Identification of key drivers for improving inventory management in pharmaceutical supply chains

Abstract

Existing literature on optimizing inventory levels in pharmaceutical supply chains has focused on a limited set of drivers. However, the global supply chain disruptions produced by the Covid-19 pandemic demonstrated the need for a more nuanced picture of the inventory management drivers in this sector to identify profitable inventory configurations while fulfilling demands and safety margins. To address this gap in the literature, this paper identifies key drivers impacting inventory levels and develops a framework for assessing inventory configurations in pharmaceutical supply chains. The framework is tested using a single case study approach. The case study showed that while external and downstream supply chain factors were recognized as being critical to pursuing inventory reduction initiatives, internal factors prevailed when making inventory management decisions. The framework developed in this paper may assist practitioners in identifying the most important factors impacting inventory levels within a specific pharmaceutical supply chain configuration and is in use in the industry today.

Introduction

Pharmaceutical supply chains are characterized by high service levels, volatile and unpredictable demands (especially in emergent markets), and complex and perishable finished-good portfolios [1]. The complex manufacturing processes and the (generally) low ratios of production sites to markets served also increase supply chain complexity [2]. These factors, combined with strict regulations imposed by governments and drug administrations, can make keeping reasonable inventory levels a daunting challenge [3]. The Covid-19 pandemic made supply chain challenges in this industry even more critical, including determining manufacturing and distribution centre locations, managing the risks associated with miscalculations of inventory levels (e.g., running out of stock), dealing with limited transparency in supply chains (making it harder to calculate lead times to ensure correct inventory levels), and overcoming long distance transportation challenges associated with requirements for pharmaceutical products (e.g. being kept at certain temperatures). However, despite the efforts of pharmaceutical companies to minimize disruptions, the prolonged recovery following Covid-19 demonstrates that the disruptive consequences of the pandemic are still being felt by the sector and indeed economies all around the world. Specifically, the fragility of supply chains was already observed before the pandemic, with shortages of medicines nationally and globally, highlighting the need to ensure that inventory levels can be correctly evaluated so that the right medication is available at the right time, in the right amount.

The topic of inventory management has been extensively investigated from diverse perspectives. Most of the research in this field includes the topics of demand planning, inventory control, and inventory planning and scheduling [e.g., 4, 5, 6, 7]. Other scholars have developed quantitative methods such as operations research, advanced mathematical models, and simulation techniques that determine optimal inventory allocations based on a number of factors [e.g., 1, 2, 6, 8, 9, 10, 11, 12]. However, the past literature has mainly focused on optimizing inventory levels in pharmaceutical supply chains by constraining a (limited) set of drivers using mathematical models and simulations [1, 12]. Moreover, previous research into this area has developed models that often use simplifications of current supply chain designs [e.g., 1, 6, 13, 14, 15]. On the other hand, a more multidisciplinary approach to determining inventory configurations and the drivers behind inventory levels is missing in this literature. Such an approach seems particularly useful for practitioners in periods of supply chain disruption, such as during and after the Covid-19 pandemic.

To address the gap described above, this paper combines two key areas that, until now, have been investigated separately in previous studies: finished-goods inventory management and inventory drivers. Thus, the aim of this paper is to develop a framework that supports the identification of the key drivers impacting inventory levels and to determine how these can be efficiently configured. By doing so, the paper aims to support companies in making more accurate decisions when determining inventory configurations.

Literature review

Factors impacting inventory levels

Inventories play a critical role in supply chains, and they are often a salient focus of supply chain managers [16], as they have a clear impact on a firm’s financial performance [17]. Inventory management ensures the right balance between demand and supply, which acts as a buffer against uncertainties [7]. To do so, inventories generally consist of safety stock and cycle stock [6, 7, 18]. Cycle stock is the portion of inventory that a company cycles through to satisfy regular sales orders [7] – it is, therefore, the mechanism that enables companies to exploit economies of scale and order in large lots; therefore, lowering purchase costs [7].

Although the advantages of holding inventory are clear, the disadvantages have been increasingly recognized in the literature [7, 19–22, 25]. Specifically, high inventory levels negatively impact cash flow and warehousing capacity [21]. In this context, there are different drivers of finished goods inventory levels. These are summarized in Table 1, in which “QN” denotes quantitative factors and “QL” denotes qualitative factors to provide an indication of the tangibility of these.

Table 1.

Summary of factors impacting finished-goods inventory levels and classification

	Driver	QN*	QL*	References (limited list)
Production / Internal operation capabilities	Order quantity or order size requirement (or lot size)	X		[7] [23] [24] [10] [18] [26] [27] [28]
	Order/replenishment policy		X	[23] [24] [10] [26] [7]
	Production capacity utilization	X		[29] [26]
	Production flexibility		X	[17] [30] [31] [32]
	Lead time	X		[23] [33] [24] [29] [26] [7]
	Lead time variance	X		[23] [33] [24] [10] [34] [7]
	Production Strategy		X	[17] [23] [24] [10]
	Inventory Space	X		[24] [29]
Product characteristics	Product customization/specialization		X	[32] [24] [17] [35] [29] [7]
	Product complexity		X	[23] [24] [18] [35] [7] [36]
	Shelf life/product durability	X		[23] [10] [26] [8]
Market factors	Average demand per product	X		[23] [10] [26] [7]
	Standard deviation of demand, or demand variability	X		[23] [37] [26] [29] [7] [8]
	Forecast accuracy	X		[23] [24] [10] [38]
	Forecast horizon/demand forecasting frequency	X		[23] [10] [29] [39]
	Service Level	X		[7] [8] [10] [23] [32]
	Sales channel type and type of contract	X		[24]
Financial factors	Material cost	X		[4] [40] [41] [32] [17] [7]
	Product price/unit	X		[40] [5] [32] [17] [18] [26] [7]
	Order cost	X		[40] [5] [7] [41]
	Holding cost	X		[5] [7] [29] [40] [41]
	Stock-out penalty cost	X		[26] [7] [41]
	Scrap or obsolescence costs	X		[40] [23] [41] [26] [7]
	Transportation cost per product	X		[23] [7] [41]
Upstream supply chain characteristics (suppliers)	Delivery performance and reliability		X	[23] [24] [17] [36]
	Delivery frequency from suppliers	X		[24] [18] [36] [7]
	Discounts from Suppliers		X	[36] [7]
	Lead time delivery from suppliers	X		[33] [26] [36] [7]
Downstream supply chain characteristics (customers)	Lead time delivery to customers	X		[33] [26]
	Decoupling point position		X	[17] [42] [43] [44] [45] [28]
	Type of transportation		X	[23] [22]
Miscellaneous 1	Outsourced logistics		X	[23] [24]
	Lean supply chain capabilities		X	[17] [18] [46] [32] [47]
	Management attitude towards inventory control		X	[23] [33] [18] [29]
Miscellaneous 2	Degree of employees training and empowerment		X	[18] [29]
	Information sharing and centralization		X	[17] [48] [37] [18] [49] [50]
	Degree of cooperation among SC player		X	[5] [51] [7] [49]
	Degree of centralization/decentralization		X	[7]
	Regulatory constraint		X	[8] [1]

* QN: Quantitative; QL: Qualitative

Inventory research in the pharmaceutical industry

Table 2 provides a summary of the identified inventory drivers by showing if they impact cycle (C) and/or safety stock (S), as well as the literature that describe these.

Table 2.

Summary of inventory drivers

Driver categories	Drivers	C/S*	Reference
			18	5	26	40 41	7	49	10	32	36	42	23	17	29	24	33	45
Production / Internal operation capabilities	Order quantity or order size	C	X	X	X		X		X				X			X
	Order/replenishment policy	S			X		X		X				X			X
	Production capacity utilization	C + S			X										X
	Production flexibility	C								X				X
	Lead time	S			X		X		X				X		X	X	X
	Lead time variance: σLT	S					X		X				X			X	X
	Production strategy	C + S							X				X	X		X
	Inventory Space	C + S													X	X
Product characteristics	Product customization	S					X			X				X	X	X
	Product complexity	S					X						X			X
	Shelf life/product durability	C + S							X				X
Market factors	Average demand per product	C			X		X		X				X
	Demand variability: σD	S			X		X						X		X
	Forecast accuracy	S					X		X				X			X
	Forecast horizon/ forecasting frequency	S							X				X		X
	Service level	S					X		X	X			X
	Sales channel type and contract	S														X
Financial factors	Material cost	C				X	X			X				X
	Product price/unit	C	X	X	X	X	X			X				X
	Order cost	C		X			X
	Holding cost	C		X
	Stock-out Penalty cost	C			X	X	X
	Scrap or Obsolescence cost	C			X	X	X						X
	Transportation cost	C + S					X						X
Upstream SC characteristics	Delivery performance and reliability	S									X		X	X		X
	Delivery frequency from suppliers	C	X				X				X					X
	Discounts from suppliers	C					X				X
	Lead time delivery from suppliers	S					X				X						X
Downstream SC characteristics	Lead time delivery to customers	S			X												X
	Decoupling point position	C + S		X			X					X		X				X
	Type of transportation	C + S											X
Cooperation and other managerial Factors	Outsourced logistics towards inventory	C + S											X			X
	Lean supply chain capabilities	C + S	X							X				X
	Personal bias	C + S
	Management attitude towards inventory	C + S											X		X
	Degree of employee training &empower.	C + S	X												X
	Information sharing and centralisation	C + S	X					X	X
	Degree of centralization	S					X

* C: Cycle stock impacts; S: Safety stock impacts; C + S: evenly distributed or ambiguous

** PSC: Relevant in pharma SC

Only a small part of the research on inventory and distribution planning directly addresses the pharmaceutical sector – and while some researchers have compared inventory performance across pharmaceutical companies [8, 12, 52], few studies seek to identify the parameters impacting inventory levels in pharmaceutical companies [1]. Another limitation of research with this industry focus is a focus on optimizing inventory levels using a limited set of drivers while omitting other important drivers. As an example, Uthayakumar and Priyan [1] and Kelle et al. [12] focus on near-optimal allocation policies in pharmaceutical supply chains under space constraints but do not consider demand variability, forecast accuracy, or regulatory constraints. Shah [8] addresses this issue by applying a more holistic view of inventory drivers, but limits the focus to intra-organizational parameters, while not considering external factors, such as disruptions caused by pandemics.

The inventory drivers matrix (IDM) framework

Based on the identified inventory drivers, as described in the previous section, this section develops an inventory driver framework, which we name the “Inventory Drivers Matrix” (IDM) framework. The IDM framework is shown in Fig. 1 and subsequently further described.

Fig. 1.

The Inventory Drivers Matrix (IDM) Framework

The framework is bi-dimensional in the sense that it distinguishes between factors according to their relative position in the supply chain (i.e., upstream or downstream SC factors) and at the same time considers whether these factors are extrinsic (inherently driven by external forces) or intrinsic (can be affected and/or modified by the company). Within each category, the areas shadowed in grey (higher position on the Y-axis) highlight the factors that have been considered intrinsic to the company.

The categories of the IDM framework can be described as follows:

• Financial factors: This category mainly includes the costs related to holding an SKU on stock and its value. Sub-factors in this category are the costs of material, holding, scrap, stock-out penalties, transportation, and order processes, as well as the final price per product stored.

• Product characteristics: The category refers to special qualities of the product being manufactured: product customization, product complexity, and shelf-life requirements.

• Upstream SC characteristics: This category entails the drivers related to the availability of the raw materials and activities that take place at the upstream node of the supply chain – hereafter, the sourcing of raw materials and supplier’s performance, which include degree of supplier orientation of the company, lead time delivery from suppliers, and discounts from suppliers, as well as suppliers’ delivery performance, reliability, and frequency.

• Production/Internal operation capabilities: This category includes the drivers related to the manufacturing of the final product, production strategy, and characteristics of the company: production strategy, replenishment policies, production flexibility, order quantity and order size, capacity utilization, production lead time and lead-time variance, and space reserved for inventory.

• Cooperation and other managerial-related factors: This category is concerned with intra-organizational factors, such as the availability of organizational structures and systems that facilitate managing inventory levels, as well as factors related to governmental and agency regulations on the product and its inventory management. Sub-factors in this category are lean supply chain capabilities, personal bias, management attitude towards inventory control, degree of employee training & empowerment, information sharing, aggregation (centralization/ decentralization), degree of cooperation among SC players, outsourced logistics, and regulation requirements.

• Market factors: This category includes the drivers associated with the end customer of the product and the predictability of customer behaviour, hereunder forecast accuracy, forecast frequency, service level commitment, sales channel type, purchase contract type, average demand, and demand volatility.

• Downstream SC characteristics: This category is concerned with proximity to the end customer and contains the following drivers: degree of customer orientation, lead time delivery to customers, decoupling point position, and type of transportation.

The framework presented does not intend to point to an optimal allocation of inventories but rather to encourage supply chain managers to reflect on contextual parameters that can affect inventory levels. It also provides ground rules to be considered when determining the company’s optimal inventory levels. The framework is operationalized in the form of the 10 steps shown in Fig. 2. Although these steps represent a logical sequence of tasks, there may be a need to revisit previous steps according to the results of certain step. In other words, the framework describes an iterative process.

Fig. 2.

The 10 Steps in the IDM Framework

The scores associated with Step 2 (the factors in the IDM framework and any additional factors identified are evaluated from 1 to 5) can obtained through surveys or workshops with relevant stakeholders. In Step 3, the impact score for each factor in the current supply chain setup of the firm (AS-IS assessment) is assigned the following values: (1) if the factor points towards high inventory levels, (0) if neutral impact or ambiguous and (-1) if the factor points towards low inventory levels. The same impact scores are used in Step 4. Step 3 and 4 can be the result of a workshop conducted with supply chain managers or assessments based on interviews with relevant stakeholders. In Step 5, the importance and impact score are multiplied to obtain the “total score” (both in the AS-IS and TO-BE situations). In summary, the scores are:

• Importance score: weight of each of the factors in the IDM framework (1–5).

• Impact score: assessment of each factor’s direction: high (1), neutral (0) or low (-1) impact.

• Global score: Impact score * Importance score.

• Potential score: Importance score * Potential impact score (all factors point towards 1).

Finally, the “total score” is compared with the “potential score” to determine positioning of the company with regards to inventory levels in comparison to the maximum overstock levels. This is described in the following formula, in which “i” is the index factor:

The formula determines the positioning of the company with regard to its inventory levels in comparison to the maximum overstock levels in two scenarios: the current situation (AS-IS) and the attainable optimal situation (TO-BE). The potential impact score considers an overstocking situation, as if all factors point towards increasing inventory levels. The expression includes normalization, so the results range between 0% and 100%.

Research method

To investigate the proposed framework, a single case study was carried out [53]. The company in focus is a multinational pharmaceutical company headquartered in Denmark. The company has several lines of business and markets its products in more than 180 countries. In the present study, only diabetes products are considered, given that the rest of products have a less mature supply chain, and their production follows an on-demand strategy. Diabetes-related drugs are the core business of the case company, which at the time of the study was currently developing processes to improve its inventory management in an international context by analysing inventory configurations in their affiliates – i.e., a goal that corresponds to the research questions presented in this study.

Seven non-recorded and four recorded semi-structured interviews were conducted, where the latter involved a production planner, a senior business analyst, a senior raw materials controller, and a senior S&OP planner. Notes were taken during the non-recorded interviews, while the recorded interviews were transcribed to facilitate later analyses. The interviews had the purpose of retrieving inputs on factors included in the framework (as described in the previous section), as well as to test the concept for scoring, i.e., “1” if the factor pointed towards increasing inventory levels, “0” if the factor was ambiguous, and “-1” if the factor pointed towards decreasing inventories. Furthermore, the interviewees were asked to explain the reasons for their ratings. Based on the findings from the interviews, a survey was conducted to further validate these results. This asked the respondents to provide similar factor scores, as used in the interviews. A double assessment was required: AS-IS (current) and TO-BE (optimal) situations.

Finally, a survey among employees in the company was carried out. Table 3 describes the identification of respondents. As seen, it was decided to exclude employees from the target sample had not been involved had not been involved in inventory management decisions and had spent less than a year in their current role in the company, which was done to ensure the validity of responses. This produced a target sample of 38, of which 16 responses were acquired implying a response rate of 43.2%.

Table 3.

Population Sample of the Survey Questionnaire

Survey respondents	Group size
Employees within the S&OP and SNP planning, Production Planning, Supply Chain Design and Analytics departments	102
Persons within the group above that had not been involved in inventory management decisions are made	87
Persons within the group above that have spent at least one year in their current role within the case study company	38
Target survey study sample	38
Number of respondents to the survey and response rate	16 (43.2%)

Test of the IDM framework: application in the case company

The study showed that the inventory configurations in the company were unstable over time, especially in clusters that present high safety stock levels (while low/high strategic stock level). The number of stock-out items was reduced during the three-year period, whereas the occurrences of above the maximum stock level target tended to increase for all the clusters. Furthermore, there seemed to be no correlation between having high safety stock and achieving higher service levels. The analysis also identified the most important factors behind each of the inventory configurations, as the demand and production throughput are the main inventory drivers. Factors belonging to the “internal operation capabilities” and “product characteristics” categories of the IDM framework were also important explanatory factors behind the company’s overall inventory levels. “Regulatory requirements” were key drivers behind inventory configurations with high strategic stock levels.

The combination of the importance and impact scores has determined the stock positioning of the company with regards to inventory levels in comparison to the maximum overstock levels in two scenarios: the current situation of the company (AS-IS), and the desired/optimal situation (TO-BE). Table 4 shows the results from the survey, which were validated through interviews.

Table 4.

Survey Questionnaire Results

It can be seen that the total/global score for the AS-IS situation is 71 and for the TO-BE situation it is 29,5. The potential score for the AS-IS and TO-BE situation is then 125,5. The results revealed that the company had a tendency to overstock finished goods in its affiliates’ warehouses (the AS-IS situation is at 57% of the maximum inventory levels attainable). This underpins the need for an assessment of the factors that can reduce the gap between the current and desired situation of the company and reduce the overall inventory levels to an optimal configuration. The use of the IDM framework for identifying the gap between AS-IS and TO-BE situations is illustrated in Fig. 3.

Fig. 3.

Use of the IDM Framework

The study showed that to a large extent, the significant gap between the AS-IS and TO-BE stocking positioning was determined by nine factors across the categories of the IDM framework: replenishment policies, internal and supplier lead times, service levels, forecast accuracy, information sharing, production strategy, personal bias, production capacity utilization, and centralization/decentralization. The study also suggests that if a company seeks to reduce the inventories held at affiliates, it should pay significant attention to external factors and downstream supply chain capabilities.

The interviewees from the case company were positive regarding the usefulness of the framework. This is further highlighted by the fact that around half a year after the study was carried out, a meeting with key stakeholders at the case company revealed they were still using the IDM framework when making inventory decisions.

Conclusion

In recent years, inventory reductions have been a key objective of pharmaceutical companies, especially within cost optimisation initiatives. However, a one-way strategy towards zero-inventory is often infeasible in practice due to the strategic nature and importance of the products being commercialized. Furthermore, the global supply chain disruptions produced by the Covid-19 pandemic crystalised the need for a more nuanced approach to inventory management in this sector. In other words, the most profitable inventory configurations must be viewed from a more holistic perspective than focusing on inventory costs. To address this issue, inventory management research has traditionally focused on studying the impact of tangible, quantitative factors on the inventory levels of a company, while more abstract parameters are often not addressed when providing methods for identifying optimal inventory levels.

To provide the pharmaceutical industry with a more holistic approach to identify optimal inventory levels, the present study developed a framework named the Inventory Drivers Matrix (IDM) framework. The framework has a bi-dimensional structure (internal vs. external and upstream vs. downstream SC characteristics) that covers the full supply chain spectrum at several abstraction levels. The IDM framework represents an extension of the models in the existing literature, which inspired the framework – in particular, the frameworks by Hoppenheit and Günthner [24] and Ekstrand and Karlsfre [23].

To test the IDM framework, a case study was carried out at a multinational pharmaceutical company, which demonstrated its usefulness. Specifically, the study showed that the IDM framework supports the identification of the most important factors impacting inventory levels within a pharmaceutical supply chain configuration, as well as determining the positioning of the company with regard to its inventory levels in comparison to the maximum overstock levels. The framework allows supply chain managers to reflect on potential factors that might not have been considered before in any of the processes established within the firm, further identifying key areas in which efforts can be made to pursue inventory reduction initiatives.

Future research should focus on (1) further refining the model to separate the factors in different abstraction levels, (2) testing the framework in other pharmaceutical companies to further develop it, and (3) adapting the framework to be used in other industries. From a more overall perspective, it would be interesting to link the IDM framework to CO2 footprint reductions, given the high environmental impact of the pharmaceutical industry, which for some part is related to inventory movements [54].

Declarations

Conflict of interest

The authors declare that they have no conflict of interest.