Exploratory Research in Clinical and Social Pharmacy 5 (2022) 100105 Contents lists available at ScienceDirect Exploratory Research in Clinical and Social Pharmacy j ourna l homepage: www.e lsev ie r .com/ locate / rcsopConsolidation or multiplicity in supply logistics for health commodities?Ebenezer Kwabena Tetteh Department of Pharmacy Practice & Clinical Pharmacy, School of Pharmacy, College of Health Sciences, University of Ghana, Legon, Accra, Ghanahttp://dx.doi.org/10.1016/j.rcsop.2022.100105 2667-2766/© 2022 The Author. Published by Elsevie 0/).A B S T R A C TA R T I C L E I N F OArticle history: Received 27 September 2021 Received in revised form 1 December 2021 Accepted 11 January 2022Public health-sectors of most low- andmiddle-income countries (LMICs) run a central medical stores (CMS)model that is monopolistic in character. Concerns raised about monopolistic CMS arrangements stress the need to encourage cost- reducing efforts and improve service levels (outputs) by having multiple competing logistics institutions. This paper examines the desirability of consolidation or multiplicity in supply logistics by focusing on the task of inventory man- agement (that is, distribution, storage andwarehousing). The paper uses theory and historical evidence to describe and suggest a desired form of multiplicity in LMICs. Consolidation shouldn't progress to the point of monopoly and multi- plicity doesn't mean having an infinite number of logistics institutions. A limited number (2−10) of logistics institu- tions, that are full-line and national in their scope and scale of operations, should be enough to provide choice, support competition and minimize the risk of supply disruptions. Health policy and planning in LMICs should explore ways of turning existing logistics institutions in the public, private and non-governmental sectors into a multiplicity of types that are capable of assuring uninterrupted supplies of health commodities.Keywords: Consolidation Distribution Logistics Multiplicity Supply Wholesaling1. Introduction Public sectors of most low- and middle-income countries (LMICs) rely on a central medical stores (CMS) supply chain model in their efforts to as- sure health commodity security. This CMS model often describes a monop- olistic logistics system with no alternative competitors to encourage cost- reducing efforts and improved service or output levels. Yet the apparent dominance of CMS monopoly in the public-sector, and conventional wisdom advocating integration and/or consolidation of logistics systems doesn't fit well with reality. What is commonly observed is a hybrid state with multiplicity of government, non-governmental and private logistics systems, not integration or consolidation.1,2 The benefits of multiplicity, it is argued, are: (1) greater flexibility to maintain supply with alternative op- tions, (2) competition to lower costs and improve service levels, and (3) op- portunities to tailor logistics systems to meet priority health program objectives and specific product needs. Multiplicity, however, has the fol- lowing downsides: (1) the need for increased oversight [regulation) and co- ordination, (2) perverse incentives to collude and increase costs, and (3) overspecialization on select geographical regions, health facilities or commodities can reduce supply flexibility.3 Clearly, how logistics systems are designed and organized in LMICs will determine whether the benefits of multiplicity (consolidation) outweighs the downsides. Beyond this gen- eral statement, the benefits and downsides listed provide no indication of what the desired form of multiplicity is or should be. That aside, there are several definitions of multiplicity. Health systems in LMICs can generally be described as having three sec- tors: government, non-governmental and private. Multiplicity could mean having one or more (competing) logistics systems serving each sector sepa- rately or serving all three sectors together. We could also have a singler Inc. This is an open access articlemonopolistic logistic system serving each sector but this will be counted as multiplicity. Besides the case of a single logistics system serving all three sectors, a blend of arrangements is possible. We could have monopo- lies in government and non-governmental sectors but two or more logistics systems in the private sector. Such an arrangement will be counted as mul- tiplicity. Within each sector, one may observe multiplicity simply because existing and emerging logistics systems only serve health facilities within some geographical locations. Or, they supply only a subset of health com- modities, for e.g., family planning products or just essential medicines. These examples are, however, not the only forms of multiplicity. A func- tional logistics system is made up of institutions undertaking the following logistics tasks: product selection, (demand) quantification and procure- ment, inventory management (warehousing, storage and distribution) and service delivery. Since what is needed for commodity security is syn- chronized execution of these logistics tasks, multiplicity couldmean having multiple logistics institutions undertaking one or more logistics tasks on be- half of affiliated actors in all three sectors. It is also important tomake a dis- tinction between (1) multiplicity in relation to the number of logistics institutions (units), where each unit is a subsystem of the logistics system; and (2) multiplicity within a logistics institution, i.e., multiple subunits or divisions that form part of a single logistics institution (unit). This is often observed for logistics institutions undertaking the task of inventory man- agement. A network of subunits could, for e.g., be organized into echelons or tiers (most commonly central, regional or district levels) to reflect public governance and administrative boundaries within a country. If logistics sys- tems can be designed in various ways that fit with the label multiplicity, what then is the desired form of multiplicity for supply logistics? According to Bornbusch and Bates,3 the challenge for LMICs is identify- ing a state of “prudent multiplicity” where the incremental costs of havingunder the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4. E.K. Tetteh Exploratory Research in Clinical and Social Pharmacy 5 (2022) 100105 Table 1 Dimensions of output. Logistics function (task) Dimensions or indicators of output Inventory management: Percentage of commodities with adequate Warehousing, storage and shelf-life distribution Percentage of logistics facilities holding stock within minimum and maximum levels Percentage of inventory counts that matches records Frequency of stockouts and/or emergency orders Average duration of stockouts Percentage of logistics facilities in compliance with guidelines Percentage of stock expired or damaged (accident rates) Percentage of orders that are filled as requested (order fill rate) Distribution lead times (versus the average) Percentage of complete reports Percentage of complete reports submitted on time Ratio of transportation cost to value of commodity Total warehousing, storage and/or distribution costs Percentage of markup on commodities (in a cost-recovery system) Source: USAID.27,28one more logistics institution or system is worth the incremental benefits. The benefits being promoting competition and choice; hedging supply- disruption risks to assure commodity security, and better health outcomes. This paper addresses that challenge by focussing on institutions executing the logistics task of inventory management (= storage, warehousing and distribution). This is because most statements about “supply logistics” are in fact references to the task of inventory management. Also, because a lo- gistics system is the sum of its component parts and multiplicity can take different forms, it is more informative to evaluate each logistics task sepa- rately from the others. For instance, the national electronic procurement service in Chile (ChileCompra or “ChileBuys”), discussed by Bornbusch and Bates,3 is an example ofmultiplicity to improve the logistics task of pro- curement. This e-procurement service was adopted in response to decen- tralization and the autonomy regional health authorities had to bypass pooled procurement run by Chile's CMS. With an electronic procurement platform, regional health authorities in Chile can still bypass the CMS but they have to procure directly from the same set of suppliers the CMS deals with. The Chilean case shows how multiplicity can help maintain, in a decentralized environment, lower prices and cost savings achieved through pooled procurement. What it doesn't show is the benefits of multi- plicity in inventorymanagement. In otherwords, a valid overall assessment of the desirability of multiplicity of logistics systems must first evaluate the desirability of having multiple institutions executing each logistics task. Health planners in LMICs can then use these assessments to build functional logistics systems with the form of multiplicity most suited to their country contexts. In what follows, this paper examines whether multiplicity in supply lo- gistics (inventorymanagement) is desirable. The paper has two aims. One is to identify and characterise a form of multiplicity that is best suited for as- suring uninterrupted supplies of all health commodities needed. Two is to show that the downsides of multiplicity (consolidation) are not universal. Much depends on whether logistics systems in LMICs can be organized, reconfigured or transformed in ways that balance the benefits of consolida- tion and multiplicity.2. What is the desired form of multiplicity? To determine whether multiplicity in supply logistics (inventory man- agement) is desirable, and in what form, this paper makes the following as- sumptions. First, each LMIC market is made up of three sectors or submarkets: public, private sector and non-governmental. In each of these submarkets, health facilities act as price-sensitive buyers or there is a strate- gic purchaser acting on their behalf. Second, health planners have identi- fied and selected a basket of essential health commodities that have positive clinical and economic value in treating the range of diseases that the general populationmight suffer from.Health planners' objective is to as- sure aggregate commodity security –which is defined as the uninterrupted supply of all health commodities in the essential basket whenever and wherever these commodities are needed. Third, health planners desire to have logistics institutions and for that matter logistics systems that are de- signed for “last mile distribution” of the basket of selected health commod- ities in whatever quantities deemed appropriate, constrained only by the number and geographical spread of health facilities. Fourth, health plan- ners have no inherent dislike for relying on private logistics institutions (private wholesalers) to undertake the logistics task of inventory manage- ment. Fifth, given the focus on inventory management (what is called pure wholesaling in private markets), the output dimensions health plan- ners care about are those shown in Table 1 below. The dimensions in Table 1 are naturally metrics to measure how well inventory management is executed. Given these assumptions, the paper describes in detail charac- teristics of what it considers as the desired form of multiplicity. These char- acteristics are: (1) logistics institutions with long-lives, (2) competition and choice; and (3) hedged supply risks.2 2.1. Logistics institutions with long lives To ensure, today and tomorrow, uninterrupted supply of health com- modities in all therapeutic categories, whenever and wherever they are needed, health planners must be concerned about the long-run survival of logistics institutions, whether public, private or non-governmental. This point is best explained using mathematical expressions. Let t refer to each short-run time period, d is the number of districts or regions within the country, n is the number of health facilities within each district or region, c is the number of therapeutic categories and j is the number of essential health commodities in each therapeutic category. To ensure long-run survival of a logistic institution undertaking the task of inventory management, the following conditions shown in Eqs. (1), (2) must be met: ( ) T D N C J   ∑ ∑ ∑ ∑ ∑ Pi i ijcndt−Cjcndt ∗Qjcndt ≥Fþ Profit (1) t¼1 d¼1 n¼1 c¼1 j¼1 Pi ¼ Pi ienduser acquisition þ Pjcndt (2) where i refers to a given logistics institutions (or group of institutions), P is the price charged for a commodity or service provided, C is the correspond- ing marginal or incremental costs (approximated by average variable costs), and Q is quantities of health commodities supplied (a product of number of filled requisition orders and the number of SKUs per filled order). F refers to the sequence of repeated sunk or (quasi)fixed cost invest- mentsmade in setting up, maintaining or upgrading the needed logistics in- frastructure (plant). Such sunk or fixed costs (term F) will cover building warehouses or distribution centres with each having adequate power sup- ply, lighting, spacing [floors and docking areas for a fleet of delivery trucks and vans etc.]; security, fire and safety devices and measures [first-aid, tracking and antitheft kits etc.]; storage facilities for commodities that are inflammable or require strict temperature control. F will also include the costs of furnishing warehouses with equipment for handling commodities (pallets, pallet racks, static shelves, lift trucks, rolling warehousing ladders etc.) as well as manual or computerized logistics management information systems (LMIS) linked to devices for automatic data collection (bar code E.K. Tetteh Exploratory Research in Clinical and Social Pharmacy 5 (2022) 100105 1 There are two instances where customer pickup arrangements make sense. One, logistics institutions cannot provide or health facilities are not willing to pay for direct (last mile) deliv- ery at any price even if this means higher order fill rates, lower direct costs and avoiding the indirect time costs of healthcare workers travelling to pick up orders when they should be car- ing for the sick. Two, close proximity to distribution centres makes it possible for health facil- ities to pick up their orders at costs that are lower than the prices charged for inventory management. It is hard to imagine the former and the latter is unsatisfactory since that argu- ment does not hold for health facilities located in remote regions.readers, radiofrequency identification technologies etc.). It will also include continuing or repeated sunk or fixed costs made (in LMIS, warehouse ca- pacity and transportation) to improve outputs (lower distribution lead times etc.). It will be necessary to do so with availability of a broader range of health commodities and increasing demands (for e.g. increases in number and geographical spread of health facilities). Without these continuing or repeated investments, commodity stockouts will be more frequent. Eq. (2) suggests the final price paid by each end-user, customer or ser- vice delivery point, in each time period must be the sum of commodity ac- quisition prices and prices for inventory management. Prices for inventory management, in the short-run, arewhat is commonly referred to as distribu- tion markups. These prices must be estimated taking into account prevail- ing LMIC demands and not just the continuing or repeated costs required to assure uninterrupted supplies. Each functional logistics institution must then aim to secure over time a sequence of short-run prices that allows re- covery or recoupment of all fixed or sunk costs plus a competitive normal profit, where relevant. If health planners prefer a policy of zero distribution markups (zero prices for inventory management), then some other source of revenue is called for to cover costs. 2.2. Competition and choice If logistics institutions are to have long-lives, they should be able to re- cover all the economic costs involved in assuring commodity security. This however should not come at the cost of monopolistic inefficiencies. In the absence of competitors, the only constraint on a monopoly in executing the logistics function of inventory management is zero demands which is at odds with the objective of assuring commodity security. Competition and choice is desirable for a number of reasons. First, it reduces how much of societal resources is devoted to inventorymanagement. In contrast to the indeterminacy of pure bargaining, competition helps reveal a wider range of feasible supply prices as suppliers (in our case logistics institutions) strive for incremental business volumes ormarket shares. Compared to a re- liance on hard (take-it-or-leave-it) bargaining skills, competition driven by price-elastic demands is akin to the bargaining tact of taking business else- where. Facedwith price-sensitive purchasers, logistics institutions will give serious consideration to cost-reducing process innovations.4 A number of preconditions must however be met or created to derive the societal bene- fits of competition. One, there must be an adequate (not an infinite) number of competing units offering a homogenous or near-homogenous service levels (outputs). That is, the competing units must be similar, and carry out the same set of logistics activities. Since the planner's objective is to assure uninterrupted last-mile deliveries of all health commodities, whenever and wherever they are needed, the competing units (logistics institutions) should be de- signed for and capable of fulfilling that task. Two, unlike theoretical ideals of perfect competition where prices charged must be equal costs, short-run prices for inventory management (the pure wholesaling function) can be higher than the corresponding marginal or incremental costs. This is in fact optimal and necessary in the presence of (continuing or repeated) fixed and sunk costs, although the optimal deviation from incremental or marginal costs depends on how price sensitive purchasers are.5 A simplified but useful representation of real-world imperfect competition is the gener- alized “many-firms” Cournot model.6,7 In this model, the excess of prices over costs is given by Eqs. (3), (4).   Li ¼ Pijcndt−Cijcndt =Pi ijcndt ¼ S =η (3)  2 L ¼ ∑ Si =η ¼ HHI=η (4) i¼1 where Li is the Lerner index (ameasure of deviation of prices from costs) for a given firm (logistics institution), S is the share of the aggregate demand facing a given logistics systems or supply chain, L is the sum of quantity- or share-weighted Lerner indices, HHI is the Hirschman-Herfindahl index3 and η is the price-sensitivity of demands of health facilities or the strategic purchaser acting on behalf of health facilities (in each sector). Eqs. (3), (4) suggest (1) policies, guidelines, regulations or rules that en- courage collusion between the competing logistics units and (2) a very small number of logistics institutions will generate a sequence of higher prices over the relevant time periods than is socially optimal or necessary. Conversely, a higher number of logistics units will not lead to competitive pricing without purchaser price-sensitivity. Thus price-sensitivity by healthcare purchasers' demands is far more important in providing ade- quate market discipline and incentives for efficiency. To be specific, what is needed is at least two competing logistics institutions with near- homogenous outputs (see Table 1) plus price-sensitive purchasers. Being price-sensitive means being able to credibly exercise the bargaining threat of taking business elsewhere such that lower prices for inventory manage- ment are rewarded with more than equiproportionate increase in demand volumes or market shares.8 Striving for custom (incremental demand vol- umes or market shares) is the hallmark of real-world imperfect competition4 – and it can easily be translated into a standard operating pro- cedure. It certainly doesn'tmean giving all business or contracts to the logis- tics unit offering the lowest price; just that prices should be inversely related to demand or business volumes. See van Valen et al.9 for a matrix that can be used to share demand or business volumes amongmultiple win- ners (competitors). The expectation here is: competition will get logistics institutions to make the continuing or repeated sunk-cost investments needed to maintain excellent outputs and provide adequate incentives to lower the costs involved either by adopting cost-reducing process innova- tions, wider short-run spreading of fixed sunk costs over larger demand vol- umes and exploitation of long-run economics scale and scope. Competition spurred on by price- and output-sensitive purchasers will initiate and en- force constant experimentation and a search by logistics institutions for the optimal scale of operations, especially with regards to the number and size of distribution centres. Multiple competing logistics institutions (units) should put the brakes on unrestrained multiplicity of distribution centres (subunits) within each unit. We want to emphasize that, to assure secure supplies of all health com- modities in the basket of essentials, the competing units (public, non- governmental or private logistics institutions) must be full-line and na- tional, not full-line subnational or short-line national or short-line subna- tional entities. Full-line here means the competing units supply, at least, everything included in the basket of essential health commodities. These units must be national (not nationalized) in that all health facilities includ- ing those in remote areas have secure lines of supply. For competition to thrive, none of the competing units should have exclusive (geographical) distribution ormarketing rights for any of the health commodities included in the basket of essentials. That is, full-line national logistics institutions must be protected from competing units that engage in what is known as cream skimming and skimping. Cream skimming arises from a specialized focus on, for e.g., fast-moving, profitable health commodities and/or on serving only selected districts or regions within any LMIC, perhaps those with higher relative incomes. Skimping refers to attempts to avoid the con- tinuing or repeated sunk-cost investments required to maintain outputs at desired levels. For instance, a competing logistics institution might opt not to supply “unprofitable” vaccines requiring stricter temperature control to avoid investing in cold-chain storage facilities. Or, instead of direct (last mile) delivery, it will only offer customer pickup arrangements.1 The issue is: under imperfect competition, logistics units might not necessarily earn zero excess profits: there will be some winnings (winners making excess profits) and losses (losers making less than normal profits) even though E.K. Tetteh Exploratory Research in Clinical and Social Pharmacy 5 (2022) 100105firm-specific (market-wide) profits maybe close to zero. Cream-skimming and skimping creates opportunities for logistics units will smaller scale or scope of operations to erode “profits hidden beneath the zero” that are cru- cially needed for long-run survival of competing full-line national units.5 Cream-skimming and skimping should be barred or at least discouraged since these practiceswork against the objectives of ensuring aggregate com- modity security and having long-lived logistics institutions. The sequence of fixed or sunk-cost investments required for direct (last mile) delivery of all health commodities to all health facilities in different regions in any given LMIC could be in thousands or millions of dollars. But these costs spread over larger demand volumes and multiple time periods will be com- paratively small. In a number of scenario analyses of direct (last mile) deliv- ery of vaccines in Guinea, Madagascar and Niger, Prosser et al.10 report that, although the sum of additional start-up and operational costs were in millions of US dollars, this amounted to less than $1 per dose of vaccine delivered over a 1-year period. As shown in Eqs. (1–2), each competing lo- gistics units has to secure a sequence of prices low enough to allow full re- covery of all (continuing or repeated) sunk costs made over the long-lives of these institutions. Prices charged for inventory management then need not be frontloaded in attempts to recover the costs incurred within the shortest possible time. The costs involved should be recovered over the long-run, not after a few years. Considering low household incomes and low per capita expenditures on healthcare in LMICs, health planners should con- sider putting limits on the number and type of competing logistics units li- censed to operate. Barring overspecialization (skimping and/or cream- skimming), for e.g., means competitive entry will be observed only by full-line national entities. This together with limits on the number of logis- tics institutions will consolidate larger business volumes (Qi) for each com- peting unit.11 Indeed, to consolidate even larger business volumes (and a steady stream of cash flows to support capital investments), competing lo- gistics institutions, in LMICs, should serve all three sectors: public, private and non-governmental. This is particularly important where the costs of raising capital is high. In addition, volume guarantees may be considered but only as the outcome of a competitive process in which the size of these guarantees are inversely related to prices charged for inventory management. 2.3. Hedged supply risks In Section 2.2,we argued that competition and choice only needs at least two full-line, national logistics units. The question is: are these limits on multiplicity desirable in a real world with diverse risks and uncertainty, where a lot can go wrong or deviate from expected outcomes. Instead of a stable source of supply, one could have variable outputs (throughput or ser- vice levels), issues with quality of health commodities and unreliable sup- pliers. These are the opposites of what characterises stable supply (dependable lead times, stable outputs, infrequent breakdowns, reliable suppliers, diverse set of suppliers and minimal constraints on supply capacity).12,13 In this section, we examine whether an infinite number of (competing) logistics institutions is necessary for mitigating supply risks. Wedo soby assuming each competing logistics unit has the sameprior prob- ability of a breakdown but these probabilities are independent. If these probabilities are only conditional on LMIC context, then the joint probabil- ity that all competing units will breakdown is∏Probi, where i refers to each logistics unit in a set ofN units. Conversely, the probability of assuring com- modity security is 1 − (∏Probi). If there are two logistics units with prior probabilities of a breakdown equal to 0.1, then there is only 0.01 (= 0.12) chance that both logistics units will breakdown. On the other hand, if the prior probabilities are 0.9, there is a 0.81 (=0.92) chance that both logistics units will breakdown. If the prior probabilities are 0.6, one will need ten competing logistics units to attain a 0.99 chance of assuring commodity se- curity. This outcome can be achieved with just three units if prior probabil- ities are 0.2. These simplified examples suggest the benefits aremultiplicity are lower when the probabilities of breakdown approaches one. Generally, one's ability to hedge against supply-risks via multiplicity di- minishes as the number of logistics units increase. But it does so more4 quickly when there is more than a 50:50 chance of a supply breakdown. Fig. 1 provides a graphical illustration for a selected set of prior probabili- ties of supply disruptions. The picture changes somewhat if the prior prob- abilities of a breakdown or supply interruptions are dependent. If prior probability of a supply-breakdown is 0.1 for one logistics unit but for an- other the probability of a breakdown conditional on the first unit breaking down is 0.4 (perhaps because both units rely on the same set of institutions undertaking one or more logistics functions other than inventory manage- ment), the joint probability of both logistics units breaking down will be higher. That aside, the probability of a single logistics unit experiencing a disruption will, in part, be determined by the number of subunits or divi- sions (in each echelon or tier). The reason is: for each logistics unit to be functional, all of its component parts (subunits) must be functional. If the number of subunits is 2 and the probability of each subunit experiencing a fault is 0.1, then the probability of the logistics institution being func- tional (as a sum of its component parts) will be 0.81.With 12 (30) subunits, the probability that the logistics unit will remain functional is 0.28 (0.04). So besides cost inefficiencies, multiplicity within a logistics unit makes it more difficult to achieve a 100% order fill rate, for example. The additional benefits of risk-hedging viamultiplicity doesn't just diminishwith increases in the number of logistics institutions, but also with interdependent risks and increases in the number of subunits (per tier or echelon) within each lo- gistics institution. Whilst these observations hold inmost cases, the answer to the question of whether multiplicity is desired also depends on health planners' aversion to supply risks and their prior subjective estimates of the probability of dis- ruptions; and whether these expectations are confirmed or refuted using historical data or events experienced. Based on Fig. 1, we believe 2–10 (competing) logistics units should be adequate for risk mitigation. A more risk-averse health planner, however, may prefer more than 10 logistics units even though the incremental reduction in disruption risk with each additional unit gets smaller and smaller. Note also the magnitude (of the consequences) of a supply breakdown is as important as the risk or proba- bility that it occurs. Rare (low probability) catastrophic events that lead to a steady fall in order fill rates to zero or complete exhaustion of stock on hand (afire outbreak for e.g.) can be adequately hedged viamultiplicity. Effective internal, day-to-day management of logistics activities should be enough to hedge against low and high probability events with minor im- pacts (such as small variations in demand, staff absenteeism, work-related injuries, staff strike actions). High probability high impact (catastrophic) events, on other hand, cannot be adequately hedged via multiplicity. In such situations, it is more difficult to assure commodity security since what is needed is frequent and costly redesign or reconfiguration of logis- tics units.14 3. Parallels with historical evidence From the above, the desired form of multiplicity in supply logistics in LMICs should be one characterised by a limited number of competing full-line, national (not necessarily nationalized) logistics institutions doing business with price- and output-sensitive purchasers in all three sectors. This section draws parallels between that proposition and how private pharmaceutical wholesalers in Europe and the United States (US) have evolved over time. 3.1. Pharmaceutical wholesaling in the US Over the period 1700–1929, there were in the US mostly small short- line regional wholesalers located near seaports since this made the business of importation easier. Settling in new territories and new transportation routes (roads and canals) encouraged entry by broker-middlemen whole- salers who had to travel from the interior of the country to port cities in order to procure their products and/or assess the quality of products pur- chased. Growth in numbers of wholesalers including broker-middlemen was fuelled by growth in healthcare demands. By 1929 there were approx- imately 55,000 drug stores compared to 25,000 drug stores in 1880, E.K. Tetteh Exploratory Research in Clinical and Social Pharmacy 5 (2022) 100105 0.30 0.25 Prob = 0.1 Prob = 0.5 0.20 Prob = 0.7 Prob = 0.9 Prob = 0.3 0.15 0.10 0.05 0.00 2 3 4 5 6 7 8 9 10 NUMBER OF LOGISTICS UNITS (BEYOND MONOPOLY) Fig. 1. Limits to hedging supply disruption risk via multiplicity.Notes: The incremental (marginal) reduction in risk of disruptionwas computed as the difference between the probabilities of assuring commodity security (1− (∏Probi)) as the number of (competing) logistics units increase by one. The starting point is the case of monopoly, and it is assumed the prior probability of any logistics unit breaking down is independent of other units. INCREMENTAL REDUCTION IN DISRUPTION RISKalthough the number of drug stores relative to population size remained at roughly one store for every 2000 people. In addition, the number of hospi- tals increased from 178 in 1872 to 4000 in 1910. Competition between the broker-middlemen was however confined to geographically distinct mar- kets (territories). Fears of excessive and unfair competition led to the crea- tion of the National Wholesale Drug Association (NWDA) in 1882. The NWDA had 201 active members in 1886. The period 1929–1978 saw entry by the first national wholesaler, McKesson & Co., after mergers of the parent company in 1929 with 64 other wholesalers in 31 cities. Given increasing healthcare demands and fierce price competition between broker-middlemen, the emergence of a national wholesalers can be seen as an attempt to exploit in the long-run economies of scale and scope in order to survive competitive pressures. Forming a national wholesaler out of smaller regional wholesalers helped avoid some or all of fixed or sunk costs (capital investments) needed to enter new previously unserved mar- kets. It also helped build a large customer base (as it is much easier to cap- ture loyal customers of regional wholesalers through mergers or acquisitions than trying towin over their loyal customers as a separate busi- ness unit). By 1961, McKesson served 33,000 drug stores and 5000 hospi- tals. Over the same period 1929–1978, there was also the emergence of other large wholesalers (formed through acquisitions and mergers of small regional wholesalers). These larger national wholesalers including McKesson switched from short-line to full-line distribution – selling both pharmaceutical and non-pharmaceutical products.15,16,17;18 Over the period 1978–1996, the number of private wholesalers fell dra- matically. In 1970, there were 144 wholesalers with 372 distribution cen- tres (145 wholesalers with 395 distribution centres in 1975) but between5 1975 and 1995, the number of private wholesalers dropped from 145 to 63 with 224 distribution centres. By 1996, there were 55 wholesalers with 233 distribution centres. In the year 2000, there were fewer than 50 private wholesalers with the big three accounting for 88% of the US whole- saling market. The dramatic reduction in multiplicity was mostly due to wholesaler mergers and acquisitions for the purpose of serving previously unserved geographical markets and increasing sales at existing locations. But consolidation across geographical markets, and the incentive for whole- salers to form and operate national distribution networks in the USwas also riven by the emergence of price-sensitive, cost-conscious hospital or retail pharmacy chains or large group purchasers acting on behalf of individual hospitals or retail pharmacies located in the different geographical regions. These healthcare purchasers (providers), themselves were faced with in- centives to be cost-conscious through the introduction of prospective pay- ment systems and managed care competition. The various cost- containment practices adopted by healthcare purchasers made the demand curves facing wholesalers more price elastic. And because these healthcare purchasers (providers) were located in different regions, their selected prime vendors were national full-line wholesalers. Regional or local whole- salers could not supply all health facilities thatwere located inmultiple geo- graphical regions and belonged to the same purchasing group.15–18 Not all (smaller) wholesalers responded to these incentives to create na- tional distribution networks, but the ones that responded were able to aug- ment economies-of-scale and scope with cost savings achieved via the introduction of cost-reducing process innovations. For example, the use of order pick labels, barcodes and lightweight wearable computers to locate stocks in large warehouse space and for sorting; “night picking” of products E.K. Tetteh Exploratory Research in Clinical and Social Pharmacy 5 (2022) 100105in warehouses to ensure orders were ready for delivery in the morning of scheduled delivery dates; and batch-order picking, i.e., picking all products listed on a given batch of orders received together so as to reduce the re- peated tasks of picking the products ordered individually (thereby reducing intra-warehouse travel time per order item). These private wholesalers also offered value-added service: electronic ordering and managing computer- ized logistics information systems.153.2. Pharmaceutical wholesaling in Europe Traditional private wholesaling in Europe is one that relies on full-line logistics units to supply a broad range of products. According to Walter et al.,19 full-line wholesalers bundle, on average, medicinal products of roughly 18 manufacturers together per delivery, selling 74% of all medi- cines in Europe to mostly retail pharmacies (93%), hospital pharmacies (4%), dispensing doctors (2%) and drug stores (1%). These private full- line wholesalers deliver within 3 h, 16 times a week on average.19 How- ever, the full-line operations of these private wholesalers is in part deter- mined by regulations of the mix of active therapeutic molecules and products containing these molecules on a country's market. In Norway, known to strictly regulate the mix of products available, full-line whole- salers in the mid-1990s supplied 4000 products containing 800 active ther- apeutic molecules. In Germany, full-line wholesalers supplied 70,000 products containing more than 3000 active therapeutic molecules because less strict regulations on productmix.20 As in the US, private wholesaling in Europe went through phases of consolidation or erosion of multiplicity. In France, for e.g., the number of wholesalers fell from more than 150 at the start of the 1960s to 13 regional and national full-line wholesalers in 2002. In 1989, there were 2 wholesalers in Norway, 2 in Sweden, 4 in Denmark, 7 in Ireland, 20 in France, 25 in Germany, 30 in the Netherlands, 33 in the United Kingdom, 60 in Belgium, 210 in Spain and 290 in Italy. In 2001, there were 346 full-line wholesalers in the EU com- pared to 600 in the early 1990s, with 70% of these wholesalers based in Italy and Spain.20,21 Consolidation or erosion of multiplicity in Europe was driven by a con- fluence of factors: (1) creation of a single European market; (2) traditional full-line wholesalers facing increased competition from foreign whole- salers, short-line wholesalers, retail pharmacy cooperatives integrating backwards into wholesaling; and (3) reduction in gross operating margins – as healthcare purchasers, who had to work with tight budgets, pursued cost-containment strategies.20,21 Notably, creation of a single European market led to cross-country specification of minimum standards for phar- maceutical wholesaling – and the costs of meeting these regulatory stan- dards were more easily borne by partnering wholesalers than by small- size short-line independent ones. Full-line wholesalers who chose to be local (national) faced declining demands in contrast to pan-European wholesalers. The reduced gross operating margins of these national full- line wholesalers led to poor quality in distribution services (mostly in term of lead times and frequency of deliveries). Private wholesaler compe- tition in Europe was such that some traditional full-line wholesalers, stock- ing a broad range of health commodities suffered losses. And, because 10% of all products distributed by full-linewholesalers, accounted for 85%of the profits, this created perverse incentives to switch from full-line to short-line wholesaling.20,21 But cream-skimming (i.e., supplying only a select range of fast-moving profitable products) does not sit well with effective competi- tion in supplying all health commodities needed. Indeed, more recent liter- ature report that short-line wholesaling is prohibited in France and Spain.19 Note that cream skimming wasn't the only way European wholesalers responded to low profit margins. They (1) increased their productivity by improving the accuracy of demand forecasts, upgrading transport infra- structure, automating warehousing activities; (2) increased cost-efficiency by closing down distribution centres; and (3) expanded their business port- folios by diversifying into non-traditional goods and services, for e.g., supplying non-drug products [cosmetics, toiletries, video tapes etc.], supplying new healthcare providers and providing financial services.206 4. The importance of similarity In Section 2, this paper argued that, to assure aggregate commodity se- curity, competing logistics units undertaking the task of inventory manage- ment must be full-line, national entities. The word national refers generally to the geographical or administrative boundaries of the relevant market, withinwhich the intended service delivery points including health facilities are located. In Section 3, the paper provided historical evidence from the US and Europe showing the importance of government regulation and price-sensitive healthcare purchasers (providers) in generating high- powered incentives to lower costs. In this section, we stress the importance of ensuring competing logistics institutions are similar or homogenous in terms of their outputs or services provided (i.e., there is no room for cream-skimming and skimping).We do so by looking at challenges of assur- ing health commodity security in China's Hubei province. In what follows, we consider the set of primary healthcare facilities within the boundaries of Hubei province in China as the relevant (“national”) market. In Hubei province, public-sector competitive tendering involves a two- envelope system under which suppliers of health commodities send bids for the procurement of health commodities whilst private distributors (wholesalers) send bids for the business of delivery (inventory manage- ment). Suppliers then select their preferred distributors from the set of bid winners. Final price to the end-user is as shown in Eq. (2). A supplier could select any number of distributors out of a total of some 305 (in 2015). This is not necessarily a bad policy: it only requires prices for inven- tory management to be inversely related to business volumes and that se- lected logistics units offer near homogenous outputs. The problem was out of the 305 distributors in Hubei province (most of which are only li- censed to operate within a particular county), only one or two of these dis- tributors could serve all primary healthcare facilities in the province. A large number of logistics units competing in smaller submarkets (counties of the province) meant inadequate revenues to support investments in warehouses and delivery vehicles to serve all local areas. This was thought to have contributed to poor delivery rates. In response, three alternative ar- rangements for supply logistics were introduced in 2012. Under these ar- rangements, the number of bid winners suppliers can select for deliveries within a county, was restricted.22 The first alternative delivery mode trialled was a medicine-tied distribution in which one distributor was se- lected for one medicine. The second was a recipient-tied distribution under which primary healthcare facilities selected one out of four distribu- tors nominated by suppliers and chosen by government. The third was a recipient-medicine distribution under which primary healthcare facilities were served by one of two distributors selected by a supplier. The medicine-tied and the recipient-medicine distribution models seem de- signed to accommodate short-line logistics units, whilst the one-recipient- one-distributor model comes closer to the full-line logistics units this paper recommends. Yang et al.22 evaluation of these arrangements showed introduction of the one-distributor-one-medicine model led to a drop of 7.78 percentage points in delivery rates for rural facilities and even worse the one-distributor-one-recipient model led to a drop of 19.85 percentage points in delivery rates. These findings suggest that simply limiting the number of distributors (winners) did not solve the problem of unreliable supplies. Attempts to exploit economies-of-scale and scope from consolida- tion may inadvertently undermine delivery rates and lengthen lead times. This conclusion runs contrary to theory and historical evidence. Limiting the number of logistics units effectively increases market size or business volumes (per unit), but it still requires distributors to make in- vestments in logistics infrastructure for deliveries to rural and not just urban areas. Until that happens, delivery rates are likely to fall if the contracted distributors do not offer much in terms of geographical cover- age. Concentration of contracted distributors in urban areas meant urban primary health facilities were not severely affected. However, delivery rates fell in rural areas with no alternative options. Rural facilities responded to erratic supplies by making larger requisition orders but then the negative impact of delivery delays is greater with larger orders.22 The key policy lesson here is: logistics units selected during competitive E.K. Tetteh Exploratory Research in Clinical and Social Pharmacy 5 (2022) 100105tendering must be those capable of serving all intended health facilities. Although selective contracting with only 2 distributors capable of serving all primary healthcare facilities may appear as a near-monopoly situation, it may be enough to hedge against supply risks and support competition if the probability of disruptions is low (< 0.1) and healthcare payers are sen- sitive to both prices and outputs (lead times, service levels etc.). A more risk-averse health planner may find this argument unconvincing; in which case, efforts must be made, over time, to encourage (smaller) distributors with limited geographical coverage to grow ormerge into the kind of logis- tics units capable of serving both rural and urban facilities. The alternative arrangement of allowing short-line subnational distributors to coexist with full-line national entities will erode the “profits beneath the zero” that logis- tics units need to cover costly investments for direct (last mile) delivery of all health commodities to all health facilities, rural or urban, within 72 h. That said, the administrative and transaction costs of switching between competing logistics units should be seen as costs to be incurred for provid- ing choice and for competition to thrive. Besides, switching costs can be minimized by inviting and selecting logistics institutions who are similar in terms of their outputs or ability to serve all delivery points in the relevant market. 5. Conclusion Reducing communicable and non-communicable disease burden in LMICs requires uninterrupted supplies of medicines and other health com- modities. Assuring uninterrupted supplies of all health commodities, when- ever and wherever they are needed, means these commodities must be available, acceptable, affordable and accessible. The logistics tasks of prod- uct selection, demand quantification, procurement and inventory manage- ment are therefore critical for the success of various vertical and horizontal health programs. Since warehousing and distribution costs constitute up to 16% of medicines budget in LMICs,1 it is important to find ways of putting societal resources to its best possible use. One way of doing this is: reconfiguring or transforming current arrangements for inventory manage- ment in LMICmarkets into one characterised by a limited number (2–10) of full-line national logistics institutions serving governmental, public or non- governmental actors. Such an arrangement is most suited for assuring ag- gregate commodity security in LMICs with relatively lower household and national incomes. Limitedmultiplicity should offer adequate insurance against disruptions related to storage, warehousing and distribution. How- ever, other sources of supply risks (regulatory, political, financial [taxes, tariffs, currency conversion and interest rates etc.]) will have to be hedged using different strategies.23,24 There are limits to the risk-reduction benefits of multiplicity. And, for choice and competition, one doesn't need a large number of logistics units than is necessary or sufficient. A large number of competitors is no substitute for price- and output-sensitive purchasers.11 Thus consolidation shouldn't progress to the point of monopoly and multi- plicity doesn't mean having an infinite number of logistics institutions. If there exists a state of prudent multiplicity, it will be at the point where con- solidation and multiplicity almost mean the same thing. This paper set out to examine whether multiplicity in supply logistics (inventory management) is desirable – and if so, what form of multiplicity will help meet social objectives of assuring commodity security. The pre- ceding sections described a form of multiplicity that offers the dual benefits of consolidation and multiplicity. Historical evidence from the US and Europe show consolidation towards limited multiplicity improved service delivery (i.e., national coverage for a full-line range of products) with little or no harm to competition. European and US private wholesalers, as logis- tics institutions, responded to price-sensitive demands and cost- containment incentives to adopt cost-reducing process innovations and, in the long run, exploit economies of scale and scope. The paper, however, leaves health planners in LMICs to make judgements as to whether the costs are worth the benefits of having 2–10 full-line national logistics insti- tutions serving all sectors. If health planners favour multiplicity but not the7 suggestion to build 2–10 competing units, they will have to make judge- ments as to whether additional multiplicity offers more benefits than it adds to costs. It is unlikely that health planners in LMICs have stacks of hard data to make such difficult choices, which means they should give themselves ample time and work with a longer-time planning horizon in their attempts to implement reform change. Having settled on N number of competing logistics units, the next challenge is transforming current structures within LMICs into the “state of prudent multiplicity” chosen. The initial drive to do so may come from (a coalition of) entrepreneurs, who understand the essence of aggregate commodity security to population health, getting together to form a national full-line wholesalers out of smaller regional short-line wholesalers. The initial drive may also come from public-sector outsourcing or contracting out distribution of health commodities in urban (rural) areas to private wholesalers or non- governmental entities whilst channelling public efforts towards supply se- curity in rural (urban) areas. An example is initiatives by VillageReach (a nongovernmental agency) in Mozambique. Another example is Uganda's National Medical Stores opting for in-house supply to district stores whilst outsourcing last mile delivery from district stores to third-party logistics providers who use trucks, bicycles, boats and even head-loading to supply entire districts within 15 working days.25,26 The drive for change may also come from the creation of a private (quasi-government) logistics insti- tutions (similar to Zambia's Medical Stores Limited) and/or healthcare pur- chasers demanding lower prices for inventory management and higher output levels (for e.g., door-to-door last mile distribution within the shortest possible time).25 Still, other government interventions may be nec- essary. In China, for instance, government introduction and nationwide im- plementation of Good Supply Practices (GSP) requirements in 2004 for private wholesalers led to a fall in the number of private wholesalers from 15,000 to 7445.1Whether this reduction in multiplicity was due tomergers and acquisitions, or some private wholesalers simply exiting the market is unclear. Perhaps exiting wholesalers in China did not have the capacity (capital) to meet these GSP requirements or were unwilling to incur the continuing or repeated outlays needed to meet these requirements, given expected business volumes. Both ways, the evidence indicates a trend to- wards consolidation. LMICs could therefore adopt a similar strategy with the explicit objective of reaping the benefits of both consolidation andmul- tiplicity. To sum up, this paper believes a balance between multiplicity and con- solidation in logistics institutions or systems (at least with respect to the task of inventory management) can be achieved in LMICs. It will require health planners tofindways of turning existing logistics institutions operat- ing in the public, private and even non-governmental sectors into types that are capable of assuring aggregate commodity security. Our ongoing re- search considers this issue, and we encourage theoretical and/or empirical work by others on what is arguably a less researched topic. These lines of inquiry could, for e.g., look at evaluating the performance of transformed logistics arrangements in LMICs in terms of how effective competition is, whether there is adequate hedging of disruption risks to support uninter- rupted supplies, and the impact on health outcomes. Declaration of interests The authors declare that they have no known competing financial inter- ests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments The statements, views and conclusions expressed in this paper are those of the author and not necessarily those of the author's affiliated institutions (in the past or present). The author takes full responsibility for all errors. No external research funding was used in support of this paper. E.K. Tetteh Exploratory Research in Clinical and Social Pharmacy 5 (2022) 100105References 1. WHO. World Medicines Situation 2011: Storage and Supply Chain Management. Avail- able at: http://apps.who.int/medicinedocs/ 2018. Accessed May 16. 2. Raja R. Commodity Security for Essential Medicines: Challenges and Opportunities. Arlington, VA: USAID | DELIVER PROJECT. 2008. Available at: https://www.rhsupplies.org/ uploads/tx_rhscpublications/CSEssMedChall.pdf. Accessed November 12, 2019. 3. Bornbusch A, Bates J. Multiplicity in public health supply systems: a learning agenda. Glob Health Sci Pract 2013;1:154. 4. McAfee RP, McMillan J. Competition and game theory. JMR 1996;33:263–267. 5. Baumol WJ. Regulation Misled by Misread Theory: Perfect Competition and Competition- Imposed Price Discrimination. Washington DC: The AEI Press. 2006. 6. Vickers J. Market power in competition cases. Eur Compet J 2006;2:3-14. 7. Clarke R, Davies SW. Market structure and price-cost margins. Economica 1982;49:277– 287. 8. Tetteh EK. Implementing differential pricing for essential medicines via country-specific bilateral negotiated discounts. Appl Health Econ Health Pol 2009;7:71–89. 9. van Valen M, Jamieson D, Parvin L, Ramirez CL. Dispelling myths about drug procure- ment policy. Lancet Glob Health 2018;6:e609–e610. 10. Prosser W, Folorunso O, McCord J, et al. Redesigning immunization supply chains: re- sults from three country analyses. Vaccine 2021;39:2246–2254. 11. Tetteh E. Creating reliable pharmaceutical distribution networks and supply chains in African countries: implications for access to medicines. Res Soc Adm Pharm 2009;5: 286–297. 12. Lee HL. Aligning supply chain strategies with product uncertainties. Calif Manag Rev 2002;44:105–119. 13. Lee HL. The triple-a supply chain. Harv Bus Rev 2004;82:102–112. 14. Wieland A. Selecting the right supply chain based on risks. J Manuf Technol Manag 2013;24:652–668. 15. Fein AJ. Understanding evolutionary processes in non-manufacturing industries: empiri- cal insights from the shakeout in pharmaceutical wholesaling. J Evol Econ 1998;8:231– 270. 16. Clemons EK, Row M. McKesson drug company: a case study of Economost - a strategic information system. J Manag Info Sys 1988;5:36–50. 17. Oswald SL, Boulton WR. Obtaining industry control: the case of the pharmaceutical dis- tribution industry. Calif Manag Rev 1995;38:138–162.8 18. Kaiser Family Foundation. Follow the pill: Understanding the US commercial pharmaceu- tical supply chain. Available at: https://www.kff.org/wp-content/uploads/2013/01/ follow-the-pill-understanding-the-u-s-commercial-pharmaceutical-supply-chain-report. pdf March 2005. Accessed March 12, 2020. 19. Walter E, Dragosits A, Said M. Access to pharmaceutical products in six European coun- tries - analysis of different pharmaceutical distribution systems. Farmeconomia (Online) 2012;13:33–41. 20. Andersson F. The distribution of pharmaceuticals in Europe - current and future trends in wholesaling. Health Pol 1994;27:271–292. 21. Taylor D, MrazekM, Mossialos E. Regulating pharmaceutical distribution and retail phar- macy in Europe. In: Mossialos E, Mrazek M, Walley T, eds. Regulating Pharmaceuticals in Europe: Striving for Efficiency, Equity and Quality. Maidenhead: Open University Press; 2004. p. 196–212. 22. Yang L, Huang C, Liu C. Distribution of essential medicines to primary care institutions in Hubei of China: effects of centralized procurement arrangements. BMC Health Serv Res 2017;17:727. 23. Jaberidoost M, Nikfar S, Abdollahiasi A, Dinarvand R. Pharmaceutical supply chain risks: a systematic review. DARU 2013;21:69. 24. Chopra S, Sodhi MS. Managing risk to avoid supply-chain breakdown. MIT Sloan Manag Rev 2004;46(1):53–61. Fall Issue:53-61. 25. GAVI. Outsourcing the Distribution Component of Vaccine and Medicine Supply chains. Available at: https://www.transaid.org/wp-content/uploads/2016/02/GAVI-Outsourcing- Report.pdf 2021.Accessed September 25. 26. McCabe A, Seiter A, Diack A, Herbst CH, Dutta S, Saleh K. Private Sector Pharmaceutical Supply and Distribution Channels in Africa: A Focus on Ghana, Malawi and Mali. Avail- able at: https://openknowledge.worldbank.org/bitstream/handle/10986/13590/ 656010WP00PUBL00PvtSectorPharma0811.pdf?sequence=1&isAllowed=y August 2011. 27. USAID. The Logistics Handbook: A Practical Guide for the Supply Chain Management of Health Commodities. 2nd ed. Arlington, VA: USAID | DELIVER PROJECT. 2011.(First edi- tion 1998). Available at: http://ghsupplychain.org/sites/default/files/2019-07/Logistics %20Handbook.pdf. Accessed June 18, 2018. 28. USAID. Logistics System Assessment Tool (LSAT). Arlington, VA: USAID | DELIVER PRO- JECT. Available at: https://publications.jsi.com/JSIInternet/Inc/Common/_download_ pub.cfm?id=14130&lid=3; Accessed May 8, 2018.