Board Characteristics and Asymmetric Cost Behavior
Board Characteristics and Asymmetric Cost Behavior
1
2
3
4
5
Board Characteristics and Asymmetric Cost Behavior:
6 Evidence from Egypt
7
8 Abstract
9
10
11 Purpose- This study aims to provide further evidence on asymmetric cost behavior (cost
12 stickiness) from one of the emerging economies, Egypt. The study provides also empirical
13 evidence on the potential impact of corporate governance on nature and extent of
14 asymmetric cost behavior.
15
16
17
Design/methodology/approach-The study estimates three multiple regression models
18 using Ordinary Least Squares (OLS) to examine the behavior of Cost of Goods Sold
19 (COGS) and the influence of board characteristics and other control variables in a sample
20 of 80 listed companies during 2008-2013.
21
22
Findings-The analysis provides evidence on COGS asymmetric behavior, where the
23
24 analysis finds that COGS increase by 1.05 %, but decrease by 0.85% for an equivalent
25 activity change of 1%, which contradicts the traditional cost model assumption that costs
26 behave linearly. In addition, the analysis finds that firm-year observations with larger
27 boards, role duality, and higher non-executives ratio exhibit greater cost asymmetry than
28 others do while firms-years with successive sales decrease, higher economic growth and
29
30
institutional ownership found to exhibit lower cost stickiness.
31
32 Originality/value-This study contributes by providing evidence on asymmetric cost
33 behavior from one of emerging economies. Further, the study extends the very few studies
34 on the relationship between corporate governance and asymmetric cost behavior. In
35
addition, the study contributes by examining a different cost type (COGS) that found to be
36
37 examined by very few studies. Finally, the study provides an evaluation of the 2007
38 Egyptian Corporate Governance Code, from the cost behavior context.
39
40
41 Keywords: Corporate Governance, Cost Asymmetry, Cost Stickiness, Egypt
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
Accounting Research Journal Page 2 of 23
1
2
3
4 1. Introduction
5
Understanding how costs behave is a vital and critical issue for managers, management
6
7 accountants, and financial analysts. Several techniques, such as Cost-Volume-Profit
8 (CVP) analysis, pricing methods, and Activity-Based Costing (ABC), need to analyze the
9 cost behavior. Almost all these techniques, among others, depend on the traditional cost
10 model.
11 The traditional cost model (y = a + bx) assumes that costs respond proportionately to
12
13
activity/volume changes. The costs are assumed to respond symmetrically (equally) to an
14 equivalent activity/volume change. If volume (x) changes by 1%, total costs (y) will
15 increase or decrease by a symmetric (equal) ratio, since the slope of the model (b) is fixed
16 and total fixed costs (a) are fixed too, within the relevant range.
17 Nevertheless, the literature provides both old and recent empirical evidence on
18
asymmetric cost response to equivalent activity changes. According to Guenther et al.
19
20 (2014), Brasch (1927) is the first that finds the cost curve when the activity increases
21 differs from the cost curve when the activity decreases, which results in an asymmetric
22 cost function. Recent evidence is provided by Anderson et al. (2003), who find that SG&A
23 costs of a USA sample increase by 0.55% per a 1% increase in demand, but decrease by
24 only 0.35% per a 1% decrease in demand. They label the costs that behave this way as
25
26
“sticky cost” or “cost stickiness.” Further, Porporato and Werbin (2012) find that when
27 demand increases (decreases) by a 1%, total costs increase (decrease) by 0.60% (0.38%),
28 0.82% (0.48%), and 0.94% (0.55%), for Argentina, Brazil, and Canada banking samples,
29 respectively. This means that costs increase higher than their decrease per a 1% demand
30 change, indicating that costs behave asymmetrically, not symmetrically, as assumed by the
31
tradition cost model.
32
33 The normal cost behavior occurs when managers respond equally to the same demand
34 changes. When the demand increases, managers should adjust the resources upward by
35 employing new resources to accommodate the increased demand, and thus, costs will
36 increase. On the other hand, when the demand decreases, managers should adjust the
37 resources downward by retiring the slack resources, and thus, costs will decrease.
38
39
Accordingly, when the demand increase equals the demand decrease, the cost response
40 should be symmetric. However, the literature finds that managers’ response to the same
41 demand change differs, and thus, cost response differs, which contradicts the traditional
42 cost model.
43 The literature explains some reasons of asymmetric cost behavior. First, Anderson et
44
al. (2003) explain that when activity decreases, managers choose between two types of
45
46 costs: holding costs if decide to operate with slack resources and adjustment costs if
47 decide to retire the slack resources. Managers will trade-off between the two costs and will
48 be more likely to choose the decision that yields the lower costs. If managers find that
49 costs needed to adjust the resources will be higher than costs incur if decide to operate
50 with slack resources, they will decide not to adjust the resources when the activity
51
52 declines, causing cost stickiness. Second, Anderson et al. (2003) explain that if managers
53 are optimistic about the demand or feel that demand decline is temporary and that the
54 demand will recover soon, they will choose to operate with slack resources and will not
55 adjust the resources when the activity declines, causing higher cost stickiness.
56 Another set of studies attributes the asymmetric cost behavior to the opportunistic
57
managers' intervention. For example, Chen et al. (2012) argue that "Empire-Building"
58
59 incentives induce managers to grow the firm beyond its optimal size. They argue that
60
2
Page 3 of 23 Accounting Research Journal
1
2
3
4 when the demand increases, empire-building managers increase SG&A costs too rapidly,
5 but decrease SG&A costs too slowly when demand declines, in order to increase the firm
6
size, which results in ''sticky cost.'' Further, Dierynck et al. (2012) find that managers
7
8 under pressure to meet or beat earnings benchmarks, increase labor costs to a smaller
9 extent when demand increases, but decreases labor costs to a larger extent when demand
10 declines, in order to save costs, which leads costs to behave as ''anti- sticky.'' Moreover,
11 Kama and Weiss (2013) suggest that researchers should exert efforts to understand and
12 investigate determinants of cost behavior in light of the managers’ motivations, especially
13
14
the agency-driven incentives that affect the resources adjustment decisions.
15 Given that the asymmetric cost behavior results mainly from managers' deliberate and
16 opportunistic intervention when demand changes, there is a need to mitigate this
17 intervention, and thus, bring the cost response closer to the optimal cost response level.
18 Corporate Governance (CG) may be a useful suggestion. An effective CG system is
19
thought to influence positively the managers' decisions and mitigate managerial
20
21 opportunism (Jensen and Meckling, 1976; Jensen, 1993; Shleifer and Vishny 1997; Chen
22 et al. 2012). Board of directors and audit committees are CG mechanisms that are
23 responsible for controlling and monitoring the managers' decisions on behalf of
24 shareholders. Accordingly, they, as CG mechanisms, could influence positively the
25 managers' decisions regarding cost behavior. Chen et al. (2012) assume that strong CG
26
27
may bring cost stickiness levels closer to the optimal cost response level. Further, Chen et
28 al. (2012) and Pichetkun (2012) find that the asymmetric cost behavior extent is less
29 intense in the strong CG sub-sample than in the weak CG sub-sample, which implies that
30 the CG system could mitigate cost stickiness.
31 Accordingly, this study aims to achieve two main objectives. The first is to examine
32
whether costs behave asymmetrically in emerging economies, compared to results found
33
34 in several developed countries. The second is to examine whether and how board structure
35 as a governance mechanism could affect cost behavior.
36 The study contributes in the following ways. First, the study enriches the literature by
37 investigating one of the possible solutions for cost stickiness by extending the extant work
38 of Chen et al. (2012) who examine how CG could affect stickiness of SG&A in the US.
39
40
This study supports the hypothesis of Chen et al. (2012) that effective CG could mitigate
41 cost stickiness. The study of Chen et al. (2012) is one of very limited studies that suggest a
42 solution for cost stickiness, where the majority found to diagnose the stickiness as a
43 problem. However, there is still a need for more empirical evidence on the effectiveness of
44 CG as a suggested mechanism for mitigating cost stickiness through affecting and
45
monitoring the managers’ decisions. The results of Chen et al. (2012) on the influence of
46
47 CG on cost stickiness are restricted to the USA and to SG&A costs only. However, this
48 study contributes by investigating this influence in one of emerging economies and by
49 investigating a different cost type (COGS). Second, the study is one of the early and the
50 few studies that examine asymmetric cost behavior in Egypt and the Arab World. Third,
51 this study evaluates the 2007 Egyptian CG Code but from a different context, the cost
52
53 stickiness context.
54 The remainder of my paper is organized in the following sections. The second section
55 exhibits a theoretical framework that summarizes the CG reforms in Egypt, reviews a set
56 of the most relevant studies classified into three groups and discusses three hypotheses.
57 The third section exhibits the methodology. The fourth section presents and discusses the
58
59
results while the last section concludes.
60
3
Accounting Research Journal Page 4 of 23
1
2
3
4 2. Theoretical Framework & Hypotheses Development
5
6
7 This study depends on two important literature arguments. The first is that asymmetric
8 cost behavior results mainly from the managers’ deliberate decisions (Anderson et al.
9 2003) and the opportunistic intervention (Chen et al. 2012) in the process of resources
10 adjustment when the activity changes, and that the managerial decisions taken to adjust the
11 resources are main drivers of cost behavior (Baumgarten, 2012). The second is that CG
12
13
could bring the cost response level closer to the optimal cost response level (Chen et al.
14 2012) and that the CG different mechanisms could affect several managerial decisions
15 including those of resources adjustments. Accordingly, this section exhibits CG reforms in
16 Egypt, provides a review of the literature classified into three groups, then formulates the
17 study hypotheses by discussing how three of the board characteristics could affect the
18
asymmetric cost behavior.
19
20
21 2.1 Corporate Governance Reforms in Egypt
22 Egypt has witnessed essential CG reforms during the last two decades, which have
23
24
affected significantly the number of listed companies on the Egyptian Stock Exchange
25 (EGX). Shehata & Dahawy (2013) state that applying the governance rules contributes to
26 decreasing the number of companies included in the stock market from 1148 companies at
27 the beginning of 2002 to 333 by mid-2009, to 240 in April 2012.
28 In June 2002, the Egyptian Capital Market Authority (CMA) issued the resolution No.
29
30, regarding rules of listing and delisting of companies on the EGX, as a first step for
30
31 setting and developing CG rules. Next, In October 2005, the Ministry of Investment issued
32 the first Egyptian Corporate Governance Code (ECGC) for corporations. The rules of
33 ECGC comply with the principles of CG issued by the OECD.
34 Next, in November 2006, the CMA announced the project of executive rules of CG,
35 which are derived from the 2005 ECGC. Moreover, the CMA considered applying these
36
37
rules one of the main requirements to stay listed on the Stock Exchange, in an attempt to
38 convert applying of CG rules from voluntary to mandatory. The CMA intended to make
39 applying of CG executive rules mandatory, starting from January 1, 2007, concurrently
40 with applying the new version of Egyptian Accounting Standards. However, many
41 companies were not ready, so the CMA allowed them a period to correct their position to
42
comply with these rules. On March 11, 2007, CMA issued formally resolution No. 11 of
43
44 2007, on the executive rules of CG that must apply on all listed companies; otherwise,
45 these companies will be delisted.
46
47 2.2 Literature Review
48
49
50 This section presents three groups of the most relevant studies. The first group presents
51 some studies that mainly provide evidence on asymmetric cost behavior.
52
First, Anderson et al. (2003) provide a pioneer model to discover whether costs behave
53
54 asymmetrically, where the majority of cost stickiness studies follow their model. The
55 authors apply this model using a sample of USA firms during 1979-1998 and find that
56 SG&A increase by 0.55 %, but decrease by 0.35 % when activity changes by 1%. Results
57 indicate also that during economic prosperity periods, cost stickiness is greater than during
58 recession periods and that extent of cost stickiness is greater in firms with higher assets
59
60
and employees intensity. One more study in USA, Subramaniam and Weidenmier (2003),
4
Page 5 of 23 Accounting Research Journal
1
2
3
4 examines a sample of 9,592 US-listed companies and find that both SG&A and COGS
5 exhibit asymmetric cost behavior, especially when activity change by more than 10
6
percent.
7
8 Further, Porporato & Werbin (2012) examine 270 observations for Argentina, 192
9 for Brazil, and 55 for Canada, during 2004-2009. They find that total costs increase by
10 0.60%, 0.82%, and 0.94% in banks samples of Argentina, Brazil, and Canada, but
11 decrease by 0.38%, 0.48, and 0.55% per 1 % sales change, which asserts cost stickiness in
12 the three samples, but with different degrees. One recent study, Abu-Serdaneh (2014),
13
14
examines all manufacturing companies listed in Jordan during 2008-2012. The author
15 finds that SG&A costs are symmetric, neither sticky nor anti-sticky. The study finds that
16 COGS behaves as anti-sticky. Further, companies with higher assets intensity show higher
17 COGS stickiness. SG&A found to show higher stickiness for - free cash flow and a less
18 stickiness for debt intensity. Moreover, Banker and Byzalov (2014) provide an
19
international study through examining a total sample of 315,967 firm-years in 20 countries
20
21 during 1988-2008. Authors provide comprehensive evidence that asymmetric cost
22 behavior is a pervasive global phenomenon, where operating costs were found to be sticky
23 for 16 countries out of 20. One more study, Via and Perego (2014), examines cost
24 behavior in samples of Italian listed and nonlisted small and medium-sized firms during
25 1999-2008. They detect cost stickiness for only total labor costs, but not for SG&A,
26
27
COGS, or operating costs. However, the stickiness of operating costs is detected in the
28 sample of listed firms only. Authors conclude that the results of investigating small and
29 medium-sized firms are not consistent with prior studies.
30 One recent study conducted by the author of this study, Ibrahim (2015), examines how
31 SG&A, COGS, and operating costs behave in Egyptian business environment during
32
2004-2011, and how economic growth affects cost behavior. The results indicate that both
33
34 SG&A and COGS behave sticky while operating costs behave anti-sticky. Besides, SG&A
35 found to be sticky during the prosperity period before the 2008 financial crisis, but anti-
36 sticky during the recession period after the 2008 financial crisis, while COGS found to be
37 sticky in both periods, but its stickiness extent was larger in the prosperity period. The
38 author concludes that economic growth could affect the cost behavior nature and extent.
39
40
The second group presents three studies that investigate the relationship between
41 managerial incentives, earnings management, and asymmetric cost behavior. The first
42 study, Dierynck et al. (2012), investigates 51,826 firm-year observations in Belgian,
43 during 1993-2006. The authors find that managers meeting or beating the zero earnings
44 benchmark increase labor costs to a smaller extent when activity increases, but decrease
45
labor costs to a larger extent when activity decreases, which makes labor costs take a more
46
47 symmetric cost behavior form. However, managers who do not face significant earnings
48 benchmark pressure were found to limit the employee dismissals, which make labor costs
49 take a more asymmetric cost behavior form. The second study, Kama and Weiss (2013),
50 examines 11,758 US-listed companies, during 1979-2006. Authors find that when
51 managers face incentives to avoid losses or earnings decreases, or to meet financial
52
53 analysts’ forecasts, they accelerate downward adjustment of slack resources for sales
54 decrease, which mitigates the magnitude of cost stickiness. The third study, Koo et al.
55 (2015), investigates the relationship between earnings management and cost stickiness in a
56 sample of USA firms during the years 1997-2007. They find that when activity declines,
57 managers cut down costs aggressively to manage earnings, which mitigates cost stickiness
58
59
while firms with fewer earnings management incentives found to show greater cost
60 stickiness.
5
Accounting Research Journal Page 6 of 23
1
2
3
4 The third group exhibits five studies that examine the relationship between CG and
5 asymmetric cost behavior. First, Calleja et al. (2006), examines 13,662 US firm-years,
6
8,659 UK firm-years, 1,694 German firm-years, and 2,968 French firm-years during 1988-
7
8 2004. Results indicate that operating costs behave asymmetrically in all examined
9 countries, but the extent of stickiness in France and Germany is greater than in the UK and
10 USA. Authors ascribe this difference to the different laws applied in those countries,
11 where both the UK and the USA are operating under the common laws while France and
12 Germany are operating under the code laws. Second, Chen et al. (2012) assume that CG is
13
14
a potential solution that could bring the cost behavior closer to the optimal cost level and
15 examine a sample of USA firms during 1996-2005. They find that SG&S behave sticky
16 and that cost stickiness is less intense in a strong CG sample, but more intense in a weak
17 CG sample, concluding that CG could mitigate cost stickiness. Results indicate also that
18 board independence, institutional ownership, and takeover threats could mitigate the
19
influence of agency problem on cost stickiness.
20
21 Third, Pichetkun (2012) assumes that CG could affect cost stickiness and investigates a
22 sample of firms listed in Thailand in 2001-2009. Results indicate that firm-year
23 observations with weak CG exhibit greater cost stickiness compared with a sub-sample of
24 stronger CG. However, the study does not examine the influence of any of board
25 characteristics on cost stickiness. Fourth, Banker et al. (2013), investigates a total sample
26
27
of 128,333 observations for 15,833 companies located in 19 OECD countries, during
28 1990-2008. Authors find that the level of strictness of the country-level employment
29 protection legislation affects the extent of cost asymmetric behavior. They conclude that
30 the stricter employment protection legislations restrict the ability of managers to cut slack
31 labor resources when the demand declines, which increases cost stickiness. Finally, one
32
recent study, Xue and Hong (2015), investigates cost behavior and the potential influence
33
34 of CG in a sample of 7,702 firm-year observations of firms listed in China during 2003-
35 2010. They find that earnings management non-suspected firms show a higher level of
36 cost stickiness, that effective CG could mitigate the extent of cost stickiness and that the
37 interaction effect between CG and earnings management alleviates cost stickiness.
38 The review of the literature concludes the following gaps. Firstly, I find the majority of
39
40
studies are conducted in developed countries (e.g., USA, UK, Germany, Canada, and
41 France). Further, only two studies, Abu-Serdaneh (2014) and Ibrahim (2015) are
42 conducted on the Middle East and the Arab Region. Secondly, I notice that few studies
43 examined COGS, compared with SG&A. Thirdly, I notice paucity in studies that examine
44 the relationship between different CG mechanisms and asymmetric cost behavior,
45
especially boards’ characteristics. Accordingly, this study contributes by providing
46
47 evidence on asymmetric cost behavior, but from different contexts, the Egyptian and
48 emerging economies contexts. Further, the study extends the extant few studies on the
49 relationship between CG and asymmetric cost behavior. In addition, the study contributes
50 by examining COGS that is examined by a few studies. Further, the study provides an
51 evaluation of the ECGC after its formal applying in 2007.
52
53
54 2.3 Hypotheses Development
55
56 This study examines how one of main CG mechanisms could influence asymmetric
57 cost behavior. The study suggests that board of directors could affect managers’ decisions
58
59
through controlling and monitoring functions, and thus could affect cost behavior, since
60 managers have to make adjustment decisions when activity changes, which will affect cost
6
Page 7 of 23 Accounting Research Journal
1
2
3
4 behavior. Baumgarten (2012) asserts that managerial decisions regarding the resources
5 adjustments when activity changes are main determinants of cost behavior. Therefore, any
6
mechanism that could affect managers’ decisions could affect also cost behavior.
7
8 Accordingly, this study examines how three boards’ characteristics could affect
9 asymmetric cost behavior: board size, role duality, and non-executive directors.
10
11 Board Size
12
13
14
The literature presents a debate on whether large boards are better than small boards. A
15 set of studies thinks that larger boards are more able to monitor, control, and provide
16 companies with diversity that could mitigate dominance of CEO and increase experience
17 (e.g. Goodstein et al. 1994; Mak and Roush, 2000). However, others favor smaller boards.
18 For example, Jensen (1993) argues that larger boards may suffer from more difficulty in
19
coordination among board members. Further, Goodstein et al. (1994) argue that members
20
21 of the larger boards are less likely to become involved in strategic decisions. Large boards
22 may also impede communication and information processing (Jensen 1993; Yermack
23 1996; Huther 1997; John and Senbet, 1998).
24 This study favors smaller boards, since I argue that several decisions' conflicts may
25 arise because of a large number of members on boards, which may decrease the CG
26
27
quality and create opportunities, such as managers' intervention in cost behavior.
28 Empirically, some studies indicate that small boards are preferable (e.g., Yermack 1996;
29 Fuerst and Kang, 2004; Bozec 2005; Haniffa and Hudaib, 2006; Cornett et al. 2007).
30 Therefore, the study first hypothesis is:
31
32 H1: Firm-year observations with larger boards exhibit greater cost stickiness
33
34
35 Role Duality
36
37
38 Role duality occurs when board Chairman and CEO are the same person. Board
39 Chairman/CEO separation is a CG mechanism that increases board independence. Agency
40
theory advocates separation of the two roles to maintain essential checks and balances
41
42 against management (Haniffa and Cooke 2002). Moreover, Gul & Leung (2004), Abdel-
43 Fattah (2008) and Chen et al. (2012) believe that role duality bans board independence and
44 reduces CG quality through concentration of decision-making power. Empirically, several
45 studies show that role duality is not an aspect of effective governance (e.g., Vafeas and
46 Theodorou, 1998; Laing and Weir, 1999; Heracleous, 2001; Kelton and Yang, 2008).
47
48
According to the Egyptian CG code, boards should appoint a chairman who is not also
49 CEO.
50 Consistent with the extant literature, I argue that board Chairman/CEO separation is an
51 indicator of an effective CG system, which creates opportunities for better alignment with
52 interests of shareholders. Further, role duality may lead to opportunistic intervention in
53
several decisions including those of resources adjustments, which also affect the cost
54
55 behavior. Therefore, the study second hypothesis is:
56
57 H2: Firm-year observations with role duality exhibit greater cost stickiness
58
59
60
7
Accounting Research Journal Page 8 of 23
1
2
3
4 Non-executives
5 Boards consist of both insiders (executives) and outsides (non-executives). It is
6
believed that the higher the ratio of outside directors on boards, the higher the board
7
8 independence. Agency theory asserts that boards should include many non-executive
9 directors, allowing them to monitor and control executives and protect interests of
10 shareholders. Moreover, Fama & Jensen (1983) argue that independent directors on
11 boards could mitigate managers' opportunistic actions through effective monitoring.
12 Forker (1992) documents that a higher ratio of non-executives on boards makes them
13
14
more responsive to shareholders’ interests and investors. Empirically, many studies
15 indicate that a large proportion of non-executives is a sign of good governance (Helland
16 and Sykuta, 2005; Haniffa and Hudaib, 2006; Cornettet al. 2007; Kelton and Yang,
17 2008). Consequently, including many non-executive directors on boards may be
18 important for effective CG.
19
Accordingly, I believe that segregation of duties and independence are critical board
20
21 characteristics that can increase the CG quality as whole, and this separation can
22 influence positively the managers’ decisions including the costs’ adjustment decisions,
23 and thus cost behavior. Therefore, the study third hypothesis is:
24
25 H3: Firm-year observations with higher executives’ ratio exhibit greater cost stickiness
26
27
3. Methodology
28
29
30 3.1 Sample and Data
31
32
Table 1 exhibits two main samples, the basic sample is to analyze COGS behavior, and
33 the CG sample is to investigate the influence of board characteristics on asymmetric cost
34 behavior. The companies listed on the Egyptian Stock Exchange during 2008-2013
35 represent the study population while the 100 listed companies that make up EGX100
36 index represent the study initial samples. The final sample comprises 80 non-financial
37 listed companies after excluding banks and other financial institutions due to their special
38
39 standards and regulations. Moreover, I depend on annual reports issued by companies and
40 disclosure books issued regularly and formally by EGX to extract any required data on
41 models’ variables. I extract data on costs and sales variables directly from income
42 statements embedded in the annual reports, which are available on the companies’
43 websites. Further, I extract data on corporate governance variables from disclosure books,
44
45
where the first page of each company in the disclosure book includes data on board
46 members, their names and positions in addition to data on executives and ownerships
47 structure. Table (2) shows source of data for each variable.
48
49 Insert Table 1 about here
50
51
52 As shown in the table (1), I apply some data screening criteria following Anderson et
53 al. (2003), Chen et al. (2012) and Dierynck et al. (2012). First, I exclude firm-year
54 observations of financial companies. Second, I discard any firm-year observation with
55 missing data on costs or net sales. Third, I exclude any firm-year observation with costs
56 exceeding net sales for the current year. Finally, I consider any firm-year observation with
57
58
a standardized residual of more than the absolute value of three an extreme observation
59 (outlier) that should be excluded.
60
8
Page 9 of 23 Accounting Research Journal
1
2
3
4 3.2 Models
5 To test the study hypotheses, three models are applied. Model (1) is to examine the
6
stickiness of COGS; Anderson et al. (2003) and Chen et al. (2012) apply this model to
7
8 examine the stickiness of SG&A costs in USA samples. Then, Model (1) is extended by
9 adding 3 corporate governance variables as interaction-terms, then added as a standalone
10 variable, as shown in Model (2). Model (3) is the same as model (2), but it includes 3
11 control variables. Year dummies are included in all regression models.
12 Anderson et al. (2003) provide a pioneer regression model to discover whether the
13
14
costs increase is different from the costs decrease for an equivalent activity change. This
15 model is able to measure the costs’ response to contemporaneous changes in sales and is
16 able to discriminate between periods of sales increase and decrease (Anderson et al. 2003).
17 The model contains a dummy variable (DecDummy) that is able to distinguish between
18 revenue-decreasing and revenue-increasing years. Therefore, the majority of studies
19
followed Anderson et al. (2003) and applied their model (e.g. Calleja et al. 2006; Kama
20
21 and Weiss, 2013; Ibrahim, 2015). To examine the potential correlation between board
22 characteristics and cost asymmetry, I extend the basic model by including board size, role
23 duality, and non-executives’ ratio and other control variables by multiplying each variable
24 by DecDummyit × Log (∆ Salesit), which creates three-way interaction terms, following the
25 relevant studies such as Anderson et al. (2003), Chen et al. (2012), Dierynck et al. (2012)
26
27
and Ibrahim (2015).
28 Nevertheless, this inclusion could increase the Multicollinearity level between the
29 model continuous variables. To avoid this problem, I follow Chen et al. (2012) and
30 undertake mean-centering for all continuous variables included in the interaction terms
31 (except for role duality as a dummy variable). Mean-centering is a statistical procedure
32
that subtracts the variable values from the variable mean. Aiken and West (1991) and
33
34 Chen et al. (2012) argue that mean-centering could reduce Multicollinearity and ease
35 interpretation of the main effects. Moreover, the models apply the ratio form and log
36 specification, since Anderson et al. (2003) suggest that these forms could alleviate the
37 potential heteroskedasticity and enhance comparability of variables across companies.
38
39
40 Basic Model (1):
41
42 Log (∆ COGSit) = β0 + β1 Log (∆ Salesit) + β2 DecDummyit * Log (∆ Salesit) + Σ
43
44
45
46 Model 2: (No controls)
47 Log (∆ COGSit) = β0 + β1 Log (∆ Salesit)
48 +β2 DecDummyit ×Log (∆ Salesit)
49 + β3DecDummyit ×Log (∆ Salesit) × Board Size it
50 + β4DecDummyit ×Log (∆ Salesit) × Role Duality it
51
52
+ β5DecDummyit ×Log (∆ Salesit) × Non-Executives it
53 + β6 Board Sizeit + β7 Role Dualityit + β8 Non-Executives it + Σ
54
55
56
57
58
59
60
9
Accounting Research Journal Page 10 of 23
1
2
3
4 Model 3: (With controls)
5 Log (∆ COGSit) = β0 + β1 Log (∆ Salesit)
6
+β2 DecDummyit ×Log (∆ Salesit)
7
8 + β3 DecDummyit ×Log (∆ Salesit) × Board Sizeit
9 + β4 DecDummyit ×Log (∆ Salesit) × Role Dualityit
10 + β5 DecDummyit ×Log (∆ Salesit) × Non-Executivesit
11 + β6 DecDummyit ×Log (∆ Salesit) × Successive Decreaseit
12 + β7 DecDummyit ×Log (∆ Salesit) × Economic Growthit
13
14
+ β8 DecDummyit ×Log (∆ Salesit) × Institutional Ownershipit
15 +β9 Board sizeit
16 + β10 Role Dualityit
17 + β11 Non- Executivesit
18 + β12 Successive Decreaseit
19
+ β13 Economic Growthit
20
21 + β14 Institutional Ownershipit + Σ
22
23
24 3.3 Control Variables
25
26 Model (3) includes 3 control variables: successive decrease, economic growth and
27 institutional ownership. The first control successive decrease is to test the hypothesis that
28
29 managers become more certain that demand decline is permanent not temporary when
30 demand declines for 2 successive periods, and thus, managers start to retire slack
31 resources, which decreases stickiness. A positive significant coefficient implies that cost
32 stickiness degree is lower during a demand-declining period preceded by demand-
33 declining periods. The second control economic growth is measured by real GDP and is
34
35
used to test the hypothesis that if demand declines during economic growth periods,
36 managers are less likely to retire slack resources, since managers consider this decline
37 temporary than if this decline happened during lower economic growth periods (Anderson
38 et al. 2003; Banker et al. 2013; Ibrahim, 2015), and thus, cost stickiness is likely to be
39 higher during the economic growth periods, since managers will not retire slack resources.
40
Finally, institutional ownership is included as a control variable in order to test the
41
42 hypothesis that institutional ownership provides better monitoring. Agency theory and
43 efficient-monitoring hypothesis suggest that institutional investors are sophisticated
44 investors who hold experience and power than individuals, and thus, could monitor
45 effectively the managerial behavior (Jensen and Meckling, 1976; Abdel-Fattah, 2008).
46 Chen et al. (2012) find that governance mechanisms such as institutional ownership and
47
48
board independence are effective in mitigating the impact of agency problem on stickiness
49 of SG&A costs. The definitions and measurements of all the models’ variables are
50 exhibited in Table (2).
51
52 Insert Table 2 about here
53
54
4. Data Analysis & Results
55
56 In this section, I provide and discuss three analyses: descriptive analysis, basic
57 regression analysis, and robust regression analysis. To prove cost stickiness, both β1 and
58 β2 should be statistically significant, and β1 should be positive and β2 should be negative.
59 To explain the results of board characteristics and control variables included in Models (2)
60
10
Page 11 of 23 Accounting Research Journal
1
2
3
4 and (3), I follow a rule set by Anderson et al. (2003), whereby the stickiness degree
5 increases (decreases) with the magnitude of the negative (positive) coefficients β2 through
6
βm.
7
8
9
4.1 Descriptive Statistics
10
11 Panel (A) of Table 3 exhibits that COGS average is 943 million Egyptian Pounds, with
12 a median of 170 million, which are lower than COGS mean of $885.48 million and a
13 median of $73.64 of the USA sample as reported by Subramaniam & Weidenmier (2003)
14 by considering the exchange rates. The net sales mean of the study sample is 1,312 million
15
16 Egyptian Pounds, with a median of 282 million, which are lower than net sales mean of
17 $5,383 million and a median of $1,433 million for the USA sample examined by Chen et
18 al. (2012), and lower than net sales means of $1,277, $1,153, and $1,294 million for the
19 USA samples examined by Anderson et al. (2003), Calleja et al. (2006), and Subramaniam
20 & Weidenmier (2003), respectively. The COGS mean as a percentage of net sales is 72%,
21
22
which is higher than mean of 63.77 %, found by Subramaniam and Weidenmier (2003) in
23 a USA sample and 64.51 % reported by Ibrahim (2015) for an Egyptian sample.
24
25 Insert Table 3 about here
26
27 Panel (B) of Table (3) shows the descriptive statistics on the corporate governance
28
29 variables. The board size mean is 7.51 and the median value is 7 members. The mean and
30 median are lower than 9.15 and 9 found by Chen et al. (2012) in a USA sample. For role
31 duality, table 2 shows a mean of 0.65 implying that 65% of the study sample boards’
32 chairmen are working as CEOs at the same time, which is slightly higher to 63% found by
33 Chen et al. (2012) in a USA sample. For a ratio of independent directors on boards to total
34
35
boards members, table 2 exhibits a ratio of 69%, which is slightly higher than 66% found
36 by Chen et el. (2012).
37 Panel (C) of Table (3) provides descriptive statistics on control variables. The mean
38 value of Successive Decrease is 0.40 indicating that 40% of firm-years experiencing two
39 successive sales declines during the last two year, which is higher than 0.23 reported by
40
Chen et al. (2012). While the median is 0 similar to Chen et al. (2012), implying that the
41
42 median firm did not experience any sales declines during the last two years. The mean
43 value of Economic Growth is 3.16, indicating that the average economic growth % during
44 the study period is 3.16%. Finally, the mean value of Institutional Ownership is 20%
45 indicating that 20% on average of the sample firms’ ownership is held by institutional
46 investors. This ratio is lower than 65.53% reported by Chen et al. (2012), indicating that
47
48
more than half of the ownership of the US sample examined by Chen et al. (2012) was
49 held by institutional investors, while less than a quarter for the Egyptian sample.
50
51 4.2 Regression Results
52
53 Evidence on asymmetric cost behavior
54
55 Table 4 exhibits results of running model 1 to examine whether COGS exhibit
56 asymmetric cost behavior. The results indicate that F-value = 765.30 (0.000) and adj.R2 =
57
58 0.84, implying that the basic model is statistically significant and explains 84 % of COGS
59 variations. Further, the coefficients β1 and β2 found to be statistically significant at 1 %
60
11
Accounting Research Journal Page 12 of 23
1
2
3
4 level; β1 (1.05) is positive, and β2 (-0.20) is negative. These results are consistent with the
5 empirical hypothesis of asymmetric cost behavior proposed by Anderson et al. (2003).
6
Moreover, the value of 1.05 of β1 coefficient indicates that COGS increase by 1.05 %
7
8 when activity increases by 1 %. However, the sum of both (β1+β2) = 0.85, which
9 indicates that COGS decreased by only 0.85 % when activity decreases by 1 %. Moreover,
10 Subramaniam and Weidenmier (2003) find that COGS increase by 1.01, but decrease by
11 0.94 for a USA sample, and Ibrahim (2015) find that COGS increase by 1.02 %, but
12 decrease by 0.57 % for an Egyptian sample.
13
14
15 Insert Table 4 about here
16
17 The regression results of estimating the basic model provide evidence on asymmetric
18 cost behavior from emerging economies. Furthermore, the results are consistent with those
19
of Subramaniam and Weidenmier (2003), Weiss (2010), Balakrishnan et al. (2014), and
20
21 Ibrahim (2015) who find that COGS behavior is asymmetric. The results assert arguments
22 of the resources-adjustment hypothesis presented by Anderson et al. (2003) and
23 Baumgarten (2012) that the managers’ decision regarding the resources adjustments when
24 activity changes is one of the main determinants of cost behavior, and with cost
25 asymmetry theory documented by Banker and Byzalov (2014).
26
27
One possible explanation for the sticky behavior of COGS is that managers are
28 regularly induced by discounts of suppliers on large quantities of raw materials. This will
29 result in several slopes and several rates per unit of raw materials, which are the main
30 components of COGS. Horngren et al. (2012) explain the influence of large raw materials
31 discounts on cost behavior and show that several slopes could exist within small ranges,
32
which results in a nonlinear cost curve. Further, the second component of COGS is labor
33
34 costs; both learning curves and intellectual capital may explain the asymmetric behavior of
35 labor costs. The notion of learning curves explains that in labor-intensive industries,
36 workers become more experienced overtime, which will improve productivity and reduce
37 labor costs. For intellectual capital, managers will not retire intelligent workers and highly
38 skilled employees when activity declines, which increases cost stickiness of COGS.
39
40
41 Board Characteristics and Asymmetric Cost Behavior
42
43 The literature examines the variables affecting cost stickiness once by including them
44 as interaction-terms only, such as Anderson et al. (2003), Subramaniam & Weidenmier
45
(2003), and Kama & Weiss (2013), and once by including them as interaction-terms and
46
47 standalone variables together at the same model, such as Calleja et al. (2006), Chen et al.
48 (2012) and Dierynck et al. (2012). Therefore, I prefer to run the models once without
49 standalone variables and once with standalone variables and before and after the control
50 variables as well. Table (5) shows the results in four columns.
51 Column (1) of Table 5 exhibits that β1 is positive and statistically significant at the 1 %
52
53 level (β1 = 1.09, t-statistic = 42.20), while β2 is negative and statistically significant at the
54 1 % level (β2 = -0.29, t-statistic = -8.25). Likewise, the coefficients β1 are positive and
55 statistically significant at 1% and the coefficients β2 are negative and statistically
56 significant at 1% even after adding control variables and standalones variables at the other
57 three columns. The results of the four columns are consistent with the empirical
58
59
hypothesis of cost stickiness, which confirms that COGS behave sticky as found when
60 running the basic model as shown in Table 4.
12
Page 13 of 23 Accounting Research Journal
1
2
3
4 Insert Table 5 about here
5
6
For board size variable, the coefficients are negative and statistically significant at 1%
7
8 as shown in the four columns of Table 5 {β3 = -0.05 (-2.89); -0.18 (-8.90); -0.06 (-2.69); -
9 0.19 (-12.04)}. Based on a rule set by Anderson et al. (2003) to explain the influence of
10 included variables, the negative and significant coefficient implies that observations with
11 larger boards exhibit greater cost stickiness. This result is consistent with my argument
12 that smaller boards seem to be a good CG mechanism, where smaller boards can reduce
13
14
the conflict that could arise between members and ease communication within the
15 company. The result is also consistent with the argument of Chen et al. (2012) that larger
16 boards could alleviate governance quality. Further, the result confirms the argument of
17 Jensen (1993) that larger boards may suffer from more difficulty in coordination among
18 board members and argument of Goodstein et al. (1994) that members of the larger boards
19
are less likely to become involved in strategic decisions. Another possible explanation is
20
21 that small boards are more likely to monitor the decisions of resources adjustment
22 effectively, with fewer disputes and a high degree of concord between board members
23 than larger boards.
24 For role duality, the results indicate that β4 coefficients are negative and statistically
25 significant at 1 % and 5% as shown in the four columns of Table 5 {β4 = -0.58 (-13.06); -
26
27
0.17 (-1.92); -0.60 (-11.29); -0.19 (-2.84)}, whether with standalone and control variables
28 or without. This result proves that role duality could influence cost behavior, observations
29 with role duality are more likely to experience greater cost stickiness. This result confirms
30 the agency theory argument that separation of board chairman/CEO roles could increase
31 independence, effective monitoring, and mitigate power concentration, which ultimately
32
will improve managers’ decisions. One explanation for this result is that when board
33
34 chairmen work as CEOs at the same time, this may create a state of power concentration.
35 My explanation is that when activity changes, board chairmen who work as CEOs at the
36 same time may allow managers adjusting the resources in any way regardless of its
37 consequence on cost behavior.
38 For the ratio of non-executives on boards, the results indicate that the coefficients are
39
40
negative and statistically significant at the 1 % level for the four cases {β5 = -1.90; (-
41 11.01); -0.96 (-4.46); -1.85 (-9.14); -0.95 (-5.81)}. This result means that cost stickiness is
42 greater in boards with a higher ratio of non-executives, which contradicts the study third
43 hypothesis. The result disagrees with Crowther and Jatana (2005) and Gul and Leung
44 (2004) who argue that a higher non-executives number on boards is a signal of strong CG
45
system that will motivate the level of effective monitoring and transparency. Moreover,
46
47 Fama & Jensen (1983) and agency theory argue that independent directors on boards
48 could mitigate managers' opportunistic actions through effective monitoring, which
49 contradicts the results found. One explanation for this result is that non-executive directors
50 may not have sufficient information and are not engaged in daily operations enough to
51 affect the decisions of resources-adjustments. Therefore, reducing the number of non-
52
53 executives may be preferable for alleviating stickiness behavior. Fuerst and Kang (2004)
54 and Bozec (2005) find similar results. Accordingly, I reject the study hypothesis H3.
55 For the control variables, Table (5) shows similar results before and after adding the
56 standalone variables. First, the coefficients of Successive Decrease are positive and
57 statistically significant at 1% {β6 = 0.38 (3.81); 0.38 (4.81)} as shown columns (2) and
58
59
(4), A positive significant coefficient implies that cost stickiness degree gets lower during
60 a demand-declining period preceded by demand-declining periods. This asserts the
13
Accounting Research Journal Page 14 of 23
1
2
3
4 argument that the successive decrease in sales makes managers recognize that sales
5 decline is permanent, and thus, they take their decision by retiring the slack resources, and
6
thus, costs decrease become similar to costs increase for the equivalent sales change,
7
8 which decreases cost stickiness. Anderson et al. (2003), Chen et al. (2012), Kama & Weiss
9 (2013) and Banker et al. (2013) find a positive and statistically significant correlation for
10 the successive decrease variable, consistent with their expectations, while Dierynck et al.
11 (2012) find a statistically insignificant correlation.
12 Second, the coefficients of Economic Growth show a positive and statistically
13
14
significant correlation at 1% before and after including the standalone variables {β7=0.10
15 (2.77); 0.10 (3.46)}, which means that during the higher economic growth periods, COGS
16 shows lower cost stickiness. Anderson et al. (2003) and Ibrahim (2015) argue that during
17 the economic growth periods, managers are optimistic and think that any sales decline is
18 temporary, and thus, do not retire the slack resources even if the sales decline, which
19
causes more cost stickiness. However, the regression result comes inconsistent with this
20
21 argument and with the results find by Anderson et al. (2003), Banker et al. (2013), and
22 Ibrahim (2015) who find a significant negative relation and with Dierynck et al. (2012)
23 who find an insignificant relation.
24 Finally, the coefficient of Institutional Ownership exhibits a positive and statistically
25 significant correlation at 1% {β8 = 0.56 (3.28); 0.57 (4.29)}, which indicates that firms
26
27
with higher institutional ownership exhibit lower cost stickiness, which is consistent with
28 the study expectations. This is also consistent with the agency theory and efficient
29 monitoring hypothesis that institutional investors hold experience, analytical skills, and
30 power than individuals that enable them to affect and monitor effectively the managers’
31 decisions (Jensen and Meckling, 1976; Abdel-Fattah, 2008). This result is consistent with
32
Chen et al. (2012) who find that institutional ownership can be an effective governance
33
34 mechanism that is able to mitigate the influence of agency problem on stickiness of
35 SG&A costs in a sample of USA firms.
36 It is noteworthy that the asset intensity was added as a control variable with a positive
37 relation expectation with cost stickiness, however, adding this variable found to increase
38 the Multicollinearity levels with other independent variables to unacceptable levels, so that
39
40
I excluded it from the models. The bottom of table (5) shows that the Multicollinearity is
41 not a problem, since VIF and condition index values are within the acceptable levels. For
42 example, the maximum VIP is 7.18 lower than 10 (Kennedy 1992; Herrmann et al. 2003;
43 Ibrahim 2015). Moreover, I find the results similar when estimating a fixed-effects model.
44 Overall, COGS insists to behave sticky and board size, role duality and non-executive
45
are found to increase cost stickiness whether before or after adding the standalone and
46
47 control variables. Further, successive decrease, economic growth, and institutional
48 ownership are found to decrease cost stickiness.
49
50 4.3 Robust Analysis
51
52
53 I divide the study’s main sample into two sub-samples using the median of each board
54 characteristic. Observations with values greater than or equal median of each board
55 characteristic make up the first sub-sample while observations with values less than
56 median make up the second sub-sample. Next, I run model (1) to examine cost behavior
57 and compare the cost stickiness magnitude in each sub-sample. Table 6 exhibits the
58
59
results.
60
14
Page 15 of 23 Accounting Research Journal
1
2
3
4 For sub-sample with board size above or equal the median, COGS found to be more
5 sticky (β1=1.02, t-statistic = 35.38; β2 = - 0.33, t-statistic= - 8.07) than its counterpart sub-
6
sample (β1 =1.10, t-statistic = 20.12; β2 =0.15, t-statistic = 2.78). The comparison result
7
8 indicates that observations with larger boards experience greater cost stickiness, which
9 asserts that results found in Table 5. For sub-sample with role duality, COGS found to be
10 more sticky (β1 = 1.10, t-statistic = 18.05; β2 = -0.17, t-statistic = -2.75) than its
11 counterpart sub-sample (β1 = 0.89, t-statistic =18.65; β2 =0.19, t-statistic = 1.78), which
12 indicates that separation of chairman and CEO jobs could mitigate magnitude of cost
13
14
stickiness, which asserts the results found in Table 5.
15
16 Insert Table 6 about here
17
18 For sub-sample with non-executives ratio greater than or equal the median, COGS
19
found to be more sticky (β1 = 0.95, t-statistic = 20.44; β2 = -0.19, t-statistic = -3.90),
20
21 compared with its counterpart sub-sample (β1 = 1.12, t-statistic = 37.19; β2 = 0.16, t-
22 statistic = 1.74), which implies that the higher non-executives on boards the higher the
23 cost stickiness, asserting the results exhibited in Table 5.
24 The overall conclusion is that COGS behave asymmetrically, not as assumed by the
25 traditional cost model that costs should behave in a linear form with equivalent activity
26
27
changes. The result confirms the cost asymmetry theory assumed by Banker and Byzalov
28 (2014) and asserts arguments of the resources-adjustment hypothesis presented by
29 Anderson et al. (2003) and Baumgarten (2012). This implies that there is a gap between
30 linear cost assumption and the real cost behavior, which implies that applying any
31 management accounting technique that depends on the linear cost assumption, such as
32
CVP, ABC or earnings forecast could present distorted results and thus lead to misleading
33
34 decisions.
35 On the other hand, board characteristics found to affect the cost stickiness extent,
36 which implies that effective CG systems may be a potential solution that could affect the
37 managers’ decisions regarding the resources adjustments, and thus, could mitigate the cost
38 stickiness extent. The results also confirm some of agency theory arguments that smaller
39
40
boards, board/CEO separation, and institutional investors are more likely to provide
41 effective monitoring and controlling for managers’ decisions including the resources-
42 adjustments decisions when the activity changes. Accordingly, these results provide
43 important implications to investors and analysts when conducting earnings forecasts, to
44 management accountants when making cost estimation and allocation, to governance
45
authorities when setting governance codes and regulations and to academic researchers in
46
47 investigating determinants of cost behavior.
48
49
50
5. Conclusion
51 This study extends the cost stickiness literature by providing new evidence from
52 emerging economies and by investigating the influence of board characteristics as one of
53
54 CG mechanisms. Results indicate that COGS behavior is sticky; it increases (1.05%) more
55 than it does decrease (0.85%) with a 1 % activity change. Further, board characteristics
56 found to affect the managers’ decisions, and thus, cost behavior, where all examined board
57 characteristics found to affect cost stickiness in some way. Results indicate that smaller
58 boards and chairman/CEO separation could mitigate cost stickiness while the ratio of non-
59
60
executives on boards found to increase cost stickiness. Further, the institutional ownership
15
Accounting Research Journal Page 16 of 23
1
2
3
4 as a control variable found to mitigate cost stickiness degree. This supports the study main
5 suggestion that effective corporate governance mechanisms could affect managers’
6
decisions and thus cost behavior. The overall conclusion is that cost stickiness is a
7
8 prevalent cost behavior in emerging economies as well as in developed ones and that CG
9 could affect managers’ decisions regarding resources adjustment when activity changes.
10 One main implication is that management accountants should estimate the different
11 cost accounting and management accounting techniques, such as standard costing, cost
12 planning, ABC, CVP, and budgeting carefully, since the slope is not always constant, as
13
14
assumed by the traditional cost model. For CG regulators, they should consider how
15 managers’ deliberate intervention could make costs behave asymmetrically, and how CG
16 could mitigate this intervention. They should consider smaller board size, chairman/CEO
17 separation and institutional ownership as variables that could mitigate cost stickiness. For
18 investors and analysts, they should take into consideration the asymmetric cost behavior
19
when conducting earnings forecasts.
20
21 This study suffers some limitations. First, the study examines only three board
22 characteristics, although there are several other variables that still need investigation.
23 Second, the study examines the potential impact of CG on COGS, although there are
24 several other costs, such as SG&A, OC, and TC, which still need investigation. Third, the
25 study samples are deemed small compared with those examined in the developed
26
27
countries.
28 Finally, future research can investigate the relationship between asymmetric cost
29 behavior and other CG mechanisms, such as audit committee characteristics, auditor type,
30 and different variables of ownership structures. Moreover, further research is still needed
31 to investigate the relationship between managerial incentives and asymmetric cost
32
behavior. I find paucity in studies that investigate this relationship. Further, future authors
33
34 should be cautious that either costs or sales figures or both could be manipulated, which
35 could mislead the results of discovering cost stickiness. They should consider this issue
36 and examine real costs and sales figures before any manipulations. Finally, future research
37 can suggest solutions, where I find that most prior studies provide either evidence on the
38 asymmetric cost behavior or a link between this behavior and other accounting issues.
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
16
Page 17 of 23 Accounting Research Journal
1
2
3
4 References
5 Abdel-Fattah, T. M. H. (2008). “Voluntary disclosure practices in emerging capital markets: the case of
6 Egypt”. Ph.D. Thesis. Durham University.
7
8 Abu-Serdaneh, J. (2014), “The asymmetric behavior of cost: evidence from Jordan”, International Business
9 Research, Vol. 7 No. 8, pp. 113: 122.
10 Ailken, L. S., and West, S. G. (1991), Multiple Regression: testing and interpreting interactions. Newbury
11 Park, CA: Sage Publications.
12
13 Anderson, M. C., Banker, R. D., and Janakiraman, S. N. (2003), “Are selling, general, and administrative
costs sticky?”, Journal of Accounting Research, Vol. 41 No. 1, pp.47–63.
14
15 Balakrishnan, R. E., Labro, E. and Soderstrom, N. S. (2014), “Cost structure and sticky costs”, Journal of
16 Management Accounting Research, Vol. 26, No. 2, pp. 91-116.
17
Banker, R. D., and Byzalov, D. (2014), “Asymmetric cost behavior”, Journal of Management Accounting
18
Research, Vol. 26 No. 2, pp. 43-79.
19
20 Banker, R. D., Byzalov, D., and Chen, L. T. (2013), “Employment protection legislation, adjustment costs
21 and cross-country differences in cost behavior”, Journal of Accounting and Economics, Vol. 55 No.
22 1, pp. 111–127.
23
Baumgarten, D. (2012), “The cost stickiness phenomenon: causes, characteristics, and implications for
24
fundamental analysis and financial analysts’ forecasts”, Gabler Verlag, Springer Fachmedien
25
Wiesdbadn.
26
27 Bozec, R. (2005), “Board of directors, market discipline and firm performance”, Journal of Business,
28 Finance, and Accounting, Vol. 32 No. 9–10, pp. 1921–1960.
29
Brasch, H. (1927).,“Zur Parxis der Unkostenschwankungen und ihrer Erfassung (2) (The practice of cost
30
fluctuation and their measurement)”, Betriebswirtschaftliche Rundschau, Vol. 4, pp. 65-73.
31
32 Calleja, K., Steliaros, M. and Thomas, D. C. (2006), “A note on cost stickiness: Some international
33 comparison”, Management Accounting Research, Vol. 17 No.2, pp. 127–140.
34 Chen, C. X., Lu, H. and Sougiannis, T. (2012), “The agency problem, corporate governance, and the
35 asymmetric behavior of selling, general, and administrative costs”, Contemporary Accounting
36 Research, Vol. 29 No. 1, pp. 252–282.
37
38 Cornett, M. M., Marcus, A. J., Saunders, A., and Tehranian, H. (2007), “The impact of institutional
39 ownership on corporate operating performance”. Journal of Banking and Finance, Vol. 31, pp. 1771–
40 1794.
41 Crowther, D., and Jatana, R. (2005), “Agency theory: a cause of failure in corporate governance”,
42 International dimensions of corporate social responsibility, Vol.1, pp. 135–152.
43
44 Dalla Via, N. and Perego, P. (2014), “Sticky cost behavior: evidence from small and medium-sized
45 companies”, Accounting & Finance, Vol. 54, No. 3, pp. 753-778.
46 Dierynck, B., Landsman, W. R., and Renders, A. (2012), “Do managerial incentives drive cost behavior?
47 Evidence about the role of the zero earnings benchmark for labor cost behavior in private Belgian
48 firms”, The Accounting Review, Vol. 87 No. 4, pp. 1219–1246.
49
50 Fama, E. F., and Jensen, M. C. (1983), “Separation of ownership and control”, Journal of Law and
51 Economics, Vol. 26 No. 2, pp. 301: 325.
52 Forker, J. J. (1992), “Corporate governance and disclosure quality”, Accounting and Business Research
53 Journal, Vol. 22 No. 86, pp. 111–124.
54
55 Fuerst, O., and Kang, S. H. (2004), “Corporate governance, expected operating performance, and pricing”,
56 Corporate Ownership and Control, Vol. 1 No. 2, pp. 13–30.
57 Goodstein, J., Gautam, K., and Boeker, W. (1994), “The effects of board size and diversity on strategic
58 changes”, Strategic Management Journal, Vol. 15, No.3, pp. 241–250.
59
60
17
Accounting Research Journal Page 18 of 23
1
2
3
4 Guenther, T. W., Riehl, A. and Robler, R. (2014), “Cost stickiness: state of the art of research and
5 implications”, Journal of Management Control, Vol. 24 No. 4, pp. 301–318.
6 Gul, F. A., and Leung, S. (2004), “Board leadership, outside directors’ expertise and voluntary corporate
7 disclosure”, Journal of Accounting and Public Policy, Vol. 23 No. 5, pp. 351–379.
8
9 Haniffa, R. M., and Cooke, T. E. (2002), “Culture, corporate governance and disclosure in Malaysian
10 corporations”, ABACUS, Vol. 38 No. 3, pp. 317–349.
11 Haniffa, R. M., and Hudaib, M. (2006), “Corporate governance structure and performance of Malaysian
12 listed companies”, Journal of Business Finance and Accounting, Vol. 33 No. 7, pp. 1034–1062.
13
14 Helland, E., and Sykuta, M. E. (2005), “Who’s monitoring the monitor? Do outside directors protect
15 shareholders’ interests?” The Financial Review, Vol. 40, No. 2, pp. 155–172.
16 Heracleous, L. (2001), “What is the impact of corporate governance on organizational performance?”
17 Corporate Governance: An International Review, Vol. 9 No. 3, pp. 165–173.
18
19 Herrmann, D., Tatsuo, I., and Wayne, B. T. (2003), “The sale of assets to manage earnings in Japan”.
20 Journal of Accounting Research, Vol. 41 No. 1, pp. 89–108.
21 Horngren, C. T., Datar, S. M., and Rajan, M. (2012), “Cost Accounting: A Managerial Emphasis”, 14th
22 edition. Upper Saddle River, NJ : Prentice Hall.
23
24 Huther, J. (1997), “An empirical test of the effect of board size on firm efficiency”, Economic Letters, Vol.
25 54 No. 3, pp. 259–264.
26 Ibrahim, A. E. A. (2015), “The economic growth and cost stickiness: evidence from Egypt”, Journal of
27 Financial Reporting and Accounting, Vol. 13, No. 1, pp. 119-140.
28
29 Jensen, M. (1993), “The modern industrial revolution, exit and the failure of internal control systems”,
30 Journal of Finance, Vol. 48, No. 3, pp. 831–880.
31 Jensen, M. C., and Meckling, W. H. (1976), “Theory of the firm: Managerial behavior, agency costs and
32 ownership structure”, Journal of Financial Economics, Vol. 3 No. 4, pp. 305–360.
33
34 John, K. and Senbet, W. (1998), “Corporate governance and board effectiveness”, Journal of Banking and
35 Finance, Vol. 22, pp. 371–403.
36 Kama, I., and Weiss, D. (2013), “Do earnings targets and managerial incentives affect sticky costs?” Journal
37 of Accounting Research, Vol. 51 No. 1, pp. 201-224.
38
39 Kelton, A. S., and Yang, Y. (2008), “The impact of corporate governance on internet financial reporting”,
40 Journal of Accounting and Public Policy, Vol. 27 No. 1, pp. 62–87.
41 Kennedy, P. (1992), A guide to econometric methods, MIT Press, Cambridge, MA.
42
43 Koo, J-H., Song, S., and Paik, T-Y. (2015), “Earnings management and cost stickiness”, Advanced Science
44 and Technology Letter, Vol. 84 (Business 2015), pp. 40-44.
45 Laing, D. and Weir, C. M. (1999), “Governance structures and corporate performance in the UK firms”,
46 Management Decision, Vol. 37 No. 5, pp. 457–464.
47
48 Mak, Y., and Roush, M. (2000), “Factors affecting the characteristics of boards of directors: an empirical
49 study of New Zealand initial public offering firms”, Journal of Business Research, Vol. 47, pp. 147–
50 159.
51 Mederios, O. R. and De Souza Costa, P. (2004), “Cost stickiness in Brazilian firms”, Working paper,
52 Universidad de Brasilia.
53
54 Pearce, J. A., and Zahra, S. A. (1992), “Board composition from a strategic contingency perspective”,
55 Journal of Management Studies, Vol. 29 No. 4, pp. 411–438.
56 Pichetkun, N. (2012), “The determinants of sticky cost behavior on political costs, agency costs, and
57 corporate governance perspectives”, Ph.D. Thesis, Faculty of Business Administration, Rajamangala
58 University of Technology Thanyaburi, Thailand.
59
60
18
Page 19 of 23 Accounting Research Journal
1
2
3
4 Porporato, M., and Werbin, E. (2012), “Evidence of sticky costs in banks of Argentina, Brazil, and Canada”,
5 International Journal of Financial Services Management, Vol. 5 No. 4, pp. 303–320.
6 Shehata, N., and Dahawy, K. M. (2013), “2013 Review of the implementation status of corporate
7 governance disclosures: Egypt”, Paper presented at the United Nation Conference on Trade and
8 Development, 6-8 November 2013, Palais des Nations, Geneva.
9
10 Shleifer, A., and Vishny, R. (1997), “A survey of corporate governance”, The Journal of Finance, Vol. 52
11 No. 2, pp. 737–83.
12 Subramaniam, C. and Weidenmier, M. L. (2003), “Additional evidence on the sticky behavior of costs”,
13 working paper, Texas Christian University, Texas.
14
15 Vafeas, N., and Theodorou, E. (1998), “The relationship between board structure and firm performance in
16 the UK” The British Accounting Review, Vol. 30 No. 4, pp. 383–407.
17 Weiss, D. (2010), “Cost behavior and analysts’ earnings forecasts”, The Accounting Review, Vol. 85, No. 4,
18 pp.1441-1474.
19
20 Xue, S., and Hong, Y. (2015), “Earnings management, corporate governance and expense stickiness”, China
21 Journal of Accounting Research (in press).
22 Yermack, D. (1996), “Higher market valuation of firms with a small board of directors”, Journal of
23 Financial Economics, Vol. 40, No. 2, pp. 185–211.
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
19
Accounting Research Journal Page 20 of 23
1
2
3
4 Table 1: The Study Samples (2008-2013)
5 Basic Sample CG Sample
6 Starting Samples 600 600
7 (-) Observations of financial institutions (120) (120)
8 (-) Observations with missing data on either costs or net sales (25) (25)
9 (-) Observations with costs > net sales (23) (23)
10 (-) Outliers (6) (10)
Final Samples 426 422
11 The 100 companies of the index (EGX100) represent the initial study sample with 600 observations during a six-year
12 study period (2008-2013). Then, I excluded 20 financial institutions with 120 firm-year observation.
13
14 Table 2: The Variables’ Definitions and Measurements
15 Variable Definition Measurement Source
16 Dependent Variable:
17 Change of Cost It is measured as the COGS of year t divided
∆ COGSit Annual Report
18 of Goods Sold by that of year t-1 for the firm i.
19 Independent Variables:
20 It is measured as the net sales of year t divided
∆ Salesit Change of Sales Annual Report
21 by the net sales of year t-1 for the firm i.
22 A dummy variable that equals 1 if the current Created based on
Dummy
23 DecDummyit year’s net sales are less than the previous data from annual
Variable
24 year’s net sales and 0 otherwise. report
25 A two-way interaction term that results from
Created based on
26 DecDummyit Interaction- multiplying the dummy variable by the natural
data from annual
27 ×Log (∆ Salesit) Term logarithm of changes in net sales for the year t
report
28 and firm i.
29 Corporate Governance Variables:
30 The total number of directors on the board of
Board size it Size of Board Disclosure Book
firm i during the year t.
31
A dummy variable takes the value 1 if the
32 Board Chairman
Role duality it board chairman and CEO are the same person Disclosure Book
33 Role Duality
and 0 otherwise.
34 The ratio of the number of non-executive
35 Non-executives
Non-executives directors on the board to the total number of Disclosure Book
36 ratio it
board directors of firm i during the year t.
37 Control Variables:
38 An indicator variable that takes the value 1 if Created based on
Successive Dummy
39 the net sales of the year t-1 are lower than the data from annual
Decrease it Variable
40 net sales of the year t-2, and zero otherwise. report
41 The percentage growth in real gross domestic
42 product during the year t, which is used as a International
43 Economic Growth it Real GDP proxy for economic growth. It was obtained Monetary Fund
44 from the website of International Monetary Website.
45 Fund.
46 The total number of shares held by
Ownership of
Institutional institutional investors divided by the total
47 Institutional Disclosure Book
Ownership it number of outstanding shares of firm i during
48 investors
the year t.
49
50
51
52
53
54
55
56
57
58
59
60
Page 21 of 23 Accounting Research Journal
1
2
3 Table 3: Descriptive Statistics*
4 Mean Median St. Deviation
5 Panel (A): COGS & Net Sales
6 COGS 943 170 2,498
7 Net Sales 1,312 282 3,012
8 COGS % 72% 60% 83%
9 Panel (B): Corporate Governance Variables
10 Board Size 7.51 7.00 2.85
11 Role Duality 0.65 1.00 0.50
Non-Executives% 0.69 0.75 0.19
12
Panel (C): Control Variables
13
Successive Decrease 0.40 0.000 0.49
14
Economic Growth% 3.16 2.2 1.42
15
Institutional Ownership 0.20 0.02 0.27
16 *The reported numbers are in millions of Egyptian pounds. The results presented in table exclude observations with
17 costs exceeding revenues for the current year, but do not exclude outliers.
18
19 Table 4: Results Running Model 1
20 Condition Year
Con. β1 β2 β1+β2 Adj. R2 F-value VIF*
21 Index* Dummies
22 COGS -0.04 1.05*** -0.20*** 765.3 Suppressed
0.85 0.84 2.14 2.06
23 Model (-0.98) (33.79) (-2.86) 0
24 Model (1): Log (∆ COGSit) = β0 + β1 Log (∆ Salesit) + β2 DecDummy * Log (∆ Salesit) + Σ
25 *Condition Indexes are less than five and VIF values are less than three, which indicate the non-existence of
26 Multicollinearity problem (Kennedy, 1992; Herrmann et al., 2003). T-values are in parentheses.
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accounting Research Journal Page 22 of 23
1
2
3 Table 5: Regression Results of Models 2 & 3
4 No Standalone Standalone
5 Before After Before After
Variables Statistics
6 Controls Controls Controls Controls
7 (1) (2) (3) (4)
8 -0.03** -0.02 -0.01*** 0.10
β0: Intercept
(-2.14) (-1.20) (-2.90) (1.60)
9
1.09*** 1.13*** 1.11*** 1.07***
10 β1: Log(∆Sales)
(42.20) (39.33) (45.87) (42.75)
11 -0.29*** -0.26*** -0.31*** -0.19***
12 β2: DecDummy×Log(∆Sales)
(-8.25) (-4.27) (-9.85) (-3.84)
13 Three-way Interaction Terms (DecDummy × Log∆Sales
14 × Variable)
15 -0.05*** -0.18*** -0.06*** -0.19***
16 β3: DecDummy×Log(∆Sales)× Board Size
(-2.89) (-8.90) (-2.69) (-12.04)
17 -0.58*** -0.17** -0.60*** -0.19***
β4: DecDummy×Log(∆Sales)×Role Duality
18 (-13.06) (-1.92) (-11.29) (-2.84)
19 -1.90*** -0.96*** -1.85*** -0.95***
β5: DecDummy×Log(∆Sales)×Non-executives
20 (-11.01) (-4.46) (-9.14) (-5.81)
21 0.38*** 0.38***
β6: DecDummy×Log(∆Sales)× Successive Decrease --- ---
22 (3.81) (4.81)
23 0.10*** 0.10***
β7: DecDummy×Log(∆Sales)× Economic Growth --- ---
24 (2.77) (3.46)
25 0.56*** 0.57***
β8: DecDummy×Log(∆Sales)×Institutional Ownership --- ---
(3.28) (4.29)
26
Standalone Variables (Variables without interaction)
27
--- 0.000 -0.01*
28 β9: Board size ---
(-0.54) (-1.76)
29 --- -0.05* -0.04
30 β10: Role Duality ---
(-1.80) (-1.78)
31 --- -0.20** -0.06
32 β11: Non-executives % ---
(-2.31) (-0. 76)
33 --- 0.02
34 β12: Successive Decrease ---
(0.57)
35 --- 0.01
β13: Economic Growth% ---
36 (0.09)
37 --- 0.03
β14: Institutional Ownership ---
38 (0.60)
39 Year Dummies Suppressed
40 963.50 454.88 599.52 400.93
F-value (Sig.)
41 (0.000) (0.000) (0.000) (0.000)
42 VIF (Max) 3.21 6.19 2.89 7.18
43 Condition Index 3.64 6.85 10.85 15.97
44 Adj.R2 0.93 0.90 0.90 0.93
45 N 422 422 422 422
46 *, **, and *** stand for significance at 10%, 5%, and 1%, respectively. T-values are in parentheses.
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 23 of 23 Accounting Research Journal
1
2
3 Table 6: Results of Robust Regression Analysis
4 Board Size Role Duality Non-Executives
5 >= Median < Median Role Duality Non-Role Duality >=Median < Median
6 Con. -0.04** -0.01 -0.02 0.02 -0.002 -0.03
7 (-2.25) (-0.27) (-0.45) (0.10) (-0.10) (-1.22)
8 β1 1.02*** 1.10*** 1.10*** 0.89*** 0.95*** 1.12***
(35.38) (20.12) (18.05) (18.65) (20.44) (37.19)
9
β2 -0.33*** 0.15*** -0.17** 0.19* -0.19*** 0.16*
10
(-8.07) (2.78) (-2.75) (1.78) (-3.90) (1.74)
11
F-value 1,320 754,28 280,65 356,50 798,25 1,244
12 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
13 Adj. R2 0.88 0.90 0.75 0.83 0.87 0.90
14 Model (1): Log (∆ COGSit) = β0 + β1 Log (∆ Salesit) + β2 DecDummy * Log (∆ Salesit) + Σ
15 *, **, and *** stand for significance at 10%, 5%, and 1%, respectively. T-values are in parentheses.
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60