Cost behavior in Dutch firms : are selling, general, and administrative costs sticky?

(1)

Amsterdam Business School

Cost behavior in Dutch firms: Are Selling, General, and

Administrative costs sticky?

Name: Vincent Dorrestijn

Student number: 10871128 Thesis supervisor: P. Kroos

Date: 16-06-2016

Word count: 7.124

MSc Accountancy & Control, specialization Control

(2)

Statement of Originality

This document is written by student Vincent Dorrestijn who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

1 Introduction

1.1 Background

Understanding cost behavior is a fundamental element of cost and management accounting. Numerous management accounting techniques are built on the assumption that costs change proportionally if the activity changes, regardless of the direction of the change (Noreen, 1991). This assumption is referred to as symmetric cost behavior. In contrast to the assumption of symmetric cost behavior, various studies claim that costs decrease less when activity declines compared to an increase in costs when activity increases (Noreen and Soderstrom, 1997; Cooper and Kaplan, 1998; Balakrishnan et al., 2004). Anderson et al. (2003) were the first to provide robust empirical evidence and concluded that Selling, General, and Administrative (SG&A) costs increase 0,55 percent for a 1 percent increase in revenue (proxy for changes in activity), but decrease only 0,35 percent when revenue decrease by 1 percent for firms in the United States (U.S.). These results assume an alternative cost behavior based on asymmetric costs. Anderson et al. (2003) label this phenomenon as sticky cost or “cost stickiness”.

The traditional cost model assumes that costs are either fixed or variable with regard to changes in activity, giving relevance solely to the magnitude of the change and not to its direction. Alternatively to the traditional cost model, cost stickiness starts from the point that some costs arise as a result of resource commitment decisions made by managers (Banker and Byzalov, 2014). Furthermore, Anderson et al. (2003) state that sticky costs arise because there are asymmetric frictions in making resource adjustments. Adjustment costs arise because firms have to match resources to activity levels, when demand declines committed resources remain unutilized unless managers make decisions to reduce them. Since activity levels are not stable managers might be inclined to not reduce committed resources when they expect activity levels to restore again. In this case managers will decide to retain unutilized resources rather than incur adjustment costs resulting in costs behaving asymmetrically (Anderson et al., 2003). Moreover, managers’ decision to maintain committed resources may also be triggered by personal considerations and result in agency costs (Anderson et al., 2003). Agency costs occur because self-interested managers make decisions that maximize their personal utility but are not in the best interest of the firms’ shareholders (Jensen and Meckling, 1976). For example, managers could maintain committed resources to avoid a loss of status when their department is downsized or the anguish of dismissing familiar employees (Anderson et al., 2003).

(5)

firms can be extrapolated to Dutch firms, given the institutional differences between both countries.

1.2 Research question

The research question of this paper is:

To what extent are selling, general, and administrative costs of Dutch firms sticky?

Consistent with prior research, the focus for the cost stickiness analysis will be on SG&A costs (Anderson et al., 2003; Calleja et al., 2006; Dalla Via and Perego, 2014; Venieris et al., 2015; Subramanian and Weidenmier, 2016;). SG&A costs are costs other than those related to the production that are needed to support revenues and overall operations during an accounting period. For example, salary expenses, advertising expenses, and other administrative expenses. According to the literature, SG&A costs are adaptable by managers and therefore influenced by managerial behavior (Dalla Via and Perego, 2014).

1.3 Motivation and contribution

Firstly, this paper contributes to the growing stream of literature regarding cost stickiness. Secondly and foremost, it contributes to the literature that examines international differences in cost stickiness. It is argued in this paper that because of differences in employee protection and governance systems the degree of cost stickiness for U.S. firms may differ from the degree of cost stickiness for Dutch firms. Thirdly, this thesis also features a societal contribution because answering the research question has implications for auditors and other finance professionals who evaluate cost changes in relation to changes in revenues (Anderson et al., 2003). For example, Messier et al. (2013) states that auditors assume that costs move proportionately with revenues when they perform analytical review procedures. Therefore analytical procedures may be improved.

1.4 Structure of the thesis

The structure of this paper will be as follows. Section two describes the prior literature and hypothesis development. The research methodology is described in section three, including a description of the sample selection and a description of the empirical models. Moreover, section four describes the results of the different analyses used. Finally, section five describes the conclusion, discussion, and the limitations of the study.

(6)

2 Literature review

2.1 Challenging the traditional cost model

Understanding cost behavior is a fundamental element of cost and management accounting. In the traditional cost model costs are described as fixed or variable with respect to changes in activity volume. Regarding this cost model, Noreen (1991) states that variable costs change proportionately with changes in the activity. In other words, the magnitude of a change in costs only depends on the level of change in the activity and not on the direction of the change.

In 1994, Noreen and Soderstrom investigated the condition of proportionality of costs as provided by Noreen (1991). They examined the proportionality of overhead costs in U.S. hospitals. However, the hypothesis was not supported by the majority of the indirect cost accounts of the hospitals, suggesting that overhead costs do not behave proportional to changes in activity. Following their prior research, Noreen and Soderstrom (1997) examined the proportionality of overhead costs based on a new model using time-series data. Nevertheless, their empirical results did not provide substantive evidence for an asymmetry in cost behavior.

2.2 Cost stickiness

Anderson et al. (2003) were first to provide robust empirical evidence on asymmetrical behaving costs. They examined the behavior of SG&A costs in relation to revenue using a sample of 7.629 industrial firms over a 20 years period, from 1979 until 1998. Anderson et al. (2003) concluded that SG&A costs increase 0,55 percent per 1 percent increase in revenue but decrease only 0,35 percent per 1 percent decrease in revenue. Their findings show that an increase in costs associated with an increase in activity is greater than a decrease in costs associated with an equivalent decrease in activity. Anderson et al. (2003) label this type of cost behavior as “sticky costs” or “cost stickiness”.

The prevalence of cost stickiness is consistent with an alternative model of cost behavior in which managers deliberately adjust resources in response to changes in volume (Anderson et al., 2003). This model makes a distinction between costs that move mechanically with changes in volume and costs that are determined by the resources committed by managers. Anderson et al. (2003) state that if there is uncertainty about future demand and firms must incur adjustment costs to restore or reduce committed resources, managers may purposely delay reductions to committed resources until they are more certain about the persistence of a decline in demand. Especially, when managers enter into contracts for resources that are costly to break or to renegotiate (Calleja et al., 2006). In contrast, resources like direct materials can be adjusted without incurring

(7)

adjustment costs (Banker and Byzalov, 2014). In addition, Calleja et al. (2006) show that sticky costs observed in one period reverses in a subsequent period and that sticky costs are less pronounced when the observation period increases.

Adjustment costs arise when firms remove committed resources and replace those resources when demand is restored (Anderson et al., 2003). For instance, if demand increases, managers increase committed resources to the level necessary to accommodate additional revenue. However, if volume decreases, some committed resources will not be utilized unless managers make the decision to remove them. Because demand is not stable, managers assess the probability that a drop in demand is temporary when deciding whether to adjust committed resources downward. Cost stickiness arises when managers rather decide to retain unutilized resources than incur adjustment costs if volume decreases (Anderson et al., 2003). Thus for example, when a manager decides to retain unutilized resources rather than incur adjustment cost due to high adjustment costs, a firm might report a drop in revenues while costs will not fall in the same proportion (Calleja et al., 2006). According to Anderson et al. (2003) cost stickiness occurs, because there are asymmetric frictions in making resource adjustments, meaning that there are forces acting to slow the downward adjustment process more than the upward adjustment process. They state that adjustment costs include things as, for example, severance pay when employees are dismissed or contract-breakage costs when contracts for resources are broken.

The decision to retain unutilized resources might also be explained by agency problems related to the agency theory. The agency theory suggests that self-interested managers make decisions that maximize their personal utility but are not in the best interest of the firms’ shareholder (Jensen and Meckling, 1976). For instance, managers could retain committed resources to avoid a loss of status when their department is downsized or the anguish of dismissing familiar employees (Anderson et al., 2003).

2.3 Variations in cost stickiness

As mentioned in the previous section, prior studies focus on economic factors and agency problems that explain the variation in cost stickiness. In this section literature with regards to variations in cost stickiness will be discussed.

2.3.1 Economic factors

Expanding the literature on cost stickiness, Calleja et al. (2006) performed an international comparison of cost stickiness. They investigated the cost behavior of operating costs, using a sample of (number) listed firms from the U.K., the U.S., French, and German markets. They

(8)

concluded that operating costs increase, on average, by 0,97 percent per 1 percent increase in revenues, but decrease by only 0,91 percent per 1 percent decrease in revenues. The countries share common characteristics: costs are sticky, nevertheless they are less sticky when aggregated over longer periods and when firms suffer larger drops in revenue. However, operating costs are stickier for French firms and German firms than for U.S. firms and U.K. firms. According to Calleja et al. (2006) this difference in cost stickiness is attributable to variances in corporate governance systems and managerial oversight.

Balakrishnan and Gruca (2008) examine intra-firm variations in cost stickiness, using department level data from Canadian hospitals. They conclude that hospital administrators are reluctant to reduce costs in core activities related to direct patient care because of the critical nature of these services to the hospital’s mission. This conclusion leads to the suggestion that the extent to which a function represents the organization’s core activities influences the cost stickiness.

Banker et al. (2013) investigate the effect of employment protection legislation (EPL) in different countries on cost stickiness, using a sample of 19 OECD countries over the period 1990 until 2008. They conclude that firms in countries with stricter employment protection exhibit a greater degree of cost stickiness. Their findings are consistent with Calleja et al. (2006), they state that firms in countries with a strong judicial system show higher cost stickiness because this forces a company into long-term commitments.

Dalla Via and Perego (2014) examine whether asymmetric cost behavior occurs in small and medium sized firms, using a sample of Italian firms over the period 1999 until 2008. They conclude that cost stickiness exhibits only for the total cost of labor and the operating costs for listed firms. Whereas SG&A costs, COGS, and the operating costs for non-listed firms are not sticky. Dalla Via and Perego (2014) state that their results are aligned with prior findings when only listed firms are investigated. However, when they examine the cost stickiness of small and medium sized firms their results are not consistent with prior literature.

Venieris et al. (2015) investigate the relationship between the level of a firm’s intangible investments in organization capital and the cost behavior of SG&A costs, using a sample of 55.769 firm-year observations of U.S. listed firms over the period 1979 until 2009. They state that firms with high organization capital exhibit SG&A cost-stickiness behavior. This paper contributes to the cost stickiness literature by recognizing a firm’s intangible investment intensity as a causal factor of the cost stickiness phenomenon.

Subramanian and Weidenmier (2016) examine whether the magnitude of the change in activity is related to cost stickiness. Particularly, they examine how SG&A costs and Costs of Goods Sold (COGS) behave for different ranges of revenue activity changes. Their study is based

(9)

on a sample of 9.562 firms and 22 years of annual data for the period of 1979 until 2000. They conclude that for small revenue changes SG&A costs and COGS do not show sticky cost behavior. However, when the change in revenue is more than 10 percent, SG&A costs and COGS do show sticky cost behavior. According to Subramanian and Weidenmier (2016) this is caused by the asymmetrical response of managers to large changes of market demand. They state that a large increase in activity causes an immediate increase of costs whereas large decreases in activity not result in an immediate decrease of costs. This is due to the fact that firms cannot reduce employees, assets, and/or other costs in the short-term. In addition, they observe inter-industry differences in cost behavior of SG&A costs and COGS.

2.3.2 Agency problems

Chen et al. (2012) were first to examine whether agency factors drive SG&A costs behavior, using a sample of 1.500 U.S. firms over the period 1996 until 2005. They conclude that cost asymmetry is positively associated with managers’ empire building incentives because of the agency problem. Besides, they conclude that the positive association between the agency problem and the cost stickiness of SG&A costs is more prominent under weak corporate governance. Thereby suggesting that strong corporate governance systems mitigates the effect of the agency problem on managers’ resource adjustment decision, which leads to a decrease in cost stickiness.

Dierynck et al. (2012) investigate the influence of managerial incentives on labor cost behavior for private Belgian firms. Using a sample of 37.880 firm-year observations over the period 1995 until 2006, they conclude that firms just meeting or beating the zero earnings benchmark actually exhibit cost symmetry. In contrast, firms that do not face pressure to meet or beat the zero earnings benchmark exhibit asymmetric labor cost behavior, which is consistent with previous literature on SG&A costs (Anderson et al., 2003). According to Dierynck et al. (2012), this difference is attributable to the underlying managerial decisions. They state that firms that report healthy profit react to changes in activity by changing the number of hours that an employee works, since such an approach limits the loss to their reputation. On the other hand, managers of firms that meet or beat the zero earnings benchmark adjust to activity changes by dismissing employees, which results in direct cost reductions.

Kama and Weiss (2013) examine the impact of incentives to meet earnings targets on cost stickiness, using a sample of all public firms covered by Compustat over the period 1979 until 2006. They conclude that when managerial decisions are driven by incentives to avoid losses or to meet earnings targets, managers adjust resources faster when activity decreases than they do when it increases. Moreover, their findings show that managers facing incentives to meet earnings targets

(10)

are predicted to decrease resources even if they expect future demand growth and these resources are likely to be required again.

2.4 Hypothesis devolpment

There are institutional differences between U.S. firms and Dutch firms that affect the cost behavior of firms. These institutional differences relate to corporate governance and employment protection.

Corporate governance

Dutch firms are operating under a two tier corporate governance system while U.S. firms are operating under a one tier corporate governance system. According to Calleja et al. (2006), the one tier system puts more emphasis on the notion of shareholder maximization and on the role of the equity market. They state that the equity market is used as an instrument through which the market for corporate control operates to discipline underachieving management. Therefore the management of U.S. firms exhibit external pressure to make decisions in the best interests of shareholders. In contrast, two tier systems are directed at a coalition of external and internal stakeholders, rather than being exclusively directed at shareholders (Calleja et al., 2006). Calleja et al. (2006) state that the acknowledgment of a wider range of stakeholders carries over to an enlarged role of co-determination between management, workers, and shareholders.

O’Sullivan (2003) investigates the two tier corporate governance system in France. She concludes that it provides more social protection to their workers compared to the one tier system. She states that French firms therefore must incur substantial costs when they want to downsize their labor forces. Calleja et al. (2006) states that due to the high costs to downsize resources, the level of external oversight, and the focus on stakeholder value managers of two tier corporate governance systems managers are inclined to rather retain resources than incur adjustment costs with cutting resources levels. Based on this statement the expectation is that Dutch firms will have a higher degree of cost stickiness than U.S. firms.

Employment protection

Dutch firms have a stricter level of employment protection legislation (EPL) compared to U.S. firms. EPL strictness is often used in prior economic research as a proxy for labor adjustment costs (Pissarides, 1999; Blanchard and Portugal, 2001; Banker et al., 2013). According to Banker et al. (2013), firms in countries with stricter employment protection exhibit a higher level of cost stickiness since stricter EPL will lead to a greater downward adjustment for labor costs. They retrieved the EPL indexes of 19 countries from the OECD Employment Outlook 2004. According

(11)

to the OECD Employment Outlook 2013, the U.S. have an EPL of 1,17 while the Netherlands have an EPL of 2,94. Thus, this implies that labor adjustment costs for Dutch firms is higher than for U.S. firms. Based on these numbers the expectation is that Dutch firms will exhibit a greater degree of cost stickiness compared to U.S. firms.

Hypothesis

Concluding, there are several institutional differences between U.S. firms and Dutch firms that affect the cost behavior of firms. These differences predict a higher degree of cost stickiness. Therefore the aim of this paper is to empirically examine to what extent SG&A costs are sticky for Dutch firms. To conclude, the hypothesis that will be tested is:

H1: The increase in SG&A costs is higher for an increase in revenue than the reduction of SG&A costs for an equivalent decrease in revenue.

(12)

3 Method

3.1 Sample selection

The initial sample of Dutch firms will be obtained from the Compustat Global database for the period 2004 until 2014. However, data from 2004 is obtained to determine the revenue growth and cost growth in 2005. The variables downloaded from the Compustat Global are: Current ISO Country Code (LOC), Standard Industry Classification Code (SIC), Data Fiscal Year (FYEAR), Total Revenue (REVT), and Selling, General and Administrative Expense (XSGA). The extraction of data resulted in an initial sample of 2.010 firm-year observations. Firstly, financial institutions are dropped using the variable SIC (standard industry classification code) since these firms have deviating financial statements (Banker et al., 2011; Kama and Weis, 2013). This resulted in a removal of 457 firm-year observations. Secondly, 371 firm-year observations are removed from the sample due to missing observations of SG&A costs, revenue, and negative revenue. Thirdly, firm-year observations where SG&A costs are greater than revenues are excluded. Subsequently, firms with headquarters outside the Netherlands are removed using the variable LOC (country code headquarters). This resulted in a loss of 198 firm-year observations. Furthermore, to extract mailbox firms from the sample, firms with their Board of Directors located outside the Netherlands are removed. This resulted in a removal of 192 firm-year observations. To remove the effects of mergers and acquisitions observations are removed in which the revenue changes by more than 50 percent in consecutive years (Calleja et al., 2006; Venieris, 2015; Subramaniam and Weidenmier, 2016). In addition, very small companies with revenues below €5 million are excluded since it is questionable that they have a well-defined cost structure (Dalla Via and Perego, 2014; Venieris et al., 2015). This leads to a final sample of 691 firm-year observations. Table 1 summarizes the sample selection process. In addition, the top and bottom 1 percent of the variables SG&A costs and revenue are winsorized.

Table 1. Sample selection

Observations

Initial sample Compustat Global 2.010

Less: observations from financial institutions (SIC 6000-6999) 457 Less: missing observations SG&A costs, revenue, and negative revenue 371

Less: SG&A costs exceed revenue 34

Less: firms with headquarter outside the Netherlands 198

Less: Board of Directors outside the Netherlands 192

Less: change in revenue of 50% 57

Less: revenue lower than € 5 million 10

(13)

3.2 Cost stickiness model

In accordance with preceding literature regarding asymmetric cost behavior, the empirical model as developed by Anderson et al. (2003) will be used for testing cost stickiness within the dataset (Chen et al., 2012; Dalla Via and Perego, 2014; Venieris et al., 2015).

Model 1: 𝑙𝑜𝑔 [ 𝑆𝐺&𝐴𝑖,𝑡 𝑆𝐺&𝐴𝑖,𝑡−1 ] = 𝛽0+ 𝛽1𝑙𝑜𝑔 [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡−1 ] + 𝛽2∗ 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ 𝑙𝑜𝑔 [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡−1 ] + 𝛽3∗ 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝐷𝑢𝑚𝑚𝑦 + 𝜀𝑖,𝑡

In model 1 log [ 𝑆𝐺&𝐴𝑖,𝑡

𝑆𝐺&𝐴_{𝑖,𝑡−1}]andlog [

𝑅𝑒𝑣𝑒𝑛𝑢𝑒_𝑖,𝑡

𝑅𝑒𝑣𝑒𝑛𝑢𝑒_{𝑖,𝑡−1}]represent the percentage change in costs and activity

levels for firm i in year t, relative to year t-1. The logarithm functions in the cost stickiness formula improve comparability of the variables across firms and reduces potential heteroscedasticity (Anderson et al., 2003). 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝐷𝑢𝑚𝑚𝑦𝑖,𝑡 is a dummy variable that is equal to one if revenue of

firm i decrease in year t and zero if the change in revenue is equal or greater than zero. 𝜀𝑖,𝑡

represents the error term. Anderson et al. (2003) state that cost stickiness occurs when the slope of an increase in revenue (𝛽1) is larger than the slope for a decrease in revenue (𝛽1+ 𝛽2), hence

𝛽2< 0.

As a proxy for change in costs is applied the change in SG&A costs. Prior literature regarding cost stickiness indicates that several cost categories behave asymmetric, however SG&A costs are used most frequently in testing for cost stickiness. Chen et al. (2012) state that SG&A costs are of substantial importance, since it represents a significant proportion of costs of business operations. They conclude that, on average the ratio of SG&A costs to total assets is 27 percent while the ratio of, for example, R&D costs to total assets is 3 percent. Furthermore, Anderson et al. (2003) state that SG&A costs make up 26,4 percent of revenue and therefore are observed to be of significant importance. As a proxy for change in activity levels is applied the change in revenue. This is based on preceding literature with regards to cost asymmetry (Anderson et al. 2003; Calleja et al., 2006; Chen et al., 2012; Dalla Via and Perego, 2014; Venieris et al., 2015).

(14)

3.3 Control variables

Consistent with prior literature asset intensity and employee intensity are included as control variables in supplemental analyses. According to Anderson et al. (2003) these economic factors increase the degree of cost stickiness. Asset intensity is calculated as the total assets divided by revenue. Furthermore, employee intensity is calculated as the number of employees divided by revenue. Since for 73 firm-year observations data on employee intensity is missing, hand collected data is used to fill the missing observations. In model 2 the control variables are added to the equation of model 1. Model 2: log [ 𝑆𝐺&𝐴𝑖,𝑡 𝑆𝐺&𝐴𝑖,𝑡−1 ] = 𝛽0+ 𝛽1log [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡−1 ] + 𝛽2∗ 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ log [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡−1 ] + 𝛽3∗ 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ log [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡−1 ] ∗ 𝐴𝑠𝑠𝑒𝑡𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 + 𝛽4∗ 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ log [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡−1 ] ∗ 𝐸𝑚𝑝𝑙𝑜𝑦𝑒𝑒𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 + 𝛽5∗ 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝐷𝑢𝑚𝑚𝑦𝑖,𝑡 + 𝛽6∗ 𝐴𝑠𝑠𝑒𝑡𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 + 𝛽7∗ 𝐸𝑚𝑝𝑙𝑜𝑦𝑒𝑒𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 + 𝜀𝑖,𝑡

(15)

4 Results

4.1 Descriptive results

Table 2 provides descriptive statistics on revenue and SG&A costs for the firms within the sample. These firms have a mean value of €3.532,89 million in revenue, where the median is €923,50 million. On average, the SG&A costs for these firms are €778,46 million, with a median of €134,36 million. The average growth in revenue (revenue of yeart divided by revenue of yeart-1) is 1,04.

However, the average change in SG&A (SG&A costs of yeart divided by SG&A costs of yeart-1) is

1,39. SG&A costs as a percentage of revenues, averages at 23 percent with a median of 19 percent.

Table 2. Descriptive results

Mean SD 10% 25% 50% 75% 90%

Revenue (€ mil) 3.532,89 7.938,29 35,84 158,05 923,50 2.572,19 8.834,77 SG&A costs (€ mil) 778,46 2.117,07 8,31 24,94 134,36 386,80 1.597,00 Revenuei,t/revenuei,t-1 1,04 0,14 0,88 0,97 1,03 1,11 1,24

SG&Ai,t/SG&Ai,t-1 1,39 4,40 0,86 0,97 1,03 1,11 1,24

SG&Ai,t/Revenuei,t 0,23 0,15 0,06 0,12 0,19 0,29 0,47

Table 3 summarizes the sample composition over industries based on a classification of SIC codes. Moreover, corresponding mean statistics on revenue, SG&A costs, and SG&A costs as a percentage of revenues are described. Within the sample there are three major industries represented: manufacturing industry (50,22 percent), services industry (24,02 percent), and wholesale-retail trade industry (15,92 percent). The sample is therefore concentrated on manufacturing firms.

Table 3. Descriptive statistics per industry (n = 691)

Industry SIC n (%) revenue Mean

(€ mil) Mean SG&A costs (€ mil) Mean SG&A/ Revenue Mining-Construction 10-19 35 (5,1) 3.971,15 531,16 0,12 Manufacturing 20-39 347 (50,2) 4.256,15 1.054,92 0,22 Transportation 40-49 33 (4,78) 1.957,70 196,54 0,23 Wholesale-Retail trade 50-59 110 (15,9) 5.072,16 1.051,44 0,21 Services 70-89 166 (24,0) 1.254,25 201,81 0,27 Total 691

(16)

4.2 Main findings

Table 4 provides the regression summary statistics of model 1. Based on the sample of 691 firm year observations over a period from 2005 until 2014 the results indicate a significant presence of cost stickiness, since𝛽1 = 1,23 (p<0,01) and 𝛽2 = -0,44 (p<0,10). Consisted with the findings of

Anderson et al. (2003), an increase in revenue of 1 percent results in an increase in SG&A costs of 1,23 percent, while a decrease in revenue of 1 percent leads to a decrease in SG&A costs of 0,79 percent. Therefore the hypothesis (H1) is supported, since an increase in SG&A costs is higher for an increase in revenue than the reduction of SG&A costs for an equivalent decrease in revenue. However, in contrast with prior studies, an increase in revenue leads to a relative higher increase in SG&A costs.

The model is highly significant as shown by the F-value of 44,48 (p<0,01). Furthermore, the model has an adjusted R2_{of 19,44 percent which implies that the model explains a small part}

of the variance in the dependent variable. Overall, the empirical analysis provides support for the hypothesis. However, the relative high increase in SG&A cost when revenue increases is in contrast with prior studies. Since, prior studies state that, when revenue increases, the increase in revenue is higher compared to the increase in SG&A costs.

Table 4. Summarized model 1 OLS regression results

Variables Predicted

sign Coef. t-stat p-value

β0: Intercept -0,02 -0,75 0,451 β1: 𝐿𝑜𝑔 [_{𝑅𝑒𝑣𝑒𝑛𝑢𝑒}𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑖,𝑡−1] + 1,23*** 7,81 0,000 β2: 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒_𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ 𝑙𝑜𝑔 [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒_𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡−1] - -0,44* -1,88 0,06 β3: 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒_𝐷𝑢𝑚𝑚𝑦𝑖,𝑡 0,04 1,23 0,22 Number of observations 691 F-value 44,48 P-value 0,0000 R-squared 0,1944 Adjusted R-squared 0,1900

I performed a range of robustness analyses.

First, I repeat the analysis on model 1, but now use robust regressions. Estimation biases due to influential statistics, with regards to outliers and high leverage data points, are limited by executing robust regression estimates for both the models. The robust regression applies the weight of observations differently based on how they behave (UCLA: Statistical Consulting Group, 2016). According to the UCLA: Statistical Consulting Group (2016), a robust regression identifies and drops the influential points in the sample using the distance of Cook. Then, cases with outsized

(17)

absolute residuals are down-weighted while using a recurrent process of calculating weights based on residuals, using Huber weighting and bi-weighting. Table 5 presents the robust regression summary statistics of model 1. The evidence found also supports the hypothesis, however, the robust regression results report coefficients not consistent with the OLS regression results. The 𝛽1 coefficient in table 5 is significantly positive at 0,78 (p<0,01), while the 𝛽2 coefficient is

significantly negative at 0,12 (p<0,05). Meaning that an increase in revenues of 1 percent results in an increase in SG&A costs of 0,78 percent, while a decrease in revenues of 1 percent leads to a decrease in SG&A costs of 0,66 percent. Thus, the robust regression coefficients of model 1 are more consistent with prior studies. Based on the large differences in betas we conclude that the dataset contains outliers. Observations are defined influential when eliminating the observation changes the estimation results. Furthermore, outliers are considered observations that are numerically distant from the rest of the data (Barnett and Lewis, 1994). Leverage is a measure of to what extent an independent variable deviates from its mean. When an observation has an extreme value on a predictor variable it is defined as a point with high leverage (Everitt, 2006).

Table 5. Summarized model 1 robust regression results Variables

Predicted sign

Coef. t-stat p-value

β0: Intercept -0,00 -0,61 0,542 β1: 𝐿𝑜𝑔 [_{𝑅𝑒𝑣𝑒𝑛𝑢𝑒}𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑖,𝑡−1] + 0,78*** 15,66 0,000 β2: 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒_𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ 𝑙𝑜𝑔 [_{𝑅𝑒𝑣𝑒𝑛𝑢𝑒}𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑖,𝑡−1] - -0,12** -2,33 0,020 β3: 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒_𝐷𝑢𝑚𝑚𝑦𝑖,𝑡 0,19* 1,80 0,073 Number of observations 691 F-value 218,91 P-value 0,0000

Second, I repeat the analysis, but now including the control variables asset intensity and employee intensity. According to Anderson et al. (2003) employee intensity and asset intensity are proxies for adjustment costs and therefore should increase the level of cost stickiness. In the robustness analysis we conclude that the dataset contains outliers. Therefore we use a robust regression to investigate to whether the control variables affect cost stickiness. Table 6 presents the robust regression summary statistics of model 2. The robust regression results are consistent with the OLS regression results of model 1. The 𝛽1 coefficient in table 6 is significantly positive

at 0,86 (p<0,01), while the 𝛽2 coefficient is significantly negative at 0,21 (p<0,01). Meaning that

an increase in revenues of 1 percent results in an increase in SG&A costs of 0,86 percent, while a decrease in revenues of 1 percent leads to a decrease in SG&A costs of 0,65 percent. Furthermore,

(18)

the robust regression results report negative coefficients for the control variables 𝛽3 and 𝛽4.

However, only the control variable asset intensity is significant. Furthermore, in contrast to prior studies, the control variable employee intensity is not significant.

Table 6. Summarized model 2 robust regression results

Variables Predicted sign Coef. t-stat p-value

β0: Intercept 0,16* 1,84 0,066 β1: 𝐿𝑜𝑔 [_{𝑅𝑒𝑣𝑒𝑛𝑢𝑒}𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑖,𝑡−1] + 0,86*** 14,16 0,000 β2: 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒_𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ 𝑙𝑜𝑔 [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒_𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒_{𝑖,𝑡−1}] - -0,21*** -2,64 0,009 β3: 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ log [_{𝑅𝑒𝑣𝑒𝑛𝑢𝑒}𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑖,𝑡−1] ∗ 𝐴𝑠𝑠𝑒𝑡𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 - -0,10** -2,43 0,015 β4: 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ log [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡−1] ∗ 𝐸𝑚𝑝𝑙𝑜𝑦𝑒𝑒𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 - -0,05 -0,01 0,991 β5: 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝐷𝑢𝑚𝑚𝑦𝑖,𝑡 0,01 1,00 0,320 β6: 𝐴𝑠𝑠𝑒𝑡𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 0,01 0,99 0,323 β7: 𝐸𝑚𝑝𝑙𝑜𝑦𝑒𝑒𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 -0,04*** -4,90 0,000 Number of observations 691 F-value 102,84 P-value 0,0000

Third, I repeat the analysis described by model 1 but now run the analysis separately for each industry. Third, Subramaniam and Weidenmier (2016) state that cost behavior varies across industries. Their results show that some industries exhibit cost stickiness while other industries do not, therewith implying that it is important to consider differences in industry when analyzing cost behavior. To control for industry differences, separate robust regressions have been estimated for the three major industries in the dataset as provided by table 3. The three major industries are manufacturing (SIC 20-39), wholesale and retail trade (SIC 50-59), and services (SIC 70-89). Table 7 provides the robust regression results of model 1. The results of table 7 indicate a significant presence of cost stickiness for the services industries, since 𝛽1= 0,80 (p<0,01) and 𝛽2 = -0,46

(p<0,01). However, no significant cost stickiness results are found for the other industries. Thereby demonstrating that that the results are mostly driven by services firms. The results are in contrast with prior literature. Subramamaniam and Weidenmier (2016) state that manufacturing firms exhibit more cost stickiness since manufacturing firms have relatively more fixed assets and therefore have higher adjustment costs. They conclude that fixed assets overall have high adjustment costs and therefore it is likely that managers are less willing to adjust fixed assets to short-term variations in revenue. In contrast, wholesale and retail trade and service firms have

(19)

relatively lower levels of fixed assets and therefore are better able to respond quickly to variations in revenue.

Table 7. Summarized model 1 robust regression results per industry

Manufacturing Sic: 20-39

Whole sale & retail trade

Sic: 50-59 Sic 70-89 Services Variables Pred. sign Coef. t-stat Coef. t-stat Coef. t-stat

β0: Intercept 0,01 0,89 0,02 1,12 -0,02 -1,03 β1: 𝐿𝑜𝑔 [_{𝑅𝑒𝑣𝑒𝑛𝑢𝑒}𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑖,𝑡−1] + 0,54*** 6,51 0,74*** 6,03 0,80*** 7,71 β2: 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒_𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ 𝑙𝑜𝑔 [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒_{𝑖,𝑡−1}] - 0,03 0,26 -0,20 -1,19 -0,46*** -2,55 β3: 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒_𝐷𝑢𝑚𝑚𝑦𝑖,𝑡 -0,01 -0,70 -0,03 -1,38 -0,00 -1,03 Number of observations 347 110 166 F-value 59,99 72,61 39,46 P-value 0,0000 0,0000 0,0000 4.3 Additional analysis

In addition, I performed an additional analysis. The analysis tests whether the level of cost stickiness differs between Dutch firms and U.S. firms. Based on differences in corporate governance and employee protection the expectation is that Dutch firms exhibit more cost stickiness. To test the difference of cost stickiness, the initial sample of Dutch firms is matched with similar U.S. firms. Data with regards to U.S. firms is obtained from the Compustat database for the period 2004 until 2014. Moreover, the sample selection process as described in paragraph 3.1 has been applied to the sample of U.S. firms. Table 8 summarizes the process. The U.S. firms are matched to the Dutch firms according to the matching criteria of Barber and Lyon (1996). They state that matched firms need to have the same two-digit SIC code. Besides, their total assets (proxy for size) need to be between 70 percent and 130 percent of a Dutch firm’s total assets. Furthermore, the firms are matched once, based on their size and SIC code in 2004. The matching procedure resulted in a sample of 1.556 firm-year observations.

(20)

Table 8. Sample selection U.S. firms

Observations

Initial sample Compustat Global 123.878

Less: observations from financial institutions (SIC 6000-6999) 38.443 Less: missing observations SG&A costs, revenue, and negative revenue 24.128

Less: SG&A costs exceed revenue 6.418

Less: firms with headquarter outside the United States 15.075

Less: change in revenue of 50% 2.602

Less: revenue lower than € 5 million 1.008

Total firm-year observations 36.204

Empirical model 3 is used for testing the first additional analysis. The interpretation of model 3 is summarized in table 9. Since, the prediction is that Dutch firms exhibit more cost stickiness than U.S. firms, 𝛽2 + 𝛽5 is predicted to be smaller than 𝛽2. Therefore 𝛽5 should be smaller than zero.

Thus, 𝛽2 + 𝛽5 < 𝛽2 therefore 𝛽5 < 0. Model 3: log [ 𝑆𝐺&𝐴𝑖,𝑡 𝑆𝐺&𝐴𝑖,𝑡−1 ] = 𝛽0+ 𝛽1log [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡−1 ] + 𝛽2∗ 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ log [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡−1 ] + 𝛽3∗ 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝐷𝑢𝑚𝑚𝑦𝑖,𝑡+ 𝛽4∗ 𝐷𝑢𝑡𝑐ℎ𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ log [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡−1 ] + 𝛽5∗ 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ 𝐷𝑢𝑡𝑐ℎ𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ log [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡−1 ] + 𝛽6∗ 𝐷𝑢𝑡𝑐ℎ𝐷𝑢𝑚𝑚𝑦𝑖,𝑡+ 𝜀𝑖,𝑡

Table 9. Interpretation of model 3

U.S. firms (DutchDummy=0) Dutch firms (DutchDummy=1) Activity increase (DecreaseDummy = 0) 𝛽1 𝛽1+ 𝛽4 Activity decrease (DecreaseDummy=1) 𝛽1+ 𝛽2 𝛽1+ 𝛽2+ 𝛽4+ 𝛽5

Table 10 provides the regression summary statistics of model 3. Based on the sample of 1.556 firm-year observations over a period from 2004 until 2014 the results, as expected, indicate a negative coefficient for 𝛽5. However, it is not significant. Concluding, in contrast to the

(21)

Table 10. Summarized model 3 regression results

Variables Predicted sign Coef. t-stat p-value

β0: Intercept 0,02 1,54 0,124 β1: 𝐿𝑜𝑔 [_{𝑅𝑒𝑣𝑒𝑛𝑢𝑒}𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑖,𝑡−1] + 0,58*** 7,25 0,000 β2: 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒_𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ 𝑙𝑜𝑔 [_{𝑅𝑒𝑣𝑒𝑛𝑢𝑒}𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑖,𝑡−1] - -0,32*** -3,42 0,001 β3: 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒_𝐷𝑢𝑚𝑚𝑦𝑖,𝑡 -0,02 -0,79 0,428 β4: 𝐷𝑢𝑡𝑐ℎ𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ log [_{𝑅𝑒𝑣𝑒𝑛𝑢𝑒}𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖,𝑡 𝑖,𝑡−1] + 0,59*** 3,66 0,000 β5: 𝐷𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ 𝐷𝑢𝑡𝑐ℎ𝐷𝑢𝑚𝑚𝑦𝑖,𝑡∗ log [ 𝑅𝑒𝑣𝑒𝑛𝑢𝑒_𝑖,𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒_{𝑖,𝑡−1}] - -0,23 -0,89 0,372 β6: 𝐷𝑢𝑡𝑐ℎ𝐷𝑢𝑚𝑚𝑦𝑖,𝑡 -0,02 -0,85 0,397 Number of observations 1.556 F-value 45,19 P-value 0,0000 R-squared 0,1714 Adjusted R-squared 0,1676

(22)

5 Conclusion

This paper investigates whether and to what extent SG&A costs are sticky for Dutch firms. Prior literature with regards to cost stickiness mainly focusses on data from U.S. firms. But, there are reasons to expect that these results will not hold in the Netherlands. These reasons relate to differences in employee protection and corporate governance systems. Based on the literature it is expected that Dutch firms exhibit cost stickiness. In addition, it is expected that Dutch firms exhibit a higher degree of cost stickiness compared to U.S. firms.

The findings are consistent with the prediction. Using a sample of 691 observations over a period from 2005 until 2014 the results show that SG&A costs exhibit cost stickiness. An increase in revenue of 1 percent results in an increase in SG&A costs of 0,78 percent, while a decrease in revenue of 1 percent leads to a decrease in SG&A costs of 0,66 percent. However, in contrast to prior studies, the results show that only asset intensity increases the level of cost stickiness. Furthermore, the additional analyses results show that, in contrast of expectation, Dutch firms do not exhibit more cost stickiness than U.S. firms.

This paper contributes to the growing stream of literature with regards to cost stickiness. As far as known to the author, this study is first to investigate cost stickiness in the Netherlands. Thereby, extending prior literature on cost stickiness that investigates international differences. Furthermore, this study features a societal contribution since the findings have implications for managers and corporate decision makers. For instance, decisions based on the traditional cost model will underestimate or overestimate the sensitivity of SG&A costs to changes in revenue. Moreover, an understanding of asymmetric cost behavior can lead to a better planning and control system since it increases the knowledge with regards to general behavior of costs. The effects of cost stickiness can be diminished when managers are able to identify and manage unused resources and unused capacity. Other ways of avoiding asymmetric cost behavior are focusing on the marketing to increase demand or move unutilized resources to other activities.

Results of this paper are subject to a few limitations. Firstly, the sample used in this study consists of 100 Dutch listed firms and therefore the results cannot be extended to non-Dutch firms or non-listed Dutch firms. Secondly, the use of revenue as a proxy for activity is a common approach in prior studies, however, the results should be interpreted carefully. Revenue is not only influenced by management decisions, but also by changes in price as well as by other factors. Thirdly, the time horizon of the data is at best 10 years. Since cost stickiness appears over time, the time horizon of this study could be a limitation affecting the results.

In order to improve the literature on cost stickiness, a change in empirical approach is recommended to examine the factor causing cost stickiness and the managerial actions inducing

(23)

from it. Hence, the use of field studies or experimental methods depending on interviews to managers that establish this type of behavior. A different empirical approach may lead to a better understanding of how managerial actions and attitudes affect cost behavior. Furthermore, a different empirical approach would also be useful to examine whether the determinants investigated by prior studies address the underlying causes of asymmetric cost behavior.

(24)

References

Anderson, M. C., Banker, R. D., & Janakiraman, S. N. (2003). Are selling, general, and administrative costs “sticky”? Journal of Accounting Research, 41(1), 47-63.

Balakrishnan, R., & Gruca, T. (2008). Cost stickiness and core competency: A Note. Contemporary

Accounting Research, 25, 993-1006.

Balakrishnan, R., Petersen, M. J., & Soderstrom, N. S. (2004). Does capacity utilization affect the “stickiness” of cost? Journal of Accounting, Auditing & Finance, 19(3), 283-300.

Banker, R. D., & Byzalov, D. (2014). Asymmetric cost behavior. Journal of Management Accounting

Research, 26(2), 43-79.

Banker, R. D., Byzalov, D., & Chen, L. T. (2013). Employment protection legislation, adjustment costs and cross-country differences in cost behavior. Journal of Accounting and Economics,

55(1), 111-127.

Barber, B. M., & Lyon, J. D. (1996). Detecting abnormal operating performance: The empirical power and specification of test statistics. Journal of financial Economics, 41(3), 359-399. Blanchard, O., & Portugal, P. (2001). What hides behind an unemployment rate: comparing

Portuguese and US labor markets. American Economic Review, 187-207.

Calleja, K., Steliaros, M., & Thomas, D. C. (2006). A note on cost stickiness: Some international comparisons. Management Accounting Research, 17(2), 127-140.

Chen, C. X., Lu, H., & Sougiannis, T. (2012). The Agency Problem, Corporate Governance, and the Asymmetrical Behavior of Selling, General, and Administrative Costs*. Contemporary

Accounting Research, 29(1), 252-282.

Cooper, R., & Kaplan, R.S. (1998). The design of Cost Management Systems: Text, Cases and Readings. Englewood Cliffs, NJ: Prentice Hall.

Dalla Via, N., & Perego, P. (2014). Sticky cost behaviour: evidence from small and medium sized companies. Accounting & Finance, 54(3), 753-778.

(25)

Dierynck, B., Landsman, W. R., & Renders, A. (2012). Do managerial incentives drive cost behavior? Evidence about the role of the zero earnings benchmark for labor cost behavior in private Belgian firms. The Accounting Review, 87(4), 1219-1246.

Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of financial economics, 3(4), 305-360.

Kama, I., & Weiss, D. (2013). Do earnings targets and managerial incentives affect sticky costs?

Journal of Accounting Research, 51(1), 201-224.

Messier, W.F., Simon, C.A. & Smith, J.L. (2013). Two Decades of Behavioral Research on Analytical Procedures: What Have We Learned? Auditing: A Journal of Practice &

Theory, 32 (1): 139–81.

Noreen, E. (1991). Conditions under which activity-based cost systems provide relevant costs.

Journal of Management Accounting Research, 3(4), 159-168.

Noreen, E., & Soderstrom, N. (1994). Are overhead costs strictly proportional to activity?: Evidence from hospital departments. Journal of Accounting and Economics, 17(1), 255-278. Noreen, E., & Soderstrom, N. (1997). The accuracy of proportional cost models: evidence from

hospital service departments. Review of Accounting Studies, 2(1), 89-114.

OECD (2013). Indicators of Employment Protection. Retrieved from OECD: http://www.oecd.org/els/emp/oecdindicatorsofemploymentprotection.htm

O'Sullivan, M. (2003). The political economy of comparative corporate governance. Review of

International Political Economy, 10(1), 23-72.

Pissarides, C. A. (1999). Policy influences on unemployment: the European experience. Scottish

Journal of Political Economy, 46(4), 389-418.

Subramaniam, C., & Weidenmier, M. (2016). Additional evidence on the sticky behavior of costs. In Advances in Management Accounting, 26, 275-305.

Venieris, G., Naoum, V. C., & Vlismas, O. (2015). Organisation capital and sticky behaviour of selling, general and administrative expenses. Management Accounting Research, 26, 54-82.

Cost behavior in Dutch firms : are selling, general, and administrative costs sticky?