Pollcy, Research, and External Affalrs J WORKING PAPERS.- Population, Health, and Nutrition Population and Human Resources Department The World Bank August 1991 WPS 759 The Profam lla Family Planning Program, uolinibia An Economic Perspective Jesus Amadeo Dov Chernichovsky and Gabriel Ojeda Profamilia, an affiliate of the International Planned Parenthood Federation, provides n,ore than 60 percent of Colombia's family planning services. In 1986, Profamilia recovered more tlhan half of its costs, which is rare for family pianning services. But it could have provided more protection for the same amount of money. The Policy. Rescarch, and Extema! Affairs Compicx disnhbutes PR I: Working I'dpers to dtsscmmalc thc findingiol scirk in progress and to encourage the cxchangc of idcas among Ilank staff and all otlhcrs inicrctcd mn devclopmcnt issues I hsc papers carry thc ndnfle. of the authors, reflect only thcir views, and should hc used and citLd accordmngly T'he finduLgs, intcrpretrions. and conclusions arc thc authors' own '[hey should not be attnhutci to the W'orld lBank, its Board of Directors, its managcncnt, or any of its mcmcbr countrncs Plc,Research, and External Affairs Population, Health, and Nutrition WPS 759 This paper - a product of the Population, Health, and Nutrition Division, Population and Human Resources Department- is part of a largereffort in PRE to examine the relative importance of constraints of demand and supply on the usc of contraception. Copies are available free from the World Bank, 1818 lI Street NW. Washington, DC 20433. Please contact Otilia Nadora, room S6-065, extension 31091 (113 pages, with tables). Profamilia, an affiliate of the Intemational run than this study found to be truc for the short Planned Parenthood Federation, provides more term.) than 60 percent of Colombia's family planning services. o The clinical program (delivering mainly the IUD) and the outreach program (delivering Profamilia's outreach effort (CBD) dclivers mainly the pill) are the most cost-effective. The mainly pills in rural and outlying urban areas, voluntary sterilization program is the least cost- through 100 field workers. Its two clinic-based effective because of the higher cost of steriliza- programs provide (1) voluntary sterilization and tion, the hcavy subsidy for sterilization, and the (2) clinical scrvices: gynecological consultation, higher mean age of clients who are sterilized. It intrauterine devicc (IUD) services, and over-thc- might be more efficient to shift emphasis from counter sales of contraceptives. sterilization to the other two programs. In 1986, these threc programs delivcred o Fees for service should be seriously consid- more than I million "couple years of protection" ered, and more research done on the issue. More (CYP) at a cost of about US$6.43 million. The demand could be met with more workers, and sterilization program provided the most protec- higher prices - particularly for slerilization- tioIn. The clinical and CBD programs each might not reduce revenucs. provided about 43 percent of revenues. Tlhe outreach programn accounted for 31 percent of e More resources should be targeted to areas costs, the clinical program 39 percent, and the where there are proportionately more mothers voluntary sterilization program 30 percent. and where people are better educated (and hence more receptive to family planping). Aniadeo, Chemichovsky, and Ojeda address the question: Could Profamilia have provided o Experienced and married workers sell more more protection with the same resources? They in the outreach program than their junior, found that: unmarried colleagues. Experienced workers tend to be paid more than inexperienced workers, but o Operations tend to be constrained by limited married workers tend to be paid less than unmar- personnel and supplies. With more of each, ried workers. It would pay to retain experienced more protection could be delivered. staff (who are more likely to be married). o The labor costs and unit costs of contracep - In both the clinical and surgical programs, tion are lower in the outreach and clinical output would increase if there were proportion- programs, which can be expanded with available ately more nurses and fewer doctors. infrastructure. Thc marginal unit cost of volun- tary sterilization is higher partly because sur- The underlying hypothesis of this study geons are paid "by the piece." (But the effects of (which remains untested) is that there is suffi- educating the people about sterilization may cient demand for the various operations to make sterilization more cost-effective in the long expand. The PRE Working Paper Series disseminates [he findings of work under way in thc Bank's Policy, Rcsearch, and Extemal i AffairsComplcx. An objecctivc ofthc scries is to get thesc findings out quickly, even ifpresentations arc less than fully xolished. Thc findings, interpretations, and conclusions in thcse papers do not necessarily r.present official Bank policy. Produced by the PRE Dissemination Center TABLE OF ONTF[ E=aXUrIVE SUMMARY PREFACE 1. INDU ON .. .. .. ....... ...** 0 1 2. PLATIONAD FALYPLLANNING IN C4BIA . ...0...... 3 2.1. The Population: Size, Grwth, ar Distribution 2.2. Poplxation Policy and Family Planning 3. PROFAXELIA . . .......... . . . . . . . . . . 0 8 3.1. Brief History 3.2. The Commmity-Based Distribution (CD) Program 3.3. The Clinic-Based Progranm 3.3.1. The Clinical Program 3.3.2. The Voluntary Sterilization Program 4. PROGRAM EFFICIENY: TE ISSUJES AND ANALYTIC FRAMEWORK .. ...... 21 4.1. Introduction 4.2. The Efficiency Issue 4.3. Costs, Reource Productivity and Internal Allocation of Resurces 4.4. Alloation Across Programs and Across Methods within Programs 4.5. Statistical Approach 5. IIE BD . . .. . . . . .. . . . . 37 5.1. Objectives and Framwork of Analysis 5.2. Field Worker Operations: A Model and Hypotheses 5.3. Data 5.5. Productivity and Effectiveness of Field Workers 5.6. Contraceptive Prices 5.7. Program costs and wage [etermination 5.8. Cost-effectiveness of Field Wrker Operations 5.9. Method Mix 5.10.Conclusions: the CBD Sub-Program 6. I{ CLMIC-BAS PRAS . . . . . . . . .. ......... . . . 73 6.1. Objective anid Franrk of An&aysis 6.2. Clinics Resources, Envirament, and ProductivitV: Data 6.3. Costs of Clinic-Based Operations 6.4. Estimticn Procedure 6.5. Ptriuctivity and EffectiveneSS of Clinical Operations 6.6. Variable Labor Costs and EfficiencY of WOrker Allocation 6.7. Fixed Capital Costs and Scale 6.8, Method Mix 6.9. nclusions: The Clinic-Based Sub-ProgramB 7. P-WTIVE (SUB) PRGRA EFFICIENCY .0. .0. .0 . .0 . . . . . . . . . 97 8. KCLJIIS .* * 0 0 . 0 . 0 . . . . . . . . . . 99 List of Tables and Figures Table 2.1: Pop.latiotr Characteristics by Region . . . . . . . . . . . . 4 Table 2.2: Distribution of Contraceptive Users by Source of Supply . . . 7 Table 3.1: Basic Characteristics of Field Wbrkers . . . . . . . . . . 11 Table 3.2: CbD Program Prices to Consumer . . . . . . . . . . . . . . . 11 Table 3.3: Profamilia CBD Program, 1986 . . . . . . . . . . . . . . . . 13 Table 3.4: Clinic Area, Consultations and Surgeries by Type of Clinic, 1986. . . . . . . . . . . . . . . . . . . . . 17 Table 3.5: Estimated Prices for the Surgical and Clinical Sub-Programs, 1986 . ......... 17 Table 3.6: Profamilia Clinic-BEzsed Sub-Programe, 1986 ... ... . . . 18 Table 5.1: List of Variables by Operational Category and Coxnptual Relationship .. . . . . . . . . . . . .53 Table 5.2: Regression Coefficients (Natural Logarithm) Quantity of Sales of Contraceptives, Regional-Level Estimates for CBD Program . 54 Table 5.3: Regression Coefficients: (Natural Logarithm) of Contraceptive Sales, Individual Worker-Level Estimates for CBD Program . . 56 Table 5.4: Regression Coefficients (Natural Logarithm) of Worker Wages 66 Table 5.5: Adjusted Tbtal CYP Gained by Investment in any Method in the CBD Program . . . . . . . . . . . . . . . . . . . . . 70 Table 6.1: Cost Components of Clinical operations . . . . . . . . . . . 77 Table 6.2: Regression Coefficients (Natural Logarithms) of Clinical Sub-Program. . . . . . . . . . . . . . . . . 83 Table 6.3: Regression Coefficients (Natural Logarithms) of Surgical Program . . . . . . . . . . . . . . ....... . 84 Table 6.4: Percentage Changes in Clinical Personnel to Produce a 10% Increase in Output with No Budgetary COmsequences . . . . 87 Table 6.5: Regression Coefficients of Program Oost . . . . . . . . . . 89 Table 6.6: Marginal Oost and Marginal Revene by Method ....... . 93 Table 6.7: Clinical Program: Adjusted Tbtal CYP and Relevant Data by Method in Clinic-Based Programs (per 100 pesos invested) . . . 94 Table 7.1: Adjusted Total CYP and Relevant Data for Profamilia Sub- Programs . . . . . . . . . . . . . . . . . . . . . . . . . . 98 Figure 4.1: Shares in CYP, Gross Value of Sales and Costs, by Sub-Program 23 Figure 4.2: CYP, and Value of Sales (x 1000) per 1000 Pesos by subp . . . . . 34 Figure 5.1: Field Worker Productivity: Interaction between Supply and Demand in a Catchmeent Area . . . . . . . . . . . . . . . . . 45 EFAN N ESU1 . . . . . e . . . . . . . . . . . . . . . . . . . . * . . . . 104 ANNEX :Clinics ty . . . . . . ......................... 105 AM 2: The Worker's Optiml Time Allocation betwemn Delivery and Resource Mobilization .. . . . . . . . . . . . . . . 106 ANNEX 3: Potential Bias in Dend EAsticity Estimates . . . . a . . . . 108 ANNEX 4: Allocation of Nurses to Clinical Operations . .. . . 109 ANNEX 5: Reallocation of Ms and Nurse for Higher otput withot Budgetary Cmsequeanos. . . . . . . . . . . . . . . . .111 ANNEX 6: Marginal Consultation Cost of IUD, Surgeries and Sales of Contraoeptives . . . . . .. . a . . . . . . . . . 113 i EXBUEXVE SUMMR Profamilia, an affiliate of the International Planned ParEnthood Federation (IPPF) in Colaibia, is a migvemental, nt-for-profit organizaticn providing more than sixty percent of the country's family planning services. fhe ireminder is provided by the governmeqt and the private swtor. Profamilia's program is vertically organized; it provides mainly family plannng ard closely related services, and does nat provide services sucih as maternal and child care. Profamilia's outreadi effort, the Crmunity-Based Distribution (D) sub- program, delivers mainly pills in rural and outlying uran areas. It is run by about 100 field v -cers. Te agency has two clinic-nised (sub-) programs, administered throuh sam 38 clinics with a stf of abt 500. ¶he first is the Clinical Sub-program which provid gynecological osultatiors, IUD servioes and over-the-mter 1=sales of cmitrae The d is the Voluntary Sterilization Sub-program. In 1986, Profamilia's three sub-pdelivered approxiitely 1,267,919 oaple Years of Protection (CYP) to the population of Colazbia at a cost of 1,252.8 million Colcabian pesos ($US 6043 million) ¶th program recovered about 647.3 million pesos (SUS 3.32 million), or jre than 50% of its cost. in this regard Profamilia is ocuoaratively unique family planing ii Aording to Profamilia's acoounts, the sterilization program provided in 1986 the larqeSt share of CYP, 61%, followed by the CED program, 27.6%. Ihe CBD and the Clinical Sub-Programs provide: the major share of revenue, about 43% each. In terms of total costs, the CBO progran acconmted for the siallest share, 31%; the Clinical Sub-Program 39s, and the Voluntary Sterilization Sub-Program for 30%. These data reflect key policy and manag decisions cacerninag fees charged to clients, allocation of resources between sub-programs, and allocation of resorces within sub-proralnD. Although Profamilia is conidered one of the best run programs, the question is, ncrxtheless: caold Profamilia have done better with its resources by providing more protection than it did? to answer this question, we examine Profamilia's resource allocation ani costs of operations in relation to output (volume and value of cartraceptive sales) and in conjunction with population characteristics, nmethod mix, and the manner in which resources are allocated. We study the CYP unit cost of Profamilia's overall service delivery, and of each sub-program separately, to identify ccrparatively cost-effective services. The study employs a cross-sectional analysis of the operations of 97 field wrkers and 38 clinics in 1986. The data show a positive correlation betwen labor output and input in all operations. Profamilia's operations mist therefore be following, althcugi not necessarily satisfying, demnnd for family planning. The indication is iii that operations are, on the average, constrained by labor and supplies availability: with more of these resources the program could deliver more ptuwUn. The data indicate that the CBD and the Clinical Sub-program can be expanded with the available clinical infrastructure. For this reascn, and because of the quasi-f ix4 nature of labor cost in these two sub-programs, higher levels of outpxts are associated, on the average, with lower unit costs of contracption. The same does not hold for the Surgical Sub-progrw. Unit cost of sterilization rises with output levels because sterilization has a caomaratively high marginal cost, which is in part associated with the payment method "by piece" to surgeom. Of the different sub-proas, the Clinical sb-program delivering mainly the RID, and the CBD Progrm delivering mainly the pill, are the most cost- effective. Per peso invested in each, these programs yield mnt CYP, adjusted for cost-recovery potential and users' age. me Voluntary Sterilization Sub-program is the least cost effective because of the omparatively high cost of sterilization, the high subsidy to clints who obtain this rethod, and their high mean age relative to the clients of other methods. Given the ,WParative efficiency of the sub-program and the viability of expanding each sub-prograw's operations, overall efficiency might be impoved by studying the possibility of shiftin resourcs, within the existing institutional onstraints and policies, from the Voluntary Sterilization Sub-program to the CBD and Clinical sub-program. iv Fees for service should be studied further as a vehicle to impove Profamilia's performnce. Since the evidence suggests that more demand could be met with more workers in all sub-programs, and that higher prices way not necessarily reclwe revenues especially in the sterilization Sub- Program, there may be scope to raise prices of sterilization in order to finance additional staff and supplies in the other sub programs, and increase overall contraceptive delivery. This issue merits, hcwever, more research about demand for contraception in the different programs, and about the factors which influence this demand. There is scope to icease the cost-effectiveness of any of the three sub-programs individually by more careful targeting of operations, better mix of labor inputs, and improved use of community resources. Productivity of outreach operations is higher where there is a higher concentration of mothers in the population, and productivity of clinical operation is higher where the population is more educated. At the nurgin, targeting or shifting limited resources towards those populations could therefore inrease the cost-effectiveness of the different progras. In the CBD Sub-Program, experienced and married workers sell more than their junior and unmarried colleagues. While experienced workers are paid more than inexperienced workers, married staff are paid, on the average, less than unmarried staff. Retaining experienced staff (who are also mDre likely to be married), should therefore increase productivity and possibly cwst-effectivoass. In the Clinical-based sub-program the ratio of nurses v to physicians crrelates with outptt of clinics; higher nurses to physiciars ratios are associated, on the average, with higher oatput- gmr is scope, thamfore, in both the Clinical and Surgical sub progre to i n OtPUt wi tRt additional cutlays-, eqg. iq=Ne efficierny, ty tradingr pWicians f mwxz nf trme wifhn a re&aScble ranc. c%tIMnity resRcas tead to augnmit p resouro in the CMD outreac activity. The nunme of pouits of sale adinist by a field wrker in the Lmuity, CTrrelates strorly with a v e sales. Wile thre are ro data available on the costsr of these sale points,, thir iulat on worker produtivity suggests they my be a majr meam to inCrease cost- effectiveass in the CBD progrm. The firndig of this study sugest that ther is scope "co ineasre the efficieny of Profamilia's operaticns thro4u varginal dchrxes in allocation of resoces across and within its different p 19# IrlYirq hypothesis, whichd rmaim to be tested yet, is that thre is inaed sufficient deimx far the %xMtraceptives offered by the different opwatios that are re cxmorW for exparDion. Ibreover, the study stressed shot term eic diucnios of ProfagLiliats delivery efforts. Lan term effects, such as -I - tirq the poplation about sterilization, presently the least ost effective mentod, may render this mthod wre cost effective in a long run perspective than is sxjgested here. vi PREFACE cost-effectiveness in family planning delivery has becm especially important in recent years in view of the slowdown in the growth of public resources for family planning, at a tim when the need for it rmins rr&;sing. This study is part of an operational research program in family plannng initiated by the World Bank. Uder this program, a quantitative eomiic approach with a cost-effectiveness arientation has been developed to assist policy--makers and mnagers of developed programis to learn frxn their own program experience about resource aLlocation, cost and finance, and ho these relate to program objectives. This process should considerably enhance efforts to improve resouroe mnbilizaticn and internal program efficiency. The present work program included developrent of guidelines for operational research with a caost-effectiveness orientation, and studies of t distirct family planing programu: the Inxanesian National Family Plaming Program and the C(oloubian Profamilia Program. This study of the Colcubian program was undertaken in collaboration with the International Planed Parerthood Federation (IPPF) and Profamilia. We twhank Ms. Susan COxrane for helpful and insightful aments. Profamilia, an affiliate of the IPPF in Coloitmia, South America, is a not-for-profit rxo-,overnmnt organization (NGO) providing family planning services. Profamilia has several important features. First, it is an NGo) providing nore than sixty percent of total family planniin services available in colombia. Second, Profamilia recovers about one-half of its recurrent costs. Third, its program is vertically organized; nest resources are invested in provision of family planing services rather than in related areas, such as maternal and child health. Ihe objective of this study is to examine whether and hMw Profamilia can do better with the resources available to it. lo that end, we analyze the cost-effectiveness of Profamilia's service delivery, which includes the Community-Based Distrib.ticn (CBD), Clinical, and Voluntary Sterilization Programs. F,r all of these progrSam cninued and for each separately, costs ard resource productivity are compared while oonsidrin poplation characteristicsz and metod mix, in arder to establish relatively efficient operations. The discussion is based an Profamilia's service statistics, including operations of 97 field workers and 38 clinics. These data conern sales and distribution of cmntraceptives, input ard costs. Sam of this informaticn is summarized in VillAmil (1986) and Ojeda (1986); other inforution is 2 drawn froni Profamilia files. All data ssrces are discussed in the following chapters. This study falls into three major parts. In the first part, Chapters 2 and 3, we provide a brief suumary of population policy and family planming in colcmbia, and an introduction to Profaamilia. In the seconrd part, Chapter 4, we present the issues and analytical franework guiding the discussion. In the last part, Chapters 5 through 7, we exwaiine resource allocation and cost effectiveness in Profamilia's CBD outreach program, the clinical based programs, and all progranm combined. Conclusions are drawn in Chapter 8. 3 2. POPUaTCa AND FAMILY PLANNIl IN CfUBLIA 2.1. The Population: Size, Growth, and Distribution olctbia is divided into five geographical regions and twenty-three or departments. Accordirng to the 1985 cernus, Coliibia had a populaticn of 30 milliacn The rate of popalaticn grWth for 1973-1985 was 1.8% anmually. Ihis growh rate is significantly lor than the high of 3.73% arnnally for 1965=1973. The crude birth rate in 1984 was 28 per tlwsarnd, dam frcm 45 in 1965. ¶¶a infant mrtality rate per thousand live births was 48. Average life at birth was 65 years in 1984. The largest population aticn, 27.4% of the total, is f=& in the Central region, which also apears to be the wmt affluent (Table 201). Seventy-two per-oent of the population lives in urtln areas; the av annual gmwth rate of Colcmia's urban population was 2.9% bten 1973- 1984o 4 Table 2,1: Ppulation Chuaracteristics by Region Atlantic Bogota Central Oriental Pacific Tbtal ibtal 5,678,001 3,982,941 7,643,553 5,214,400 8,887,741 31,406,636 'opulation (18.1) (12.7%) (24.3%) (16.6%) (28.3%) (100.0%) Average 4ousehold (noome 1981 16,086 12,673 22,501 15,278 17,136 17,374 (Oolontian esos) Urban 76.8% 72.5% 72.8% 62.4% 73.5% 72.0% )wellers !Wmen Aged '1 2-59 With Wo or Primary 60.1% 37.5% 59.4% 63.9% 61.3% 58.0% Education only source: Corporacion Centro Regional de Poblacion et al. (1986) 5 2.2. Population Policy and Family Planning The government of Colombia does not have an explicit population policy defined in term of fertility levels and demrographic growth targets. In 1969, the goverrment introduced a developnent plan which included a set of proposals for family planning within maternal and child health program. Ihese proposals did not specify demographic objectives. The attemt to formulate a population policy granted de facto legitimacy to private organizations providing family planning services, stimulating growth in the scope and volume of services they provided. Since 1969, all gverient administrations have included a family planning cxmoent in their economic developmnt plans. The 1986 Denograpthic and Health Survey (CMS) indicates that 99.4% of married women of fertile age (15 to 49) were aware of at least one method of family planning (see Table 4.1, Corporacion Centro Regional de Publacion, 1986). Of these, 64.8% were actually using scme cxntraceptive method, with female sterilization being the most prevalent at 18.3%. Oral amtmceptives were used by 16.4%, intra-uterine devices (IUDs) by 11%, rhythm and withdrawal by 5.7%, injectables by 2.4%, vaginal tablets by 2.4%, condoms by 1.7%, vasectcxuy by 0.4%, and other methods by 0.9%. The level and volume of family planning servioes offered by the public sector has been inconistent over time and across region. It has varied historically as the relative importanoe plaed an the subject by different 6 admstraticns has varied. The gverrment provides its services through local, regicnal and university hospitals, health ceters, and "health psts". The volume and scope of family planing services is determined independently by the service directors in each locality, leading to large geographical variances. No central coordinating or administrative unit nmitors and evaluates delivery and izpact of family planng servios. TZhis ladc of central coordination cntribted to expansion of the private sector in family planming services. The private sector, ixmluding not-fo-profit organizations, delivers the majority of the family planing services provided in Colcmbia. Mst services are subsidized. Private sector family planning services are provided by Profamilia, snall private clinics, private physicians practices and pharmcies. Family planing service in the private sector are finanoed by international and national doatins, and by fees collected for services in private clinics and private physicians practices. Irsurance policies, with the exception of those of sc large loyers, do not uually cover ses associated with family plaming. e mmin soue of suaply of family planing services is Profamilia, which provided coerage to 38.7% of those wa using sce cotraCeptive method (Table 2.2). Profamilia's activity is most notable in the provisicn of sterilization. The next not iuwtant sourc of ontraceptive method is camercial dn* tares, whidc provide 33.3% of all coerage. Profamilia's actual coverage ext1nds wll beycnd the prpoti receiving family planning servicas dixectly through its program, as it supplies mare than 80% of the 7 drgstores and pharmcies in olcabia with subsidized omtrative pcts. Table 2.2: Distributicn of cntra ve Users (%) by Soure of Supply Method Source Pill IUD Inj. Speridc. Ca-Km Steril. Total Hospital/ Health Post 12.2 38.3 8.8 2.2 5.2 10.8 16.3 Clinic/ Private Hospital 0.8 4.8 3.1 0.0 0.0 9.1 4.6 Profamilia clinic 2.3 42.3 1.7 7o0 6.5 74.4 36.3 Profamilia Distributicn Outlet 6.2 - 1.0 4.9 8.5 0.0 2.5 Private Doctor 8.7 10.1 7.1 3.3 7.5 1.7 6.2 Pharmacy 63.1 - 73.4 75.5 60.9 0.0 28.2 Halth Worker 1.6 - 2.7 1.2 0.0 0.0 0.7 F:riend/Relative 0.4 - - - 0.0 0.0 0.1 otherl 4.7 4.5 2.2 4.9 3.6 4.02 3.5 No respmise 0.3 0.0 1.0 0.0 7.8 0.0 0.0 Number of current users 522 332 78 72 44 5733 1,623 1 c1wzues private and pjblic social seacity system 2 Private social Secuity SYSteu aily 3 Irludes s e ies sor!oe: poaci C;etro Regicmal de PClacian et al. (1986) 8 3. PROFAMILIA 3.1. Brief History Profamilia was founded in 1965 to offer family planning information and services, primarily to families with limited economic means. Its first service delivery outlet was in a private Oiysician's office. In 1966 it became an affiliate of the Western H0nisphere Region of the IPPF. That same year, Profamilia founded its pilot clinic in Bogota. During the following years it opened clinics in the principal cities of Colombia, establishing a network of forty-tw clinics and family planring centers. Profamilia's provision of voluntary surgical sterilization was initiated in 1970 with a vasectomy program. Female sterilization was included as part of the Volutary Sterilization Progran in 1971. In 1971, Profamilia inaugurated its (snuuunity-Based Distribition PrOgram to local comwmities to provide family planming informtion and contraceptive services which do not require strict medical supervision. This program is run by about 100 "instructors", each responsible for a paticular jurisdiction. Ihe activities of the Informaticn, ducation, and (aunication (IEC) Program are closely associated with service delivery. IBC activities include: preparation and publication of panolets, training manuals, and posters, production of family planning prnotional radio spots; and sponrship of 9 public family planning conferences at clinics, schools, and other public and private institutions. IEC activities played an important role in the initial expansion of Profamilia in the early 1970s, when much effort was devoted to informing the public of the existence of family planning services. Currently, Profamilia reinforces the importance of family planning in the public oonsciousness, but is increasingly concerned with emphasizing the quality of that service. Profamilia emphasizes mnmitoring and evaluation of its delivery of family planning servioes. Explicit output and oost-effectiveness targets are set by senior management, and efforts to meet them are guided and monitored by Profamilia's Planning and Evaluation Departnent. Although organized as three separately administered prograrm, the discussion below deals with the outreach activities, the CQmmunity-Based Program, on the one hand, and the clinic-based activities, the Clinical and Voluntary Sterilization Proqrams, cn the other hand.1 3.2. The Community-Based Distribution (CBD, Program Profamilia inaugurated its CBD Program in 1971 only in rural communities as part of a cooperative agreement with the National Ooffee Growers Federation. Based on the initial suocess of the rural program and on economies of scale, Profamilia created a separate urban CBD Program in 1974. 1 They are referred here as sub-program. 10 Because it was not cost-effective to have two separate management and support structures for program which shared the same philosophy, objectives, ard ,rocedures, the rural and urban CBD Program were merged in 1981. The most recent structural change in the CBD Program was made in 1986, when the cxmercial marketing program was merged with the traditional CBD Program. The resulting program, in which CBD Program instructors wre allowed to sell Profamilia products to private sector outlets (e.g. pharmacies, etc.), is called the Cammity Marketing Program.2 Ihe basic objective of the (umiunity-Based Distribution (COD) Program is to provide family planning information and services to those sectors of Colombia's population which cannot or do not wish to use the services provided by the Clinical Program or are unable to do so. The COD Program operates primarily in rural and outlying areas of urban centers, and the mthods it distrikes - condom, pills, and spermicides - do not cenerally require direct medical supervision. The current CBD Program identification, training, motivation, and logistical support of over 3,600 distributors and over 8,400 pharmacies, cooperatives, and other outlets where Profamilia's contraceptive products are sold. This work was carried x3t in 1986 by ninety-seven "instructors", or field workers (Table 3.1). 7tese workers to not sell contraceptives directly to users. Ihe COD nrhrk has achieved national coverage, distributing 2 Te term "CBDI is noethles retained thrm*hot the di8sion. catraoeptives in all t-ly-threa of Oolacbia's provie. Ihe C Program is daigned to capitalize on local camtmity facilities and personalities for the prcoticn of family planning and the distributicn of onbtaceptives. Selection of distributors is made y the field workers, who also deliver supplies to their points of sale. Infoiiational meetimxs are held periodically in bth new and presently active camnities. Table 3.1: Basic characteristics of Field Workers Mean Age 34.4 years Percent Males 20% Percent Married 35.8% Mean Years of Sctooling 12.5 years Percent let High Shol 86.3% lHaan Experiene 6.1 years Mean Number of Children 1.0 children Source: Profamilia internal documents Table 3,2: CSD Program Prices to Cmw mmX_a_ SMea S.D. Pill 52.5 2.69 CXsrgbm 11.6 0.73 Spermicides 18.4 15.72 scure: Pofmilia internal dments During these mwtings, aity leader, qualified to distrihbtors and effective noti of family plannin, ae id:fied. Atw selecticn, 12 new distributors are traine3l to aoounsel family planning, to identify side effects of the oontraceptives, and in basic screening of candidates for cxintraceptive use. CBD Program field workers use tw modes of supply delivery: conigrment basis and cash basis. Distribution outlets which receive supplies on a consignment basis must sell the contraceptives at Prfamilia's established prices, which allow a small profit nmrgin as an incentive to the distributor. Outlets operating on a cash basis purchase the cceptives at the tie of delivery, and may then set a selling price higher or lower than Profamilia's retail price (but no higher than the ceiling price set by the governIUent). lable 3.2 shows method prices. The data indicate substantial variation in the prices of the three methods, and of spermicides in particular. The output in CYP, units of cxntraeoptives sold, and revenues are shown in Table 3.3. In 1986, Profamilia spent about 389 million colcobian pesos on the CBD Program, which generated some 276 million pesos in revemnes. Pills constitute the most ivportant form of contraception delivered through the CBD Program, both CYP and revenues.3 CYP fram pills is 84% of the total CYP, and 78% of total revenues. The relative contributions of the Sub-Program to condoms and spemicides are minor. 3 The following CYP coefficients were assigned to methods: pills, 0.077; condoms, 0.010; spermicides, 0.118; IUD, 2.500; and sterilization, 12.500. IThese are based cn Profamiliats data. Slight variations in implied ooefficients may exist in some omputaticns, due to rounding. 13 Table 3.3: Profamilia CBD Program, 1986 ^. 11 esc;.^.S.;\. ~3,818,413 acx"M (trnidS) 3,240,992 Spermicides (Units) 199,769 _oule-Yes of rMtection (¶btal) 349,731 (100.0%) Pills 293,724 (84.0%) Oandas 32,410 (9.3%) Spermicides 23,597 (6.7%) Values of Gross Sales in 1.00Q ColoQban 275,709 (100.0%) Pills 216,373 (78.4%) 30,082 (10.9%) Spemicides 29,254 (10.6%) Total Program Rwrent Costs (in 1,000 Oolcmbian pesos) 388,863 Note: 'he CED and ome ial Marketing Progrm are -czbined under "WCE. Distrilztion of sales by method is based an data from the field. Sources: Ojeda (1987), tables 30,31a,32,37,40; Villamil (1987), tables 7-9 14 3.3. ihe Clinic-Basd Progras ¶W distinct program are clinically oriented: the Clinical Program and the Voluntary Sterilization Progrm. 3.3.1. The Clinical PrOgram The cbjective of the Clinical Program is to provide lowinoome groups with l-cost family planig services ancl information about effective contraneption. Profamilia's thirty-eight clinics (1986) also offer tests for the early detecticn of cervical cancer, infertility treatent, gyological examinations, prenancy tests, aid some general mndicine. The clinics also provide the surgical setting for the Volutary Sterilization progrm, wuidh is considered a separate program providing Just sterilization and are administrative and logistical bases for the CBD Progra. Beause of their location in principal cities, 93% of Clinical Program acCeptos are residents of urban areas. Profamilia divides its clinics into fotW major grOUps: large clinics, uidium clinics, mll clinics, and male clinics4. Table 3.4 provides the nub,er, aneage area, cxr.tatios and surgies by type of clinics. The large clinics have an aveage area of 1,576 square maters (M2), the Msdium, 476 m2, ard the smEll clinics, 344 m2. The large clinics suPPly cmsiderably mm sevices than the medium and small cm.. Szdi services include even 4 Separate data for male clinics has bee available only for cre clinic. This clirhic ws included with the grop of large clinics. See Anne 1 for the list of clinics by g 15 medical gyneoological, pregnancy tests ard even pertinent legal advice. The Clinical Program is run by five major types of perscmnel5: physicians, auxiliary wrkers, assistants, office workers and administrators. At the end of 1986 there were 538 full time equivalent (FrE) positions.6 These positions were distributed as follos: Physicians: 33 F1rE Auxiliary workers: 238 FTE (141 non-certified and 106 certified nurses) Assistants: 118 FrE Office workers: 55 FrE Administrators: 16.50 The remaining staff are administrative help. Tme physicians were further distributed between physicians in the Clinical Program (53% of total), physicians in the Surgical Program (31.1%), and managers (15.9%). No explicit data on prices of contraceptives to clients, are available for the Clinical Program. It is Profamilia's policy that for the clinical procedures, including IUD insertion and sterilization, poor clients pay less. Prices of contraceptives sold over-the-coanter (pills, condoms, and spermicides) were assuned to be the sam- in the clinics as in the 5 Based on Profamilia irl-ernal documents. 6 An FrE is based on 8 hours per day as a full-time position. Thus two workers with 4 h3urs per day each, are equivalent to one FTE. 16 corrdsrxdiqn CBD distribition outlets for that clinic.7 m1he clinics also charge for coxxsultations, for laboratory tests and for surgeries. Table 3.5 indicates the average price for each service. Table 3.6 shos the r".tput of the two clinic-based programs in units of contraceptives, CYP, and revenues in 1986. In 1986, delivery of contraceptives through the Clinical Program generated 917,918 CYP.8 The major activity of the Clinical Sub-Program is its family planrning consultation service, which provided 328,283 consultations, of which about 37% were new users of the service. Most of the consultations were for IUD users (53%), about 30% were for sterilizations, 8% for pill users, and about 9% for other methods. When the two Clinic-based Sub-Programe are separated, the great mejority of Clinical Program total CYP, about 80%, was generated through IUDs. Of the CYP from the other ontraceptive methods pr3vided, pills were predcuinant with 11.1%, and condos and spermicides together provided less than 9.5%. The remining consultations were medical in nature and some resulted in referrals to the Surgical Sub-Program. 7 Tee is substantial price discrimination in clinical operations, and gross revenue data were unavailable. It was hard, therefore, to establish unit costs for those contraceptives. 8 lhe data here follow Profamilia's practice to report CYP only for new IJD users. This practice is not followed later in the discussion. Ech IUD insertion, whether to an old or new user, is considered as delivering 2.5 CYP. 17 In 1986, Profamilia inmested 484.5 million pesos in its Clinical Sub- Progrm, and recovered 237.1 million pesos in revemies. Of those 237.1 millicn pas, over 61% wre for ocultaticn fees. F'ee for laboratory services, sui as pregnar:y tests, acomted for 32.1% of all revenues. Pills, coxkm, and spermicides provided the remaining 6.5% of revenues. Table 3.4: Area, Annual clwltatico. and Surgeries by Type of Clinic, 1986 Large Medium 1 il Number of Clinics 4 16 18 Average Area (m2) 1576 476 344 Average (nultations 47,951 5,123 3,025 Average Surgeries 5,522 1,841 460 source: Profamilia internal dcmints Table 3.5: Estimated Prices and Total Revenues for the Surgical and Clinical Sib-Programs, 1986 Average Price (Colcmbian pesos) Pills* 52.50 Condoms* 11.60 Spermicides* 18.40 Consultations 443.24 aSrgeies (per cperation) 1,555.82 Laboratory Tusts 514.36 *Prices are assumed as C averages for pills, ocdma, and spemicides. soce: profamilia interJal cumnts 18 Table 3.6: Profamilia Clinic-Based Sub-Programs, 1986 Pill (Cycles) 209,214 condomm (individual units) 250,261 Spermicides (units) 95,697 IUD* (insertions) 45,906 Family Planning Conultations* 282,377 Male Sterilizations 2,201 Female Sterilizations 59,681 ouple-Yearx of Prot.ectieo (Thtal) 917,918 (100.0%) Pills 16,093 (1.7%) Oondoms 2,503 (0.3%) Spernmicides 11,302 (1.2%) IUD* 114,765 (12.5%) Male Sterili2ations 27,837 (3.0%) Femle Sterilizations 745,418 (81.2%) Values of Sales (Ibtal) (in 1.O0 Ckilcmb,ia-n pesOga 366,930 (100.0%) Pills, Condoms, and Spermicides 17,800 (4.8%) Consultations* 168,145 (45.8%) Laboratory Services 87,906 (23.9%) Male and Female Sterilizations 93,079 (25.4%) Total Sub-Program's Remwrent Costs* 863,974 (in 1,000 Colonbian pesos) of hiich sterilization 379,500 I Tis figure refers just to new acceptors, per Profamilia reporting. ** Does not include consultations acioxmanied by IUD insertions. Souroes: Ojeda (1987), Tables 1,26,28,29; Villamil (1987), Tables 7-9 19 3.3.2. The Voluntary Sterilization Program The basic objective of the Voluntary Sterilization or Surgical Sub- Progrm is to offer irreversible surgical sterilization to users who have achieved their desired number of children and wish to cease reproducing. Requirements for voluntary sterilization are a minimum age of twenty-five for females and thirty for males, and a minimum of three live children. The Surgical Program was initiated in 1970 with a vasectomy servioe. Femle sterilization was added to the program in 1971, and has had the largest impact of any family planning method in reducing the population growth rate in coloambia. Profamilia offers four types of surgical sterilization: laparoscopies, mini-laparotomies, post-partum, and vanctomies. Sterilizations are mostly performed in Profamilia's clinics. However, Profamilia also offers sterilizations via mobile units and through agreements with private clinics, physicians and government outlets. Ihe mobile units travel to areas not served by clinics, and perform surgical sterilizations in government and private clinics. Agreements with other public and private sector clinics oaver sterilizations in areas not covered Oy the mobile units and Profamilia's clinics. For this reason, many of the sterilizations shan by the contraceptive prevalence survey to have been Provided by government or private clinics were actually provided indirectly by Profamilia. Aocrding to the 1976 DHS (Corporacion Centro...1986), there were awroximately 750,000 male and female sterilizations perforwd in 0olambia by the end of 1986. Profamilia directly provided almost 500,000 of these through its Voluntary 20 Sterilization Program. In 1986, 773,255 CYP were delivered through the 61,882 surgeries performed by the Voluntary Sterilization Program. Over 96% of the surgeries performed were female sterilizations. Profamilia spent approximetely 379.5 million Oolombian pesos in 1986 and recovered about 93.1 million pesos in surgical fees. 21 4. PRO)PM EFFICIENCY: IH ISSUES AND ANALYTIC FRAMEWOK 4. 1. Introduction Prof amilia is a not-for-profit organization whose ultimate goal is to prnte family planing. Other activities as well as cost recovery efforts aim to serve this goal. In 1986, Profamilia's three progrem delivered aproxivetely 1267 million CYP to the population of Cbloatia at a ost of 1252.8 million Colcuwbian pesos ($US 6.43 million). The program recovered about 642.7 million pesos ($US 3.32 million), or about 50% of its costs. In total CYP, the share of cmmunity-Based Distribution Sub-Proam ms 27.6%, of the Clinical Sub-Program 11l4%, and of the Voluntary SterilizatiOn or Surgery Sub-Program 61.0% (Figure 4.1). The shares in reveues from these programs were 42.9%, 42.6% and 14.5%, resp-tively. In costs, the cmmmity-Based Distribition Sub-Program's share was 31.0%, the Clinical Sub= Program 38.7%, and the Voluntary Sterilization Sub-Program 30.3%.lO er, the different sub-prgrau service populations of different ages, leading to variations in potential demograPhic inqact. Thes data reflect key policy and uunagmnt decisions: allocation of resources angst sub-progms, fee setting, and allocation of resouros 10 These data are based on Profamilia*s financial reors, whidc may "over-repor" the cst of the clinical program aid tuheCt" s of the other two progrn because the clinical sub-progra ser89e the other su-rg0 22 within program The basic e,ficiency questions are whether and ho Profamilia might do better with the resources available to it, or alternatively, where and how it stuld allocate the resourc it has, or additional resourca donated to its operations, to maximize the contraceptive protection it offers. Figure 4.1. Shares in CYP, Gross Value of Sales and Costs, By Sub-Program Share 70% 61% 60% - 50% 43- 43t , / ,1 ~~~~~~~~~~~39-'7 3100; jil 10 L°l N~~~~~~~~~~~0 40%- 30%- 20%~~~~~~~ 10% 0% CBD Clinical Surgical PROGRAM CYP O Sales Cost 24 4.2. The Efficiency Issuel1 Efficiency in operations calls for prvdutican of maximm contraceptive prtection subject to: a) the budget available to Profamilia fro s aue m m itg onost reoery effortst b) patterns of populaticn use of contracepticn; and c) equity considerations. Because of a lack of suitable data about the population, incluiMng its incone levels and distributicn and family plannig practices, we fooE on pogrm efficiency with regard to only the first cantaint.12 In aition, profamilia's arrent fee strucue is taken as given.13 11 Ibis section is based cn Chernichosky (1991) and ChernichxYskY and Anson (1991). 12 lbs Colatbian D--aphidc and Helth Suvey (;) data for 1986 wold be useful to fill this inforaticn gap. Teee data we not available for this analysis. scheduled work will take advantage of this data. 13 A discussion irnolving canges in the fee structure wld couider conar resporse to fees, whidh has ficatii for both use of oot'c io and cst recmery. 'ue in clear ncdl guiding fee-setting under the circumtano5 discussed he. Such a mode1 awld balance the two wideratics as well as equity in delivery. Fees - at leat beycnx sme level - are detrimetal to overall caitr e use, bt gnerate revenues wihd can be used to furthr pmote contr to o thrkogh izproved access. Reues collected thugh fees shozld be used to inrve service so that total protection delivered wz1d irease without harm to oveall equity. 25 maximum protection delivered with a given "external" txdget means that the program delivers protection at minimel unit cost, which is considered a prime measure of operational efficiercy. Various methods have different attributes in tenrs of: (a) efficacy coefficients (or CYP), (b) efficiency levels even when properly used, (c) cost recovery coefficients (or prices to cost ratios), (d) potential inpact, due to age of users, and (e) costs. Crosequently, with a given external budget and input prices, unit costs of protection can be controlled through i.-provenrnts in: a) input levels or scale of operation; b) combination of inputs and worker attributes; and c) method mix. We divide the efficiency issue here into two cxmgoneTtns: internal efficiency and allocative efficiency. Internal efficiency issues relate to unit cost of protection for a gin metd mix. These issues concern two questions. The first is: which inputs (e.g. nurses, MDs) and worker attributes (e.g. age, sex, experience) should be expanded at the margin, possibly at the expense of others, in order to increase the efficiency within each sub-program or part of it? The second is: should the entire scale of the operations be expanded to increase efficiency? Allocative efficiency issues relate to the question to what sub-program or method shculd an additional peso be allocated. The allocation question 26 stera primrily from varying levels of efficacy (in terms of CYP), w;t reoovery potential (in term of prioes or fees relative to cost), and different consumer deand patterns (in terms of age of users of particular plgY Xrampthods) . 14 Terefore, we divide the discussion into two major parts. In the first part we exanine separately each of Profamilia's sub-programs, both outreach and clinic-based, in order to identify the neans by which Profamilia may increase the internal efficiency of each sub-progrm or oAsonents thereof individually. In the second part we ocpare the two types of progrme, studying the ways in which Profamilia might increase overall efficiency by shifting resotwoes between its program, or by prcmoting particular prograsr. Within the clinic-based operations we ccapere Profamilia's Clinical and Surgcal SUb-Pograms. Any progrwaitic change suggested should be cosidered marginal within the realma of crrwnt cxmtraceptive technology, consumer demMd patterns, and Profamilia's aorret structure and delivery patterns. The t1 analyses may suggest prcotion, at the margin, of one family planning method at the expense of another, and the serving of different populatior. It nurt be borne in mind, haver, that no data are available an denmd for alternative methods. Ihat is, stggestions fran a cost-effectivenss perspective about a "preferred" method (cm mnode of delivery) and program frau a cost- 14 Method mix optimization &-d unit cost minimization through better use of res_atzs are inter-related, because the efficiency of a method is in pa related to the cost of dei -aring it. The tw are dealt with sepaately here for the sake of clarity and sinplicity of discussion. 27 effectiveness perspective mist ultimtel.y consider conmmr prefere. A program my be cost-effective but s:cially inefficient when consumer preferences are disregarded. That is, the program may deliver a mix of nathods with a given bidget so that total di , y per CYP, are minimized. Yet, deman patterns may not match this mix; scae methofd may be oversupplied while others are undersupplied. 4.3. cts, Resource Productivity ar. Internal Allocation of Resources A program or operation can be identified at arny particular time by tb types of inputs: those which hange with otput level - variable inputs - and those whidh do not - fixed iputs. In a clinic, the building size and amount of equiprnt are unlikely to chne with the nuuter of visitors. Levels of supplies (e.g. of contr ves) certainly dcnge, probably in proportion to the nuntr of visitors. In an outreach operation with a single field worker promoting, coordinating, and delivering comtraceptives, the wrker may be considered the fixed input, if fployed full-time, although his or her tim input may Change in relation to Output levels. Hee again, supplies vary with utpu. Parallel to inpts, their costs are also divided ac~rdingly between fixed and variable. ithe unit coot o an operation is the ratio betwem total costs, detemined by the natue of inputs and their prices, and rescaos prodwutivity, mnaswed by nuter of contraceptives sold, CYP delivered, and revenues. 28 Qxieyuently, the potential for irceasing return to scale - laaer unit cost because of scale of operaticms - is higher in operatiom where the fixed cost ccxnnnt is high in relaticn to the variable cost cmixoent. Fram the viewpoint of variable costs, costs of upplies are proportional to output levels. Hence, a rise in the variable untit cost of an operaticn follos a decline in uarginal prxd=tivity of labor amuu other things. This leads to an increase in marginal cost of outpuc. fhis is to say, the rise in outpat levels is less than proportional to the leading rise in amunt and cost of labor. This may reflect, on the one hand, constraints of fixed inuts - including ngt, and on the other hand (especally in the field of family planring), the need to iarease efforts to mbilize additional users. When all costs are comidered, the decline in the fixed costs (per unit of output) my be offset, beyond a particular level of ,apt, by rising marinal costs due to fallin labor prOdUctivity within that range. In the long run, an operation my be considered inefficient in two eotr-oe situations. First, wht in caqparison with anoter adjacent or second cperation, it operates at too small scale; that is, evpasion of the secad operation at expense of the first would reduoe unit cost. Second, whn it opa:tes at too large a scale in given operation, expansion of the said operation, e.g. a new clinic, wild redc unit cost in the long nm. In the se8xid case, this may require now lao-term invesbts and 29 rearganization. Identification and su of levels and cctwsiticn of aosts in relation to labor productivity and scale of operations are, therefore, critical elemnts of a cost-effectiveness analysis. Their study in Profamilia is the porime objective of this analysis. A suply and demind framework, atined here in geeral terms, is employed to determe resource prodwivity in tems of sales of Cxmtraceptin. 15 Aoordingly, deand (a)) for coeptives in a given operaticn is depicted by: (3 = d (pop.lation size, populaticn dcaracteristics, prmoticn activities, the "full price" [FP] of service to clients) (4.1) That is, the quantity of fertility control d (Q) in a given catchwi area is a function of: a) populaticn size, which influerxs demad for on and potential scale of operatios; b) populaticn charcteristics, which determine the denmd for children, fecmdity, and attitixes tmard family planirn (Easterlin and Crimnins 1983), all of which determine dlemard for contracepon; 15 An cpration is usually characterized by sales of several comit?raptives. The fraumcrk as outlined, may relate to a particular cmitraceptive or to a cistant "rix" of contraceptives. 30 C) Errtcd activities sucfh as Information, Eiucation, and OcN1 lication t(IBC) activities; and d) the full price of service to the client, which is determirned by fees (whe applicable) and ease of to cutlets and support services.16 Effective demnd levels vis-a-vis the capacity of an operaticn are important to cost-effectiveness because of retns to scale, associated with fixed cost elents, as suggested above, are a nejor aans of reducin unit caost per user. Demand levels are therefore a major cmen to prgram plannrs and mnagerss. Demand levels can be influenced by (a) the size and nature of a cattment area allocated to an operation, and (b) by pruwtion of cansumnr dutand in that area through IBC activity, and lowering of the full price of service. Both optios require a delicate balanc amuRgt reswces allocated to delivery, IEC, and to reduction of the full price of contraception to the oonsuwuer The supply of cantraceptives in the cocimity ccwerns a progIms capability to influee and a_ te potential cliernts by affecting the full price (FP) of contraception, or the clients' perceived access to service. This price can relate to, but sbxuld not be confused with, cast of delivery. mhe price can be lowered when the program has mRo and better resorcs for delivery. Geneally, for a given external program budget, the folloing relationship holds: 16 2m "full price of service" relates to mnry aspects of clients' percptions of cost of, and oufort with, service. Therefore, the nature of in±uts, e.g. female vs. male warkes, might be considered in order to assess their ipact on the full cost of on to the onewuer, especially in the a of fees for service. 31 PP = & (O), rescure allocated to delivery, inpit prices, infrastructure, fees, naturl envirment) (4.2) That is, the price is a function of: a) quantity dwarded (Q(), which determines the actual scale of an operaticn: w it is "too high" for the (short run) supply efforts of a partialar eraticn, it may inrce a high price to commmers through queuing and high costs of operations; when it is "too lw" it mya indue high unit cost of oratiom bcam of high (log nrn) fixed costs; b) resorces allocated to delivery, which determine how well csumers can be -mK,-ated; clients can be ac- - n-ated with more and better resoaros, especially when the fees clients pay do not cove the margial costs of the service; C) input prices, which determine the level of real resouras available for delivery. Th higher the prices, the lower the level of real resorces available to the program; d) infrastructure, which enhae a program's potential by aiqmentir Eproductivity of other resources; e) fees which deter clients by inareasing the FP of service;17 f) natural envitmnmmt, which if harsh, my be detrimntal to ting clients with given bhd8ts. A third relatiorship cexns mlbilizaticn of ommity s, 17 Fees have th potential icrease availability and quality of service, and attract . This isma is rot coidered here explicitly. 'Rescurcs allocated to delivery can be regarded as incluxing reveme frcm fees. 32 which are included under infrastrucu: I = i (program inpits for resource mbilization, commnity infrastructue) (4.3) This quantity is determined by: a) the irnpts the program allocates to mobilize resources, such as points of sale or outlets in the omiuity. The higher the irnpt levels, the higher the level of resources mobilized; b) the infrastrcu, health and other ccmunity facilities, which determine the potential for resource mobilization. The discussion thus onoerns the question how the program can maximize the protection it delivers through allocation of program resources bebwan .:ian, delivery, and res#m mobiliza , so that delivery unit cost is minimized.18 4.4. Allocaticn Acmss Program and Armss Metkxds within Progrm Profamilia's program are distinct. As discussed above, they serve different poplatica and at different unit costs. A key allocation question is therefore where should Profamilia's it invest the additional or marginal peso or dollar it receives in donations? 18 b all costs snd irstitutins participating in delivery are considrd, the third relationship can be integated in relationhip 4.2. This may be warranted also for the subsequent statical analysis because it is difficult to separat statistically the inpact of the enimt and the ty's contributicn to infrastructue from the impact of the it s ity'Ps through deafor ofn. 33 Barring equity issuss, optimal allocaticn decisions call for a situaticn whewe there is no preferred gain in protection frau allocating the marginal peso in any partilar sb-progrm or method peational guidelines leading to this situaticn are not straightforward, beause each su unit of resour~ yields not just CYP through different methods, but also rwenues which can be reinsted. MIever, different methods are used by couples with different levels of pregnaxEy risk due to their age. This issue is ell illustrated in Figure 4.20 On the avere, 1000 pesos (grmss) yield the moqst CYP in the Surgical Sub-Program ard the least in the Clinical Sub-Program. This amount reovers 710 pesos in the C Program, 570 pesos in the Clinical Sub-Program, ard 230 pesos in the Surgical sub-Proam. The programs also serve different age gups.9 19 gle arrnt fee stncte is taken as given, althoug it is a cnwial policy instrument- Figure 4.2. CYP and Value of Sale8 (xlOOO) per 1000 Pesos bv Sub-Program 2.5-" 2.04 1.5 - 0.89 0.57 0.5-2 .2 0 CBD Clinical Surgical PROGRAM CYP Sales (xlOOO) 35 Each ecternal peso allocated to a particular program has a "nltiplier effect". It generates sales of cortraceptives and protection Ughlxt the program, directly in the operaticn receiving allocation, and indirectly through revenues reinvested in all operations. Adjusted *btal CYP (ATCYP) of this marginal peso is given by: AXCYP = ((ai/nci) x (&i x Si x ti) ] / (1 - Ii pi (ai Awi)]20 (4.4) where: ai = the share of the operation, characterized by method or method mix, in total cost; Mi = the arinal cost of each unit of outpxt in this operation; di = CYP or length of protection iated with a unit of outpat i; Sii = efficacy of unit when adequatly used; = estanadized "risk of prewgany" coefficient for the aveage user; pi = the average fee charged to a oasmr per unit of the opeation; The first parameter reflects the basic allocation decision of Profamilia's . Ihe seacod paran.t.r is derived fra the cost 20 Fw a full dission, Se Ornidhsky and AneS (1991). 36 functicm of Profemilia's operations. The next three paamters are determined by the method mix in each program and the fee charged for each method. The last is determined by Profamilia's t Clearly, identical reasonirn applies to allocation of resoarces among methods within each sub-program. 4.5. Statistical Approadc The statistical analysis is based on observed o-variations in output, input, and oosts across Profamilia's operational units in each of its sub- program. The basic assumption underlying this approach is that What wmics best in cne operational unit, can be adopted by anothe. Given the rnn- experimental nature of the available cross-sectional data, an atteqpt is mne to statistically oontrol for as mny social and other environm tal variables as possible, as they may affect resouroe productivity and costs across units. cOnequety, data on popilation characteristics and natural environment are included in tte analysis, in addition to data on resou and autpu.21 21 The analysis applies to entire popilations of Profamilia's field wrkers and clinics. Hence, the statistical estimates depict the actual situation. The t and F-statistics shold be used for their predictive value, and _ t of strength of the estimated relationshipd. 37 5. I1 CBD PROGRAM 5.1. Objectives and Framnrk of Analysis Profamilia's primary and almost exclusive resources in the cBD or outreach prograun are its staff of field workers and supplies. How Profamilia recruits and allocates them, who they are, and how they are supported in the coamunity, all influence contraceptive supply and denand which in turn determine the effectiveness and the cost-effectiveness of these workers' operations. The objective in this chapter is to examine field worker charactistica, allocation and support strategies, for their cost-effectiveness, and the whether and how Profamilia might improve the aost-effectiveness of its outreach operations. The chapter falls into two major parts. lhe first is an application of the conceptual fram crk discussed in the previas chapter to outreach operations, and the second cmrises a statistical analysis of pertinent data. 5.2. Field Worker Operations: A Model and Hypotheses Tbe resources available to field workers are their am time and the community resources they have available or can mobilize to assist them: medical infrastructure and distribition points. Let us assume that field worker potnial production in the omnity (Us) is a function of time spent in delivery and coordination of sales (td) and of infrastruce (I) 38 Us= f(td, I).22 (5.1) This is a tecological relationship rertestng a wrker's pote:ial in the cmmnity: the rn,er of clients or wmld be clients he or she can service, given the particLlar nature of the amuuity and the mix of methods delivered.23 This function is depicted by ojrve f() in the uPr right quadrant of figure 5 .1. It axrr to relatinhip 5.1 and irporates 403. It is fUrhe asMed that all workers share the samiM prWtion fuicticn or technology, that each strives to maximize output, and that the maximum can be reached by exha all working time. clearly, the greater the output per wrker, the lower the unit nost per average user or unit of CYP delived by the worker. A wrker is ciwidered a egquasi-fixed" input in the short rum; the marginal cost of his or her operation entails primarily costs of suplies and possibly travel costs. 24 optimal ca1tp and minima unit cmst wald be achieved at a:tpit 22 For clarity of disussicn, the na*er of variables in this and other fUn i whidh follcw in this secticn, is kept to a minimu, witkout loss of generality. 23 As rescr ubilization ehac wirk pjzodtivity by providing mvre help in the o Lmumity, f() my he viewed as the a funtatin expessing optiml soluticr for allocation of tim bem delivery and resource mobilization. See Anx 2 and arnidx,WW (1991B). It is further assud that werkers wish to deliver the mst efficacioas ethod mix within their capacity. They my be omstrained by availability of medical infra ie and comumr preferences. 24 Wags are comided a fixed cost because the Workers d not wcrk part tims and are not paid on that basis. 39 level Usm where the worker reaches maximum potential025 Let us further a&sume that the potential demand, delireated by nud3er of would-be users (Ud) in an- individua worker's Iar dint axra, is a furttion of number of eligible couples, or s (E), their soioaenomic characteristics (SE), and field worker time allocated to proEtion or IEX Ud = g(E, SE, tp) (5.2) Tiis function is delineated in the lower left quadrant of figure 5.1. It crrespod to relationship 4.1. Clearly, the number of would-be users cannot exceed the number of eligible couples (which may be estimated from popilation size) in any catchmnt area (Ud<= E). The upper limit of demand is E. Its lower limit is Udl (the intercept) or the level of latent demand that exists with no prctntion efforts, when tp=, as can be shon at the base of g(E3). lb the extent that go is indeed a function of t, it forn another production function representing a worker's ability to promote sales through better marketing in a given catdtment area delineated by E. The marginal productivity of this effort mist be falling because of the upper limit set by numbers of ELr. As in the case of f (), it is assumd that all workers share the sawe function. Individual worker productivity can mean both nar effective supply 25 The underlying assumpticn is that while an individual worker cannot be hired on a part-time basis, workers can be hired and dismissed, o that those remaining an the job would ptoduee each at point Usm. 40 efforts, shifting fo, and moxe effective promotion efforts, shifting go, per unit of time invested in any activity. Some particular personal worker traits may be useful in supply efforts, and others in promotion efforts. A more productive worker can deliver more with given resources, or meet the same demand levels with fewer resources. As under full elployment conditions, td + tp = T, (5.3) where T is total working time available to a worker, there is a trade-off between the two time allocation options. This trade-off is depicted by the 450 (negatively sloping) line in the lower right quadrant of figure 5.1. The 450 line in the upper left quadrant represents all points where supply equals demand. ¶I fundamental regimes can be idemntified in this molde. The first is depicted in figure 5.1 with the aid of g(E3). In this case the worker cannot satisfy latent deijand in his or her catchment area. Prcmotion activities would be wasteful for as long as latent dend exceeds uaxlm= potential supply (Ud, > Urm). Staff works to capacity and unit oosts are minimal, but demand is not fully exploited. This situation should be indicative of a program which does not have sufficient resources to hire enough workers and provide sufficient supplies to servioe its entire potential population. 41 The second regime is depicted with the aid of g(EO), g(El), or g(E2), indicating situations we mininum latent demard in the worker's catchment area is less than the worker's maximum capacity to serve, Udl < Usm. In this case the worker can sell more contraception than is needed to satisfy latent denand, and therefore can allocate time to deiand promtion. mbis situation should be indicative of a program which can afford to hire nmre workers than needed to satisfy latent deuand. Therefore, accounting for costs of suplies, the program should promote denand. This can be done in the case of Profamilia through IEX activity. Clearly, under such circtIStances the program may satisfy demand but risk worker uneiloyment and higher-than-warranted unit cost of delivery. The optimal situation under the second regime wmld be the singular supply and demand situation depicted by point 0°P, where a worker is assigned to a population in which he or she can allocate all wor1king time to supply and demand in a way that Udc = Us0, and td°+tp "-. No non-equilibrium situations may lprevail. The first is portrayed by point O' associated with function g(El), indicating excessively low demmnd. A worker producs nmre than the quantity demanded. This is wasteful, more resources are allocated to delivery than are used. The worker should increase promion efforts at thie expense of delivery efforts, until reaching equilibrium or close to it. The second situation is delineated by point 0", associated with function g(E2), indicating excessively low suply, Ud > Us. This is wasteful in term of program resoure; the worker spends too nuh time in prcIotion. It is also socially wasteful in terms of client 42 queuing time. Ihe worker may fine-tune time allocation; more to delivery and less to promotion, %> eping fully employed and loweringj delivery costs as well as the scial ost entailed in queuing. Given the worker's prxductivity potential, delineated by f(), there, may be only one demand function g(E O.) which makes possible a full ecuilibrium such as depicted by 00. Particular fur&ctions may be such that the worker's marginal promotion efforts do not 'match" their marginal delivery efforts. For exanple, suppose that in a given range of operations there is excess supply. A worker may decide to reduce delivery efforts by one hu, and increase promotion efforts by the same amunt of time. The two outooes ray not match; the excess supply situation may persist beause the marginal increase in demand is still smaller than the marginal decrease in supply. If orkers are assigned relatively low target popudations, their marginal productivity in promotion may begin declining at relatively lw levels of time allocation to prcmtion, and it may fall fast. The likelihood of excess supply is greater in such situaticis. If more functios permitting equilibrium exist in addition to g(E), the equilibria points wmld be traced by the line EE in the lwer left quadrant. It shuld be furtr noted that under circnWstarCe of excess supply, ther is little or no savings to Profamilia even when worker are more productive. The worker is paid in any cas. only under dmnd situations g(DEa) indicating likely queuing, a more productive worker wold produce more and benefit the program. 43 Two basic allocation prcblee are presented here. The first is the assigmment of workers to populaticn and envirorment. This is done by Dalgemento 'The second is the worker's tive allocation. This is nKst likely to be a prsonal decisicn to be nitored by t.26 It mist be rezognized that in ary of the situatiosm discussed above, workers nxy misallocate their time: too mch in delivery and too little in promation, or vice versa. Aowding to the nodel, the logical sequence of efficient resore allocation by progra mnag shld be as follcws: a) satisfy latent demard for as long as possible with available resowces by hirin workers and letting tkm work to capacty. Then, if Kr resor es are still available, b) continue hiring workers, reducing further the catduait area per worker, and start prmacn of dmmnd ewough IBC, while belancing time allocated to pruticn and tim allocated to delivery.27 The er,irical analysis aime to test tw altrative h MUsth derived frca this framework. These are: a) Profamilia's field workers are underutilized (situation suc as 0'), inplying that: o worker prodctivity is influenced by demd conditicns: 26 These allocaticn pmblem are hardled separately and in detail by C1 s'rnidDVsky (1991B). 27 It is aed that fees for svice are set. Hm, wetwge mwdize mtu ordlivey, maimKXize revemns 44 only mre denwd through a larger target population or IK activity will generate highe productivity; arnd, cprodctive works would work less, but withiout oost savings to Profamilia. b) Profamilia's field staff work to capacity (situations 0° or u5X and above), inplying that: dsad conditions do not affect worker productivity; and, relatively productive workers, including towse who nxbilize more resources in the ccmnity, would produce more and at lower unit costs. Eac hypothesis has different inplications on how Profamilia might increase the efficiency of its operatios. 0 I~~~~~~~~~~ 1~~~, X a H , , \ \ @ \~a *0 I <; -I- ----I-J _ --\ , e------------ - - V~~~~~~~~~~~~ \ '-0 C w U,~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~c C U,~~~~~~~~~-- - -- - - - - -- - - - - I&a.p4 I-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~u -------------------~ ~ ~ ~ ~ ~ ~ ~~~~~" 46 5.3. Data and Specific Hypotheses The data for the analysis concern the opeations of the ninety-seve instructors, or field workers, who coordinate the CBD Program (see Section 3.2). Two performance or output measures are available for each worker: (a) sales of contraceptives, and (b) the ynetary value of these sales. Those, coupled with cos-t data for the opeat of each worJe, are related to explanatory variables which are studied for their effect on worker output or productivity in view of the model just presented.28 Ie variables are grouped by their operational significance. These groups concern program design and targeting, personnel policy, pricing policy, and method mix. under program design we inoorporate variables that relate to the structure of the program and the organization of its resources. Tmhe are: a) size of population served per field worker;29 b) number of points of sale in the cmmmity through which any field worker operates; and, c) number of workers supported by a clinic (in a clinic's catciment area). As for the first variable, nore people per field worker my inply higher potetal denad. In a situation like UdL > UeR (figure 5.1) and above, the 28 See Table 3.2. 29 It should be noted that unlike the other dmanad-related variables, popilation size is viewed as a variable representing the uper limit to the supply efforts in a given area. Other variables are more qualitative in nature. For exaple, a higher percentage of nmthers in a given population would boost demmnd in a given catchment area, and thereby, ease the field WGwrkrs takb ocnrtnead. 47 hypothesized effect of this variab'e is nil, as demand cant be met by supply. It may be even regative, if workers tend to overly extend themselves. If the average situation is depicted by UdL < UsM, there will be an increase in output when workers are allocated a larger target population which brings about more effective demand. The effect of poplation size per worker is thus a priori unknown. Points of sale are an added resource to the field worker, part of the infrastructure available to him, augmenting his or her productivity through increasing acess to the population. Therefore, a positive association between sales of cxntraceptives and points of sale is hypothesized, when demand is not fully satisfied in the worker's catcmtent area. Number of field workers served by a clinic is meant to measure whether or not, and how, the clinical infrastructure available to the average field worker, influences his or her productivity through its effect on both workers and clients. A clinical base means availability of vehicles and other forms of support. In addition, it means wider options and some measure of msdical security for clients (e.g. referral to clinics for clinical methods and general medical advice). It is thus hypothesized, that to the extent that clinical infrastructure constrains worker productivity, the mom workers per clinic, the less each worker's productivity. 'b the aetent proximity to clinics affects demand, clinics my boost demand for worker output, on the one hand, beause of availability of medical attention, and lower it, on the other hand, throug substitution. That is to say, clients may opt for clinical methods, and even non-clinical methods through clinics rather 48 throxh oommity Ca±ets. Urder targeting, which carerns fine-tuning operations to nature of populaticn and natural environent, the followng variables are discussed: a) pertag of mothers in the population;30 b) average household m*nthly irnome; and C) peraiiage of illiteracy among w aged 12 to 59031 The first t variables are hypothesized to have a positive effect on demru for cves in a given popilaticn, and henoe on field wrker productivity by a nore omentrated danKi in the sae populaticn, requiix less effort per client. The effect of the last variable is a pricri unkrn; while educated wm may be uxre able to take advantage of a p m, they my have less need for it (Schultz ard osenzweig, 1982). For the natural e variables are eamined: a) area size in square kilcueters; and b) average altitude in mters. It is hypotheized that all other things equal, larger areas and higher 30 Dat a n nuwer of wan 15-49 married or in accsual unicns we rnt available at the time of this study. Ihe "pcntage of mothrsn" 'M chasm upmn the alternative, A of all m 12-59 in the populaticn, as representin better damrd for tractio. Be selected viable ueetimtes the demmd for ption by e:ulidingq demad of no- m,thez, but does nr irclude groups who have r demand altoge=r to the extent that the alternative variable does. 31 See Table 2.1. Another factor affecting deind for Profamilia's serices is availability of government servioes. A future discssion might us MDiS dis poplation data, and data am availability of er_xt services 49 altitudes can diminish worker productivity as he or she needs to sperd mre energy to oover a particular population. Personnel policy is discussed through the inact of worker characteristics on their effectiveness and cost. This is a critical aspect, as an estimated 61% of total operating cost (not including supplies) of the CBD Program are labor costs (Villamil, 1987, Table 7). Worker characteristics pertain to: a) age; b) gender; c) marital status; d) level of education; and e) experience. In many ways, these are the basic inputs in the D program.32 All are hypothesized to affect both suply and demand (functiors f (.) and g (.) in Figure 5.1). It is hypothesized that older and mtore experienced workers who know their population and environment better are more productive. lThw are possibly more costly. Prfamilia maintained, at least at the inCeption of the program, that female workers are more productive than their male counterparts because of the nature of family planning activity. Worker education is hypothesized to promote productivity. It is usually associated also with higher wages. It is assumed that any variable which is associated with a higher (%) gain in productivity than (%) addition in wages, increases 32 It should be kept in mind that field workers do not sell citraceptives directly, but through points of sale or outlets. orker characteristics matter, nonetheless, as wrkers are highly involved with both the sellers and hyers of ontraceptives in the cxzMInity. 50 efficincy. Prices are naturally a significant variable affecting cansumer demnd and the program's revenues. Higher prices are hypothesized to deter deuand. But they may lead to higher net revenues where consumer response to a higher price is not offset by the fall in the quantity of sales. Actual prices charged for contraceptives were not available. For e an inplicit average price was caipted by dividing the value of his or her sales of any particular method, by the quantity of sales. rEthod mix is a crucial variable, as different methods have different delivery costs and cost reoovery potential, provide varying levels of protection, ard attract users of different age groups.33 Unfortunately, populaticn and envirommental data are available only on a provincial level for twenty-two of Colombia's twenty-three provinces As a result, two levels of statistical analysis are discussed: individual- worker level, and province-level. In the latter, number of workers and characteristics of the average worker (in the province) are included in the analysis. As the individual-level analysis exclude cimmity level data cn population characteristics, this analysis is meant to establish and acnfinr the findings about worker characteristics, points of sale, and contraceptive prices. 33ConUMe preferenCe (for which no data is available here) should be considered as well. It is assumed throughout the discuss;ion that the satisfies those preference at given fees for ontraaceptives, but cn still (at the margin) methods that increase program efficiency. 51 5.4. Dstimation Procedure Information about a field wrker's allocation of time to altrative activities is missing from the data. This eliminates the possibility of fully exploring the model presented in Seation 5.2, pfrtiaclarly the ele@1nt of resource mobilization or the establishment of sales points in the community. It is possible, nonetheless, to study through corr-elation of different variables with sales, whether or not Profamilia's field workers operate under excess suply or excess deand conditions and the probable inqact of the variables on sales. In line with the discussion in Section 4.3, the following furctiaml relationship has ban used for statistical. estimation: NatJral logarithm of Mjk = Aoj + aij Natural logarithm of Xi.k + Vkj where Nj refers to sales of specific method (j), or alternatively, to total value of sales of all mthods by field worker k. AOj is a shift parar, and Xi refers to each of the explanatory variables suimrized in Tlable 5.1. The term vkj is a random error term. aij measures the influence of £ach variable on outpit: the % change in outpt associated with a given % chnge 52 in input. ihis function is a "reduced form" of relationship (4.1) inorporating relationship 4.2. as discussed in section 4.3. "Points of sale" are assumed exogenus to the field workers' decision making.34 That is, relationship 4.3 is disregarded. bTe estimated ooefficients (aij) indicate, therefore, interactions betwen supply and dynd or the "net effects" of the explanatory variables through the estimt. ooefficients reported in Tables 5.2 and 5.3.35 34 As noted earlier, pertirent data on a field worker's time allocation, is missing. 35 It can be denwasted that estimated oefficients that wauld Lrtain to struntural relationship 4.1 retain their sign; they are, h,aever, dcwnwrd biased. 53 Table 5.1: List of Variables by Operational Category and Conoeptual Relationship Policy and Program Relationship category / variable population size per field worker 4.1; f ( ) population characteristics 4.1; f ( ) points of sale 4.2; 9 ( ) no. of field workers served by clinic 4.1; f ( ) gender of field workers 4.1 and 4.2; g ( ) and f ( ) age 4.1 and 4.2; g ( ) and f ( ) marital status 4.1 and 4.2; g ( ) and f ( ) level of schooling 4.1 and 4.2;g( ) andf( ) social environment % of mothers in population 4.1; f ( ) average household income 4.1; f ( ) % of illiteracy among wmnen 12-59 4.1; f ( ) natural environment area size (sq. km.) 4.2; g ( ) altitude (m.) 4,2; g ( ) wlcing Pollc prices of contraceptives 4.2; g ( ) Mettbdt MiMx 4.4 54 Table 5.2*2 Regression coefficients (Naturl Logarithm) Qu.antity of sales of Cotraceptives, Regional-Level Estimates for CBD Program (t-Statistic in Parentheses) QUL'MT OF SALES Dtal Pills Condcs Sperm. Value (cycles) (units) (units) (peso) Program Design Population Size+ -0.94 -0.20 -0.81 -0.31 (-3.20) (-.38) (-1.09) (-2017) Number of Instructors+ 0.47 -0.64 -0.11 1.22 in the provinoe (1.81) (-1.30) (-.16) (9.35) Nmber of points. of sale+ 1.56 1.33 2.20 0.14 per instructor (4.71) (2.20) (2.45) (.87) Number of clinics+ -0.16 1.06 0.62 -0.45 in the province (-0.39) (1.47) (.63) (-2.32) Price of cOntraoeptive -5.70 -3.80 -12e09 1.97 (-2.28) (-.85) (-.55) (1.55) Natural Envirouent Area size (sq. kmi.)+ 0.31 -0.32 0.42 0.13 (1.71) (-1.06) (.86) (1.47) Averg altitude (m.)4 0.17 -0.06 0.27 0.03 (2.00) (-0.36) (1.31) (.85) Worker charactexistics % male 0.23 -0.23 0.75 0.03 (0.25) (-0.13) (.32) (.07) % Mmaied -0.13 0.99 0.59 -0.79 (-0.30) (1.32) (.55) (-3.76) Mean years of schooling4 0.16 0.036 0.26 0.03 (2.52) - (.31) (1.54) (1.04) Mean years of experienxet 0.06 -0.03 -0.07 0.05 (1.33) (-.04) (-.64) (2.62) (oatinue) 55 Population Characteristics % of mothers in pop. of 12-64 19.93 -0.42 30.10 19.21 (2.16) (-0.03) (1.32) (-0.96) Average household income+ 0.48 0.19 -4.87 0.67 (0.50) (0.09) (-2.05) (1.44) % women with no education 0.60 -7.50 4.85 -1.46 (0.10) (-1.29) (.60) (-.92) Constant 21.98 17.96 73.86 -2.38 (1.58) (.62) (1.14) (.04) N 22 22 22 22 Adj. R squared 0.8 0.34 0.36 0.97 F 6.9 1.8 18 44.7 + Natural Logarithm of 56 Table 5.3: Pegression coefficients (Natural Lgrithm), Cantraoeptives Sales, Individual Worker-Level Estimates for CBD Progran (t-Statistic in Parentheses) IINDEPENDERr Value VARIABLES Pills condms Spernicides of Sales (cycles) (units) (units) (pesos) Age -0.578 -0.57 -0.016 -0.370 (-1.95) (-0.06) (-0.02) (-1.38) Gender (male-1) -0.245 0.017 -0.024 0.004 (-1.43) (0.07) (-0.08) (0.02) Marital status 0.258 0.153 0.350 0.267 (married=1) (1.92) (0.80) (1.58) (2.20) Experience (years) 0.050 0.054 0.024 0.047 (3.42) (2.61) (1.00) (3.62) Number of 0.922 0.820 0.752 0.880 Points of Sale (9.03) (5.53) (4.46) (9.49) P It D L H-5,053 0.183 -0.102 -1.089 (-4.13) (0.12) (-0.24) (-1.07) Constant 27.750 5.380 5.867 15.563 (5.63) (1.32) (2.12) (3.99) N 95 95 95 95 Adj. R-square 0.57 0.29 0.17 0.55 Note: ntr-s-ptive price for value is defined as the average of the field worker's prices for all three methods, with each weighted by the share of that method in total CYP which he or she delivers. For number of units of individual methods sold, the price for the single method alone we used. 57 5.5. Productivity and Effectiveness of Field Workers In terms of program design, all other things being equal incaluing number of field workers, the larger the population in a provinoe, the low the volume and value of pill sales (Table 5.2). By the same token, the mom field workers in a province, c s , the mmre contraceptives sold and the noe revenues generated. ihis finding suggests that attempts to allocate, on the average, workers to "too large" poptlations my indoe negative marginal productivity, as woers are overly extinded. That is, worker praductivity and overall cost-effectiveness cannot be increased, on the average, by assigning wrkers to larg than curret populatioi. Alternatively, Profamilia has exploited this allocation criterion probably to its full potential and beyond. Profamilia's field workes are not under- worked and deand prcmotion activity (IEC) would be redundant. A significant program design variable explaining sales is the sitxe of contraceptive aotlets or points of sale (Tables 5.2 and 5.3)036 By the estimates reported in Table 5.3, a 10% increase in the number of outlets under the control of a field wrker is associated with an apprxdMtely 8% increase in his or her outpit, based on individual-level estimates. It is noteworthy that the ivpact of this variable is retained in the provincial- level estimates (Table 5.2) when population size, number of workers, and 36 Note that in Table 5.3, the individual worker is the unit of ctservation and analysis. The reader is reminded that an outlet or point of sale is a haosehold or a shop awner who either buys or accepts a cCni9ruint of contraceptives from the field wrker. Ihis is the field wrkers' d;stribztion chmel. 58 some relevant population characteristics larely affecting demand, are statistically controllei. The meaued effect is consistent with the hypothesis that points of sale enhance worker productivity through supply, in a situation where there is effective denand to exploit.37 It also suggests that one way workers can enhance program resoures is by mobilizing more ctaanity resources throgh points of sale. Here again, whatever ineass worker productivity or supply increases output as well.38 Clinical suport is not an inhibiting factor in field workers' sales. The estimated coefficients on the quantities sold (Table 5e2) are all insignificant, except total! suggesting that the same clinical infrast n way support more workers than at present, without hindering, on the av , the quantities of contraception sold by the average worker. or, in other words, there are still eonnmies to exploit as far as clinical support is -However, the more clinics, ateri , e lw value of sales. This suggests that where clinics are more readily available, field workers my refer users of relatively efficacious methods to the clinics. Clints, for their part, my opt for close-by clinics rather than for CBD outlets. As a a, clinics may also depress CBD sales39 37 The provincial data suggest that in texs of value of sales, mor posts my be associated with sarat laex prices. This may indicate camfietition betwae posts. 38 As there is no knowledge of the effort required to "produce" more sales outlets, no analysis is presented here on the optiml allocation of a worker's time betesn delivery and mobilization of points of sale. See Ane 1. An attevpt to explain nmmber of montraceptive outlets by population characteristics did not yield reaningful statistical results. 39 In this regard, the actual utput of the CBD prram is underestimated and undervalued beause its referrals to clinics and ply to mo efficacious mtods, are not comted as outp:t of this program. 59 Worker characteristics are discussed with the aid of Table 5.3. Gender has no measurable association with worker performanoe. contrary to Profamilia's original notions, male field workers do as well as femles in selling contraceptives. Marital status has an effect; merried workers sell more pills in particular and hence generate more CYP and revenues.40 These variables are probably associated with culture; married workers apparently appear more "credible" in the promotion of family planning than unmarried.41 Of the field worker characteristics studied, experience is the most significant in its influence on sales.42 The data indicate that ten percentage points in worker experience around the mean - or about seven months of addltional experience - is associated with an approximately two percent increase in all measures of output. Experience may be serving as a proxy for knowledge of the market and consumer behavior. This issue introduoes conceptual problems which have no clear solution, and which we do not introduce into the discussion. 40 With regard to the marriage status, tte results of Tables 5.2 and 5.3 conflict. As the data in Table 5.2 represent effects estimated on the basis of aggregate data, we base our conclusions on Table 5.3. 41 Note that two field worker characteristics of interest do not appear in the lists of independent variables: "years of schooling" and "number of children". These variables were available for just seventy-six of the ninety-five field workers. Regressions including these variables indicated that neither were significant for any of the output measures (with the exception of a positive influence of number of children on condom sales) and that their presence had a negligible effect on the measured influence of the other independent variables. Scme unreported variants of the estimates suggest that female field workers do somewhat better than males in the sale of spermicides. 42 The zero-order oorrelation coefficient between age and experience is relatively low (0.33), indicating that experience is not simply masking a positive age and marital status effect. 60 As for the variables for targeting, the percentage of mothers in the population has a positive effect on sales of pills, the mejor method, and insignificant statistically for the tw other minor methods, cadcms and spermicides (Table 502) ITlhe ooefficients on the composite price variable in "value of sales" equation suggest that sales my be pramoted by relatively lower prices.43 Literacy among wue appears to have no measurable effect on demand for ontraoeption in the CBD program. At the same time, hasehold inone has a negative and statistically significant effect on sales of spenndcide, and, wLth rather limited significance, on the value of sales. This finding s gests the possibility of lower prioes in high-ime areas.44 The is a positive associaticn between "altitd' and sale of pills (Table 502)0 This finding is irznsistert with the relevant hypothesis that work my be harder in hilly areas. It may well be that this variable is a poor measure of variations in altit1ude in any given area of field worker operations and it may represent a host of other factors asoiated with 43 The price varin:ble amtrolled for in this eqation is a weigted average. ReJ ateitly low prices for pills, the most ocan method, and high pric ror other less acom methods my yield the same average prices for different method mixes, and therefore induce more M because the more efficacias method, which is allocated a higber weight, has a lower prio. Indeed, there is a negative correlation between the proportion of mthers and prioes of pills (-0.43). 43 The data suggest that whre demand is higher the price is in fact lcr, and that the effect of in=mmay be associated with the effect of lower prioes. The simple correlation coefficient of inam with the mean price is -.28. This may indicate that Profamilia My be wting with othr conceivably private providers, in better-off areas. 61 given geographdcal areas. 5.6. OCntraceptive Prices mhe effect of contraceptive prices on sales is of particular interest, as prices are mare amenable to short-run policy change than ary other variable.41 ' i .netive asoiatio ririce ad iuntiy f il sales- sugMgeting that a 10t decreas in pill vixe will increase sales by a& 50%. Sales of both condom and spermicides are also negatively associated with their price, but the associaticn is imx weaker,perhaps indicating acceptor preferences for certain brands (as implied by the wide variation in spermicide prices) 46 While the estimte on sales of pills suggests a high price elasticity (or quantity respme to price) of demand for pills, the estimate on total revenues does not suggest the same as would be expected given the share of pills in sales. It is hard therefore to deduc price elasticities frem the estimates.47 Te 45 The prioes used in the analysis are implicit average prices calculated, for any one field worker, by dividing total value of sales by total quantity for ea c traeptive. This calculation leads to a built-in negative correlation between prices and quantities. 46 Attempts to estimate cross-elasticities, the effect of the price of one method on the quantity demanded of another, did not yield statistically significant results, and therefore remein unreported. Relatively higher pill prices see to depress aoind sales. Hen the prioe variable may represent some general deterrent effect of prices of the cammon method on denand for cmntraception in general. 47 The measured effect is too powerful, and may inoorporate several factors biasing it. The data suggest a negative association betwee prices and exogenous variables which have a positive influence an denjd: proportion of mothers in the population and average level of household inci. llehat is, prices are on the average lower where demand is presuwlbly higher in the first placo. The estimated coefficicr- is therefore higher (in absolute tsnis) than it shold be (see Annex 3). Lede prioes - even 62 imsignificant estimated effects of prices on total value of sales (Tables 5.2 and 5.3) in view of the built-in positive correlation beteen value (V =PQ) and prices, suggest a demand elasticity higher than 1:value of sales decline with higher prices. The number of users declines in any case. The finding concerning a negative (partial) aorrelation of "population size" with "sales of pills", on the one hand, and a positive correlation of low prices and percentage of nmthers in the population with such sales, on the other hand, suggests that workers can, on the average, handle better cOncetrated demand in given populations. This highlights a particular feature of the field worker's production technology, given their allocation. While they cannot cope with larger catchment areas, measured by population size, they can deal with higher demand in concentrated areas. This helps explain also the profound measured effect of points of sale on output. *The points are a means to augment the worker's span of operations. where coupled with higher demand - do not bring about higher revenues, as suggested by the insignificant price coefficients on total revenues. It may well be the case that in high demand areas cxcmetition with Profamilia is fiercer,, leadin the organization to lwer prices in these areas. 63 5.7. Progrm oots and Wge Determination The omqmosition of the cxost of the CBD Program is as follas: Labor 140.1 million pesos (36.0%)48 Supplies 158.9 " n (40.9%) overhead and other 89.8 (23.1%). there may be under-reporting in the last category due to the clinical support given to the CBD Program tlat is not included in these cost figures.49 Labor costs are by-and-large a fixed cost in the operations of any one field worker as discussed above. An increase in his or her productivity, getgr± ,arjlm wmd entail largely the marginal cost of supplies, and reducoe overall unit cost. of corse, any suggested change in number of worker and their "quality" would entail chtang in labor rost. Ixwledge of how different variables relate to labor oosts, on the one hand, and to output, on the other hand, is crcial to an understanding of hoiw tu influene the cost-effectiveness in the outreach progrm through personnel policy. 7Tu far we exained the effect of different variables, including num,r of wrkers and their characteristics, on output. the satm hes to be doe with regard to labor cost. 48 Thee costs include cost of medical personnl contributing to the CBD progrm. 49 Capital costs are included in Profamilia's act st 64 While wages are known, their relationship to wrkcer characteristics and operations, has yet to be established. Tb this end we estimated a wage eqution, with the natural logarithm of a wrker's wages as the dependent variable.50 The estimated coefficients are reported in Table 5.4. The level and statistical significanoe of the constant indicate that there is a relatively fixed wage bas, irrespective of any other variable.51 Moreover, most earning variations are explained by regional ditferences, which may reflect wage rates in local labor markets (that are all lower than wage rates in the capital, Bogota). As expected, seniority, measured by age and experience, contribtes to earnings. At the same time, narried warkers appear to earn less than their umrried colleagues052 Fhile marital status correlates with age (r2=.27), this is not sufficient to explain this particular finding.53 50 There may not be an explicit wage policy in Profamilia vis-a-vis the variables discussed here. Yet, co-variations betwee wages and worker characteristics do emerge in the data. Ihe may reflect systeatic decisions by managemnt that are not part of a clearly stated policy. 51 The reader is reminded that since we deal with the entire population of field wrkers, the reported statistics should be used just for their predictive value. 52 There is no information about hours wor-ked by field staff. It is assused that all work full-time. Given the unstrucured nature of this work, it is quite possible that married wocrkers work fewer hours than unmsried workers, and therefore earn less. 53 One possible explanation wild be that married workers in the CBD paogrm are secondary workers in their families. this is usually the case for maried Wvv. 1itere is, hover, no correlation in the data between marital status and gende. 65 It is also of interest that the rnml,er of points of sale is neatively asiated with wags. One possible explanation to this associaticn is that Profamiliaes _ agemt may coomider expost, relatively high riaws of points of sale and high volumes of sales as lowr need for worrs and henoe depress their wages. 66 Table 5.4: Regression coefficients (Natural Log of) Worker Wages coefficient T- Statistic Male (=1) -0.042 -0.77 Age 0.005 2.64 Married (=1) -0.110 -2.60 Experience (years) 0.018 4.10 FMgMDesign No. of Oontraceptive outlets -0.001 -1.91 Region: * Atlantic (=1) -0.217 -3.58 Central (=1) -0.363 -6.47 Oriental (=1) -0.356 -4.91 Pacific (=1) -0.412 -7.07 Constant 14.033 158.31 N 94 Adi. R-square 0.51 F 11.9 * "Bogota is the excluded region. 67 5.8. Cost-Effectiveness of Field Worker Operations The findings reported in the last three sections suggest marinal changes that can improve the cost effectiveness of the C3D Program. Relatively costless and therefore clearly cost-effective changes wuld involve: (a) a nmodified personnel policy; (b) prcmation of distribution outlets; and (c) finer targeting of field worker operations. With an imprved personnel policy, given Profamilia's current wage structure, the pxrgram could save by increasing the percentage of married field workers and introducing a long-tenr policy of reducing the worker's average age without sacrifice of average experience. That is, Profamilia's anag t might txy hiring young people and reducing turnover. It could thereby gain in efficiency by lowring the wage bill and, at the same time, increase productivity. Field workers should be encouraged to work with more points of sale in any given population. While there are no data about the cost of such a policy, the powerful association of points of sale with worker output renders such a policy likely to be oost-effective. For finer targeting, Profamilia might consider reallocating field workers from areas of relatively unfavorable demand conditions, wher the percznage of mothers in the poplation are relatively low, to ares whGre 68 those are higher. Such a policy must be handled with care, as the arginal gains in sales of contraceptives where there are relatively nore nothers ny be smell ccqpared with the lass in areas left without adequate field worker coverage altogether. The program should clearly be reorganized so that each wrker prodes U. = Udl (Figure 5.1). With the lxdget for the CBD Program given, such a reorganization wld inply that scam .areas reiin unoovered by the program. Such areas cauld be those with low dand, or areas close to clinics, as productivity there seem less than elsewhere. If the situation is indeed of "tcx large" populations (UCU> Usr,, Figure 5.1) as is suggested by the data, then the mejor challenge is to alloaate workers twards the level Usm vis-a-vis population per worker. If, in the average situation, the progran hires another full-time worker, so that each sells less than his or her ixiDm potential, say U%s, the costs to the program are the wge rate of the added worker and the loss of the ctrret worker's prodiction meawed by the segment (Usm - tP5), as the two wrker ncow share the sam population and level of effort (tod). The production gain (2UtsUsr) should be weighed against the additioral wage eoxditures. Since there is no evidence of a lack of effective demnd for Profamilia's cBD servioes, bit rather a resorce cnstraint, there is no sope to prczDte damnd in the short nrm through IBC. 69 5.9. Method Mix As alternative methods used by clients of different ages, have different CYP ooefficients and cost recavery potential, changing method mix at the margin may increase program efficiency, as outlined in Section 4.4. ro establish the relative efficiency of methods in terms of "Adjusted Total CYP" in the CBD Program, all relevant data are compiled in Table 5.5. The CBD program yields a total of 1.42 AICYP, adjusted for method efficacy and average age of users. The highest contribution is made tarog the pill. However, if all marginal funds are invested just in one method, they ought to be invested in spemicides. That is, it pays on the nergin to prolwte the most spermicides in the CBD Program and the least condca , subject to the program's ability to modify population behavior acxcrdingly. 70 Table 5.5: Adjusted Total CYP Gained by Investment in any Method in the CBD Program. Parameter* Pill Condom Spermicides Share in Tbtal Cost (%) ai 83.70 12.00 4.30 Marginal cost (pes.) mi 52.60 27.85 26.47 Price to client (pes.) Pi 52.50 11.60 18.40 CYP per unit ar 0.77 0.01 0.12 Efficacy Xi 0.97 0.88 0.79 Relative risk ri 0.96 0.91 0.91 Adjusted Total CYP ATCYPi 1.240 0.036 0.144 ATCYP if the last 100 pesos are invested the specific method 1.48 0.30 3.36 Note: AICYP are computed on the assumption that all funds are invested in all methods according to shares ai. Relative risks are based on mean age of new acceptors: pill - 26.1, condom - 29.5, and spenmicides - 29.5. For further elaboration, see Chermichovsky and Anson (1990). 71 5.10. (onclurAons and Implications. the CED Progran Several reajor conclusions and operational implications emerge from the analysis of the CBD Program: * T}he ntudber of field workers in an area as measured by the size of its population, is positively associated with cantraceptive sales. This implies that, on the average, Profamilia allocates workers according to demand and increasing the number of field workers wold enhance sales of contraceptives. * The average field worker appears to be over-extended in delivery efforts. As delivery of contraceptives is the worker's major task, it wuild therefore not be advisable to allocate additional IEC activities to them, at least in the short run. Evidence suggesting that female education levels in the target population do not have a measurable effect on sales, provides further s8p4xort for this conclusion. * mhe proportion of nmthers in a region's population is positively oorrelated with sales. Marginal gains in worker productivity can, therefore, be achieved through (marginal) allocation of workers to areas where there are higher concentrations of mothers in the population. * The number of points of sale supervised by a field worker, correlates strongly with contraceptive sales. While there are no data available on the program oosts of these sale points, their impact on worker productivity suggests they may be a major means to increase cost-effectiveness. * The experienoed and married workers sell more than their junior and unmarried colleagues. While experienced workers are paid more than inexperienced workers, married staff are paid, on the average, less than unarried staff. Reducing worker trnovwer, i.e retaning experienced staff (who are also mora likely to be married), should, therefore, increase productivity and possibly cot-effectiveness. * 'The prioe of contraceptives affects sales; relatively high prices have, as expected, a deterrent effect on sales of contraceptives, and by implication on equity of delivery. It is hard to infer from the data the sensitivity of consuemr demand to the price Profamilia's clients pay. .* Spermicides are the lost cost-effective method in the CBD Sub- Program. Subject to omwE ee preferences, current ccLtraoeptive prices, and marginal costs, Profamilia should, therefore, try to prxonte this metbod at the margin. 72 * Te clinical infrastructure available to an average worker does not have a measurable effect on worker produL.. vity. Hence, all the marginal dianges suggested abve can be achieved without expanding the clinical infrasture supporting the CB) Sub- Program. The implied inarease in nmter of field workers per clinic wold reduce average fixed aosts of almtraception in the CBD Sub-Program. 73 6. CLINIC-BASED ROGRAMS 6.1. Objective and Framewrk of Analysis Profamilia's Clinical Sub-Program and Surgical Sub-Program are both amn by ntutber of MM in each clinic. This is the "Block" of MDs and nure. At the same time, the m1 ber of nurses in the clinic deviate frcm the predicted nmber. The ratio betwe the actual mmber and the predicted mmber is assigned to the pedicted urnEw for a given activity, surgeries or onsuultations. This yields an "actual" 1n*er in the activity that is divided by Me in the activity, yielding Kj. 87 Table 6.4: Percentage C2anges in Clinical Personnel to Produce a 10% Inaease in Output without budgetary onsequences Certified Non-Certified Eysician Surgeons Nurse Nurse Clinical Pills -2.36% 7.24% Condoms -1.04% 3.19% Sperimicides -1.36% 4.17% IUDs -1.00% 3.06% consultations -2.58% 7.91% 1. Alternative 1 - 1.47% 8.05% 2. Alternative 2 - 8.55% 24.64% 88 6.7. Fixed Capital osts-; and Scale of Operations Determination of the optisal size of the average clinic is inmpotant from an efficiency perspective, as it leads to the best organization and use of overall resources, including fixed inputs and administrative costs. Te data suggest that clinics operate within a range of decreasing nmrginal returns vis-a-vis clinical labor inputs, especially with regard to the Surgical Sub-Program. This is sound eooncmic practice. At the same time, capital inputs - clinic area and equiprent value, and administrative staff - do not seam to constrain productivity. This suggests that,, on average, clinics operate in the range of decreasing average fixed costs. That is, the Clinic-based Program hav rising mxrginal oost of their prime labor inputs and presmebly decreasing average fixed costs.66 Are total unit costs per operaticn rising or falling, on the average? One way to awr the question, is to relate total costs directly to output. Cost functions have been estimated separately for the owultations in the Clinical Sub-Program, and for surgeries in the Surgical Sub-Program. Ibe general function for exanination has been tne following: Tot. cost = SO + i3 (level of outt) + 132 (level of tput)2 + I3 (Vedium size" clinic = 1) + 134 ("large size" clinic = 1) + v. 66 Csts of supplies are by definition prortional to atput. 89 Allowance is made for type of clinic, because of possible qualitative differences amnog the different types. The estimated coefficients are shan in Table 6.5. Table 6.5: Regression Coefficients of Program Cost (t-Statistics in parn ) Clinical SUb-Program Surgical Sub-Program (1) (2) (1) (2) Comultation or surgeries 1098 2901 2709 2029 (14.41) (7.94) (2.81) (1.80) (Consultations or Surgeries)2 -0.0032 -0.0226 0.6667 0.6219 (-3.71) (-2.62) (5.78) (5.01) Typel: Large Clinics (=1) - 105432 - 9517783 - (3.19) - (3.31) Type2: Medium Clinics (=1) - -1004477 - 660896 - (-0.65) - (0.42) Constant 2388481 1959507 2113436 2188365 (4.61) (1.72) (1.99) (2.35) N 38 38 38 Adjusted R-Square 0.97 0.94 0.95 F 749 333 221 The coefficient on "Cnsultatiam" is positive and cn "Cosltations"2 ergative for the Clinical Sub-Program, suggesting that total cost per unit is decreasing; the more consultation a clinic has, the lower average total cost of a consultaticn, on the average. This result is amiistent with the 9o finding discussed in section 6.4. While there are decreasing returns to 'blocks" of clinical labor inputs, there is scope for wre use of existing capital. The rising unit cost due to clinical staff inpts are offset by decreasing costs of other ir'uts. Hen, more demand can be a dated at dereasing total average costs - catbining all costs - in the Clinical Sub-Program; msr clients can be managed at lower unit oost per client. This sub-program can do so through dcges in ratios of medical staff, withct buigetary autlays, and by increasing only medical inpts, withi the apropriate outlays, in existing clinica, cn the average. For the Surgical Sub-Progrm, the coefficients on "Surgeies" and cn "Surgeries2" are positive, as are tke cefficients on @"ypl1" ad "pe2". These oefficients imply that the large the surgial opatior, the highe the average cot of surgey, and that the large clinics are mere expensive. This is consistent with the findings reported in section 6.4 of relatively strong decreasing marginal productivity of clinical staff in the Surgical Sub-Program. The implied rising mrginal and average variable costs are not offst, on the average, by any deceasing fixed costs per unit of output. This reult is probably strongly influenced by the fact that MeI in the surgical Sub-Program are paid 'ty pieoe", and therefore there is less scope in this sub-progrna than in any of the other to exploit eoonmies that would be associated with the "quasi-f ixed" nature of the cost of medical staff when paid fixed salaries. 91 6.8. Methd Mix Ihe importance of the method mix is from a oost-effectiveness perspective, discussed in section 4.4. OQarisons ammng methods within the same program, require infonration about the marginal cost and nmrginal revenue for each methode As cost data are not given separately for each nmthod, estintes are used for the marginal cost of each metod. It has been assumed that capital costs,. IC costs, and overhead oosts are fixed costs, and thus dihuld oe exluded from nMrginal aost calculation. Tfherefore, these cot elements are deducted fmrn cost of both ooultations and surgeries. The marginal costs of a method include the costs of the omvtraceptives and the labor costs of consultations.67 Costs of contraceptives are available, but those of consultations are not. To estimate the marginal cost for each nthod in consultations, a linear specification of the following function was astinated using regression analysis: Cost of cionsultalons = ao + a, x pills + a2 x oondcms + a3 x spernicides + a4 x IUDS + a5 x surgeries + w. 67 Based on Profainilia data, the costs for contraceives wre: 36.42 Oolalbian peasc per cycle of pills, 7.9 per cordm, and 6.6 per unit of spermicide. See Annex 5. 92 The margial aost of consultation for each method would be the aoefficient crrespomding to that method.68 The mazginal cost of a method include the cost of suplies plus inputed cost of acsltation. The revenues for each method are estimate on the basis for the unit price for each method plus the average wrsultation fee to the client.69 Table 6.6 shoWS the marginal costs and revenues for eact method offered in the Clinical-based Sub-Programs. lb which method within the Clinic-based Program should an additional peso be bikgeted? Ihe data in Table 6.7 provide a snuary of all relevant data per relationship 4.4. 68 Cherenichovsly and Zwra, 1986. See nun 6 for the estimates. 69 Prices wre based on the CBD prices for pills, condcm and spenuicides, and the average revenue for sterilization (see abve). Fbr IUts the price, not including consultation, was assumed to be nil or 0 93 Tu-le 6.6: Marginal Cost and Marginal Pavemue, by Method Mrinal Margil Cost Revenue Sukfidy Pill 192 109 83 CQndmn 57 33 34 Spermicides 65 39 26 IUD1* 4599 1691 2908 IUD2* 1205 595 610 Surgeries 8350 2270 6080 * IUDm uses "niew IUD consultations" as the utpa unit. IUD2 uses "tDtal IUD o1nsultations" as the output unit. 94 Table 6.7: Clinical PrOgram: Adjusted Total CYP and Relevant Data by Method in Clinic-based Programs (per 100 pesos invested) Pill QCndoms Spermicid. IUD Surgery Share in Tbtal Cost% - ai 4.14 1.86 2.61 26.68 64.70 Marginal cost (pesos) - u=i 192 67 65 1205 8350 Marginal Revenue (pes.) - pi 109 33 39 595 2270 CYP per unit - a . 0.077 0.010 0.118 2.500 12.500 Efficacy - Xi 0.970 0.880 0.790 0.970 0.990 Relative risk - ri 0.96 0.91 0.91 0.96 0.89 Adjusted Tbtal CYP A¶CYPi 0.0022 0.0003 0.0047 0.06703 0.1.100 ATCYP if the last 100 pesos are invested the specific method 0.055 0.017 0.180 0.251 0.170 Note:AMCYP are computed on the assumption that all funds are invested in all methods acxrduinq to shares ai. Relative risks are based on mean age of new acptors: pill - 26.1, cmr-dcn - 29.5, spermicides - 29.5, IUD - 26.1, and feanle sterilization - 30.4. Foc further elaboration, see Chernichovsky and Anson (1990). The method with the largest protecticn per 100 pesos allocated to the clinical-based activities is sterilization.70 However, an the basis of "objective efficiency" sterilization is a relatively "poor" method, being even less efficient than spermicides, for two reasons: it has the highest subsidy, both in absolute and relative terms, and it has a poor "relative risk" coefficient because of the relative high age of users. The marginal peso has the highest return in IUD, followed by Spexmicides. 70 All IUD consultations, old and new, were assumed to produrce identical CYPs of 2.5 CYP per insertion. This is in contrast to Profamilia's assmmptions that only new acceptors of IUn "produce" CYPs. For 1986 Profamilia reports 45,906 insertion of IUDs to new users and 129,311 to old ones (Ojeda, 1986; Table 1, p.2; Table 10, p.10). 95 6.9. ConclusioLns: The Clinical-based Sub-Programs Several corclusions arnl operational implications emerge for the Clinical- based Sub-Programs. * The size of population in the average clinic's operaticnal area, dam not correlate with the clinic's output. Output does correlate, however, with number of clinical staff. Hence, there is sufficiernt demand for clinical operations; they are constrained by availability of such staff. * The relatively edwated and =-- less effluent populations appear to take =re advantage of Profamilia's clinic-based operations than less educated and better off populations. Therefore, better utilization especially of the Clinical Sub-Program, can be acheved throgh improved targeting of relevant operations in favor of the fonrer populations. r * The narginal costs of the Clinical Sub-Program, delivering mainly ITJD, are lowr than the marginal costs of the Surgical Sub-Program providing sterilizations. The activities of the faor can be expanded within the same clinical inf:rastructure, and hence at lowe than current unit cost of mainly IUD sevices. Tere is, therefore, sc;pe to acIjI-ate use of the services of the Clinical Sub-program at relatively low marginl costs and lwer overall wuit costs. * The clinical sub-progrm is more cost effective than the Surgical Sub-program because the latter involves a relatively high subsidy to sterilization. Moreover, the acoeptors of this method have a higher mean age, and conequaiztly a lower risk of prEnancy. * The ratios of nurwse to physicians correlates with output ot clinics: highe nurses/ptysicians ratios are associated, on the average, with higher aotput. There is scope, therefore, in both the Clinical and Surgical Programs to increase outPut with no additional outlays, and therefore umpe efficiency, by trding physicians in favor of numes. * The fees for service affect the numbe of acceptors at least in the Surgical Sub Program; fees are nrgatively associated with nmber of surgeries. Tlhe sensitivity of this number to the fee appears low, hawver. FAhile direct measuereEnt of the ptential effect of fees in the Clinical Sub-Program was not possible, the data suggest that relatively high levels of sales are supported by low price levels. Profamilia may, therefore, be able to finance a necessary expansion of its relatively efficient Clinical Sub-Program by raising its average fee for service, especially in the Surgical Sub-Program uhere the subsidy elemet is high, and the demand elasticitY appears low. Wtile prices will deter sc2e demand, revenues Wald 96 inease ard enable to service new ard possibly outlyin pqpulatirns. 97 70 RELATIVE (SUB) PROGRAM EFFICIENCY 'lhe discussion thus far has treated the different sub-prograis separately, studying ways to improve each sub-program's internal efficiency. mhe question of how to gain in efficiency by allocation across programs still revains to be answered. To establish the efficiency of allocation across programs, including the CMV, Adjusted Total CYP (ATCYP) for the entire Prof amilia prg have been caalculatd (lable 7.1) on the basis of relationship 4.4. For each sub-program the ATCYP is the average of the ii1lividual methods wighted by the share of the metthd in the cost of that sub-program. That is, the relevant ATcYP figures represent ATcYP that can be achieved with 100 pesos invested across programs and in each sub-program. Wen all activities are looked at ocrt,ined, pill delivery through the CBD program, is most efficient, follawed by sterilization and IUD. Vte ocmparin sub-programs, the calculations show that the CBD sub- progrm relatively the most cost-effective operation. It is followed by the Clinical Sub-Program. The Voluntary Sterilization is the least oost-effecdive sub-program, once cost recovery and mean age of users is considered. ! ~~~~~~~~~~~~~~~~9B Mt2e 7.1: d)mb WI1 C:P ad Ieat P fc amfii Sb- S~~~~~~~~~~~~~Mia-w P;m"C -C E2n [B -_ C_D PWill an . - ID 9I.I PiUll Cn g S:e in '1 st 2.07Y 0.9A 1.31t 13.34% 32.36t 41.8% 6.OM 2. L% ftwlimi mr(P.) 192.00 67.00 65.00 1205 MM0 52.60 27.85 X.47 !9giml 1Eme (P.) 109.OD 33.00 39.00 595 2270 52.50 11.60 18.40 'W pE mit 0.77 O.01 0.12 2.b0 12.50 0.77 0.0 0.12 EffiHy 0.97 0.88 0.79 0.97 0.99 0.97 0.88 0.79 EiPative Ri& 0.96 0.91 0.91 L.96 0.89 0.96 0.91 0.91 AMP, 0.00X 0.0003 0.0)42 O.t5%2 0.20 1.240 0.006 0.144 AtYPfIzrutn 0.28 0.16 1.42 0.17 -1.42 NP&Esm taU 5.5 ad 6.7. 99 8. CONDSICNS In 1986, Profamilia's three sub-program delivered approximately 1 267,919 Couple Years of Protection (CYP) to the popilation of Colombia at a cost of 1,252.8 million Colombian pesos ($US 6.43 million). Ihe program recovered about 647.3 million pesos ($US 3.32 millicn), or about 50t of its costs, which makes it comnparatively unique amng family planning program. According to Profamilia's account, the sterilization pmgram provides the largest share of CYP, 61.0t, followed by the CBD Prgram, 11.4%, and the clinical provides the remaining 27.6y6. The CBD ard the Clinical Sub-Programs provide the major share of revenue, 42.6* arnd 42.3%, respectively. In terms of total costs, the CBD Program acmounts for the smllest share, 9.9%; the Clinical 47*3%, and the Voluntary Sterilization Sub-Program 32 8%. The data reflect key policy and anagm decisions: allocation of resources a1mong progranm, fee setting, and allocation of resources within prograqn. A major is5ue is wtether Profamilia can do better with the resources available to it. Eor example, how should it allocaSte reso es to maximize the protection it offers, thus increasing the cost-effectiveness of its operations by reducing the unit cost of protection? To ansrr these questions, we examine Profamilia's resource allocation and costs in relatiuL to oaut (quantity and value of contraceptive sales) in conjunction with population characteristidcs, uthod mix, anid program 100 design. We study the cost-effectiveness of Profamilia's overall service delivery and each sb-program separately, emloying a cross-sectional econcmic analysis of the operations of 97 field workers and 38 clinics comprising the program. The data show a positive correlation betwen labor input and outpxt in all operations. Profamilia's operations are thus mainly constrained by resource availability, with nmre resources the program could deliver more protection. No program activity, particularly in the Clinical Sub-Program and the CBD Sub-Program, appears bound by a lack of effective denEmd. Moreover, these t: sub-programs can be expanded with the available clinical infrastructure. For this reason and the quasi-fixed nature of labor cnst in these two sub-prorams, higher levels of output are associated, on the average, with lwer unit costs of contraception. The same does not hold for the Surgical Sub-Program because of its comparatitely high narginal cost in part associated with the paymant method "by surgery" to surgeans. Of the different sub-pograms, the Clinical Program, delivering mainly the IUD, and the CBD Program, delivering mainly the pill, are the mos cost- effective. The Voluntary Sterilization PrograM is the least cost effective because of the relatively high cost of sterilization, the high subsidy element, and the high mean age of acceptors. Given the relative efficiency of the sub-.prograns and the viability of eqxpnding each program's operations, overall program efficiency could be improved as a result of studying the client needs and preferences and considering within this context, a shift of resornes frcm the Voluntary Sterilization Program to the CBD and Clinical 101 Pragrals- Given that the program uses fees for service extensively, these shwld be considered as a vehicle to improve its performnce. Since the evidence suggests that nmre demancd could be satisfied with more orkers in all sub- progrm, and that lower prices may not necessarily reduce revenues, theire my be soope to raise prices in order to finance additional staff, and increase overall sales. Thus would be the case if access rather than prices is a barrier to higher levels of sales. TIhe evidence about the positive impact of points of sale in the coamunity on sales of contraceptives in the CBD Sub-Program, supports the noticn that access is important. In that event, higher fees may even improve equity. Profamilia could raise fees for sterilization at same loss of acceptors of this method, and thereby shift resources to the other sub-program. That is, by reduing the current cross subsidy from the other programs to sterilizatiom, Profamilia should be able to improve its overall efficiency. The issues cancening fees merit more research, especially in conjunction with data on consmer demand. here is scope to increase the cost-effectiveess of any of the three sub-program individually by more careful targeting of operations, better mix of labor inputs, and imrved use of canmmity resouces. 102 Prcductivity of outreach operaticns is higher where there is a higher concetration of mothers in the population, and of clinical operations where the population is more educated. At the margin, targetin or shiftiug limited resources towards those populations could therefore increase the cost-effectiveness of the different p:o.ram. In the CB1D Sub-Program the experienced and married workers distribute (through outlets) more contraceptives than their junior and unmarried oolleagues. tWile experienced workers are paid more than inexperienced workers, married staff are paid, on the average, less than unmarried staff. Retainin experiencd staff (who are also more likely to be married), should, therefore, increase productivity and possibly cost-effectiveness. This could be done by raising wages. In the Clinical-based Sub-Programa th ratio of nurse to physicians correlates with output of clinics; higher nurse/physician ratios are associated, on the average, with higher output. There is scope, therefore, in both the Clinical and Surgical Sub-Program to increase output without aditional outlays, and thereby improve efficiency, by trading ptysicians in favor of nurses. Cmmity resources tend to augment urgram resources in outreach activity. The rnuter of points of sale administered by a field worker correlates strongly with contrceptive sales. kthile there are no data available on the pr=grm costs of these sale points, their impact r. wrker productivity suggests they may be a major maans to increase cost- effectiveness in the CBD Program. 103 The implications of this study suggest marginal changes in resource allocation ard management of the Profamilia program. They must stard a nore refined evaluation of consumer demard. 104 REFEE Chernichovsky, D. and Zmora I. (1986) "A Hedonic Prices Approach to Hospitalization Costs", J. of Health EoorAi , 5: 179-191. dernichovsky, D. (fortdxning, 1991) "Effectiveness, Cost, and Cost- Effectiveness of Family Plannng Program; Methcdolugy and Operational Guidelines". wrnTidwovsky, D. and Anson J. (1990) "Cost Recxery and the True Cost- Effectiveness Ratio of Contraceptive Delivery". Mimo. World Bank. Washington, D.C. Chernidwsky, D. (forthorning 1991B) "0ptiml Allocaticn of ctension Workers in Family Planing .treach operations" COrporacicn Centr Regional de Pablacion (1986), ECUeta de O=vl=ia [DaKgrafia y Salud~ I96 Easterlin A.R. and Crinuirs M.E. (1985) g F ilityA- Dud Aaalysis. University of Chicago Press, Chicago & Lcrkn. Ojeda, G. (1986), "Informe de Actividades de Servicio 1986", BoleFEn de Evaluacign Y _tadistica. Profijj, Vol. 44. Ojeda, G., J. Amdeo and A. M?ry (1981), 5t r M21eYea of Pote=tim: 'fle Cas of Profaiilia 1977-19. Rlaeweig, M. and T. Sdcultz (1982), "Child Mortality and Fertility in Cocabia: Individual and mmouity Effects", Hea Poli a, 2: 305-348. Villamil, R. (1986), "Informs Finrgciero", Boletin do 1abid prgtii, Vol. 15. ANNEC 1 Clinic by Type Bogota Male Clinic Cali (including the Male Clinic) Medellin (including the Male Clinic) garmnia Barraqilla Ccuta Manizales Monteria Neiva Painira Pasta Pereira Santa Marta Sincelejo Tulua Valledupar Apartado Bello* Barrancabemeja* Bueamventura Caldas-Antioquia Castilla-Antioquia Florencia Girardot Kennedy-Bogota Ocana* IPaxyan Quirigua-Bogota Quiroga-Bogota Riaha Riomegro San Andres* Soledad-B/quilla TumaO TUnja Villavioencio *No data for the analysis are available an these clinics, and they ar not include in the analysis. 106 ANNEX 2 The Worker's Wtimal Time Allocation bee Delivery ard Resa-ce Mbilization. Let us assum that a wrker's roduction funcsticn is denoted by: UOf(td, I, PC) whiere: U = measue of outpit td = time allocated to delivery I = cxmumity and other infrastrure PC = worker characteristics. Let us further assume that: I I(tpj, CC) where: tm = time allocated to mhilizaticn of rescwces in the omaunity C = cimunity disarateristics A worker can allocate his or her total tim (T) so that: T = td + tm + te where: 4 = time allocated to dewnd p-zKticn. Let us assum that te is a ooi3tant. Henae dtd = -dtm. That is, whatever time is allocated to delivery is withdrvn frci resmaze xbilizaticn. The optimal allocation rule bebm the tw activities is: 71 See Q1ernidhivsky (19913). 107 dU/dtd U/&td - (&U/&I)DItd = 0, Nauly, the gain in mrgimal pOdCtivity dUe to aditional ti in delivery sk"uld rcatch the loss in productivity due to less activity in rxl mbtilizatia3n. 108 ANNEX 3 Potential Bias in Dsmand Elasticity Estimates Supose that we have a family of demarnd curves for contraception characterized by D1 and D2 in the figure below. D2 indicates that higher demand is associated with higter inxne. The data indicate that prices are lwer in higher inrxxe areas; P1 is associated with D1.and P2, with D2. cmseguently, the estimated price elasticity is influenoed by the slope of AB rather than AC. The demand elasticity is nuch higher along AB for the relevant ranxe, indicating the bias in estimates. 109 ANNEX 4 Allocation of Nures to Clinical Operaticns F t*aile information is available separately for physicians in the Surgical and Clinical Sub-Programr, no such data exist for urns; only the total number of nurses in each clinic Li available for the to sub-prorams. As there is a high correlation between certified and non-certified nurses, and pysicians, it is assumed that proportions betwee nurses and pysicians in each program follow sone fixed ratio, but there is a varianoe about this ratio across clinics.72 conseuently, to estimte the allocation of nwses between the Surgical and Clinical Sub-Program, the following function has bee estinated: Nurse typek = (a, * MDs in consultattcnk) + (a2 * MDs in surgeryk) (k = 1....37) Ihe results are reported in the TaDle below. 72 The first order correlation mtrix is: Ms es- . m Cert. Nurs 0.95 0.97 ° Non-Cart. NUrses 0.84 0.93 0.88 110 Regression Coefficients of Nurse to Physician Ratio (t-values in parentheses) Non-Certified Certified Nurse Nurses MLs in 1.6720 0.5194 Surgical Program (6.12) (4.13) MDs in -0.2262 0.2455 Clinical Program (-1.61) (3.79) N 37 37 Adj. R-square 0.8837 0.9571 F 141.83 414.47 Accordingly, there are about 1.7 certified nurses and 0.5 non-certified n-rses, on the average, for every MD in the Surgical Sub-Program, and 0.25 non-certified nurses to every MD in the Clinical Sub-Program.73 These estimates yield predicted values for nurses in each sub-program for each clinic on the basis of numbers of M?s in each sub-program. The sm of these values naturally deviate in most instances frmm the actual nummer of nurses. Consequently, the numder of nurses was adjusted in each clinic so that the ratio of numbers of nurses in the different programs is the ratio of the predicted value. 73 As the ooefficient for MDs and non-certified nurses in surgery is negative but not significantly different from zero, it was assumed that no- certified nurses did not work in the Clinical sub-program, bit only in the surgical pro . illl ANC 5 Reallocation of MDs andN trses for Higher Output witout 8udgetary_ Let us denote: K = N/M and Y =AKa, where y = level of ouutpt, N = nurber of nmze, MH number of Me, and a = the p d in outpt due to a percage dg in K. Henoe, (&dK) = (dYA)/a. Let's denote: t(dy/y) ]R r = the rate of change in no. of nwses, t - the rate of dhange in no. of Ms. Henoe, 112 (dk/k) = r-t R/a. (1) As in our case a is known and >0, suppose we wish raise output of R% by increasing the number of nurses (N) by r% and decreasing the number of sysicians (M) by t% with no budgetary conseuences. We need to establish r and t. In order to retain the same budget: (r) (N) (WN)=(t) (M) (WH) (2) r-(t) (MIN) (WM/V?N)-tf 1/K) (WtWt) Let us denote B=(1/K) (WHAfN) By substitution of (1) into (2): r-rB = R/a r = R/[a(l-B)] 113 ANNMX 6 Marginal QCosultation Cost of IUD, urgeries and of Sales of Oontraceptives Regression coefficients: Marginal Corsltation Cost of IUD, Surgeries and of Sales of o Ve s Marginal Oost of Consultation t-value Pill 167.94 1.50 Condom 377.84 2.79 Sperwicide -280.36 -1.79 IUD 4266.03 4.68 Surgery 1724.98 3.06 Oonstant 1005878.58 1.25 N 38 Adj. R-square 0.9356 F 109.09 Regression coefficients: Marginal Oost of Surgery and NWi,e of Sugeries Marginal Cost of surgeries t-value surgery 7689.22 18.78 Constant -2000448 -1.91 N 37 Adj. R-square 0.9071 F 352.51 PRE Working Paper Series Contact Z ~~~~~~~~~AutAhorX for paper WPS742 The Cost of the District Hospital: A. J. Mills August 1991 O. Jadora A Case Study from Malawi 3 091 WPS743 Antidumping Enforcement in the Ange '