Dynamic resource management and slack resources

Abstract

Purpose

Organisational slack has been widely considered in strategic management, but there is a gap in understanding the process of accumulation and application of slack resources. From a dynamic perspective and over an extended period of time, this paper analyses the management of slack resources and evaluates whether the different behaviours, in relation to the accumulation and consumption of slack resources, have any effect on performance.

Design/methodology/approach

The resource-based view and the dynamic extension of this theory, i.e. resource management and resource orchestration, were analysed in order to evaluate how slack resources can be managed and generate value. Assuming a configurational approach, the analysis was structured into two stages to answer the proposed hypothesis. The first stage studied whether there were different patterns of management of slack resources over time using the DistatisR package. The second stage evaluated which behaviours had the greatest impact in terms of profitability by using a dynamic panel data regression.

Findings

Three different types of slack resource management were found in companies: efficient, effective and erratic. Different types do not have the same impact on performance.

Originality/value

The dynamic management of slack resources has scarcely been considered, even during periods of crisis and economic expansion. This research advances the understanding of how firms transform slack resources into performance from a dynamic perspective.

Citation

Agustí, M.A., Aguilar-Caro, R., Galán, J.L. and Acedo, F.J. (2024), "Dynamic resource management and slack resources", Management Decision, Vol. 62 No. 13, pp. 223-242. https://doi.org/10.1108/MD-01-2023-0119

1. Introduction

Dynamic resource management has become a fundamental challenge for strategic management (D'Oria et al., 2021), as it is the management of resources rather than their ownership that enables the firm to generate profitability and create value (Lavie, 2012). The resource orchestration approach, a theoretical extension of the resource-based view of the firm, has sought to understand how managers transform resources to create value (e.g. Helfat and Winter, 2011; Helfat et al., 2009; Sirmon et al., 2007). In fact, few studies have attempted to empirically analyse the resource management process (e.g. Sirmon et al., 2007), focussing primarily on human capital (Chadwick et al., 2015; Andersen, 2021) or management teams (Domínguez-CC et al., 2023). However, organisational slack or slack resources, have been a widely studied phenomenon within strategic management (Carnes et al., 2019; Daniel et al., 2004) and are considered a source of profitability.

The dynamic management of these resources remains rarely studied (Agustí et al., 2020). This management involves a deliberate process of accumulating and applying slack resources under certain circumstances and is particularly important in times of crisis (Alessandri et al., 2014; Bourgeois, 1981; Godoy-Bejarano et al., 2020) or when the firm is trying to address important strategic challenges (Bourgeois, 1981; Hernandez-Vivanco and bernardo, 2022; Wasiuzzaman et al., 2022). Recently, aspects such as the generation of slack resources (Titus et al., 2022), their association with crises (Flammer and Ioannou, 2021; Kim et al., 2021), how these affect performance (Lefebvre, 2021) or the assessment of their consumption (Namiki, 2015) have become key research topics.

However, to the best of our knowledge, the ideas of dynamic resource management (Sirmon et al., 2007) or resource orchestration (Helfat et al., 2009; Sirmon et al., 2011) have not yet been applied to this type of resource (Agustí et al., 2020, 2021), although it is an appropriate theoretical approach to analyse how firms accumulate and use their slack resources (Conz et al., 2023). Consequently, this study aims to investigate how companies manage their slack resources in both crisis and expansion periods and how different forms of management affect performance.

Studies analysing the relationship between slack resources and performance have been characterised by their cross-sectional nature and by treating this relationship separately, considering each type of slack resource independently (see Carnes et al., 2019, for a review). Dynamic resource management necessarily implies a temporal analysis and the treatment of the different types of resources together (Du et al., 2022). As a result, understanding the processes of slack consumption (Namiki, 2015) and generation (Titus et al., 2022) and their relationship to performance (Carnes et al., 2019) requires broadening theoretical and empirical approaches beyond what has been applied so far so that the accumulation, structuring and application processes can be understood. Furthermore, following the approach of some authors, this study considers that the effect of slack resources on performance does not occur in isolation and individually, but through the configurations of these resources (Agustí et al., 2022; Geiger et al., 2019; Marlin and Geiger, 2015b) and from a dynamic, rather than static, perspective. Therefore, we study the process by which firms change or accumulate slack resources through the theoretical lenses of dynamic management and resource orchestration, answering the call made by Bentley and Kehoe (2020) to deepen knowledge on the relationship between slack and value creation.

In addition to the theoretical challenge of applying these theories to slack resources, the study also presents an important methodological challenge. Thus, from an empirical point of view, there is a need to introduce a methodology that considers the study of trend and evolution beyond the mere relationship between variables (Swider et al., 2023).

Longitudinal clustering and panel data regression analysis are methods not yet used in studies on slack resources but that allow a proper analysis of the dynamic management of this type of resource.

This study contributes to the literature on slack resources by applying the theoretical approach of resource orchestration to the dynamic management of slack resources in a novel way, not yet studied in the field. It also contributes to the literature on dynamic resource management by providing empirical evidence on a different type of resource from those analysed in existing research, which tends to focus on human capital. From a practical point of view, the results shed light on the types of slack resource management that lead to higher performance and provide guidance to managers on how to jointly manage their slack resources.

2. Theoretical framework

Organisations can be seen as a set of resources and capabilities that managers combine and use to create value and generate profitability (D'Oria et al., 2021). Amit and Schoemaker (1993) defined resources as those productive factors which a firm owns or controls, such as physical assets, human capital and financial capital, while capabilities are the ability to combine resources to create value. According to the resource-based view (RBV), not all resources are equal. Thus, for a firm to achieve competitive advantage, its resources and capabilities must be valuable, rare, inimitable and nonsubstitutable (Barney, 1991). However, the mere possession of such resources does not guarantee the possibility of generating competitive advantage or value creation (Barney and Arikan, 2005; Sirmon et al., 2007), but firms must be able to accumulate, combine and exploit resources (D'Oria et al., 2021; Grant, 1991; Sirmon and Hitt, 2003).

Due to the absence of theoretical approaches on how these resources are transformed in the value creation process, Sirmon et al. (2007) proposed the need to structure the value creation process based on given resources. Almost simultaneously, Helfat et al. (2009) produced their theory of “asset orchestration” (2009: 24), which considered the process of selection, investment, deployment and reconfiguration of assets, affecting the “firms’ abilities to adapt to changing conditions in their industry environments” (Sirmon and Hitt, 2009, pp. 1376) and, therefore, explicitly recognising the importance of managerial action in managing resources and in shaping the performance of the company.

In this sense, the works of Sirmon et al. (2007), Helfat et al. (2009), Sirmon et al. (2011) or Lavie (2012) constitute the different proposals for the development of the RBV that contemplate the necessary modification of resource stocks (Dierickx and Cool, 1989) and their effect, not only individually but also in combination, on business profitability (D'Oria et al., 2021; Hitt et al., 2021). Therefore, the mere possession of resources is a necessary, but not sufficient, condition to achieve competitive advantage, and it is necessary to understand the processes through which these resources are used to assess their effect on performance (Kraaijenbrink et al., 2010; Mallon and Lanivich, 2023; Conz et al., 2023) according to environmental conditions (Martin and Cuypers, 2024).

Barney (1991) claimed that firms need to use their strategic resources to achieve a competitive advantage. This need to modify the stock of resources and their management can be observed in the study of slack resources (Bourgeois, 1981). However, most of the research to date has concentrated on assessing the effect of owning slack resources on performance (Carnes et al., 2019), leaving aside the possible joint effect, management or orchestration, and their application over time.

2.1 Configurations in the management of slack resources

Bourgeois (1981) defined them as that cushion of current or potential resources that allows a firm to successfully adapt to internal pressures, such as adjustment or to external pressures, such as a change in policy or a crisis and to initiate changes in strategy with respect to the external environment.

Although this definition is the most widely accepted, it refers to resources indiscriminately. Bourgeois and Singh (1983, p. 43) describe three types of slack, depending on the degree of availability (available, recoverable and potential). Most research that has studied the effect of slack resources on performance has considered the effect of each of the types of slack independently (Lecuona and Reitzig, 2014; Tsang, 2006; Tyler and Caner, 2016). This is one of the possible causes of the contradictory results found in such a relationship (Carnes et al., 2019).

However, several authors (Marlin and Geiger, 2015a, b; Mousa et al., 2013) have proposed that it is the joint effect of the combination that determines firm business performance, thus following the initial ideas of the RBV (Amit and Schoemaker, 1993) and the later developments of resource orchestration (Helfat et al., 2009). Studies have confirmed the existence of these configurations of slack resources but identified them transversally, thereby focussing on the possession of the resources and not on how those resources are managed (Marlin and Geiger, 2015b; Geiger et al., 2019).

The literature has associated the accumulation of slack resources with two main motives (Bourgeois, 1981): (1) to have the necessary means to develop aggressive strategies or to protect against adverse situations or shocks, both external and internal (Conz et al., 2023; Du et al., 2022; Flammer and Ioannou, 2021; Godoy-Bejarano et al., 2020; Kim et al., 2021) and (2) to maintain performance or ensure the survival of the firm (Carnes et al., 2019; Paeleman and Vanacker, 2015). Bourgeois (1981, p. 37) argued that managers can generate their own slack through deliberate action (“slack gainers”) (Titus et al., 2022) or as a determinant of strategic behaviour as outcomes to the application of slack (“slack losers”) (Bentley and Kehoe, 2020). Similarly, businesses can also modify the quantities of each type of slack resource. However, a possible joint management of these types of resources (configuration) is envisaged, as it is foreseeable that the indicators co-vary or compensate each other (Bourgeois, 1981; Conz et al., 2023).

Therefore, businesses can achieve greater profitability through the application and redistribution of slack resources (Abukari et al., 2024; Conz et al., 2023; Greenley et al., 1998), it being necessary to pursue a combination of slack resources to maximise the level of competitiveness of the business (Dolmans et al., 2014) or even a reallocation of these resources allowing the pursued objectives to be achieved (Helfat and Peteraf, 2003), being able to facilitate a strategic transition (Mallon and Lanivich, 2023). These processes are assimilable to those proposed by resource management approaches (Helfat et al., 2009; Lavie, 2012; Sirmon et al., 2007, 2011). Thus, following the ideas of resource orchestration (Helfat et al., 2009), managers must obtain and modify the slack resources of their business (structuring), combine and integrate them so that they are available when needed (bundling) and use them to create value or ensure the survival of the business (leveraging). The application of these ideas to slack resources is original in that it involves considering the different types of slack together and analysing how the combination of these resources is managed through their accumulation and application to address different situations and strategies. There are studies that analyse the dynamic relationship between each slack type and performance (Agustí et al., 2020) and the relationship between resource configurations and performance (Geiger et al., 2019; Marlin and Geiger, 2015b), but these are from a static perspective. The present study combines both approaches using a highly developed theoretical approach in strategic management and a novel methodology in the field (Swider et al., 2023; Conz et al., 2023).

Consequently, it is possible to consider that the presence of slack resource configurations can not only be identified in the possession of these resources (Marlin and Geiger, 2015a, b; Mousa et al., 2013), but also in their management, that is, in the dynamic processes of accumulation and consumption as a reaction to unforeseen situations or proactively, for strategic reasons (Bourgeois, 1981). Following these premises, we propose the following:

H1.

There are differentiated groups of firms according to the dynamic slack management strategies.

2.2 Slack resource management and performance

To date, research has proposed that different slack configurations are associated with different levels of profitability (Mousa et al., 2013). Previous studies noted that slack configurations influence performance and also recognised the existence of equifinality, in which different configurations could lead to the same result (Furnari et al., 2021; Geiger et al., 2019; Marlin and Geiger, 2015b; Mousa et al., 2013). However, the effect of accumulation or consumption, that is, the dynamic management of slack resources, on performance remains unexplored (Namiki, 2015), especially in smaller companies (Conz et al., 2023).

Assuming that performance contributes to slack generation (Titus et al., 2022) and that the effect of slack resources on performance occurs when they are consumed (Carnes et al., 2019), it is interesting to perform a dynamic analysis of this relationship, taking into consideration the different slack configurations. In this sense, firms that are able to develop aggressive strategies or protect themselves against adverse external and internal shocks (Flammer and Ioannou, 2021; Godoy-Bejarano et al., 2020; Kim et al., 2021) will perform better when a reduction in environmental munificence occurs.

However, accumulating these resources in excess can have negative connotations for performance (Nohria and Gulati, 1996). Justification is associated with agency problems (Tan and Peng, 2003). In the same vein, some authors have argued that many organisations perform better with fewer resources (Baker and Nelson, 2005). This is explained by managing resources more efficiently, potentially showing the absence of essential rigidities (Leonard-Barton, 1992) and the alignment of managerial and shareholder interests (Jensen and Meckling, 1976), which is necessary in these cases to have a “resource orchestration capability in the form strategic deployment of resources” (Abukari et al., 2024). This constraint-focused view of resource management requires an examination of the source of scarcity. Thus, firms may have less slack through their own will (George, 2005) or in response to overconsumption to cope with hostile and resource-scarce environments (Baker and Nelson, 2005). Under the resource orchestration approach (Helfat et al., 2009; Sirmon et al., 2007), managers structure, combine and leverage the slack resources of their businesses to create value or ensure the survival of the business (Conz et al., 2023; Mallon and Lanivich, 2023). Managers will make different orchestration decisions that will lead to different resource management strategies.

Consequently, considering the existing literature on the relationship between slack and performance, using a transversal approach (Marlin and Geiger, 2015b; Geiger et al., 2019) and the recognition that the influence of (slack) resources on performance occurs when these are managed (Sirmon et al., 2007; Carnes et al., 2019), we propose the following hypothesis:

H2.

Different slack management strategies will lead to different levels of performance.

3. Methods

3.1 Data sources and sample

The sample was drawn from the Spanish SABI database (Bureau van Dijk) and covers the period 2006–2019, which includes a situation of generalised crisis, specifically the global financial crisis of 2008, and a period of recovery, which in Spain began in 2013. We selected those firms that had undergone an audit process, retrieving those with unqualified and unqualified Emphasis of Matter (EOM) audit opinions (Azim, 2013) in each of the years of the period under consideration. To avoid problems produced by sectoral heterogeneity (Gral, 2014), our study focused only on industrial companies. Therefore, our sample included companies from sectors ranging from NACE 20 to 38, both inclusive, reducing the sample to 631 companies. The identification of slack resources followed the approach that has dominated the literature (Bourgeois, 1981), which uses financial measures.

In general, most studies (Daniel et al., 2004; Gral, 2014; Carnes et al., 2019; Karacay, 2017) point to the need to consider different types of slack (available, potential and recoverable). Current ratio (CR) (current assets/current liabilities) and quick ratio (QR) (cash/current liabilities) are the most used indicators when measuring available slack. Although the QR can be measured considering other assets associated with liquidity, in our case, cash was considered due to its immediacy (Deb et al., 2017). For the other types of slack, the literature has used the debt/equity (DE) ratio to measure potential slack and the selling, general and administrative (SG&A) expenses/sales ratio for recoverable slack. The Spanish accounting system does not allow the identification of these types of expenses through financial information, so another of the most used expenses was Personnel Expense/Operating Income (PE) (Lecuona and Reitzig, 2014; Mishina et al., 2004). The consideration of this variable is justified in view of the context, as personnel expenses were one of the key elements in the 2008 financial crisis.

Among the diverse set of measures regarding firm performance, the most common way to measure profitability has been through economic profitability, i.e. the return on assets (ROA) (Carnes et al., 2019). ROA reflects the effectiveness of management in utilising the assets of the business to generate earnings (Adler and Kwon, 2002). Additionally, ROA is a reliable measure of the efficient utilisation of an organisation’s resources (Krishnan and Park, 2005).

To control for possible effects not related to slack resources, three variables were selected. The age and family character of the firm and the size of the firm in terms of the number of employees (George, 2005). The inclusion of the size of the firm, in terms of the number of employees and the age of the firm, is justified because larger and older firms exhibit varying degrees of resource slack (Lefebvre, 2021; Elbanna, 2012). The inclusion of the family character variable is warranted due to the unique characteristics of family firms, which enable them to effectively utilise their slack resources in pursuit of substantial growth (Minola et al., 2022).

3.2 Analyses

The analysis was structured into two stages to answer the proposed hypothesis. The first stage studied whether there are different patterns of use of slack resources. These patterns group different ways of managing slack resources over time. The longitudinal clustering was performed with the DistatisR package (Beaton et al., 2019) in R language.

The second stage of the analysis evaluated which behaviours have the greatest impact in terms of profitability. It was decided to employ panel data regression analysis, given the structure of the database, which consists of 631 companies tracked over a period of 14 years, resulting in a strongly balanced panel. Furthermore, taking the dynamic nature of the dependent variable into consideration, specifically the return on assets (ROA), is crucial in the analysis (Arellano and Bond, 1991). This analysis was developed using xtabond2 (Roodman, 2006) in Stata 17.

3.2.1 First stage: identification (cluster analyses)

The identification of possible dynamic processes in the use of slack resources requires different steps. First, the within-firm distance was calculated using the dynamic time-warping method (Giorgino, 2009). Then, DistatisR was performed to obtain a large 631x631 distance matrix. Finally, the large-distance matrix was clustered around the medoids. The data were partitioned (clustered) into k clusters around medoids (PAM), as it is a more robust version of K-means (Kaufman and Rousseeuw, 2009; Schubert and Rousseeuw, 2019).

The clustering process offers a range of optimal solutions between 3 and 4 clusters. After a closer look at the results, it was decided to retain three groups, as these offered greater theoretical consistency (Solano, 2021). Additionally, the validity of the cluster was evaluated. In this sense, the average silhouette width provided a good indicator of the validity of the observation classification. Thus, observations with an average silhouette width (SI) > 0.20 are considered to be well clustered (Kaufman and Rousseeuw, 2009). In our case, the average silhouette width = 0.37 ensured a good classification. Table 2 shows the main descriptive statistics for the cluster.

Table 1 and Figure 1 illustrate the physiognomy of the three groups of companies. It could be said that the first group is composed of companies that make efficient use of their slack resources. The second group is not as efficient as the first but makes effective use of its slack resources and the third group shows erratic slack resource management. The companies in the first group have a greater range of all types of slack resources, accumulating available slack in the years prior to the crisis, using it during difficult times and building it up again, especially in the case of liquidity, during the recovery period. Although the companies in group 2 have fewer slack resources, they utilise and manage the available slack in a similar way to those in group 1. In both groups, potential slack is used as a reserve, showing an increase throughout the period, taking into consideration the inverse nature of the ratio used (debt/equity), which somewhat ensures a response to possible economic downturns. Lastly, the companies in group 3 have the least amount of available, recoverable and potential slack, managing the latter erratically with significant changes over the analysed period.

Furthermore, ANOVA and MANOVA analyses were performed. The MANOVA analysis ensures that the use of slack resources is more similar within the cluster than between clusters (F = 840.85 p-value = 0.000). The ANOVA analysis, in turn, indicates that the use of slack resources is different between clusters from variable to variable (CR: F = 3678.43 p-value = 0.000; QR: F = 1424.37 p-value = 0.000; DE: F = 131.19 p-value = 0.000; PE: F = 135.33 p-value = 0.000). Table 2 shows the within cluster correlation.

To ensure the decision to keep three groups, each clustering variable (CR, QR, DE and PE) was regressed over the variables cluster membership and year. For instance, in the case of the current ratio (CR), the equation would be:

CRit=β0+βclusteri+βyeari+εit

The differences between coefficients were tested using the lincom (linear combinations of parameters) post-estimation test. The Stata 17 lincom analysis gave a confidence interval as well as a test of the null hypothesis, where the difference between the coefficients was zero.

H0:βCi−βCj=0; ∀i≠j∈[1,2,3]

From the results obtained, it can be affirmed that, for each clustering variable and between all groups, there is no equality of coefficients.

All in all, it can be said that each cluster represents a different configuration of joint management of the four slack variables. These results allow confirmation of H1.

The three clusters show three different behaviours, which could be associated with three different postures towards slack resources and their management. Thus, the first group can be associated with companies that understand that slack resources are positive and useful for the development of the company. Therefore, when possible, these companies tend to accumulate slack resources. The second group is composed of firms that, unlike the former, try to be efficient and keep slack resources to a minimum. Finally, the third group is composed of firms with an absence in the management of slack resources, according to the theories.

3.2.2 Second stage: evaluation (dynamic panel data regression)

Following previous research, to test the effect of the slack management strategy, its possible influence on performance was evaluated.

Table 3 shows the mean values, standard deviations and intercorrelations of the variables in our study. As Table 3 shows, all variables are negatively correlated with the dependent variable. All correlations between pairs of variables are less than 0.3, so it was evident that there were no multicollinearity problems (Kalnins, 2018).

To evaluate this effect, a panel data analysis was performed. Specifically, the following model was estimated:

ROAit=∝i+β1age of firm+β2family firm+β3size of firm+β4cluster membership+εit

Due to potential endogeneity issues (Roodman, 2006), Arellano–Bond GMM estimators (xtabond2) were applied. Specifically, to control for endogeneity, the lagged-dependent variable, with a maximum of three-year lags, was included as an independent variable (Roodman, 2006). Finally, a two-step estimator with a robust option was used (Arellano and Bond, 1991). The robust option provides standard errors that are robust to heteroskedasticity and arbitrary patterns of autocorrelation within individuals. Applying the Windmeijer correction, it is also possible to correct for the part of downward bias that can appear in standard errors when samples are small (Windmeijer, 2000, 2005).

Table 4 presents the results of the panel data analyses and the quality of the estimation, including the identification of the models. The regression models include control variables (Model 1) and the main effect of cluster membership.

The quality of the estimation is manifested in five conditions that could be verified from the results shown in Table 4. First, the number of groups is greater than the number of instruments in the seven models. Second, the Wald test is statistically significant in all three models (all p values = 0.000). Third, there is no second-order autocorrelation in any of the seven models (AR (2) p-values> 0.05). Fourth, the Hansen test for over-identification is accepted (as the null hypothesis is that the model is identified, and the p-values are between 0.1 and 0.8 in all models). Finally, the instruments are exogenous in both the GMM and IV estimations. Overall, the quality of the model continues to improve as the full model is approached.

The results obtained support Hypothesis 2, indicating that the companies in Cluster 3, composed of erratic entities, demonstrate a lower performance (−0.183***) than the companies in Cluster 1 (efficient) and Cluster 2 (effective). In addition, the findings indicate that the control variables used, namely firm age (−0.0428***) and family business (−0.104***), exhibit significant negative associations. However, in the case of company size (0.00159), evidence of statistical significance is not observed (Elbanna, 2012).

To confirm the performance differences associated with these different management behaviours, a post-estimation test (lincom – linear combinations of parameters) was conducted, which involved processing the linear combinations of coefficients. This analysis gave a confidence interval as well as a test of the null hypothesis, where the difference between the coefficients was zero. Table 5 shows the results of the lincom test.

It can be confirmed that the performance of ‘efficient companies’ (C1) and ‘effective companies’ (C2) has similar profitability (ROA), whereas ‘erratic companies’ (C3) demonstrate lesser profitability. Certainly, the application of slack resources has an impact on the performance of companies.

4. Discussion and conclusions

The study of resource management has developed considerably in recent years. Since the seminal work of Dierickx and Cool (1989), various research approaches in strategic management have argued that the impact of resources on performance should be analysed on the basis of the dynamic management of these resources rather than on the basis of their availability. Following the many authors who have proposed the importance of analysing the consumption or use of slack resources (Love and Nohria, 2005; Tan and Peng, 2003), this article examines how businesses manage their slack resources by analysing a broad time span that includes periods of deep crisis and recovery.

Similar approaches, such as resource orchestration (Helfat et al., 2009) and dynamic resource management (Sirmon et al., 2007), consider that managers have discretion to manage their business’s resources in ways that allow them to achieve competitive advantage and create value. As a result, it is possible to find different ways of managing the same types of resources, which can lead to different results between firms. When these ideas are applied to the management of slack resources, the results of this study show, as suggested in H1, that firms have different approaches to resource management. Thus, some firms manage their slack resources by maintaining high levels and accumulating them quickly after use (efficient firms) others manage them by trying to maintain low levels (effective firms) and finally, a third group does not appear to present a strategy towards slack management (erratic firms). The first two assume an intentional and relatively similar management of the different types of slack. However, these differ in the level of slack that managers consider appropriate for their companies as well as in the way and speed of the accumulation of those resources. This is especially notable in the slack available in the run-up to the crisis and in the period of economic recovery. Consequently, following the ideas of resource orchestration (Helfat et al., 2009), managers in both types of firms use their slack resources in similar ways (leveraging), but maintain notable differences in the level of resources these accumulate and dispose of (structuring) and in how these are combined and integrated for use when needed (bundling).

The third group of companies does not appear to develop a coherent management of slack resources, especially in the case of potential slack, which shows notable changes throughout the period analysed. In other types of slack resources, the levels are so low that management must have been clearly reactive before the arrival of the crisis, not carrying out a strategy of increasing levels when the period of economic recovery occurred.

Indeed, if firms have not accumulated significant amounts of slack resources (structuring), the orchestration approach suggests that dynamic management of these resources, i.e. combining them and using them when needed, is not possible.

Although it is difficult to know, in detail, how firms combine their slack resources, the availability of a long time span, during which firms went through a period of very deep crisis, makes it possible to describe how these managed their slack resources. From the graphical analysis, it can be inferred that, in a situation of generalised downturn, the most available slack, represented by QR, is of crucial importance. Companies with high slack drastically reduce QR to cope with the first shock caused by the crisis and, subsequently, keep it at stable levels to “protect the cash box” to ensure the manoeuvrability and capacity of the company during the crisis period (Deb et al., 2017; Kim and Bettis, 2014; Jung et al., 2020; Martin and Cuypers, 2024), as Conz et al. (2023, p. 976) show for family businesses: “in times of crisis, slack, especially financial, is the key resource to leverage to build resilience and preserve the continuity of the family firm”. These behaviours confirm the existence of a minimum or survival level of the most available slack (Campello et al., 2011), although this level will differ according to the efficiency orientation of the firm (Deb et al., 2017; Kim and Bettis, 2014).

On the contrary, the other types of slack do not have a buffering function against the initial jolt, as their levels do not decrease during the crisis period. In fact, a slight increase in these resources can be observed. The management of CR is particularly paradoxical since, as a type of available slack, a reduction of this indicator could be expected in the first moments of the crisis and a recovery with the change of cycle (Bourgeois, 1981). However, as soon as the crisis begins, the CR presents an increase, not due to the company’s actions but from an accumulation of stocks and credits to customers, including unpaid accounts, due to the effect on sales of the economic crisis. In summary, available slack resources do not present the same level of availability and special attention should be paid to the consideration of cash as a strategic asset (Deb et al., 2017; Kim and Bettis, 2014; Martin and Cuypers, 2024). This underlines the importance of the interpretation of the variables chosen for the analysis and, possibly, of the interpretation of the discrepancy between some of the results obtained in the previous literature (Carnes et al., 2019).

Regarding potential slack, some authors have pointed out the need to consider the joint level of financial flexibility (Arslan-Ayaydin et al., 2014), availability and leverage of the firm to understand the success of firms in crisis situations. In our case, most companies appeared to increase their potential slack during the crisis by improving (reducing) their leverage ratio (D/E), with greater intensity in those companies with lower levels of slack, which are, therefore, relatively more indebted. This trend responds both to the desire of the company to have a margin of indebtedness in the event of greater financial difficulties and, above all, to the difficulty or impossibility of accessing external sources of financing (Campello et al., 2011).

Regarding recoverable slack, as noted in previous studies (Carnes et al., 2019), this type of absorbed resource is not easy to recover in times of crisis, as the cost of reducing it can offset the benefits generated (Lecuona and Reitzig, 2014). However, from a slack resource perspective, those companies with high levels of slack resources have been unable to “recover” them to increase their efficiency and improve their situation against the crisis.

Consequently, in a crisis, managers are able to “orchestrate” (Helfat et al., 2009; Conz et al., 2023), in the short term, the most available slack and potential slack. The former is to maintain the company’s operations and viability and the latter is to ensure continuity in the event of a worsening of the crisis. Discretion in the management of the other types of slack resources analysed is considerably less and only begins to yield results after a longer period of time, provided that the company has a chance of recovery.

The second question was whether management or lack of management as occurs in cluster 3 (erratic firms), had any effect on performance. Our results show that firms that do not develop coordinated slack resource management (Hitt et al., 2021) show lower performance than those that carry these resources. However, it is also interesting to note that there is equifinality (Furnari et al., 2021) in the management style (Marlin and Geiger, 2015b), as the companies that carry out the management of these resources, regardless of the orientation of that management, show similar effects on performance. Although efficient and effective firms achieve similar levels of profitability, on average, the dispersion of the effective cluster shows that it is more plausible to obtain higher profitability for those companies that maintain higher levels of slack resources (efficient).

In summary, the results show, as the resource orchestration approach points out, that managers manage resources differently and that these differences have an impact on value creation and performance (Helfat et al., 2009; Sirmon et al., 2011). Therefore, firms should strategically manage these resources to achieve superior performance (Mallon and Lanivich, 2023; Martin and Cuypers, 2024), in particular when facing an external jolt (Conz et al., 2023; Du et al., 2022).

5. Theoretical and practical implications

5.1 Theoretical implications

First, this paper applies the ideas of resource orchestration (Helfat et al., 2009) and dynamic resource management (Sirmon et al., 2007) to the study of slack resources in a novel way to answer the requested question of how slack resources are consumed or applied (Flammer and Ioannou, 2021; Paeleman and Vanacker, 2015). The results show that management of resources is necessary (Hitt et al., 2021) and that firms must follow a rationale in their accumulation (Titus et al., 2022) and their application (Flammer and Ioannou, 2021). However, types of resources and measures commonly used in the literature should be used with caution when analysing the function of slack resources, as not every type of slack resource can serve as a cushion against an external or internal shock (Bourgeois, 1981), and each type plays its own role in the company.

As the resource orchestration approach suggests, our study shows it is necessary to develop a slack resource strategy (Lefebvre, 2023). Companies must go beyond merely accumulating valuable and rare resources, as steps to bundle and leverage them are also required to gain the desired benefits.

Second, through a fact-based approach and by focussing our study on documenting the impact of the complex phenomenon of the 2008 economic downturn on firm-level decision-making, we have striven to answer the call by Agarwal et al. (2009) for studies on these rare phenomena that contribute towards a better understanding of their effects.

Furthermore, our study is longitudinal and uses a dynamic approach not previously used in the analysis of slack resources and, therefore, contributes to the analysis of the consumption or application that companies make of these resources, an issue scarcely addressed in the literature (Swider et al., 2023). The methodology mainly used in the literature is focused on panel analysis. These models offer no clear path to causal inference as a temporal process (Zyphur et al., 2020).

The use of resource orchestration theory and longitudinal methods opens up new and interesting avenues of research in the field of slack resources, in analysing more confidently how these resources are managed and their impact on performance.

5.2 Practical implications

The findings of this study provide important implications for practice. First, the finding suggests resource coordination is important in leveraging value from slack resources against environmental jolts (Conz et al., 2023; Mallon and Lanivich, 2023). Thus, beyond previous studies that indicated performance could be affected by keeping slack resources, we propose that managers need to carefully orchestrate these resources in the pursuit of competitive advantage. Therefore, managers must improve and reconfigure the internal process to create value through slack resources.

Second, the results show that managers cannot manage their available resources with absolute discretion, as there are limits to the use of some slack resources, which cannot be easily used, when the company is facing a deep crisis. However, the results also suggest that it is important to have some level of slack resources available, even if minimal, to allow managers to manage these resources efficiently or effectively, as such management has clear implications for performance.

Consequently, the results obtained point to the need to explicitly develop a “slack strategy”, which incorporates the different sources of slack and indications for its use (Lefebvre, 2023; Mallon and Lanivich, 2023). Companies that develop a clear strategy for scarce resources have been shown to achieve better results than those that treat them erratically. This idea complements the work of Fadol et al. (2015) and Rau et al. (2021), which consider slack resources as enablers of the effect of strategic planning. In our case, we propose the need to plan slack resources with an active vision and not as a mere facilitating instrument.

Finally, with regard to the specific management of resources, managers know that, in the face of a deep crisis, cash must be secured and solvency maintained by improving the debt ratio. However, it would appear that the possibility of managing other slack resources is overestimated, such as the less liquid items of current assets and recoverable slack. It would, therefore, be advisable to keep these resources at a reduced level unless important strategic changes are to be undertaken (Bourgeois, 1981).

6. Limitations and future research

This study is not without limitations. Although the methodology reflects changes with much more precision than most of the studies carried out to date, only a few slack indicators have been used, demonstrating that slack resources associated with the same type can present radically different behaviours. It would, therefore, be necessary to introduce new measures to verify the results obtained. Furthermore, despite the identification of configurations, the possible existence of sequential management within each configuration is also possible. The question is whether the management of each type of resource is intertwined with the management of the other slack resources and their configuration. Furthermore, the causality of the relationships has not yet been addressed this would require a study of a totally different nature. Finally, the analysis was carried out on Spanish industrial companies and any generalisation of the results obtained could be questioned. However, the economic crisis of 2008 affected almost all countries worldwide and the studies that have been carried out on other countries and contexts have shown similar results (Flammer and Ioannou, 2021; Godoy-Bejarano et al., 2020; Namiki, 2015; Wenzel et al., 2020; Campello et al., 2011).

Furthermore, it should be studied whether different stakeholders constitute an important factor that could explain the allocation and use of slack resources within an organisation (Lefebvre, 2023). In this sense, agency theory may play an important role in the choice between different configurations of slack resources.

Similarly, it is necessary to advance the dynamic analyses suggested by the theoretical lens employed. Further research into the way in which sequencing and balancing assets affect asset orchestration and linking it to different outputs, beyond economic performance (innovation policies, types of strategies, etc.), are lines for the future that have yet to be explored.

Figures

Figure 1

Means of clustering variables

Table 1

Main descriptive statistics of the clusters

Cluster	Variables	Mean	SD	Min	Max
Cluster 1 (n = 133)	Age	35.252	14.021	2	112
	Size (n employees)	4.197	0.689	1.386	6.469
	Family firms	0.526	0.499	0	1
	Current ratio	4.814	2.864	0.498	27.524
	Quick ratio	1.116	1.409	0.000	17.469
	Debt equity	0.430	0.456	0.023	7.754
	Personnel	22.715	11.045	0	83.716
	ROA	7.836	7.814	−42.11	101.888
Cluster 2 (n = 236)	Age	32.424	16.827	0	113
	Size (n employees)	4.558	1.000	0.693	7.796
	Family firms	0.297	0.457	0	1
	Current ratio	2.257	0.922	0.340	13.442
	Quick ratio	0.243	0.378	0	10.058
	Debt equity	0.910	1.629	−42.449	42.121
	Personnel	19.165	11.473	0	85.945
	ROA	7.922	11.859	−278.398	181.094
Cluster 3 (n = 262)	Age	30.553	17.927	0	102
	Size (n employees)	4.812	1.170	0	9.580
	Family firms	0.309	0.462	0	1
	Current ratio	1.252	0.479	0.151	6.287
	Quick ratio	0.091	0.162	0	2.160
	Debt equity	2.804	9.171	−164.632	133.284
	Personnel	17.503	10.874	0	89.633
	ROA	4.294	8.866	−113.075	58.161

Source(s): Authors’ own creation

Table 2

Correlation within cluster

Cluster 1	CR	QR	DE	PE
CR	1
QR	0.1960*	1
DE	−0.1752*	−0.0520*	1
PE	0.0656*	0.0053	−0.0259	1
Cluster 2	CR	QR	DE	PE
CR	1
QR	0.5795*	1
DE	−0.2931*	−0.1395*	1
PE	0.1442*	0.0492	−0.0793*	1
Cluster 3	CR	QR	DE	PE
CR	1
QR	0.2259*	1
DE	−0.0467*	0.004	1
PE	0.0441*	−0.0219	−0.1121*	1

Note(s): *p < 0.01

Source(s): Authors’ own creation

Table 3

Statistics descriptive and correlations

N	Variables	Mean	SD	Min	Max	1	2	3	4	5
1	ROA	6.397	10.061	−278.398	181.094	1
2	Age of firm	32.243	16.847	0	113	−0.0537*	1
3	Family firm	0.35	0.477	0	1	−0.0784*	0.1064*	1
4	Size of firm	4.587	1.046	1	9.579	−0.0193	0.1364*	−0.2258*	1
5	Cluster membership	2.204	0.764	1	3	−0.1530*	−0.1037*	−0.1486*	0.2195*	1

Source(s): Authors’ own creation

Table 4

Dynamic regression model with dependent variable and quality of estimation GMM regression results predicting ROA (change and experience effects)

Variables	Model 1	Model 2
Past ROA	0.394***	0.412***
	(0.0775)	(0.0789)
Age of firm	−0.0340**	−0.0428***
	(0.0158)	(0.0153)
Family firms (yes)	−0.104***	−0.104***
	(0.0323)	(0.0304)
Size of firm (N employees)	−0.0152	0.00159
	(0.0145)	(0.0140)
Cluster 1 (base)
Cluster 2		0.0137
		(0.0351)
Cluster 3		−0.183***
		(0.0430)

Quality of estimation	Model 1	Model 2
Observations	8,146	8,146
Number of groups	630	630
Number of instruments	38	40
Wald χ2	328.95	546.10
p-value	0.000	0.000
AR(1)	−3.25	−3.24
p-value	0.001	0.001
AR(2)	0.92	0.97
p-value	0.356	0.330
Hansen test of override (H0: The model is identified)	27.84	28.73
p-value	0.145	0.121
GMM instruments for levels	11.19	11.74
p-value	0.427	0.383
IV (Exogeneous variables)	15.30	16.61
p-value	0.430	0.481

Note(s): Robust standards errors in parentheses

***p < 0.01, **p < 0.05 and *p < 0.1

Variables have been standardized

Time dummies and constant are included in the models

Source(s): Authors’ own creation

Table 5

Post-estimation lincom test

Cluster	Coefficient	SE	z	P>|z|	[95% conf. Interval]		Results
C1–C2	−0.014	0.035	−0.390	0.697	−0.082	0.055	C1=C2
C1–C3	0.183	0.043	4.260	0.000	0.099	0.268	C1>C3	C1=C2>C3
C2–C3	0.197	0.046	4.300	0.000	0.107	0.287	C2>C3

Source(s): Authors’ own creation