Corporate governance and real activity manipulation: the
case of
advertising spending
Master Thesis
Student Christian d’ Hooghe, 6156509
University of Amsterdam
August 2014
ABSTRACT
This study examines the role of board of directors in constraining questionable advertising spending levels. Extant research on earnings management indicated that independent directors reduce accounting accrual manipulation. However, there is little evidence on the effectiveness in limiting real earning management practices. Using a sample of US companies, I study whether independent boards and or higher degree of woman on boards are more likely to constrain (1) advertising decreases around benchmarks and (2) abnormal levels of advertising spending, both negative and positive. The results indicated that a higher board independence constrain advertising cuts and negative abnormal levels of advertising. However, they don’t constrain positive abnormal levels of advertising; actually they use it more than less independent directors. With respect to a higher proportion of female on boards, the results indicated that woman only constrain advertising cuts. The evidence support the emphasis that recent policy statements have put on increasing independent directors on corporate boards.
Keywords
Independence, female, board of directors, Agency theory, real earnings management,
advertising spending
Supervisor
Alexandros Sikalidis (UVA)
CONTENT
I Introduction pp. 3-4
II Literature and hypothesis development
2.1 Earnings pressure and earnings management p.5-6
2.2 Optimal level of advertising p.6-7
2.3 Advertising decreases around benchmarks p.7 2.4 Advertising increases around benchmarks p.7-8
2.5 Abnormal levels of advertising p.8
2.6 Advertising decreases p.9
2.7 board monitoring and advertising spending p.9-12
III Research design
3.1 Empirical model pp.12-16
3.2 Sample selection Procedure pp.16-19
IV Results
4.1 Descriptive evidence p.19-28
4.2 Main regression results p.28-30
V Conclusion p.31
VI References pp.32-35
VII Attachments p.36
I INTRODUCTION
According to Osma (2008,p.116) who refer to shleifer and Vishny (1997) corporate governance encompass all the provisions and mechanisms that guarantee the assets of the firm are managed efficiently and in the interests of the providers of finance, mitigating the inappropriate expropriation of resources by managers or any other party to the firm. Osma (2008,p.116) suggest that the board of directors is at the center of this decision making and control system and, therefore plays a fundamental role in the corporate governance of large companies (Fama and Jensen, 1983).
According to Osma (2008,p.116) who refer to Weisbach (1998) and Byrd and Hickman, (1992) extant research on board of directors’ influence and characteristics confirms that independent directors influence board decisions and referring to Dechow, Sloan and Sweeny, 1996; Peasnell, Pope and Young, 2005) are capable of detecting and constraining earnings management practices. Although there is general agreement that independent and female boards limit accounting accruals manipulation (Bedard, 2001), there exist less evidence that they also limit real activity manipulation. The only evidence so far came from Osma (2008, p.116) who provide evidence that independent boards are efficient at detecting and constraining myopic R&D cuts. This paper extends previous research in this area, by explicitly analyzing abnormal advertising spending levels, looking to how independent and female directors have a effect on this.
According to Osma (2008, p. 116) prior work on the association between board monitoring and accounting quality has focused on analyzing accrual accounting decisions. However, there are other instruments that manager may use to meet short term goals. Although Osma (2008, p.116) refer to myopic R&D cuts, I refer to opportunistic shaving of advertising spending, especially to Cohen (2010). He finds that managers reduce advertising spending to avoid losses and earnings decreases. Trimming advertising expenditure can be an attractive earnings management alternative, because one prevailing characteristic of advertising activity is the information gap what the manager knows about the investment opportunity of the firm and what is disclosed to corporate boards and other parties. Corfman and Lehmann (1994, p.35) stated that the setting of advertising budgets encompass a very complex strategic task. This information asymmetry may originate from, according to Cohen (2010), the difficulty to estimate (1) the actual response of sales to advertising and (2) the
lasting effect of advertising on future sales or carry over effect.
This asymmetry may reduce the effectiveness of the monitoring exerted by corporate boards. Further on, considering the board dual duty of advising and monitoring, it is not necessarily obvious that independent and female directors will consider in the primary duty to question managerial advertising decisions, nor they have the required expertise to do so. The same stated Osma (2008, p. 116) in prior research, but than in the case of managerial R&D expenses. In addition Osma (2008, p. 116) stated, referring to Adams and Ferreira (2007) that it is likely that managers disclose less information to boards that are perceived to be less friendly. Therefore because of this pervasive information asymmetry, the technical difficulties and the specialized knowledge required to distinguish opportunistic and efficient advertising decisions, it is possible that managers can use their superior knowledge in convincing boards that less advertising spending is optimal, even when they are not, for example in the long run. Although Corfman (1994, p.36) who refer to Cyert and DeGroot (1973); Eliashberg 1981; Hotelling 1929; Moorthy 1985) stated that considerable past research and decision making under uncertainty had described what managers ought to do in competitive situations and therefore the board can know which behavior is the “good” behavior, this serves only as guidance.
This paper analyzes if independent and or female directors are effective in curbing advertising spending abnormal spending levels in the presence of earnings-related incentives. According to Osma (2008, p. 117) short term earnings pressures are expected to entice management into behaving opportunistically, as managers become overly concerned about the value effect of earnings disappointments and lower their compromise with the firm long term performance. If independent and or a higher proportion of female on boards have sufficient expertise to see through this type of real earnings management, it is expected that they will constrain it, reducing the incidence of opportunistic advertising spending
The role played by governance mechanism, such as board of directors in curbing managerial advertising decisions., is a relevant issue that has not been addressed by previous research (e.g. only on R&D decisions) and that has significant policy implications.. This study seeks to fill this relevant gap.
II LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT
2.1 Earnings pressure and earnings management
According to Osma (2008,p.116) managers face pressures to report earnings that meet certain targets. Extant research, according to Osma (2008,p.116) , documents that an unusually high (low) number of firms report small profits (losses) or small increases (decreases) in earnings (Burgstahler and Dichev 1997; Gore, Pope and Singh, 2007) Also she stated that earnings management is widely accepted as the cause of this discontinuity in the earnings distribution around the zero reference points, as well as that this manipulation of accounting income is perhaps unsurprising, as there exist a market premium (penalty) for meeting (missing) expectations (Barth, Elliot and Finn, 1999, Kasznik and Ncnichols, 2002) that persist even when the targets are likely to have been met through manipulation (Bartov, Givoly and Hayn, 2002). Even Gunny (2010) stated that real earnings management used to achieve benchmarks will lead to better future performance. Finally according to Graham et all (2005) missing targets breed uncertainty about a firm’s future prospects, which managers believe hurts stock valuation.
According to Osma (2008,p.118), following Schipper (1989), it is broadly accepted that earnings management activities can be classified into two levels: (i) purely accounting decisions, such as accrual accounting manipulation (Dechow et al ,1995; Peasnell et al, 2000 ; Garcia Osma and Gill- de- Albornoz Noguer, 2007); and (ii) real earnings management actions that affect the activities of the company. Also she stated that accounting accruals manipulation is generally regarded as less visible and costly method and, thus as the one preferred by management to meet their income targets. However, she also stated that recent survey evidence questions this view and suggest that the manipulation of real activities is also widespread (Graham, Harvey and Rajgopal, 2005) Specifically Cohen (2010, p. 811) refer to Roychowdhury (2006) who find evidence that firms trying to avoid losses in several ways using real earnings management: (1) boosting sales through accelerating their timing: (2) generating additional unsustainable sales th rough price discounts (3) overproducing and thereby allocating less overhead to cost of goods sold and (4) aggressively reducing aggregate discretionary expense (sum of research and development, advertising and selling general and administrative expense). Cohen (2011) stated that these events are most likely if such expenses do not generate immediate revenues and income.
Although R&D expense easily be called expenses directed to the long term, also stated by Osma (2008), and therefore any decrease around benchmarks can be seen as opportunistic (bad for the firm in long term), off course controlling for other variables that drives this cuts, it is another case when we talk about advertising expense. Cohen (2010, p.812) stated “because advertising have a more immediate impact on sales than for example R&D, this can lead to the possibility that managers may increase advertising to generate a positive short term response in revenues and earnings”.
2.2 Optimal level of advertising
As mentioned before, managers can increase of decrease advertising spending to meet benchmarks. However, only if it deviates from optimal advertising spending (=best for the firm), than it can be seen, according to Cohen (2010), as real earnings management. According to A*(optimal advertising level) he stated the following:
“Before managers alter current advertising to influence financial reporting outcomes, they were committed to spending an optimal amount predetermined level of advertising” (A*) given the prevailing business conditions and information.”
(Cohen, 2010,p.813)
Further on he stated that the optimal level maximizes total expected life time earnings of the firm : II1 (current period earnings) + II2 (sum of earnings in future periods beyond this point). Given that managers objective is to increase current earnings, and the maintained assumption, real earnings management results in a reduction in life time earnings earnings and firm value.
According to Cohen (2010);
Theory (Adec <A*)
A manager interest in increasing current reported earnings would reduce advertising only if the decrease in earnings resulting from declining sales does not offset the increase in earnings from the direct effect. (time lag between advertising and the reaction in sales is relatively large, or elastisticity to advertising is not high) Given that you advertising level first was at optimal level, the fact that you now increase II1, means that life time earnings will be
decreased more than the increase in II1 , so in the end this means lower life time earnings. (bad for the firm in the long term)
Theory (Ainc> A*)
A manager interest in increase current reported earnings would increase advertising only if the impact on sales and earnings is offset by the reduction in earnings (cause by more advertising expenses) (lag between advertising and sales is short or the elasticity of advertising is high or both). However, if managers choose higher but suboptimal level of advertising, then the increase in current reported earnings resulting from an increase advertising must be offset by a larger reduction in future earnings.
2.3 Evidence relating to advertising decreases around benchmarks
Beside Roychowdhury (2006), also Graham et all (2005) find evidence that managers reduce discretionary expenses to achieve benchmarks. According to Osma (2008) , in Graham et all. (2005) a large number of respondents admit to reduction of discretionary expenditures and/or capital investments than engaging in other real activity manipulation. According to Zhang (2010, p. 743), the survey of Graham, Harvey, and Rajgopal (2005), indicated that 80% percent of the respondents (CFO’s and financial executives) would decrease discretionary spending to meet an earnings target. Because advertising expense is only a part of discretionary expenditures, we can only partially use this percentage, as evidence found by prior literature (Graham, Harvey, and Rajgopal , 2005, p.33), to assist our assumption that a relative high degree of advertising spending framing is used, relative to other tools to meet
earnings targets.
Cohen (2010) stated this “ while the implied conclusion from prior literature is that documented aggregate effects can be attributed to each of the components equally; he also stated that there is no direct evidence on this” . Cohen (2010, p.808) provided in his research direct evidence whether advertising is used as a tool for real earnings management activities. He found that managers on average, reduce advertising spending to avoid reporting a quarterly loss and meeting earnings from the same quarter in the previous year.
2.4 Evidence relating to advertising increase around benchmarks
Beside evidence of advertising decreases around benchmarks, Cohen (2010) also find evidence that managers increase advertising to achieve benchmarks. Firms in late stages of their life cycle increase advertising to meet earnings benchmarks, and they find evidence that firms increase their advertising in the third month of the fiscal quarter to meet 7
earnings from the same quarter in the previous year. Finally they find that suspect firms are more likely to increase their advertising spending in the fourth fiscal quarter compared with other quarters.
2.5 Abnormal levels of advertising
Cohen (2010) used a time series approach to estimate normal advertising spending or with the maintained assumption, to estimate A* or optimal advertising level. To estimate normal advertising level I use modification of the quarterly earnings model of Foster (1997). Cohen (2010,p.818) used this model to compute monthly advertising outlays. However, I use it to compute annual advertising outlays:
E(ADSt) = θ1 ( Ann advertising t-1) + θ2 (annual advertising t-1 – annual advertising t-2) Where E(ADSt) is expected advertising outlays in year t; annual advertising t-1 is an autoregressive term; and (annual advertising t-1 – annual advertising t-2) is a drift term. Further on, I also assume that advertising outlays follow a first order autoregressive process in annual differences. Estimating advertising by taking prior year advertising spending level and adding up the drift term, I find further in this research that E(ADSt) is almost never equal, in 99% of the cases, to actual advertising spending level in year t. This indicates in my opinion that most firms fail to determine initially an optimal advertising spending level. I don’t have already include control variables for these findings, but I think I can consider the assertion as taken for granted because the founded percentage, 99 percent, is very high. The reason that Cohen (2010) find other results (less deviations with respect to optimal level of advertising spending, in this case E(ADSt)) I think is due to the fact he looked to monthly advertising changes, something which I can’t investigate because I can’t use the specific database he used. Because almost every firm deviates from A* in my sample, it seems as a result that everyone engage in real earnings management. However, this is due to the fact that I had to exclude a lot of observations due to missing values. If I was able to include this observation there was less real earnings management founded. However, I am not, therefore instead I look to both negative and positive abnormal advertising levels separately. Secondly I look whether certain board characteristics constrain either types (positive and negative abnormal advertising spending levels) or just one type, for example only negative abnormal spending levels.
2.6 Advertising decreases
Although one can say that decreases can be part of a strategy and only abnormal levels of advertising can harm the business (Cohen, 2010, p. 818), in my opinion decreases in advertising spending can also be opportunistic. According to Osma (2008,p.117) who refer to David, Hitt and Gimeno (2001) managers investing in R&D face a temporal tradeoff: R&D expenditures are incurred in the short term while payoffs are often only over the longer term. While the payoff relation to advertising spending is less long term than for example R&D, I think it works in general the same for advertising spending. The payoff of increased advertising is often unsecure; you expense but do not know whether you get an extra payoff or not. According to Cohen (2010) it is only useful if there is a high advertising elasticity. However, this is difficult to estimate beforehand. Actually the only thing that you know, in case of increased advertising spending, is that your expense increase. You don’t know whether you get a payoff of it. In turn, decreasing your advertising spending means that your expense decrease and that at the same time it immediately leads to a higher current income pretax income, regardless their outcome. Although according to Cohen (2010, p.815) a decrease can also harm you short term result, if carryover effect is low for example, however, this isn’t sure. The only thing you know is that you decrease your expense and that it will be at some point bad for the firm. As a result, I think that more independent boards and a higher proportion of female on boards will constrain advertising cuts, but beside, also more likely to constrain negative abnormal advertising spending levels than abnormal advertising increases.
2.7 Board monitoring and advertising spending
As mentioned before, real earnings management occur when advertising level deviates from optimal advertising spending A*. Hence, independent and female boards likely have to question managerial advertising expenses deviations from A*, distinguishing a opportunistic from a efficient one can be difficult, determining the optimal amount (A*) of advertising is often a very complex task.
As stated by Corfmann and Lehmann (1994, p.35), the dominant solution for a single play is not that difficult: choose high advertising budgets. However, often there are a lot more players active in the market. As a result board members have to judge about manager’s judgment relating to the prediction of (1) the actions which the opponent will take (2) the length of time the competitor was expected to remain in the market (3) the influence which competing firms’ market shares have. Further on, if the optimal amount is known and the element aren’t determined correctly in the view of the board member, they have to consider
(4) the term of the subjects profit objectives (short vs. long) to look whether the advertising level optimal or not. However, according to Cohen (2010) often the optimal amount is not known because of uncertainties about what the concurrent is doing for example. Individual characteristics of the manager play in that case a role (Corfmann and Lehmann ,1994, p.35) In my opinion a manager can use his knowledge about individual board characteristics to cover opportunistic behavior. For example in the case he state that some things are unsecure, while actually they aren’t. Describing the following situation: If a manager can convince the board of a more “risk averse” play, than this can be either (1) a opportunistic one (manager don’t tell all information, as a result a investment seems risky, while actually it is not) or (2) a efficient one (managers tells all information, as a result the board is more aware the full amount of uncertainties which the firm face and therefore it is in the best interest of the firm to avoid the risk. As a result, this makes it very hard for the board to explore the truth.
What one can say is that A* is very difficult to determine. However, given the maintained assumption that initial A spending is A*, and secondly using the time series approach, the boards of directors can note any changes and as a results can question this changes. Looking to the mentioned literature I think that board independence has a moderating (constraining) effect on the amount of earnings management. In developing my hypotheses I first consider earnings pressure. Because earnings pressure will lead to a higher possibility that managers are more short term focused, I assume that they are more forced to gain immediately gains than gambling on later payouts or gains. So my first hypothesis, which results I will explain in the descriptive statistics part, is the following:
H1 Managers who do face previous earnings losses and earnings decreases decrease advertising relatively more than firms which do not.
But also
H2 Managers who do face previous earnings losses and earnings decreases are more likely to engage in negative and positive abnormal advertising levels that those which do not.
Relating to the moderating effect of the board of directors, I assume the following.
H3: A more independent board use decreased levels of advertising relatively less than independent boards when facing short term earnings pressures
But also
H4 A More independent board use negative abnormal level of advertising relatively less than less independent boards will do when facing short term pressures.
And
H5 A more independent board don’t constrain positive abnormal levels of advertising when facing short term pressures.
Looking to the above two hypotheses, however, I have to consider the statement of Osma (2008, p.118) “having more independent directors, which is positive, may be compromised by their lack of specialized technical expertise”. (e.g. dependent directors possibly pose more expertise) so possibly they act not like what is expected from them. According to Osma (2008, p.118) who refer to Beekes, Pope and Young (2004) , two conditions must be met for independent directors to be efficient monitor: (i) they must possess sufficient incentives to monitor, and (ii) they must understand the consequences of managerial actions over the financial reporting system. According to these two conditions, I assume US settings creates sufficient incentive for directors to monitor management (Peasnell et al, 200o). However, if mentioned before, there may be some doubt on whether the second condition is always met. For my following two hypotheses. I will look to another variable “board gender diversity” Previous literature (Krishnan and Parsons) founded evidence that earnings quality is positively associated with board gender diversity in senior management, which means that there is less accrual earnings management found at companies where senior management consist out of a higher degree of woman. As a result I am wondering whether a higher proportion of females in board also contribute to constraining earnings management, specifically now considering the case of discovering and decreasing real earnings management, especially looking to advertising expense. As a result I introduce two additional hypotheses:
H5: Higher proportion female boards use decreased levels of advertising relatively less than lower proportion female boards when facing short term pressures
But also
H6: Higher proportion female boards use negative abnormal level of advertising relatively less than less independent than lower proportion female boards when facing short term pressures.
And
H7: Higher proportion of female boards don’t constrain positive abnormal levels of advertising when facing earnings pressure
III RESEARCH DESIGN
I follow extant research (Low and Mohr, 1999, p.3) in this to isolate situations where advertising cuts are likely reactions to short term pressures. Managers face pressures to meet simple earnings benchmarks, as explaned in the above section. Particularly whenever these benchmarks are missed, managers are expected to feel pressure into avoiding a second earnings disappointment. Thus, I consider two main earnings targets that can potentially create incentives to shave advertising expenditures: avoiding reporting (1) consecutive losses, and (2) consecutive earnings decreases.
To isolate firms that face pressures to prevent a repeated disappointment, I follow the method used by Osma (2008,p.118) and Garcia Osma and Young (2007). Firms that report a loss or an earnings decrease in the previous period are identified and labeled as Miss (Zero) and Miss (Growth) respectively. Miss (zero) takes the value of 1 if the firm reported a loss in the previous period; 0 otherwise. Miss (Growth) takes the value of 1 if the firm reported an earnings decrease in the previous period; 0 otherwise. All things being equal, it is expected that firms that missed their earnings targets will face additional pressures to cut advertising spending in the current period.
Once earnings pressures are identified , I model the probability that firms will decrease adverting, conditional on the existence of target beating incentives and board monitoring, controlling for other factors that drive advertising spending deviations. Formally:
Model 1
P(CUT) = f( p0 + p1 (BOARDINDEPENCE orFEMALE BOARD)+ p2 (EARNGOAL) + p3 (BOARD INDEPENDENCE or FEMALE BOARD)* (EARNGOAL) + controls
Where f(-) is the logistic cumulative density function. CUT equals 1 if advertising will be reduced; 0 otherwise. CorpGov (IND) is a measure of board independence , alternatively:
BDIND, defined as the percentage of independent directors on the board; or INDOM, a
dummy variable that equals 1 if the board is dominated by insiders, defined as having 1 or no independent directors, BDIND measures board independence, i.e., higher values are associated with more independence, whilst INDOM = 1 signifies reduced independence. EARNGOAL is a dummy variable that equals 0 if the firm faces pressures to meet an
earnings target; 0 otherwise, and is defined, alternatively as Miss (zero) and Miss(growth). If these variables capture short-term pressures to reduce advertising, it is expected that p2 will be positive. If independent directors are effective in constraining advertising cuts in the presence of earnings pressures, p3 will be negative when Earngoal is interacted with BDIND, and positive when interacted with INDOM.
Although advertising cuts can be opportunistic, because a reduce means a immediately gain, while an increase not, they can however also be part of a strategy. Therefore, as mentioned earlier, I use a time series model to calculate the normal advertising level. Estimating normal advertising level spending or estimating A* (optimal advertising level), I will run the following model
Model 2
E(ADSt) = θ1 * ( Ann advertising t-1) + θ1 (annual advertising t-1 – annual advertising t-2) + εt
Where E(ADSt) is expected advertising outlays in year t; annual advertising t-1 is an autoregressive term; and (annual advertising t-1 – annual advertising t-2) is a drift term.
Further on, I also assume that advertising outlays follow a first order autoregressive process in annual differences. Estimating advertising by taking prior year advertising spending level and adding up the drift term. Actually I have derived this model from another model which is introduced by Foster (1977). Cohen (2010) derived his monthly time series model also from his model. For more background information about how he is doing that, I can refer to Cohen (2010,p.818) My own results suggest that almost each firm deviates from A*. (If A* has to be equal to E(ADSt). From the 1097 observations, 11 observations have the outcome of zero, if you subtract E(ADSt) from actual advertising spending level in year t. This indicates that achieving benchmarks will be in general preferred, even if these have to be achieved by real earnings management. I think this is due to the fact that missing a target has big consequences, while achieving them can be even in the long term beneficial, also if this is achieved by real earnings management. For example, Gunny (2010), provided evidence that real earnings management lead to achieving benchmarks and in turn to better future performance.
So considering this, it will be possibly appropriate to make a distinction between negative abnormal levels of advertising (A<E(ADSt)) and positive abnormal advertising levels of advertising (A>E(ADSt)). So I introduce two new models, one for negative abnormal advertising spending and positive abnormal advertising spending around benchmarks, looking whether board characteristics have any moderating effect on this. As mentioned earlier, I expect that a higher board independence and higher proportion in boards only constrain negative abnormal levels of advertising.
Model 3
P(ABN (ADV) (-)) = f ( p0 + p1 (BOARDINDEPENCE, FEMALE BOARD)+ p2 (Miss Zero,
Miss growth) + p3 (BOARD INDEPENDENCE, FEMALE BOARD) or * (Miss Zero, Missgrowth) + controls
Where f(-) is the logistic cumulative density function. ABN (ADV)(-) equals 1 if advertising level is abnormal negative; 0 otherwise. (other variables are the same as in model 1) Model 4 is for positive abnormal advertising levels, formally
Model 4
P (ABN _ADV_+) = f (f( p0 + p1 (BOARDINDEPENCE, FEMALE BOARD)+ p2 (Miss
Zero, miss growth) + p3 (BOARD INDEPENDENCE, FEMALE BOARD) or * (Miss Zero, Missgrowth) + controls
Where f(-) is the logistic cumulative density function. ABN (ADV)(+) equals 1 if advertising level is abnormal positive; 0 otherwise. (other variables are the same as in model 1)
Controls
Model 1, 3 and 4 includes a vector of control variables that drives advertising cuts and positive and negative abnormal levels of advertising spending, partly the same as used in the research of Cohen (2010, p.822-823)
_AACC (CS: AT, RECT) is abnormal accruals calculated using the modified Jones model (Dechow et al, 1995) and is a proxy of an alternative to real earnings management. If management favors the use of AACC , increased accounting manipulation will lead to lower probability of deviations from A*. (TA it / A it-1 = α0 (1/Ait-1) + α1 (delta Revert - delta RECit)
/ Ait-1+ α2 PPEit / Ait-1 + α3 IBXIit-1 / Ait-1+εit))
_MB, (CS: MKVALT / (AT-LT)) = the book to market ratio at the beginning of the fiscal year. The book-to-market ratio attempt to identify undervalued or overvalued securities. If securities are overvalued (ratio is below zero) than it is more likely that advertising spending will be reduced. However, if securities are undervalued (ratio is above zero) than it is more likely that advertising spending will be increased.
Table 1
_________Variable Definitions and Expected Relation to advertising cuts __________
Variables Definition ____________
CUT _AD = 1 if firm cut advertising in the current period; 0 otherwise
ABN_AD (-) = 1 if the firm use negative abnormal advertising levels; o
otherwise
ABN_AD (+) = 1 if the firm use positive abnormal adverting levels;
0 otherwise_________________________________________ CorpGov Variables_________________________________________________________
BDIND = 1 if the proportion of independent board members is greater than 50 percent; 0 otherwise.
INDOM = 1 if the board has 3 or less independent directors; 0 otherwise.
BDFEM = 1 if the proportion of female on boards is greater than median; 0 otherwise.
MALEDOM = 1 if the board has 0 or 1 female directors; o otherwise.____
_EarnGoalvariables___________________________________________________________ Miss (Zero) (1) = 1 if the firm reported a loss in the previous period; 0 otherwise
Miss (Growth)(2) = 1 if the firm reported an earnings decrease in the previous period;
0 otherwise.
_________________________________________________________________________ Control variables__________________________________________________________ AACC = Abnormal accruals, if this is high, advertising cuts are less likely because another form of earnings management is already used MB = Market to book ratio, if this ratio is lower than 1,
this means that a specific firms is overvalued and thus
advertising cuts are more likely ____________________________
Sample selection procedure (1)
Data for all US non – financial firms that expense on advertising and have at least three years of consecutive data to compute annual advertising account movements, are compiled from Compustat alive and dead stock files. This procedure avoids survivorship biases. Financial
companies are excluded (8) because their accrual generating process differs significantly, altering their incentives and opportunities for successful earnings management (Ahmed, Takeda and Thomas, 1999; Young 1999).
The sample starts in 1996 and spans all fiscal years up to 2006. Firms years with missing data or witch changes in fiscal year end are excluded (9) otherwise accounting figures might not be comparable. This procedure results in 1097 firm years (Table 2) that spans 76 Industries (table 3)
_________________________________________________________________________ Table 2
___________________Sample selection North America_ ___________________________ 1996-2006 Compustat
All data items initial observations 130052 firm year observations 1. Excluding financial companies 92860 firm year observations 2. Spending on advertising greater or equal to zero 23008 firm year observations 3. Excluding missing values 3848 firm year observations
(1291 companies)
4. Exclude those with less than 3 consecutive years 3455 firm year observation
(1151 companies)___ __________
1996-2006 Risk Metrics
6. Only include those companies which
have data available on risk governance data 2047 firm year observations (Board independence, Female, board size) (320 companies)
7. Excluding the first two years of each firm
(to compute values of Miss (growth)) 1407 firm year observations
(320 companies)
8. Only include those companies years where
risk governance data is available 1097 firm year observations _____________________________________________ (final)_______________________
Because there a lot of missing values, many observation are eliminated. Further, it was a pity that a lot of information wasn’t available on risk metrics database.
_________________________________________________________________________ Table 3
_____Compustat Sample Industry composition North America___ ________________ Code Description Sample % Code Description Sample % _ ____________________# firm years____ __________________________________ 120 Automobiles (12) 1,09 345 Machinery (2) 0,18 130 Auto Parts & Equipment (19) 1,73 355 Manufacturing(div)(16) 1,46 135 Trucks & Parts (5) 0,46 357 Manufacturing (spe.) (31) 2,83 140 Beverages (Alcohol) (7) 0,64 370 Office equipment (5) 0,46 145 Beverages (Non-alcohol) (18) 1,64 382 Oil and Gas (ref.) (2) 0,18 147 Biotechnology (20) 1,82 385 Oil (4) 0,36 155 Building materials (6) 0,55 395 Oil and Gass (dri.) (5) 0,46 165 Chemicals (diversified) (3) 0,27 400 Paper & Forest (3) 0,27 175 Services (Comm. & Cons.) (4) 0,36 403 Photography (6) 0,55 180 Communication Equip. (10) 0,91 410 Publishing (9) 0,82 185 Computers (Software) (54) 4,93 415 Publishing (news) (9) 0,82 188 Computers (Peripherals) (13) 1,19 420 Restaurants (55) 5,01 190 Computers (hardware) (17) 1,55 426 Retail (comp.) (10) 0,91 202 Construction (cement) (2) 0,18 430 Retail (dep.) (19) 1,73 203 Consumer (Jewelry) (3) 0,27 432 Retail (disc.) (21) 1,91 210 Containers (Paper) (3) 0,27 435 Retail (drugs) (13) 1,19 215 Personal Care (7) 0,64 440 Retail (food) (9) 0,82 217 Distributors (Food & Health) (9) 0,82 445 Retail (merch) (14) 1,28 220 Electrical Equip. (23) 2,10 447 Retail (home) (1) 0,09 222 Electronics (component) (6) 0,55 449 Retail (Specialty) (63) 5,74 283 Healthcare (drugs) (15) 1,37 450 Retail building Sup. (16) 1,46 230 Electronics (Instrum.) (18) 1,64 452 Retail (apparel) (22) 2,01 235 Electronics (Silicon.) (31) 2,82 453 Services(data) (2) 0,18 245 Entertainment (5) 0,46 454 Services (comp.) (1) 0,09 247 Equipment Semiconductors (19) 1,73 455 Footwear (29) 2,64 250 Foods (62) 5,65 457 Services (Employ.) (12) 1,09 262 Gaming and Lottery (3) 0,27 458 Specialty printing (8) 0,73 270 Hardware and Tools (24) 2,18 460 Iron & Steel (3) 0,27
280 Health care (diversified) (12) 1,09 462 Telecom. (Wire) (1) 0,09 283 Healthcare (drugs-generic) (15) 1,37 463 Telecom. (Long) (7) 0,64 285 Healthcare (Drugs-major) (9) 0,82 465 Textiles (App.) (26) 2,37 289 Healthcare (long term c) (5) 0,46 467 Textiles (H.F.) (4) 0,36 292 Healthcare (specialized) (6) 0,55 470 Tobacco (4) 0,36 300 Healthcare (medical pr.) (32) 2,92 475 Leisure Time (45) 4,10 305 Homebuilding (12) 1,09 605 Airlines (17) 1,55 310 Lodging Hotels (7) 0,64 615 Truckers (2) 0,18 315 Household furniture (21) 1,91 620 Air freight (6) 0,55 320 Household products (20) 1,82 715 Telephone (26) 2,37 325 House wares (12) 1,09____________________________________ Note: COMPUSTAT composition data is based on data downloaded on July 2014
IV RESULTS
4.1 Descriptive Statistics
Descriptive statistics of main variables appear on table 4. Out of the full sample, 35% of firms are classified as cutting advertising, 46% of firms are classified as engaging in negative abnormal advertising spending levels and On average, independent directors represent more than half of the board composition (70,6%) with slightly over one fifth (21,1%) of the board being dominated by independent directors. On average, woman directors represent more than one tenth of the board composition (11.7%) The average firm in the study has a nine-member board, including six independent directors and three insiders or including one female director and eight male directors. Only 0,36% of the boards are almost entirely dominated by insiders (INDOM =1). However, 67,6% of the boards are almost entirely dominated by male directors (MALEDOM =1)
_______________________________________ Table 4
______ Descriptive Statistics_________ _____ Variable Mean Min Med Max_
CUT 0,35 0,00 0,00 1,00
ABN (-) 0,46 0,00 0,00 1,00
ABN (+) 0,53 0,00 1,00 1,00__
______________________________________
Variable Mean Min Med Max_
Corp Gov variables_______________________________ BDIND 0,68 0,00 1,00 1,00 INDOM 0,10 0,00 0,00 1,00 BDFEM 0,12 0,00 0,00 1,00 MALE________ 0,63__0,00 1,00 1,00_ Control variables_______________________________ AACC 2,00 -1,00 0,00 3,00 SIZE 8,00 1,00 8,00 13,00_ ____________________________________________________________________ Table 5
____________________Descriptive statstistics cont ______________________ PANEL A1 Last year’s earnings target: advertising decrease versus no advertising decrease______________________________________________________________
Last year’s earnings relative to target Chi-square statistic Achieved Missed (probability value)
(PREVIOUS LOSS)
Target: Earn t-1>0 (loss) 18,09 (<0,01) No advertising decrease 662 (68%) 52(47%)
advertising decrease 324 (32%) 59 (53%)
Sample size 986 111
(PREVIOUS EARNINGS DECREASE)
Target: ΔEarnt-1 >0
No advertising decrease 501 (71%) 213 (54%) 31.02 (<0,01) Advertising decrease 204 (29%) 179 (46%)
Sample size 705 392 ________________
Table 5 present results of a test of differences in the proportion of advertising
cuts and abnormal spending levels of advertising across sub samples. Panel A1 shows that firms that missed last year’s targets are significantly more likely to reduce advertising spending (53 and 39 per cent for target zero and growth, respectively) than firm that met their targets (30 and 29 per cent). The results indicates that firm reporting a loss or earnings decrease in previous period are more likely to reduce advertising spending. Alternative explanation for the founded outcome (possibly also more likely in the case of advertising spending instead of R&D spending), is the possibility that previous advertising spending has not the intended result (meeting the benchmark), and as a result it is more likely that advertising expense will be decreased. Panel A shows also that firms that achieved last year’s targets are significantly more likely to increase advertising spending (68 en 71% for target zero and growth, respectively) than firms that miss their targets (47% and 54%) Possibly this means that the advertising spending has the intended result (meeting the benchmark), and thus they are more likely to increase the advertising expense further. _____________________________________________________________________ PANEL A2 Last year’s earnings target: abnormal advertising decrease and abnormal advertising increases decrease____________________________________________
_____________________________________________________________ Last year’s earnings relative to target Chi-square statistic
Achieved Missed (probability value) (PREVIOUS EARNINGS DECREASE)
Target: ΔEarnt-1 >0
No abn. AD decrease 391 (55%) 202 (52%) 1,56 (0,10) Abnormal decrease 314(45%) 190 (48%)
Sample size 705 392
PREVIOUS EARNINGS DECREASE)
Target: ΔEarnt-1 >0
No abn. AD increase 321( 46) 194 (49) 1,56 (0,10) Abnormal increase 384 (54) 198 (51)
Sample size 705 392 _____________________
Considering abnormal advertising levels, Panel A2 given the fact that 99% of the companies years deviate from normal advertising level spending, calculated with a time series model, it is highly unlikely that firms deviate from normal advertising spending caused by differences in earnings pressure. The above panel looked whether abnormal advertising spending will be used more or less in the presence of earningspressure. De results indicate that firms that report a earnings decreases in the previous period are more likely to engage in negative abnormal advertising levels and less likely to engage in positive abnormal levels of advertising. This was also expected, because a decrease immediately results in fewer expenses, while looking to advertising increase, you have to gamble on future payoffs .
_____________________________________________________________________ PANEL B1: Deviations or abnormal levels by types of boards ________________
Board type 1
BDIND>0,5 BDIND<0,5 *INDOM = 0 **INDOM =1 All firms
No advertising decrease 601 (64%) 113 (59%) 659 (67%) 55(50%) Advertising decrease 304 (36%) 79 (41%) 327 (33%) 56 (50% Sample Size 904 193 986 111 Chi square statistic
(probability value) 3,65 (<0,05) (13,90 <0,01)
No abn decrease A 499 (55%) 94(48%) 541 (55%) 52(47%) abn decrease A 405 (45%) 99 (52%) 445(45%) 59 (53%)
Sample Size 904 193 986 111
Chi square statistic
(probability value) 2,70 (0,05) 2,58 (0,06)
No abn increase A 413 (46%) 102 (53) 455 (46%) 60 (54%) abn increase A 491 (54%) 91 (47) 531 (54%) 51 (46%)
Sample Size 904 193 986 111
Chi square statistic
(probability value) 3,28 (<0,05) 2,25 (<0.10)
Table 5 Panel B1 shows that firms with less independent boards, more insiders
(BDIND<0,5 and INDOM =1) are more likely to reduce advertising spending (41 and 50 per cent) than outside dominated boards (36% and 33%). This implicates a negative unconditional association between advertising spending and the percentage of inside directors. Further on, the table shows that less independent boards are more likely to decrease advertising. Alternative explanation for the founded outcome is the possibility that previous advertising spending has not the intended result (meeting the benchmark), and as a result it is more likely that advertising expense will be decreased. However, this table doesn’t take into account this. Further on this table shows that firms with less independent boards, more insiders (BDIND<0,5 and INDOM =1 ), are more likely to use negative abnormal levels of advertising spending, while more independent boards, more outsiders, are more likely to use positive abnormal levels of advertrising spending. So what one can say is that less independent boards are more likely to use negative abnormal levels of advertising spending and less likely to use positive abnormal levels of advertising, as expected.
_____________________________________________________________________ PANEL B2: Cuts by types of boards___________________________________ _____________________________ Board type 2_________ _______________ BDFEM>MED BDFEM<MED *MALEDOM = 0 **MALEDOM = 1 All firms
No AD dec. 369 (68%) 345 (63%) 281 (70%) 433 (63%) AD dec. 177 (32%) 206 (37%) 122 (30%) 261(37%) Sample Size 546 551 403 694
Chi square statistic
(probability value) 2,98(<0,05) 6,04 (<0.01) ____
Table 5 B2 shows that firms with more females on board (BDFEM>MED and
MALEDOM =0) are less likely to reduce advertising spending (32% and 30%) than boards with less females on board, (BDFEM<MED and MALEDOM1, both 37%) . Both results are significant. This implicates a positive unconditional association between advertising spending and the percentage of female directors. Although the results also show that less females are also likely to engage in less negative abnormal advertising spending, and more likely to engage in positive abnormal advertising, these results are not significant.
_____________________________________________________________________ Panel C1 Effect of board independence (BDIND) on decision to decrease or abnormally decrease advertising spending after missing target ________________ No significant results!
Focusing on BDIND (panel C1) I analyzed the impact of board independence on the decision to decrease advertising spending after missing a target (Earn t-1>0 and ΔEarnt-1 >0) and the decision to use positive and negative abnormal advertising spending after missing a target. Although more independent boards are less likely to decrease advertising spending after missing a target (reporting a earnings decrease, or loss), as expected, these results are not significant. Also, as expected more independent boards are less likely to engage in negative abnormal levels of advertising spending after missing a target, However, also these results are not significant. Finally a more independent board is not more or less likely to use positive abnormal increases after missing a target. .
_____________________________________________________________________ Panel C2 Effect of female board (BDFEM) on the decision to decrease or use negative abnormal levels of advertising spending levels after missing a target___________
Last year’s earnings relative to target Chi-square statistic Achieved Missed (probability value)
Target: Earn t-1>0 (loss) 2,16 (0,06)
Advertising decrease and 169 (52%) 37 (63%) 206 BDFEM <MED
Advertising decrease and 154 (48%) 22 (37%) 176 BDFEM>MED
Sample Size 323 59 382_____________
Target: Earn t-1>0 (loss) 1,58 (0,10) abnormal decrease and 232 (51%) 29 (60%) 261
BDFEM <MED
abnormal decrease and 224 (49%) 19 (40%) 243 BDFEM>MED
Sample Size 456 48 504_____________
Focusing on BDFEM, see panel C2, I analyze the impact of proportion of female on boards on decision to decrease advertising and using negative abnormal levels of advertising spending after reporting a loss in the previous period. As expected, the higher the proportion the less likely that advertising spending will be decreased. Surprisingly, when there are earnings pressure, a higher proportion of female makes it less likely that negative abnormal levels of advertising spending will be used even when a target is missed in the previous period. Earlier, I found there was no difference in the case you only look to the proportion of female, but now take into accountant whether there is a earnings pressure or not, you can see that woman constrain abnormal decreases when there are earnings pressures. _________________________________________________________________ PANEL C3: Effect of insider dominated boards (INDOM) on the decision to decrease advertising___________________________________________________________ Last year’s earnings relative to target Chi-square statistic
Achieved (0) Missed (1) (probability value)
Target: Earn t-1>0 (loss) 3,06 (<0,05) Advertising decrease and 281 (87%) 46 (78%)
INDOM =0
Advertising decrease and 43 (13%) 13 (22%) INDOM =1
Sample Size 324 59 ________________
_____________________________________________________________________ Target: Earn t-1>0 (loss) 4,97(<0,025)
abnormal decrease 407 (89%) 38 (79%) INDOM =0
abnormal decrease 49 (11%) 10(21%) INDOM =1
Sample Size 456 48
Target:Δ Earn t-1>0 (loss) 3,61 (0,03)
abnormal decrease and 284 (90%) 161 (85%) INDOM =0
Abnormal decrease and 30 (10%) 29 (15%) INDOM =1
Sample Size 314 190 ________________
Panel C3 analyzes the potentially detrimental effect of having an insider
dominated board. The results stated that firms with insider dominated boards (INDOM = 1) are more likely to reduce advertising after missing a target. This confirms my expectation. Further on, earlier there was found that more independent boards constrain negative abnormal advertising spending. Now considering earnings pressure, so taking into account that managers are more short term focused, we see that negative abnormal decreases will be used more by INDOM =1. Also this confirms the expectation. Given the fact that I found earlier a positive relation between a more independent board and positive abnormal increases, I am not considering anymore whether earnings pressure has an influence on this.
________________________________________________________________
PANEL C4: Effect of MALE dominated boards (MALEDOM) on the decision to decrease advertising__ ___________________________________________
Last year’s earnings relative to target Chi-square statistic Achieved Missed (probability value)
Target: Earn t-1>0 (loss) 5,60(<0,01) Advertising decrease and 111 (34%) 11 (19%)
MALEDOM =0
Advertising decrease and 213 (66%) 48 (81%) MALEDOM =1
Sample Size 324 59
Target: ΔEarnt-1 >0 1,75 (<0.10)
Advertising decrease and 71 (35%) 51 (28%) MALEDOM = 0
Advertising decrease and 133 (65%) 128 (72%) MALEDOM =1
Sample Size 204 179
Target: Earn t-1>0 (loss) 4,99(<0,025)
abnormal decrease 169 (37%) 10 (21%) MALEDOM = 0
abnormal decrease 287 (63%) 38 (79%) MALEDOM =1
Sample Size 456 48
Target: Earn t-1>0 (loss)
abnormal increase 203 (39%) 15 (24%) 218 5,21 (0,01) MALEDOM = 0 abnormal increase 317 (61%) 47 (76%) 364 MALEDOM =1 Sample Size 520 62 582 27
Panel C4 analyzes the potentially detrimental effect of having an male
dominated board. The results stated that firms with male dominated boards are more likely to decrease advertising after missing a target. This confirms my expectation. Furthermore, as mentioned earlier I didn’t found whether female on boards in general has an influence on abnormal advertising levels. However, after missing a target, when firms face earnings pressure (short term focused)), I find that more females on board has a constraining effect on negative AND positive abnormal advertising spending levels.
4.2 Main regression results
_____________________________________________________________________ PANEL A CorpGov: BDIND, their influence on advertising cuts and negative abnormal levels of advertising spending _________________________________
CUT Coefficient Chi Square
Intercept
BDIND (-) -0,274 2,82* (p= 0,046) AACC (-) -0,271 4,45* (p=0,018)
MB (+) -0,019 0,30
ABN_AD (-) Coefficient Chi square Intercept
BDIND (-) -0,261 2,69* (p=0,05)
AACC (-) -0,111 -1,10
MB -0,01 0,06______________________________
Controlling for other factors that drive advertising cuts, you can see that a higher board independence (BDIND>0,5) constrains advertising cuts. BDIND is significiant. Controlling for other factors that drive advertising cuts, you can see that a higher board independence (BDIND>0,5) constrains abnormal advertising cuts. BDIND is significant.
_____________________________________________________ PANEL B CorpGov: BDFEM, their influence on advertising cuts ____
Model 1
Coefficient Chi Square
CUT Intercept
BDFEM (-) -0,22 2,89*(p = 0,045)
AACC (-) -0,27 4,54*(p=0,017)
MB -0,02 0,23____________
Controlling for other factors that drives advertising cuts, you can see that a higher proportion of female in boards (BDFEM>MED) in general constrains advertising cuts. (unconditional on the fact whether there is earnings pressure or not)
_____________________________________________________________________ Panel C: CorpGov = INDOM, influence on advertising cuts and positive abnormal advertising level spending ____________________________________________
CUT Coefficient Chi Square
Intercept
INDOM (+) 0,64 9,55* (p=0,001)
AACC (-) -0,81 37.57*(p=0,000)
MB -0,00 0,00
ABN (+) Coefficient Chi square
Intercept
INDOM -0,28 1,93* (p=0,082)
AACC 0,40 11,29* (p=0.0005)
MB -0,01 0,04_____________________________
Controlling for other factors that drives advertising cuts, you can see that a reduced board independence (INDOM =1) have a positive relation with advertising cuts. The
result is highly significant.
Controlling for other factors that drives abnormal advertising levels, you can see that a reduce board independence (INDOM =1 ) have a negative significant relation with positive abnormal advertising levels. This result means that less independent boards constrain positive abnormal levels of advertising.
_____________________________________________________________________ Panel D: CorpGov = MALEDOM, influence on advertising cuts, negative abnormal advertising level spending_ _____________________________________________
CUT Coefficient Chi Square
Intercept
MALEDOM 0,30 5,11 (p = 0,012)
AACC -0,25 3,96 (p= 0,023)
MB -0,02 0,22 ____________________________
Controlling for other factors that drive advertising cuts, you can see that the existence of male dominated boards (MALEDOM) shows a positive relation with advertising cuts.
V CONCLUSION
This paper analyzes how effective independent and female boards are at constraining real earnings management as measured by the incidence of advertising cuts and abnormal advertising spending levels. According to Osma (2008,p.129) Recent literature document that independent directors are effective monitors at constraining accrual earnings management. (Dechow et all, 1996, Peasnell et all, 2000, 2005). I started this research because I was wondering board characteristics also constrain real earnings management practices. Analyzing advertising spending practices, I found that more independent boards (BDIND>0,5, and INDOM = 0) constrain advertising cuts in general. Moreover, I found that they also constrain negative abnormal advertising. However what they don’t constrain are positive abnormal levels of advertising spending. Even, instead they are more likely to use it than less independent directors, controlling of course for other factors that can be possible drivers. Looking to the degree of female on boards (BDFEM>MED, and MALEDOM = 0), I found that a higher degree constrain advertising cuts in general. However, there are no significant results found related to abnormal levels of advertising spending.
VI REFERENCES
Adams, R.B. and Ferreira, D. (2007) A theory of friendly boards. The Journal of Finance,
Vol. 62, pp. 217 - 250
Ahmed, A.S. Takeda, C. and Thomas, S. (1999) Bank loan loss provisions: A re- examination of capital management, earnings management and signaling effects, Journal of Accounting and Economics, Vol. 28, pp. 1-25.
Anthony, J., & Ramesh, K. (1992). Association between accounting performance measures and stock prices: A test of the life cycle hypothesis. Journal of Accounting and
Economics, Vol 41, pp.47-63.
Arora, R. (1970). How promotion elasticity’s change. Journal of advertising Research, Vol. 19, pp.57-62.
Barth, M. E., Elliot, J.A. and Finn, M.W. (1999) Market rewards associated with patterns of increasing earnings, Journal of Accounting Research, Vol. 37, pp.387-413
Bartov, E. (1993) The timing of asset sales and earnings manipulation, The Accounting
Review, Vol. 68, pp.840-855.
Bartov, E., Givoly, D. and Hayn, C. (2002) The rewards to meeting or beatings earnings expectations, Journal of Accounting and Economics, Vol. 33, pp. 173-204.
Bédard. J. Courteau, L. (2001). Corporate governance and earnings management. Social Science Research Network (SSRN), pp.1-39.
Beekes, W., Pope, P. and Young, S. (2004) The link between earnings timeliness, earnings conservatism and board composition: Evidence from the UK, Corporate
Governance: An International Review, Vol. 12, pp.47-59.
Beidleman, C.R. (1973) . Income Smoothing: The Role Of Management. The Accounting
Review, Vol. 48, No. 4 , pp. 653-667.
Byrd, J.W. and Hickman, K.A. (1992) Do outside directors monitor managers? Evidence from tender off bids, Journal of Financial Economics, 32, pp.195-221
Burgstahler, D. and Dichev, I. (1997) Earnings management to avoid earnings decreases and losses, Journal of Accounting and Economics, Vol. 24, pp.99 – 126
Bushee, B.J. (1998) The influence of institutional investors on myopic R&D investment behavior, The Accounting Review, Vol. 73, pp. 305-333.
Ceyert, Richard, M. and Morria H. Groot (1973), “An Analysis of Cooperation and Learnings in a Duopoly Context, ”American Economic Review”, Vol. 63, pp. 24-37.
Cheng, S. (2004) R&D expenditures and CEO compensation, The Accounting Review, Vol. 73, pp.305- 333.
Corfman, P.K. and D.R. Lehmann (1994). The Prisoner’s Dilemma and the Role of information in setting advertising budgets. Journal of advertising, Vol. 23, No.2, pp. 35-48.
David, P., Hitt, M.A and Gimeno, J. (2001) The influence of activism by institutional investors R&D, Academy of Management Journal, Vol. 44, pp.144-157.
Dechow, P.M., Sloan, R.G. and Sweeney, A.P. (1995) Detecting earnings management. The
Accounting Review, Vol. 70, pp 193-225.
Dechow, P. M, Sloan, R.G. and Sweeney, A.P. (1996) Causes and consequences of earnings manipulation: An analysis of firms subject to enforcement action by the SEC, Contemporary accounting research, Vol. 13, pp. 1-36
Eliashberg, Jehoshue (1981), “An Investigation of Competitive Preference Structures and Posterior Performance Through a Bayesian Decision: Theoretic Approach, “Management Science”, pp. 785-801.
Foster, G. (1997). Quarterly accounting data: Time-series properties and predictive ability results. The Accounting Review, Vol. 52, pp.1-21.
Fama, E.F. and Jensen, M.C. (1983) Separation of ownership and control, Journal of law and
economics, 26, pp.319-337
Garcia Osma, B. and Gill- de- Albornoz Noguer, B. (2007) The effect of the board composition and its monitoring committees on earnings management: Evidence from Spain, Corporate Governance: An International Review, Vol. 15, pp.1412- 1427.
Gore, P. O.,Pope, P.F. and Singh, A. (2007) Earnings management and the distribution of earnings relative to targets: evidence from the UK, Accounting and Business
Research, Vol. 37, pp.123-150
Graham, J.R., Harvey, C.R. and Rajgopal, S. (2005) The economic implications of corporate financial reporting, Journal of Accounting and Economics, Vol.40, pp. 3-73
Hotelling, H. (1929), “Stability in Competition”, Economic Journal, Vol. 39, pp. 41-57. Kasznik, R. and McNichols, M.F. (2002) Does meeting earnings expectations matter? Evidence from analyst forecast revisions and share prices, Journal of Accounting
Research, Vol. 40, pp. 727- 759.
Krishnan, G.V., Parsons, L, M. ((2008). Getting to the bottom line: An exploration of of
gender and earnings quality. Journal of Business Ethics, vol. 78, pp.65 – 76. Low, George S. and Jakki J. Mohr (1992). “The Advertising Sales- Promotion Trade-Off: theory and Practice, “Working Paper, pp. 92-127, Marketing Science Institute, Cambridge, MA
Low, George S. and Jakki J. Mohr (1999). Setting advertising and promotion budgets in multi-brand companies. Journal of Advertising Research, pp. 67-78.
Montgomery, David B. and Charles B. Weinberg (1979), “Toward Strategic Intelligence Systems” ,Journal of Marketing, vol. 43, pp.41-52.
Moorthy, K. Sridhar (1985), “Using Game Theory to Model Competition”, Journal of
Marketing, vol. 32, pp. 41-52.
Opler, T., Pinkowitz, L., Stulz, R. and Williamson, R. (1999) The determinants and implications of corporate cash holdings. Journal Of Financial Economics, Vol. 52, pp.3-36
Osma, B.G. (2008). Board independence and Real earnings management: the case of R&D expenditure. The Author Journal Compilation. Vol. 16, No. 2, pp.116,131. Parker, P.M., & Gatignon, H. (1996) Order of entry, trial diffusion, and elasticity dynamics: an empirical case.
Parsons, L. (1975). The product life cycle and time –varying advertising elasticity’s , Journal
of Marketing Research, Vol .12, pp.476-480.
Peasnell, K.V. Pope, P.F. and Young, S. (2000) Accrual management to meet earnings targets: UK evidence pre- and post-Cadbury, The British Accounting Review, Vol. 32, pp.415- 445.
Peasnell, K.V. Pope, P.F. and Young, S. (2005) Board monitoring and earnings management: do outside directors influence abnormal accruals? Journal of Business Finance and
Accounting, Vol. 32, pp. 1311-1346
Porter, Michael (1980) Competitive Strategy, New York, Free Press. Rockness, H. O. and Shields, M.D. (1988) An Empirical Analysis of the expenditure budget in research and development, Contemporary Accounting Research, Vol. 4 pp. 568 – 581.
Roychowdhury, S. (2006). Earnings management through real activities manipulation. Journal of Accounting and Economics. Vol. 42, No.3, pp. 335–370.
Schelling, T.C. (1960) The strategy of conflict, Cambridge, MA: Harvard University Press. Schipper, K. (1989) Commentary on earnings management, Accounting Horizons,
Vol.3, pp.91-102
Shah, S.Z.A. (2001) Disclosure of advertising expenditures in annual reports, pp.1-29 Schleifer, A. and Vishny, R.W. (1997) A survey of corporate governance, Journal of
Finance,52, pp. 737-783.
Weisbach, M.S. (1988) Outside directors and CEO turnover, Journal of Financial Economics, 20, pp.431-460
Wiedman, C.I. (1996) The relevance of characteristics of the of the information environment in the selection of a proxy for the market’s expectations for earnings: An extensions of Brown,
Richardson, and Schwager (1987), Journal of Business Finance and Accounting, Vol. 26, pp. 833-862.
Winer, R.S. (1979). An analysis of the time – varying effects of advertising: The case of Lydia Pinkham. Journal of Business, vol. 52, pp.563 – 576.
Xie, B., Davidson, W.N. and Dedalt, P.J.(2003) Earnings management and corporate governance, the role of the board and the audit committee.
Young, S. (1999) Systematic measurement error in the estimation of discretionary accruals: An evaluation of alternative modeling procedures, Journal of Business Finance and
Accounting, Vol. 26, pp.833-862.
Zhang, Y. (2010) Earnings pressure and competitive behavior: evidence from the US electricity industry. Academy of management journal, vol. 53, No.4, pp. 743 -768
Attachment 1: Chi-square statistic , probability value table
