana Policies Affect Economic Growth 37946 and Capital Accumulation in Developing Countries WPS626 The Macroeconomics of the Public William Easterly March 1991 R. Luz Sector Deficit: The Case of 39059 Colombia WPS627 The Role of Institutions in Poverty Sharon L. Holt March 1991 WDR Office Reduction: A Focus on the 31393 Productive Sectors WPS628 The Indonesian Family Planning Do\ Chernichovsky March 1991 0. Nadora Program: An Economic Perspective Henry Pardoko 31091 David De Leeuw Pudjo Rahardjo Charles Lerman WPS629 An Atheoretic Evaluation of Success Pat: ick Conway March 1991 D. Ballantyne in Structural Adjustment 37947