Procurement's role in resolving demand–supply imbalances: an information processing theory perspective

S&OP from the information processing theory perspective

The tasks related to S&OP typically suffer from significant uncertainty because it can be said that the inputs, processes and outputs of such a work system lack predictability (Griffin et al., 2007). Indeed, customer demand is typically irregular, production lead times and yields may vary, and suppliers' operations and deliveries may deviate from expectations in an unpredictable operating environment. S&OP may therefore be framed as an information processing theory (IPT) problem (Galbraith, 1974; Tushman and Nadler, 1978) because “… the greater the task uncertainty, the greater the amount of information that must be processed among decision makers during task execution in order to achieve a given level of performance” (Galbraith, 1974). Usefully, IPT suggests that organisational effectiveness may be achieved by accomplishing a fit between the needs or requirements for information processing and the capacity for information processing in organisations. This places an emphasis on the drivers of the information processing need (uncertainty and its causes), as well as the mechanisms through which the capacity for information processing comes about (Tushman and Nadler, 1978), or more accurately, the fit between need and capacity can be managed.

Regarding the underlying dimensions and causes of uncertainty, the literature, affiliated with IPT suggests the following. The seminal work of Duncan (1972) distinguishes between complexity and dynamism, and these prominent high-level contingency factors may plague the internal task execution or the external task environment. In the supply chain context, complexity-driven uncertainty may arise due to the nature of a task (scale, variety, novelty and product characteristics), supply source (location, length of relationship and process characteristics) and supply chain (horizontal, vertical and spatial complexity; Bode and Wagner, 2015; Busse et al., 2017). In addition to this kind of detail complexity, both downstream and upstream dynamism, or dynamic complexity (e.g. stochastic demand levels and supplier lead times), have been shown to have a negative effect on manufacturing performance, suggesting uncertainty and a need for information processing (Bozarth et al., 2009). Essentially, these contingency factors (internal/external complexity and internal/external dynamism) serve as a means of characterising the contexts in which procurement seeks to contribute to S&OP from the point of view of information processing.

In terms of managing the fit between information-processing needs and capacity, the literature suggests two principal methods. First, the need may be reduced by managing the environment, creating slack resources and creating self-contained tasks through isolation (Galbraith, 1974). In the supply chain disruption-management context, Bode et al. (2011, p. 836; see also Meznar and Nigh, 1995) suggest buffering as a general means of reducing information-processing needs in, for example, supplier relationships because a “firm can build up slack resources to act as ‘shock absorbers’, such as larger inventories, flexible production processes, redundant suppliers, and product designs that are not dependent on a specific supplier”. Drawing on Galbraith (1974), the buffering construct is further elaborated here and it is suggested that, in the procurement context, the reduction of information-processing needs related to the S&OP task may be achieved through redundancy (akin to creation of slack resources) and demand change (akin to managing the environment). Procurement may achieve buffering by redundancy via, for example, securing flexibility for a production capacity increase in the supply base (establishing second sources or maintaining options for extra capacity) and maintaining safety stocks internally or at suppliers (cf. Azadegan et al., 2021). In addition to figuratively throwing money at the problem, procurement may seek to implement fundamental changes in demand or address the root causes for the need to process information. This may be achieved by reducing dependence on the supplied item (e.g. reduced demand), changing item specifications in order to allow sourcing from more favourable (predictable) markets (design-for-sourcing; Schuh et al., 2009) and levelling manufacturing requirements collaboratively with production and sales (Olhager et al., 2001).

Second, the fit between information-processing needs and capacity may be managed by increasing capacity (Galbraith, 1974; Daft and Lengel, 1986). For example, in the inter-organisational-relationship context, the multiplicity of information channels, the frequency of information exchange, joint action and commitment, as well as information system linkages may increase such capacity (Bensaou and Venkatraman, 1995). In a supply chain disruption management context, Bode et al. (2011, p. 836; see also Meznar and Nigh, 1995) suggest bridging to be “associated with investments in collaborative structures or initiatives such as joint risk management systems, or with scanning approaches such as monitoring or intensifying information exchanges”. In this vein the current paper suggests that, in the procurement context, an increase in information-processing capacity related to the S&OP task may be achieved by bridging the gap between the buyer and the supplier by the means of collaboration and monitoring. In more detail, procurement may engage in collaboration with suppliers in terms of capacity management, production and delivery scheduling, risk management and information sharing (Kaipia et al., 2017; Cao and Zhang, 2011) in a two-way bilateral fashion. In contrast, procurement may also seek to reduce uncertainty by increasing the capacity to monitor suppliers' delivery performance, production scheduling and capacity utilisation in a more unilateral manner (Maestrini et al., 2018), for example, using information systems means in the S&OP (Schlegel et al., 2021). Essentially, the above discussed mechanisms for managing the need-capacity fit help characterise the methods with which procurement seeks to contribute to S&OP imbalance resolution from the point of view of information processing.

Hypotheses

The previous section framed S&OP from the IPT perspective and identified several relevant constructs for our research. First, it is suggested that the IPT constructs of buffering (via redundancy and demand change) and bridging (via collaboration and monitoring) serve as the independent constructs enabling procurement to contribute to favourable S&OP outcomes. Second, it is suggested that these outcomes, as the dependent constructs, may be defined in a dichotomous fashion as related to either cost or delivery performance (Srinivasan and Swink, 2018). In more detail, the nature of procurement's contribution is the ability to resolve demand–supply imbalances, when they occur, efficiently, meaning without cost escalation, or effectively, meaning without disruptions in deliveries. For example, if, in the S&OP process, demand is forecasted to increase to an unexpected level, procurement is called to establish balance by perhaps confirming additional supplier capacity for the plan, optimally doing so without budget overruns or jeopardizing the delivery and production plans. Third, all of the above actions take place in various procurement contexts, such as were defined above as the contingency factors. Additionally, it is useful to consider the co-management of S&OP as procurement contexts (Wagner and Eggert, 2016). In the following, a set of hypotheses is developed reflecting this general set up.

Procurement may increase its ability to resolve demand–supply imbalances by engaging in buffering by demand change. By fundamentally changing the nature of demand, procurement addresses the root causes of the occurring imbalances by reducing demand for the supplied item, reducing demand volatility (Olhager et al., 2001) and seeking to support design-for-sourcing (Schuh et al., 2009). Essentially, buffering by demand change thus enables the minimisation of demand–supply imbalances, both in terms of frequency and degree and, furthermore, allows procurement to operate with a greater degree of freedom and relational power in supply markets. For example, the changed design and specification of the requirement may allow the buying of standardized, instead of supplier proprietary components, with low switching costs and several alternative suppliers in the market (Cox, 2015). Buffering by demand change therefore implies the resolution of demand–supply imbalances cost-efficiently due to occurrence minimisation and an increase in relational power of the buyer, as well as fewer delivery delays or disruptions due to occurrence minimisation and a greater degree of item interchangeability and flexibility. The following hypotheses are stated:

However, the beneficial effect of buffering by demand change may vary depending on the context. Sourcing managers often face internal complexity within their area of responsibility, such as a commodity group or purchasing category, involving multiple items, perhaps even in the hundreds. The category management task may therefore be characterised as complex (detail) due to the number of decision-making factors (Duncan, 1972), the number of stock-keeping units that must be managed (Bozarth et al., 2009) and the number of subtasks not easily factored into independent parts. The resulting information diversity and cognitive load (Campbell, 1988) implies reduced efficiency and effectiveness in imbalance resolution, and therefore, efforts toward demand change, involving for example standardisation and reduction of items, may be expected to have particularly beneficial effects in such complex contexts. Furthermore, the dynamic variation of demand serves as a source of significant dynamic complexity, causing stock-outs, obsolescence, the bullwhip effect (Bozarth et al., 2009; Lee et al., 1997) and general uncertainty. By standardising and simplifying categories, as well as by levelling dynamic customer demands or production volumes, uncertainty and information processing need is reduced with buffering by demand change, and the imbalances may be resolved more efficiently and effectively. The following hypotheses are posited:

Procurement may increase its ability to resolve demand–supply imbalances by engaging in buffering by redundancy. By securing flexibility in the supply base with second sources and options for extra production capacity at suppliers or by maintaining higher safety stocks internally or at suppliers (cf. Azadegan et al., 2021), procurement may essentially insulate against uncertainty and reduce the information-processing need for S&OP and the successful realisation of the plan. While both the use of backup suppliers and flexibility through options are potent buffering practices, it has been shown in a simulation study that the use of second sources may be a preferable practice in comparison to capacity flexibility in terms of both cost reduction and service level improvement (Kamalahmadi et al., 2022). However, the net effect of buffering for demand–supply imbalance resolution by the means of redundancy may also be paradoxical, or have the opposite of the intended beneficial effects, because the fragmentation of spending across multiple suppliers implies the foregoing of volume discounts (due to smaller volumes allocated for each supplier), and also as it may involve a heavy reliance on and investment in safety stocks. While safety stocks may allow for the resolution of demand–supply imbalances without disruptions and delays in deliveries (Chung et al., 2018), the cost of buffering with inventory may be significant (Fredriksson and Jonsson, 2009; Raman and Kim, 2002). The following hypotheses are stated:

Context may play a significant role in terms of how these generally hypothesised associations play out. Internal category complexity, i.e. the management of large numbers of stock-keeping units, may amplify the negative relationship with a cost-efficient resolution of imbalances because product variety is bound to increase the overall inventory for market mediation (Randall and Ulrich, 2001). Essentially, non-standardised categories with many SKU variants imply higher safety stocks and therefore aggravate the cost implications of resolution through redundancy. However, a mature S&OP process in which procurement is internally integrated and co-managed (Wagner and Eggert, 2016) enables the proactive planning of resources and better inventory management, and therefore, such an environment may alleviate the negative relationship with the cost-efficient resolution of imbalances (Bower, 2006) because less inventory is needed with better resource utilisation. Regarding the ability to resolve imbalances without disruptions or delays in deliveries, buffering by redundancy may be a particularly potent remedy in unpredictable contexts, such as those characterised by dynamic variation in demand and also of supply (Bozarth et al., 2009, p. 81). By isolating procurement from dynamic complexity on both the demand and supply sides by the means of inventory buffers, uncertainty and information-processing needs may be reduced, and delivery performance may be effectively secured. The following hypotheses are posited:

Next, it is examined how procurement may resolve demand–supply imbalances by means of bridging by collaboration. The evidence suggests that buyer–supplier collaboration plays a significant role in the buyer's efforts to draw on the supplier's flexibility and responsiveness capabilities, as well as achieving responsiveness in its own operations (Squire et al., 2009). Warning capability with collaborative information sharing across the dyad may reduce the effects of supply disruptions (Craighead et al., 2007), both in terms of service level and cost because, with early warning, the length of business interruption can be minimised (Norrman and Wieland, 2020). Indeed, the essential mechanism at play draws on the bilateral nature of the collaborative relationship, involving for example direct contact, use of boundary spanning integrators (e.g. on-site supplier development engineers; Hartley and Choi, 1996) and group meetings with several functions participating across the buyer–supplier interface (Cooper et al., 1997). With broader bandwidth (Aral and Van Alstyne, 2011), such information processing mechanisms facilitate “information richness” for clarification of issues and learning in a timely manner (Daft and Lengel, 1986, p. 560), for example about changes and threats regarding supplier's capability to supply and trends and opportunities regarding demand. Fundamentally, bridging by collaboration allows the minimisation of the imbalances, both in terms of degree and frequency, enabling service levels, and furthermore, supports the crucial initiation of early and proactive measures for imbalance resolution, including, for example, the cost-efficient and early reservation of suppliers' production capacity (late-reservation more costly) or the use of normal transport modes instead of faster and more expensive modes for rush orders. The following hypotheses are posited:

Considering the contexts of these associations, the following may be noted. Collaboration, in terms of information sharing in a two-way manner, is likely to allow procurement to better handle variability and unpredictable patterns in both demand and supply, as well as the consequent bullwhip and ripple effects (Lee et al., 1997; Dolgui and Ivanov, 2021). More specific evidence suggests that there are broad beneficial effects on the part of involving suppliers in even the higher levels of decision-making and planning, particularly in contexts with high demand uncertainty (Ambulkar et al., 2023). Indeed, suppliers benefit from customer forecasts and planned order information distinctly when demand is non-stationary (Jonsson and Mattsson, 2013). Regarding the supply side, Wong et al. (2011) show that supplier integration under uncertainty is associated with higher delivery performance; however, this association is absent for production cost. Nevertheless, the cost-efficient resolution of demand–supply imbalances, especially in uncertain conditions, may be expected to be more likely in the presence of collaboration due to early warning and response (Craighead et al., 2007). Furthermore, mature S&OP processes with advanced elements for external information sharing and co-management engage the supplier in collaborative planning, facilitate the supplier's access to forecast or point-of-sale data and, thus, endow it with an improved ability to respond to customer requests (Kaipia et al., 2017; Wagner and Eggert, 2016). Furthermore, the external co-management of S&OP provides an information-processing infrastructure for leveraging knowledge gained from external sources (Schoenherr and Swink, 2012). The achieved supplier responsiveness and the customer infrastructure for leveraging knowledge support the cost-efficient (with fewer rush orders and earlier capacity reservation) and timely (with early warnings) resolution of demand–supply imbalances on the customer side. The following propositions are stated:

Finally, procurement resolving demand–supply imbalances by the means of bridging by monitoring is considered. The literature on the topic suggests inconclusive results regarding the benefits of supplier monitoring. On one hand, it has been suggested that supplier monitoring positively affects the suppliers' operational performance (Maestrini et al., 2018; cf. also Croom et al., 2018), while on the other hand, there are indications that monitoring practices have little effect on the performance of the supplier (Akamp and Müller, 2013; Cousins et al., 2008) or of the buying organisation, particularly in a context of an arms–length relationship, i.e. when the other side of the coin, collaboration, is missing (e.g. Ittner et al., 1999). In this vein, Joshi (2009, p. 145) reveals the nuances involved in monitoring and controlling suppliers: the buyer's attention should be focused “on the supplier’s underlying capabilities rather than solely on the tangible and concrete activities and outcomes”. Joshi also acknowledges that these “underlying capabilities are difficult to access and may be even more difficult to discern” without integration and socializing (cf. Akamp and Müller, 2013; Cousins et al., 2008). In conclusion, it may be postulated that as the effects of supplier monitoring may be limited (Shafiq et al., 2022), buying firms should emphasise collaborative assessment over potentially ineffective hands-off monitoring to ensure the full transparency of supplier operations in order to assess the supplier's true capabilities and foster continuous improvement.

When one transitions from the normal steady-state conditions to our research context, i.e. whether procurement is better able to efficiently and effectively resolve unexpected demand–supply imbalances (with time constraints) when relying heavily on bridging by monitoring, it is possible to hypothesise even for a paradoxical outcome. In contrast to collaboration strategy in supplier relationships, in which there are opportunities for capability control (Joshi, 2009) and equivocality reduction through rich information (Daft and Lengel, 1986), allowing an early understanding of potential supply capability issues, the monitoring strategy in supplier relationships may have to rely mostly on more superficial control measures, as well as “explicit questions” with limited potential for deep insights and the early detection of supply issues (Daft and Lengel, 1986). In essence, it may not be simply possible to fully bridge the asymmetric information between the buyer and the supplier with monitoring (Boström, 2015). Therefore, the resolution of demand–supply imbalances, when they occur, is likely to be more reactive in nature, with fewer options on the table for cost-efficient outcomes. This is because with time-constrained reactive rush orders, transportation modes may need to be switched to faster and more expensive alternatives and capacity reserved in short notice and therefore at higher rates, suggesting inefficiency of imbalance resolution in terms of cost. In contrast, the monitoring strategy may be beneficial in terms of effectively safeguarding deliveries due to the slight edge in warning capability (Craighead et al., 2007). The on-time delivery of supplies may be secured even on short notice, but at high cost.

It is perhaps easy to accept that supplier monitoring has its limits, or indeed, diminishing returns (Shafiq et al., 2022) also in the time-constrained imbalance resolution context. However, instead of assuming that more intensive monitoring contributes at least somehow to cost efficiency, there may also be grounds for hypothesizing a paradoxical association, i.e. the more one emphasizes the monitoring strategy, the lower the cost efficiency of imbalance resolution. This hypothesis may be based on two logics. First, “looking harder” may not resolve the problem of asymmetric information (Boström, 2015): there are simply limits to what you can learn by examining supplier KPIs, and thus one may not be able to expand the narrow time window in which resolution must be achieved, and so there are only few available and costly options. Earlier warning and mitigation cannot be achieved and therefore the preferred long lead time resolution options, such as proactive supplier capacity risk mitigation and onboarding a new supplier, are off the table. Second, similarly to redundancy with safety stocks, high levels of monitoring “are likely to yield exponentially higher costs to the buying firm” (Shafiq et al., 2022, p. 690), as this implies for example ever higher levels of inspection and auditing activity, investment into data sharing systems and portals, and indeed diseconomies of scale, due to intensive monitoring related non-value-added work (Liu et al., 2009). In summary, while the cost of monitoring increases, the already limited time window for response does not expand for less costly resolution options, suggesting a negative association of monitoring with cost-efficiency of imbalance resolution. We propose that this short-term and reactive nature of bridging by monitoring has an important positive effect on the delivery performance in the demand–supply imbalance resolution, but that it does not have a positive effect on cost-efficiency. Monitoring directly increases the short-term operational costs and will most likely indirectly lead to reducing some long-term demand–supply imbalance-related costs. It is, however, hard to estimate this potential long-term cost effect. Therefore, we suggest a negative association of monitoring with cost-efficiency of imbalance resolution.

The following hypotheses are posited:

Again, context matters and may indeed affect the hypothesised relationships. When supply dynamism, in the form of, for example, unreliable lead times (Bozarth et al., 2009), is high, monitoring may offer beneficial effects through timelier reactions and thus alleviate the problem in terms of delivery performance. In contrast, such a dynamic context is likely to further aggravate the inadequacy of monitoring for cost-efficient resolutions because larger and more frequent imbalances imply higher costs, given an increased number of reactive and late efforts for capacity reservation, for example. Furthermore, supply base complexity within a synergistic procurement category may imply the existence of several suppliers per item, suggesting diversification as a risk management strategy (Chod et al., 2019) and method of securing deliveries. However, paradoxically, it has been shown that increases in horizontal complexity, with the addition of second sources, have the most significant negative effect on the frequency of supply disruptions because, with more suppliers, it is more likely that something will go wrong; the population is difficult to monitor and map beyond the first tier (Bode and Wagner, 2015). Increased and better resourced monitoring efforts may therefore work particularly well in such contexts and increase the prospects for demand–supply balance resolution in terms of maintaining delivery performance. In this vein, the maturity of S&OP in terms of the external co-management of procurement (Wagner and Eggert, 2016), focused on market information provisioning is likely to have similar effects (cf. Wagner et al., 2014) due to its synergy with monitoring. The following hypotheses are proposed:

The above development of hypotheses has identified a group of contextual variables which may, for example, amplify or reduce the relationships between bridging and/or buffering and demand–supply imbalance resolution. However, these contextual variables and their direct relationships with cost-efficient and disruption-free demand–supply imbalance resolution are interesting in their own right. For the sake of technically testing these direct relationships, a set of hypotheses are stated that fundamentally draw on the above discussion of the contexts involved. In a nutshell, complexity and dynamism, both internally and externally, make it difficult to resolve imbalances, whereas the maturity of S&OP in terms of both the internal and external co-management of procurement (Wagner and Eggert, 2016) supports such efforts. A notable exception to this pattern is the redundancy-inducing effect of supply base complexity, which supports the disruption-free resolution of imbalances. The following hypotheses are posited:

Panel A of Figure 1 provides an overview of the research model, and Panel B summarizes the detailed research hypotheses.

Measurement

The measurement items for buffering and bridging are based on the work of Bode et al. (2011), who identified buffering and bridging as alternative coping strategies for supply chain disruptions. Originally representing generic responses to supply chain disruptions, buffering and bridging were adapted to the procurement context. Viewed within the IPT, buffering reduces information-processing needs by establishing safeguards against risk and uncertainty in the environment (Bode et al., 2011; Manhart et al., 2020). Buffering by demand change (BufDC) attempts to address the root cause of supply-demand imbalances by managing the environment, while buffering by redundancy (BufR) aims at reducing exposure to imbalances through the creation of slack resources. Bridging strategies, in turn, attempt to create linkages with the environment through boundary-spanning activities to increase information-processing capacity (Bode et al., 2011; Manhart et al., 2020). Bridging by collaboration (BridgC) increases the information-processing capacity between the buyer and the supplier by means of collaboration, and bridging by monitoring (BridgM) aims to achieve the same via supplier monitoring.

The outcome side follows Srinivasan and Swink (2018), who examined the relationship between visibility and operational performance improvement. Following their operationalisation reflecting cost and delivery performance, the ability to resolve demand–supply imbalances from the procurement perspective was divided into two constructs: cost efficiency (Cost) and delivery performance (Delivery) in demand–supply imbalance resolution. Furthermore, the researchers used a group of contextual variables, which may amplify or reduce the relationships bridging and/or buffering and demand–supply imbalance resolution. Three single items were used: category complexity (CC) was measured as the approximate number of active parts (Bozarth et al., 2009), demand dynamism (DD) as the stability of manufacturing plans (Bozarth et al., 2009) and supply base complexity (SBC) as the approximate number of first-tier suppliers (Bode and Wagner, 2015). Single-item measures are acceptable if the object of the construct is “concrete singular” and the attribute of the construct is concrete and easily and uniformly imagined (Bergqvist and Rossiter, 2007, p. 176). In this study, single items are deemed unambiguous and concrete. For supply dynamism (SD), the respondents were asked to indicate whether the delivery performance of their suppliers varied a great deal and whether supplier lead times were too long (Bozarth et al., 2009). Procurement's involvement in S&OP was measured using two constructs inspired by the co-management concept of Wagner and Eggert (2016) and S&OP frameworks (Grimson and Pyke, 2007; Thomé et al., 2012a; Tuomikangas and Kaipia, 2014; Wagner et al., 2014; Noroozi and Wikner, 2017). The first (with four items) is called internal co-management of procurement in S&OP (Int-CoMgmt). It adopts an internal co-management perspective on procurement and measures the maturity of the fundamental characteristics and design parameters of S&OP. The second (with two items) is called external co-management of procurement in S&OP (Ext-CoMgmt). It adopts an external co-management perspective and measures the extent of supply market information in the S&OP.

The operationalisation of all constructs is shown in column A of the Appendix. Means and standard deviations for each item in the sample are shown in columns B and C. All constructs had the areas of supply management responsibility (categories of items) as the unit of analysis. Apart from some of the above-described single items, the measures used a seven-point Likert scale ranging from (1) completely disagree to (7) completely agree. Turnover was also included as a four-category control variable: (1) 0–2 million, (2) 2.1–10 million, (3) 10.1–50 million and (4) over 50 million EUR. Furthermore, the industry sectors were divided into three broader categories: the process industry; the assembly industry and trading, services and other. To control for potential differences between industry sectors, the two first categories were used as dummy variables in the regression analysis.

Data collection

To promote content validity, the survey instrument was first tested with three industry experts, each with more than 10 years of experience in procurement. Paying particular attention to the newly developed measurement scales, the industry experts were asked to critically review each construct and item, and they were also able to suggest additional or redundant items. As a result, improvements were made to the questionnaire concerning wording, clarity and practical relevance.

Data were collected in Finland and Sweden. Finland and Sweden are both Nordic countries with high levels of political stability, competitiveness, education and technology. Moreover, supplier–customer relationships are similar, for example, in terms of collaboration (Lindblom et al., 2009). The business culture and business structure in both countries are very similar, and the business environments are well developed and stable. Results may thus be cautiously generalisable across other developed economies and contexts. In Finland, the online questionnaire was launched in June 2021 and in Sweden in November 2021. In Finland, the invitations to participate were sent to (1) experienced procurement professionals in the authors' professional networks (N = 72, response rate 76.4%), (2) a sample of procurement personnel with contact details (email addresses) on a platform of a commercial provider of company contact details (N = 104, response rate 13.5%) and (3) as part of a newsletter for the procurement forum of LOGY (Finnish Association of Purchasing and Logistics). In Sweden, the invitation was sent to (1) experienced procurement professionals in the authors' professional networks (N = 44, response rate 56.8%), (2) a sample of PLAN (a Swedish logistics association) members with a position in purchasing or supply chain management (N = 235, response rate 14.9%) and (3) as a part of a newsletter for Silf (Swedish Association of Purchasing and Logistics). The respondents could choose whether they wished to receive a report and an invitation to a webinar in which the results would be discussed. The majority of respondents provided their contact information. Given that the quality of the report and webinar depend on the input of the respondents, it can be assumed that the respondents took the questionnaire seriously. The duration required to fill in the questionnaire was also recorded. No abnormally short response times were detected, with the median duration being 11 min.

The use of single respondents was justified because the unit of analysis was each respondent's area of supply management responsibility, such as the category of spend (Krause et al., 2018). There informants were procurement professionals who are knowledgeable about their own micro-level procurement practices and their ability to resolve demand–supply imbalances from the procurement perspective (Montabon et al., 2018). The present research question focused on the respondents' own domain of management in one functional area of a firm, and monadic constructs were employed. Furthermore, the relationship between the respondents' skills and expertise and the research constructs were aligned and relevant. Thus, the single-source research design is less likely to suffer from respondent bias; in other words, the respondents are expected to occupy a role that makes them knowledgeable about the issues that are investigated (Flynn et al., 2018).

After removing incomplete responses, the final number of responses was 150. The sample covered a variety of industries, with the majority representing manufacturing (Table 1). Respondents show sufficiently high levels of procurement experience. A priori and post-hoc power analyses were performed using the G*Power software (Faul et al., 2009). With the a priori test with a medium effect size f² = 0.15, α = 0.05 and power = 0.80 (Cohen, 1988), the minimum sample size required for testing the full model was 45. In the post-hoc test, the power obtained for our sample size of 150 was 0.999, indicating that the sample size is sufficient.

To reduce the likelihood of consistency motive bias, the independent and dependent constructs were separated and placed in different phases of the survey. To avoid social desirability bias, the respondents were assured of their confidentiality and anonymity and could either complete the survey anonymously or reveal their email address to receive an invitation to a practitioner webinar in which the results were discussed. The comparison of early and late respondents across theoretical constructs and demographic variables with an independent-samples t-test (Armstrong and Overton, 1977) showed no significant differences (p-values between 0.167 and 0.837), indicating that a significant nonresponse bias did not influence the results (Anderson and Gerbing, 1988).

Common method variance

Collecting data on dependent and independent variables via a single respondent may raise concerns regarding common method variance (CMV). As a procedural remedy, independent and dependent variables were places in different sections of the questionnaire (Podsakoff et al., 2003). Years of professional experience was used as a theoretically unrelated marker variable to test for CMV. A modified Lindell–Whitney test was performed in which the mean correlation between the marker variable and other variables (0.058) was used to adjust the correlation coefficients and their significance (Lindell and Whitney, 2001). The modified approach was chosen so as to reduce the risk that the smallest possible correlation had occurred by chance. All significant coefficients remained significant after the adjustment, which suggests that CMV is not likely to substantially reduce the validity of the results (Table 2).

It is also reasonable to assume that different capabilities in terms of demand–supply imbalance resolution may affect information-processing needs and capacity. To test the extent to which the results could be affected by reverse causality, it was tested whether cost efficiency and delivery performance affect buffering by demand change, buffering by redundancy, bridging by collaboration or bridging by monitoring. The results suggest a lack of significant associations for other variables apart from the relationship between delivery performance and buffering by redundancy. To further assess the potential endogeneity of the exogenous variables in the research model, a Durbin–Wu–Hausman test (Lu et al., 2018) with the work experience of the respondent as an instrumental variable was performed. The parameter estimates for the residuals were not significant, which supports the assumption of exogeneity. Thus, although reverse causality cannot be fully excluded, it does not seem to drive the results. Finally, Little's MCAR test was conducted. Based on the results, the missing values in the data can be assumed to be missing completely at random (χ² = 346.495 df = 325, p = 0.095).

Measurement model

Confirmatory factor analysis (CFA) was performed to test the psychometric properties of the multi-item scales of buffering by demand change (bufDC), buffering by redundancy (bufR), bridging by collaboration (bridgC), bridging by monitoring (bridgM), cost performance (Cost), delivery performance and the external (Ext-CoMgmt) and internal (Int-CoMgmt) co-management of procurement in S&OP. The results (Table 3) suggest that the measurement model fits the data adequately (Χ²/df = 1.543, CFI = 0.911, IFI = 0.948, TLI = 0.906, RMSEA = 0.061, p-value = 0.050). All items loaded on their respective constructs. The standardised factor loadings were significant and ranged from 0.653 to 0.927, mostly exceeding the recommended threshold of 0.7 (Hair et al., 2010). Three items were slightly below this value and thus retained. All the constructs demonstrated acceptable reliability, consistency and convergent validity (average variance extracted, AVE >0.50; alpha >0.70, composite reliability >0.70) (Fornell and Larcker, 1981; Garver and Mentzer, 1999).

Two tests were performed to evaluate discriminant validity. Firstly, the square root of AVE was compared with measurement error-adjusted inter-construct correlations between each construct (Voorhees et al., 2016). The square root of AVE (displayed on the diagonal of Table 4) was greater than the zero-order correlations with other constructs (lower part of Table 4), thereby implying discriminant validity (Fornell and Larcker, 1981).

The heterotrait-monotrait (HTMT) test has been represented as a superior method for assessing discriminant validity as compared to the traditional constrained Phi approach (Henseler et al., 2015). Recently, Roemer et al. (2021) introduced the HTMT2 as an improved version of the HTMT. The HTMT2 test ratios are presented above the diagonal in Table 4. All of them are below 0.85 (Henseler et al., 2015), which provides evidence of discriminant validity.

To ensure measurement equivalence across data collected from Finland and Sweden, three steps outlined by Wiengarten and Pagell (2012) were used: calibration, translation and metric equivalence. The measurement units were on a seven-point Likert scale, which does not require calibration across countries. Their associated explanations, ranging from completely disagree to completely agree, are universally understood. For translation equivalence, the questionnaire was simultaneously developed in Finnish and English. Versions were compared and refined when this was deemed necessary.

One-way ANOVA was performed on key constructs to test for potential differences between Finland and Sweden. There were no statistically significant results (p < 0.01) for other variables, except for delivery performance, which had a statistically lower mean value in the Swedish sample. To assess metric equivalence, Cronbach's alpha values were calculated separately for each construct in the Finnish and Swedish sample. Consistent scoring across countries was supported because the variance of the alpha values was below the threshold of 0.10 (Wiengarten and Pagell, 2012).

Hypothesis tests

Table 5 shows the results of the hierarchical regression models. The assumptions of the regression analysis were first tested. The absolute values for univariate skewness and kurtosis were below the recommended thresholds of 2 and 7, respectively (Curran et al., 1996). The linearity and equality of variances were supported by plotting standardised residuals against the standardised predicted values (Hair et al., 2010). To avoid multicollinearity, variables were mean-centred before being entered into the regression model. Variance inflation factors (VIFs) (Table 4) are below the commonly accepted threshold of 5, apart from item MATU1. The Breusch–Pagan test detected the existence of heteroscedasticity, and therefore, robust error terms were used.

There are eight models: cost-efficiency is the dependent variable in Models 1–4, and delivery performance is the dependent variable in Models 586. In Models 1 and 4, the control variable of turnover and two industry dummies (process industry and assembly industry) were entered. In Models 2 and 5, the independent variables of BufDC, BufR, BridgC and BridgM were added. In Models 3 and 7, the context constructs of Ext-CoMgmt, Int-CoMgmt, SD, DD, CC and SBC were entered as independent variables. Models 4 and 8 add the interactions between BufDC, BufR, BridgC and BridgM, as well as context variables.

The results for cost-efficiency show that BridgC has a significant positive association (p < 0.05) and BridgM has a significant negative association (p < 0.05) with cost-efficiency (M2–M4), supporting H3a.i and H4a.i. BufR has a hypothesised (H2a.i) negative relationship with cost-efficiency in Model 2, but it disappears when context variables are added. Int-CoMgmt is also positively related to cost performance (p < 0.10) (H5a.vi). Regarding the other context variables, SD (H5a.iv) and DD (H5a.iii) are negatively related to cost (M3–M4). Adding the interaction terms in Model 4 did not significantly add to the variance explained in cost-efficiency, rejecting H1b, H2b and H3b. Interaction effects in Model 4 may indicate associations between BufDC and/or BufR and cost-efficiency in contexts characterized by category complexity (H1c.i and H2b.i), but these findings must be confirmed in a future study.

Concerning delivery performance, Models 7 and 8 show that BufDC has a significant positive relationship with delivery performance (p < 0.05) (H1a.ii). Models 7 and 8 also imply that there is a positive relationship between BufR and delivery performance (p < 0.05) (H2a.ii). Interestingly, Ext-CoMgmt shows a positive relationship in this regard (M7 and M8) (H5b.vii). Furthermore, SD (p < 0.01) (H5b.iii), DD (p < 0.05) (H5b.ii) and CC (p < 0.05) (H5b.i) are negatively related to delivery performance, while SBC (H5b.v) is positively related to delivery performance. BufDC works particularly well when aiming to resolve demand–supply imbalances without delivery disruptions in contexts characterised by DD, supporting H1c.i. BridgC improves delivery performance, combined with external co-management (H3c.ii). However, the F-change is significant at p < 0.10, so the interaction effects should be interpreted with some caution and confirmed in a future study.

Theoretical contributions

This research drew on information processing theory (Galbraith, 1974; Tushman and Nadler, 1978) for the highly relevant concepts of buffering and bridging in order to capture the key strategy alternatives available for procurement to cope with the information-processing needs inherent in the task of demand–supply imbalance resolution (cf. Bode et al., 2011; Schlegel et al., 2021). Furthermore, information processing theory allowed us to identify and link several contextual factors with these coping mechanisms so as to characterise the task environment of procurement both internally and externally. This general set-up led us to put forth several literature-based hypotheses, the testing of which has brought clarity to the picture thus far provided by the somewhat fragmented literature regarding how procurement can engage with and enable demand–supply balancing in the S&OP process (e.g. Roscoe et al., 2020; Jonsson et al., 2021). Within the contextual perspective of the second research question, this research also contributes to the contingency perspective on the S&OP process research (Kristensen and Jonsson, 2018), with indications of when and how procurement's involvement in S&OP generates favourable outcomes. The operationalisation of the internal and external co-management of procurement into S&OP constructs and the subsequent empirical testing show how S&OP can be a mechanism for procurement management. This provides operational details for the procurement-focused literature on demand and supply integration (Wagner and Eggert, 2016).

It is worth re-emphasising the significance of the findings that suggest the need for integration both internally (for demand change) and externally (for collaboration) in order to effectively and efficiently resolve demand–supply imbalances, respectively. With the evidence regarding the superiority of collaborative demand change over more independent redundancy measures in terms of enabling effective responses and securing the delivery of key inputs, additional perspectives are brought to the literature on the benefits of the cross-functional integration of procurement (Foerstl et al., 2013). Mitigating the root cause with joint efforts is more effective than attempting to isolate oneself from dynamism of supply. The evidence regarding the primacy of hands-on supplier collaboration over relatively more hands-off monitoring further clarifies the contexts in which collaborative or even deep supplier relationships (Kim and Choi, 2015) may be the relational strategy of choice (Terpend and Krause, 2015), that is, when resolving input shortages cost-efficiently is of primary importance. Because the literature suggests that supplier collaborations can draw on complex social interactions and a long history (cf. Foerstl et al., 2010), such potentially valuable and unique resources may help manufacturing firms create sustained competitive advantages in the key domain of matching demand with supply. The big picture therefore suggests that the more demanding mechanisms bring in relatively higher benefits.

Considering our findings from the resilience perspective, the extant literature on resilience response describes, for example, COVID-19-related responses to supply chain disruptions (van Hoek, 2020) and strategies of food processing SMEs in coping with the COVID-19 crisis, emphasising the time and cost of reactions (Ali et al., 2021). Lapide (2022) suggests the importance of a quick response with S&OP, and Dube et al. (2022) identify responses by public organisations in terms of ventilator procurement during COVID-19. While some generalisable propositions have been put forth regarding, for example, the association between collaboration among supply chain partners and shorter response times (Hohenstein et al., 2015), our research provides more comprehensive insights into response as an outcome because it places the efficiency and effectiveness of imbalance resolution in the role of dependent constructs. This somewhat rare research design emphasises that the efficiency and effectiveness of a response are fundamentally different from more general outcomes, such as, for example, the cost or delivery performance of a firm or a procurement function. Thus, as the findings suggest, for example, that supplier monitoring paradoxically makes a purchasing manager worse off in the heat of an immediate imbalance response because it is more likely to result in reactive and expensive measures constrained by a limited number of alternatives in a short timeframe, whereas in a normal situation and in the long term, monitoring may well allow for the timely management of supplier delivery and cost (Maestrini et al., 2018). With the above-described results, an important contribution to the broader supply chain resilience literature on response can be made, in addition to the procurement and S&OP literatures.

Managerial implications

The results of this study highlight the importance of procurement in the S&OP process, as well as calling for several tactical measures that can be taken to resolve and reduce the effects of supply and demand imbalances. However, the way in which procurement participates in and provides input for the S&OP process is typically not as advanced as the methods for modelling and analysing in-house production capabilities. One way of developing a better understanding of the risk of shortages in the supply network would be to add another pre-meeting to the supply side of the S&OP process. The objective of this meeting would be to take the unconstrained demand plan and load it into a supplier network model in which each supplier is represented according to contracted capacity or communicated and confirmed capacity. Such a novel approach would indicate the quantified load on the supplier network, with the resulting “heatmap” showing which suppliers are overloaded, by how much they are overloaded and when this occurs. The “when” must be outside the shortest reaction lead-time to enable proposing actions that will reduce the risk of a future imbalance, and therefore, the planning horizon and time fence may have to be increased. Focusing on this process of facilitating procurement's ability to resolve imbalances can help implement the various tactical actions that this paper has highlighted, such as more proactive collaboration to support timely decision-making, even with relatively powerful suppliers.

As the importance of supply availability has increased, the focus of S&OP has, in many cases, changed from the unconstrained demand forecast to a constraining supply forecast. Supply forecasting may thus become as important as demand forecasting in more companies' S&OP processes than this is true of today. As concluded in the study, both the internal and external co-management of procurement are keys to resolving imbalances because they are conducive to a better awareness of the challenges involved.

Limitations and future research

Our study adopts a cross-sectional research design with data collected from Finland and Sweden, which may limit the generalisability of the findings to other geographical contexts. While the focus of this research was not on demand–supply imbalances caused by disruptions due to COVID-19, the data were collected while the pandemic was still ongoing, which may have affected how the respondents perceived they had been able to resolve imbalance situations. Future research may also control for the size of imbalance situations and potential differences in demand patterns.

In terms of further research, a more detailed investigation of the best practices in reducing information-processing needs (buffering) and increasing information-processing capacity (bridging) is suggested, as well as how to better integrate procurement and supply base information into the S&OP process. This, for example, includes exploring the specific mechanisms and outcomes of buffering by demand change, the proactive interventions and cost implications of bridging by monitoring and the role of S&OP as a process for enhanced procurement co-management, especially in dynamic environments. The proposed procurement-specific S&OP interventions in the managerial implications section should be further developed, explored and assessed in future research. This includes the design and outcome of a supply pre-meeting and extending the supplier capacity visibility from a first tier to a multi-tier focus.