Analysts' earnings forecasts: implications for managed earnings via pension expense.

Parker, Paula Diane

INTRODUCTION

Analysts' earnings forecasts create capital market incentives for firms to manage bottom-line reported earnings. This study examines whether or not pension expense is strategically used by firms to manipulate reported earnings in the direction that will move them closer to their analysts' earnings forecasts than they would otherwise be without the earnings manipulation.

The primary motivation for this study is the integrity of financial statement reporting because it is vitally important to capital markets. Various stakeholders, such as investors, creditors, directors, auditors, regulators, standard setters, and academicians rely heavily on the integrity of financial statement reporting in assessing firm value and in making a wide range of business decisions. Therefore, when the true economic condition of a firm is distorted by financial statement manipulation the ultimate outcome is poor decisions based on flawed information. Capital markets are weakened and public confidence in the accounting profession is impaired as a result of financial statement manipulation. For these reasons, this study is relevant to decision makers in today's business environment and makes an important contribution to accounting literature.

This study extends earlier research by not limiting the sampling technique to only those firms with actual reported earnings in the vicinity extremely near their analysts' earnings forecasts. This allows for a broader array of firms to be included in this study and not just those firms expected to exhibit the strongest sensitivity to manage earnings.

The research design raises public awareness and provides pertinent information about the predicted directional change in pension expense that is crucial in detecting and preventing future earnings management of this kind. This study provides basic information and practical analyses for stakeholders, particularly standard setters, to more carefully monitor the changes in pension expense to reduce future financial statement manipulation.

A common obstacle associated with attempting to identify financial statement manipulation is that of determining what a firm's financial statements would report absent the manipulation. The Statement of Financial Accounting Standards No. 87, Employers' Accounting for Pensions (SFAS No. 87), provides a unique measure of what pension expense should be from year to year based on its built-in corridor smoothing (1) technique. Firms are allowed to smooth pension expense to avoid the immediate recognition of wide swing market fluctuations that affect pension investments. The rationale behind the allowed smoothing of pension expense is a long-term perspective where market fluctuations are expected to average out smoothly over the long-term. This technique allows for the reasonable estimation of what a firm's pension expense would be absent manipulation.

A basic characteristic of the research design is modeling the behavior of pension expense to identify its discretionary and nondiscretionary components. This study builds on an approach similar to the random walk approach whereby the prior year's pension expense is assumed to be the most relevant and reliable approximation for predicting the current year pension expense. Theoretically, pension expense is expected to be the same from year to year. Therefore by design, any change in pension expense from year to year is considered discretionary and is the primary focus of explanation in this study. In addition, the specific accruals research design is used because it is powerful in detecting earnings management because the explanatory factors for the discretionary portion of pension expense are tested directly.

An earlier study by Powall et al. (1993) finds evidence that earnings forecasts are value relevant, and thus, establishes their importance in capital markets. Investors often use analysts' earnings forecasts in assessing firm value rather than using more costly and complex valuation tools. According to Collinwood (2001), firms convey good news by meeting analysts' earnings forecasts and firms convey bad news by missing analysts' earnings forecasts. Roen et al. (2003), in studying the effect of preliminary voluntary disclosure and preemptive preannouncement on the slope of the regression of returns on earnings surprise, find when firms manage earnings by attempting to inflate them; the response to negative earnings surprise is stronger than the response to positive earnings surprise. Therefore, managers are motivated to meet analysts' earnings forecasts to avoid stock price penalties and to receive stock price rewards.

Most prior studies are unable to provide convincing evidence that pension expense is used as an earnings management strategy to manipulate earnings. This lack of empirical evidence is astonishing because auditors as well as many others perceive pension expense as being a frequently used earnings management strategy to manipulate earnings. Parker and Sale (2007) and Parker (2009) suggest that most prior studies are unable to detect earnings management via pension accounting for two fundamental reasons. The first reason is that most prior studies focus on contracting incentives rather than on capital market incentives for explaining earnings management. The second reason is that most prior studies focus on the manipulation of pension rates rather than on the direct manipulation of the pension expense amount. Therefore, following Parker and Sale (2007) this study focuses directly on the manipulation of pension expense in response to capital market incentives.

GAAP AND PRIOR LITERATURE

In 1966, shortly after 4,000 auto workers lost their promised retirement benefits (2), the Accounting Principles Board (APB) issues APB Opinion No. 8, Accounting for the Cost of Pension Plans. This opinion is issued to avoid possible government intervention in the financial reporting and disclosure process as well as to address public demands for pension reform.

In 1980, the Financial Accounting Standards Board (FASB) issues Statement of Financial Accounting Standards (SFAS) No. 35, Accounting and Reporting by Defined Benefit Pension Plans, for the purpose of providing additional pension information to help interested parties determine whether pension plans are funded in a manner adequate to provide for payments of retirement benefits when due. In 1985, the FASB issues SFAS No. 87, Employers' Accounting for Pensions, which remains the primary standard influencing pension expense measurement for defined benefit pension plans. In 1998, the FASB issues SFAS No. 132, Employers' Disclosures about Pensions and Other Postretirement Benefits, which is intended to make pension disclosures more informative.

Then again in 2006, the FASB issues SFAS No. 158, Employers' Accounting for Defined Benefit Pension and Other Postretirement Plans, which improves financial reporting by requiring an employer to recognize the overfunded or underfunded status of a defined benefit plan as an asset or liability in its statement of financial position and to recognize changes in that funded status in the year in which the changes occur through comprehensive income of a business entity or changes in unrestricted net assets of a not-for-profit organization. Although SFAS No. 158 is an amendment of SFAS No. 87, 88, 106, and 132 (R), SFAS No. 87 is not amended for the calculation of pension expense. The changes in SFAS No. 158 represent the first phase of the Board's planned two-phase project to reconsider the accounting for pensions and other postretirement benefits. The second phase is expected to be a multiple-year, comprehensive review of the fundamental issues underlying SFAS No. 87 and 106, including measurement of liabilities and the determination of pension expense. As a result, the public can expect more pension accounting changes to be implemented in the near future.

VanDerhei and Joanette (1988) show earnings management incentives are correlated with the permitted actuarial cost method choices made by sponsors in the pre-SFAS No. 87 era. The findings lend credibility to the FASB's decision in SFAS No. 87 to mandate a standardized actuarial cost method for the purpose of averting sponsors from manipulating pension expense through the strategic choice of different actuarial cost methods.

Kwon (1989) focuses on the explanation of the pension discount rate. The results provide evidence that managers use the assumed discount rate to manipulate financial reporting. The finding highlights policy implications in connection with the two opposing schools of thought on strict FASB guidelines. One school asserts the assumed discount rate should be elastic in order to reflect the characteristics of different pension plans. The other school advocates strict FASB guidelines in establishing specific benchmark (ceiling and floor) rates for all pension plans in order to stop rate manipulation by managers.

Blankley (1992) investigates incentives for managerial selection of pension rate estimates by incorporating two distinct paradigms, efficient and opportunistic behavior (3), rather than assume one or the other applies to accounting choice. A learning effect is discovered, whereby as managers get more familiar with SFAS No. 87 opportunistic incentives play a greater role in the choice of pension rates.

Weishar (1997) focuses on the explanation of the simultaneous effects of the three pension rates (i.e., discount rate, compensation rate, and assumed rate of return on plan assets) and finds pension rates are not changed independent of each other. Brown (2001) not only focuses on explaining the three pension rates but somewhat changes the direction of research by using a market valuation model.

In an auditing survey paper, Nelson et al. (2000) find twenty-three potential areas where managers attempt earnings management along with several factors that affect the frequency of decisions of managers and auditors with respect to earnings management. Pensions are included as one of the twenty-three potential areas where managers attempt earnings management. Results indicate managers attempt earnings management to increase earnings, however, forty percent of the determinable current year income effects are income decreasing. Evidence supports income-decreasing earnings management attempts are more likely to occur with respect to imprecise financial standards such as SFAS No. 87.

Parker and Sale (2007) use a specific accrual model with a sample screening technique to investigate whether or not firms use pension expense as an earnings management tool to maintain a steady stream of earnings. The results indicate that pension expense is an active tool used by firms to manage actual earnings when the firm would otherwise miss achieving its current year earnings target (i.e., prior year earnings).

Parker (2010) uses a specific accrual model without the sample screening technique to investigate whether or not firms use pension expense as an earnings management mechanism to maintain a steady stream of slightly increasing earnings. The results indicate that pension expense is again an active mechanism used by firms to manage actual earnings when the firm would otherwise miss achieving its current year earnings target (i.e., prior year earnings).

In recapping post-SFAS No. 87 pension research, contracting variables are primarily used in attempting to explain manager choice in selecting particular levels of pension rates rather than trying to explain pension expense manipulation taken as a whole. A paradigm shift in pension research is expected to occur because SFAS No. 132 and 158 require disclosure sufficient for financial statement users to recalculate pension expense using the disclosed pension rate information.

In looking at the benchmark literature, Burgstahler and Dichev (1997) theorize investors in publicly traded firms use simple low-cost heuristics (4), more specifically earnings-based benchmarks, in determining firm value. Burgstahler and Dichev (1997) use frequency distribution as a method for demonstrating the existence of earnings management. Evidence indicates a disproportionally low incidence of firms reporting small decreases in earnings and small losses relative to a high incidence of firms reporting small increases in earnings and small positive earnings.

DeGeorge et al. (1999) use a similar research design as Burgstahler and Dichev (1997) and report earnings are the single most value relevant item provided to investors in financial statements. Earnings are used as performance measures that provide the enticement for managers to manipulate earnings. Empirical evidence reveals how efforts to exceed thresholds, that is, to sustain recent performance, to report positive earnings, and or to meet analysts' expectations, induce particular patterns of earnings management. Clearly emerging patterns show earnings falling just short of thresholds are managed upward. Whereas, earnings falling far from thresholds, regardless of the direction, call for the thresholds to be adjusted for future ease of attainment.

In recapping a number of other studies evidence indicates firms are managing earnings to continue a steady stream of earnings (Burgstahler and Dichev 1997, Barth et al. 1999, DeGeorge et al. 1999, Moehrle 2002), to avoid reporting a loss (Burgstahler and Dichev 1997; DeGeorge et al. 1999), and or to meet analysts' earnings forecasts (DeGeorge et al. 1999, Brown 2001). In addition, Matsunaga and Park (2001) show evidence of manager compensation-based incentives to avoid earnings declines and to meet analysts' earnings forecasts.

Therefore based on the logic of these prior findings, this study examines whether or not pension expense is strategically used by firms to manipulate reported earnings in the direction that will move them closer to their analysts' earnings forecasts than they would otherwise be without manipulation.

RESEARCH DESIGN

The aggregate accruals method, the specific accruals method, and the earnings-based distribution method are the three research designs prevalent in the earnings management literature (McNichols 2000). Each particular research design has its own advantages, disadvantages, and tradeoffs. The common themes of these designs are the discovery of how managers manipulate earnings, what motivates managers to manipulate earnings, and the costs and benefits associated with earnings management.

According to Healy and Wahlen (1999), future research contributions in the earnings management area are expected from documenting the extent and magnitude of the effects of specific accruals and from identifying factors that limit the ability of managers to manage earnings. Therefore, this study uses the specific accruals research design which is a disaggregated research method.

This approach advocates the examination of an individual accounting item that is subject to substantial managerial judgment and is capable of significantly impacting reported earnings. One advantage of this research design is the ability for directional predictions based on researcher knowledge, skill, and scrutiny of the individual accounting item being examined. However, this research design lacks the ability to analyze simultaneously aggregated effects of multiple accounting items used by managers in managing earnings (McNichols 2000, Fields et al. 2000, Francis 2001, Parker and Sale 2007).

The study examines whether or not there is an association between the change in pension expense and the amount by which firms would otherwise miss or beat their targeted analysts' earnings forecasts. This study extends earlier research by not limiting the sampling technique.

The theoretical concepts discussed above are formalized in alternate form in the following hypothesis.

[H1.sub.A]: Pension expense is managed to meet analysts' earnings forecasts.

The estimated cross-sectional regression model is presented below.

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)

* Diff_PE is the change in pension expense equal to current year pension expense minus prior year pension expense all scaled by lagged assets.

* Miss_Dummy is a dummy variable that equals 1 if the continuous variable, Probe < 0, and 0 otherwise.

* Probe is a continuous variable equal to pretax income absent manipulation minus the applicable benchmark all scaled by lagged assets.

* Interact is an interaction variable equal to Miss_Dummy times Probe.

* [DELTA]Staff is a control variable equal to the number of employees for the current year minus the number of employees for the prior year all scaled by lagged assets.

* [yrD.sub.t] is a dummy variable for each applicable year 1995-2001 with the 1995 dummy effects captured in the intercept.

* [indD.sub.i] is a dummy variable representing 61 industries.

* [[alpha].sub.0] is the intercept for Probe [greater than or equal to] 0 where Miss_Dummy = 0.

* [[alpha].sub.0] + [[alpha].sub.1] is the intercept for Probe < 0 where Miss_Dummy = 1.

* [[alpha].sub.2] is the incentive slope for Probe [greater than or equal to] 0 where Miss_Dummy = 0.

* [[alpha].sub.2] + [[alpha].sub.3] is the incentive slope for Probe < 0 where Miss_Dummy = 1.

As is the case in all earnings management studies, a reasonable proxy for earnings management must be developed. In this study, the regression analysis incorporates Diff_PE as the earnings management proxy which is the dependent variable. The proxy development is accomplished by using the unique smoothing feature of SFAS No. 87 whereby the prior year pension expense provides a logical approximation for the firm's premanaged pension expense. Assuming the number of employees remains unchanged, current pension expense should be approximately the same as the prior year pension expense. Diff_PE is defined as the current year pension expense minus the prior year pension expense all scaled by lagged assets. Thus, Diff_PE is a proxy for the extent of manipulation in pension expense after controlling for the change in the number of employees. Therefore, earnings management is measured by Diff_PE.

Premanipulation earnings relative to analysts' earnings forecasts represent the level of capital market incentives for earnings management. The capital market based incentive measure to manipulate earnings is represented by the variable called Probe. The independent variable, Probe, is a continuous scaled variable calculated as the difference between pretax earnings absent pension manipulation and the analysts' earnings forecasts.

Following Burgstahler and Eames (2002), a benchmark representing target earnings is necessary. The benchmark for target earnings in this study is pretax analysts' earnings forecasts. Pretax analysts' earnings forecasts are used for consistency because pension expense is reported in the financial statements on a pretax basis. Earnings absent pension manipulation are constructed using pretax income adjusted for the change in pension expense and is called PIAM. The measure for pension expense absent pension management is, therefore, the prior year pension expense.

A dummy variable (i.e., Miss_Dummy) for hypothetically missing analysts' earnings forecasts is included in the analysis. Miss_Dummy is coded zero for firms that hypothetically beat their analysts' earnings forecasts using premanaged earnings. Whereas, Miss_Dummy is coded one for firms that hypothetically miss their analysts' earnings forecasts using premanaged earnings. If [[alpha].sub.1] is significant and positive, firms missing their analysts' earnings forecasts have a higher intercept than the other firms. If [[alpha].sub.1] is significant and negative, firms missing their analysts' earnings forecasts have a lower intercept than the other firms. If [[alpha].sub.1] is insignificant, there is no difference between the two groups of firms.

After controlling for the change in the number of employees, the association between Diff_PE and the level of capital market incentive (i.e., Probe) for earnings management constitutes this study's test of interest. Because both smoothing (8) and benchmark (9) incentives exist and may not be equally important, the slope coefficient on Probe is allowed to vary with the prediction on Interact (i.e., [[alpha].sub.3]) being nondirectional.

The dependent variable, Diff_PE, is expected to be positively correlated with the incentive variable Probe. The slope coefficient for the group of firms (i.e., smoothing group) that hypothetically beat their analysts' earnings forecasts is represented by [[alpha].sub.2]. The slope coefficient for the group of firms (i.e., benchmark group) that hypothetically miss their analysts' earnings forecasts is represented by [[alpha].sub.2] + [[alpha].sub.3]. Thus, I predict that [[alpha].sub.2] > 0, and that [[alpha].sub.2] + [[alpha].sub.3] is > 0.

The logic behind the predictions for [[alpha].sub.2] and [[alpha].sub.2] + [[alpha].sub.3] is that the dependent variable, Diff_PE, is expected to move in the same direction as the independent incentive variable, Probe. For example, if a firm has premanaged earnings equal to $.50 per share and analysts' forecasted earnings equal to $.48 per share, the firm is expected to manipulate actual earnings by increasing pension expense by $.02 in order to offset the $.02 excess in premanaged earnings. In this situation, there is a positive $.02 excess in premanaged earnings and the change in pension expense (i.e., Diff_PE) is expected to move $.02 in a positive direction as well. The variable Probe (i.e. [[alpha].sub.2]) captures the positive $.02 excess in premanaged earnings. Therefore, because Diff_PE and Probe move together in the same direction, a positive correlation is predicted.

On the other hand, if a firm has premanaged earnings equal to $.48 per share and analysts' forecasted earnings equal to $.50 per share, the firm is expected to decrease pension expense by $.02 to offset the $.02 negative premanaged earnings. The variable Probe (i.e., [[alpha].sub.2] + [[alpha].sub.3]) captures the negative $.02 deficiency in premanaged earnings. Here again, because Diff_PE and Probe move together in the same direction, a positive correlation is predicted.

Since the coefficient on Interact (i.e., [[alpha].sub.3]) is predicted as nondirectional, it will be interpreted as follows. If [[alpha].sub.3] is positive, this will indicate that firms hypothetically missing their analysts' earnings forecasts are actually decreasing pension expense (i.e., increasing earnings) more, to avoid missing their analysts' earnings forecasts, than firms hypothetically beating their analysts' earnings forecasts are actually increasing pension expense (i.e., decreasing earnings) to smooth income downward in the direction of their analysts' earnings forecasts. On the other hand, if [[alpha].sub.3] is negative, this will indicate that firms hypothetically missing their analysts' earnings forecasts are decreasing pension expense (i.e., increasing earnings) less, to avoid missing their analysts' earnings forecasts, than firms hypothetically beating their analysts' earnings forecasts are actually increasing pension expense (i.e., decreasing earnings) to smooth income downward in the direction of their analysts' earnings forecasts.

In other words, if [[alpha].sub.3] is significant and positive, firms missing their analysts' earnings forecasts have a steeper slope than the other firms. Whereas, if [[alpha].sub.3] is significant and negative, firms missing their analysts' earnings forecasts have a flatter slope than the other firms. However, if [[alpha].sub.3] is insignificant, then both groups of firms have the same slope.

In summary, analysts' earnings forecasts create incentives for firms that are in opposite directions depending on the level of premanaged earnings relative to their earnings targets. So that, if firms hypothetically miss their analysts' earnings forecasts they are expected to exhibit benchmark behavior by manipulating pension expense to increase actual earnings in order to reach their benchmark. On the other hand, if firms hypothetically beat their analysts' earnings forecasts they are expected to exhibit smoothing behavior by manipulating pension expense to decrease actual earnings so that their actual earnings are closer to their analysts' earnings forecasts than they would otherwise be.

[DELTA]Staff is a control variable to account for any variation in the dependent variable (i.e., Diff_PE) caused by the change in the number of employees from year to year. [DELTA]Staff is calculated as the current year number of employees minus the prior year number of employees all scaled by lagged assets. In addition, the inclusion of the control variable, [DELTA]Staff, should lessen confounding results attributable to changes in organizational structure such as mergers and acquisitions. A positive relationship is expected between the change in pension expense (i.e., Diff_PE) and the change in the number of employees from year to year (i.e., [DELTA]Staff). The reasoning is likely because an increase in the number of employees is expected to result in an increase in pension expense, whereas a decrease in the number of employees is expected to result in a decrease in pension expense. Therefore, a positive slope coefficient is predicted for [DELTA]Staff.

On the other hand, if an economy of scale exists, then a negative slope may occur for [DELTA]Staff. For example, when a higher paid employee is replaced by two new lesser paid employees and the overall pension expense is less for the two new employees than it was for the one higher paid employee, an economy of scale occurs. In this situation, the addition of one new employee (2 - 1 = 1) actually decreases pension expense; whereas, adding an additional employee would normally be expected to increase pension expense. A merger or acquisition may also cause an economy of scale for [DELTA]Staff. Another possible scenario is where the actuarial assumptions are different for the acquiring firm's pension plan and the purged plan automatically becomes overfunded as a result of using the acquiring firm's actuarial assumptions.

Two additional control variables ([indD.sub.i] and [yrD.sub.t]) are included in the model. These are intended to control for industry and time fixed effects. Multicollinearity (7) and heteroscedasticity (8) diagnostic tests indicate these common regression problems are not present in the current study. Outlier observations are addressed by windsorizing variables so that large and small outlier values are still large and small values within the dataset but are less likely to disrupt the mean, standard estimates, and other statistics that depend on them. The top and bottom one percent of the Compustat variables are windsorized to dampen their effects without eliminating the observations from the sample.

Other studies (Schwartz 2001, Dhaliwal et al. 2002) indicate managers may attempt to guide analysts' earnings forecasts in order to then meet the analysts' forecasts. Therefore, if managers do not manage pension expense or do effectively guide analysts' earnings forecasts, there should be no association between the change in pension expense (i.e., Diff_PE) and the amount that firms hypothetically miss or hypothetically beat their analysts' earnings forecasts (Dhaliwal et al. 2002, Parker and Sale 2007).

RESULTS AND INTERPRETATIONS

Table 1 summarizes the sample selection information. The final sample consists of 2,904 firm observations representing 61 industries for the period 1995-2001. The large sample provides information on a wide range of industries which is desirable in this study. The data set is very cost effective for the researcher as it corresponds with another study. The data set is from the Compustat files and includes all the firm observations available for the applicable time period. For a firm to remain in the final sample there must be two years of consecutive firm data available because of scaling by lagged assets.

Table 2 reports the results of the regression analysis. The rationale for explaining Table 2 results is based on the belief that pension expense manipulation is a function of the value of the magnitude of hypothetically missing or hypothetically beating the benchmark earnings (i.e., analysts' earnings forecast) based on premanaged earnings.

The economic substance is captured by the regression main effects of the incentive variable for the two distinct groups (i.e., benchmark and smoothing) of firms. The results of the control variables are not reported as they are not important for interpretation.

Diff_PE, representing firm manipulation, is expected to be positively correlated with Probe, the incentive variable of interest. The incentive slope is captured in the model for the firms that hypothetically beat their benchmark by [[alpha].sub.2] and for the firms that hypothetically miss their benchmark by [[alpha].sub.2] + [[alpha].sub.3]. The slope on Probe (i.e., [[alpha].sub.2], and [[alpha].sub.2] + [[alpha].sub.3]) represents the estimated average change in pension expense when the applicable incentive variable increases or decreases by one unit. If managers are more concerned with reaching their benchmark than smoothing, then the prediction is that [[alpha].sub.3] > 0.

The slope coefficient (i.e., [[alpha].sub.2] > 0) for the firms that hypothetically beat their benchmark is expected to be statistically significant and is tested with a t-test. The slope coefficient (i.e., [[alpha].sub.2] + [[alpha].sub.3]) for the firms that hypothetically miss their benchmark is expected to be statistically significant and is tested with an F-test.

Table 2 reports the association test results which indicate a significant regression (F-statistics p-value = .0001). There is strong evidence that the linear relationship between the change in pension expense (i.e., Diff_PE) and the independent explanatory variables does, in fact, exist as expected. The [R.sup.2] and adjusted [R.sup.2] are .1142 and .0923 respectively, which indicates a high proportion of the change in pension expense (i.e., Diff_PE) is explained by the combination of independent variables.

As predicted the slope coefficient (i.e., [[alpha].sup.2]) for the group of firms that hypothetically beat their benchmark is statistically significant (p-value = .0001) and the sign is positive. The inference is that for every $1 that premanaged earnings are above the benchmark earnings (i.e., analysts' earnings forecasts) firms increase pension expense $.55 to move actual reported earnings downward closer to their analysts' earnings forecast than they would otherwise be without the manipulation.

As predicted the slope coefficient (i.e., [[alpha].sub.2] + [[alpha].sub.3]) for the group of firms that hypothetically miss their benchmark is statistically significant (p-value = .0001) and the sign is positive. The inference is that for every $1 that premanaged earnings are below the benchmark earnings (i.e., analysts' earnings forecasts) firms decrease pension expense $.52 to move actual reported earnings upward closer to their analysts' earnings forecast than they would otherwise be without the manipulation.

In summary, the results indicate a consistent pattern of association between the change in pension expense (i.e., Diff_PE) and the incentive variable, Probe. The pattern of evidence indicates both distinct groups of firms (i.e., smoothing and benchmark) are strategically manipulating pension expense in the direction that will move their reported earnings closer to their analysts' earnings forecasts than they would otherwise be without the manipulation.

Overall, smoothing behavior is stronger than benchmark behavior. One likely explanation is that auditors may be more vigilant in constraining upward earnings (i.e., benchmark behavior) manipulation than downward earnings (i.e., smoothing behavior) manipulation. This is likely because litigation exposure is more probable and costly with upward earnings manipulation than with downward earnings manipulation. The rationale is that upward earnings manipulation uses resources that belong to future periods, whereas downward earnings manipulation stores up hidden reserves in the current period to be expended in future periods.

It is interesting to note the findings in the Nelson et al. (2000) survey study suggests income-decreasing earnings management attempts are more likely to occur with respect to imprecise financial standards. The results in this study support that more actual manipulation is occurring in financial statement reporting in the direction of income decreasing earnings management via pension expense. Again assuming the incentive to manipulate earnings upward to meet the benchmark earnings (i.e., analysts' earnings forecast) is at least equal to the incentive to manipulate earnings downward to meet the benchmark earnings (i.e., analysts' earnings forecasts), the pattern of evidence suggests auditors are less vigilant in constraining downward earnings management than upward earnings management.

SUMMARY CONCLUSIONS

Managers have strong incentives to manage earnings to achieve analysts' earnings forecasts in order to reap stock price advantage and to avoid market devaluation. In addition, many contracting incentives are tied directly or indirectly to earnings based measures which also provide strong incentives for managed earnings.

This research study contributes to the literature by providing evidence that managers are, in fact, using pension expense to manipulate reported earnings in a predictable rational economic manner. The research provides evidence that analysts' earnings forecasts create capital market incentives in opposite directions depending on the economic status as measured by whether or not firms will miss or beat their analysts' earnings forecasts based on premanaged earnings.

By using "what if analyses, firms that hypothetically miss their analysts' earnings forecasts are shown to strategically manipulate actual pension expense downward to increase actual reported earnings; whereas firms that hypothetically beat their analysts' earnings forecasts are shown to strategically manipulate actual pension expense upward to decrease actual reported earnings. As predicted, both groups of interest are strategically manipulating pension expense in the direction that moves their actual reported earnings closer to their analysts' earnings forecasts than they would otherwise be without the manipulation. The results suggest that smoothing behavior is stronger than benchmark behavior. One reason may be that auditors are more diligent in constraining efforts to manage earnings upward than in constraining earnings downward.

This study is timely because it has important implications in support of FASB's planned upcoming phase two project to comprehensively review the determination of pension expense measurement. As a result of the completed phase one project, FASB issued SFAS No. 158 addressing pension reform exclusive of pension expense measurement.

This study is relevant as capital markets and the U.S. economy are heavily influenced by the integrity of financial statement reporting. When the true economic condition of a firm is misrepresented by financial statement manipulation the ultimate outcome is poor decisions based on flawed information. Capital markets are weakened and public confidence in the accounting profession is impaired as a result of financial statement manipulation. For these reasons, this study is important to decision makers in today's business environment and makes an important contribution to accounting literature.

REFERENCES

Accounting Principles Board. Accounting for the Costs of Pension Plans. Accounting Principles Board Opinion No. 8, 1966.

Ali, A. and K. Kumar. 1993. Earnings Management Under Pension Accounting Standards: SFAS 87 versus APB 8. Journal of Accounting, Auditing & Finance (Fall): 427-446.

Bergstresser, D., Desai, M. and Rauh, J. (2006). Earnings Manipulation, Pension Assumptions and Managerial Investment Decisions. Quarterly Journal of Economic (Feb 2006) 121, no 1: 157-195.

Blankley, A. 1992. Incentives in Pension Accounting: An Empirical Investigation of Reported Rate Estimates. Dissertation, Texas A&M University, College Station, Texas.

Brown, S. 2001. The Impact of Pension Assumptions on Firm Valuation. Dissertation, Northwestern University, Evanston, Illinois.

Burgstahler, D. and I. Dichev. 1997. Earnings Management to Avoid Earnings Decreases and Losses. Journal of Accounting and Economics 24 (1): 99-126.

Collingwood, H. 2001. The Earnings Game. Harvard Business Review 79-6:65-72.

Dhaliwal, D., C. Gleason, and L. Mills. 2002. Last Chance Earnings Management: Using the Tax Expense to Achieve Earnings Targets. University of Arizona, Working Paper.

Degeorge, F., J. Patel and R. Zeckhauser. 1999. Earnings management to Exceed Thresholds. Journal of Business 72 (1): 1-33.

Fields, T., T. Lys, and Linda Vincent. 2001. Empirical Research on Accounting Choice. Journal of Accounting and Economics 31: 255-307.

Financial Accounting Standards Board. 1980. Accounting and Reporting by Defined Benefit Pension Plans. Statement of Financial Accounting Standards No. 35.

Financial Accounting Standards Board. 1985. Employers' Accounting for Pensions. Statement of Financial Accounting Standards No. 87.

Financial Accounting Standards Board. 2006. Employers' Accounting for Defined Benefit Pension and Other Postretirement Plans. Statement of Financial Accounting Standards No. 158.

Financial Accounting Standards Board. Disclosure Effectiveness. 1995. Financial Accounting Board Prospectus.

Financial Accounting Standards Board. 1998. Employers' Disclosures about Pension and Other Postretirement Benefits. Statement of Financial Accounting Standards No. 132.

Financial Accounting Standards Board. 2006. Employers' Accounting for Defined Benefit Pension and Other Postretirement Plans. Statement of Financial Accounting Standards No. 158.

Francis, J. 2001. Discussion of empirical research on accounting choice. Journal of Accounting and Economics 31, 309-319.

Healy, P. and J. Wahlen. 1999. A Review of the Earnings Management Literature and Its Implications for Standard Setting. Accounting Horizons 13-4: 365-38.

Kwon, S. 1989. Economic Determinants of the Assumed Interest Rate in Pension Accounting. Dissertation, University of Oklahoma, Norman, Oklahoma.

Matsunaga, S. and C. Park. 2001. The Effect of Missing a Quarterly Earnings Benchmark on the CEO's Annual Bonus. The Accounting Review 76-3:313-332.

McNichols, M. 2000. Research Design Issues in Earnings Management Studies. Journal of Accounting and Public Policy 19 (2000): 313-345.

Moehrle, S. 2002. Do Firms Use Restructuring Charge Reversals to Meet Earnings Targets? The Accounting Review 77-2:397-413.

Nelson, M., J. Elliott, and R. Tarpley. 2000. Where do Companies Attempt Earnings Management, and When Do Auditors Prevent It? Cornell University and Washington University, Working Paper (October Draft Copy).

Parker, P. and M. Sale. 2007. Using Pension Expense to Manage Earnings: Implications for FASB Standards. Academy of Accounting and Financial Studies Journal, 11-3: 109-123.

Plummer, E. and D. Mest. 2000. Evidence on the Management of Earnings Components. CUNY-Baruch College and Rutgers University, Working Paper.

PBGC. 2000. About PBGC - Mission and Background. [email protected] (Version update 7-19-2000).

Powell, G. C. Wasley, and Waymire. 1993. The Stock Price Effects of Alternative Types of Management Earnings Forecasts. The Accounting Review 68-4:896-912.

Roen, J., T. Roen, V. Yaari., J. 2003. The Effect of Voluntary Disclosure and Preemptive Preannouncements on Earnings Response Coefficients (ERC) When Firms Manage Earnings. Journal of Accounting, Auditing, and Finance (Summer) 18-3:379-410.

Schwartz, W. 2001. Managing Earnings Surprises. Dissertation. University of Iowa. Iowa City, Iowa.

VanDerhei, J. and F. Joanette. 1988. Economic Determinants for the choice of Actuarial Cost Methods. The Journal of Risk and Insurance (March) 59-74.

Weishar, J. 1997. A Cross-Sectional Evaluation of the Ability of Authoritative Standards to Influence the Use of Defined Benefit Pension Plan Actuarial Assumptions as an Income Smoothing Technique. Dissertation, University of Arkansas, Fayetteville, Arkansas.

Paula Diane Parker, University of Southern Mississippi

ENDNOTES

(1.) The term smoothing is used in this paper in two very different contexts. First, here smoothing indicates spreading evenly over time. So that no significant different should occur from year to year. Second, later in the paper, smoothing is used to identify firm behavior when the firm unjustifiably increases pension expense to reduce reported earnings in order to move their reported earnings closer to their analysts' earnings forecasts than they would otherwise be without manipulation.

(2.) The Financial and Estate Center published this information at www.worldtraffice.com in an article titled All About Pension Plans.

(3.) Efficient behavior proxies for the three pension rates are: (1) the Pension Benefit Guaranty Corporation's (i.e., PBGC's) published discount rate, (2) the industry average compensation rate, and (3) the firm's actual rate of return on plan assets. Opportunistic behavior proxies for the three pension rates are: (1) the firm's discount rate adjusted for the PBGC's published discount rate, (2) the firm's compensation rate adjusted for the industry average compensation rate, and the (3) the firm's expected rate of return on plan assets adjusted for the actual rate of return on plan assets. The theory is that firms are simultaneously influenced by both efficient and opportunistic behavior. Therefore, Blankley's study controls for efficient behavior and attempts to explain opportunistic behavior in terms of the independent variables which are cash constraints, debt-covenant constraints, monitoring by union concentration, tax management incentives, and the number of analysts covering the firm.

(4.) When it is expensive for investors to retrieve and process detailed information about earnings, it is conjectured that investors use information processing heuristic cutoffs, such as zero changes in earnings or zero earnings, to assess firm value.

(5.) Smoothing incentives result in smoothing behavior which is where a firm unjustifiably increases pension expense to decrease actual reported earnings in an attempt to store up hidden reserves and move closer to their analysts' earnings forecast than they would otherwise be.

(6.) Benchmark incentives result in benchmark behavior which is where a firm unjustifiably decreases pension expense to increase actual reported earnings in an attempt to reach their analysts' earnings forecast.

(7.) Multicollinearity is a common problem that affects regression analysis when two or more of the independent variables are highly correlated.

(8.) Heteroscedasticity is another common problem that affects regression analysis when the variances of the regression errors are not constant. TABLE 1: Sample Selection Firms in original sample covering 1995-2001 21,608 Firms that do not have defined -18,704 benefit plans and firms with missing observations Firms in the final sample 2,904 Table 2: Cross Sectional Pooled Effects Estimation With Time and Industry Fixed Effects Variable Prediction Coefficient One Tail p-value intercept + -0.00878 .0316 miss_dummy - 0.00397 .0091 probe + 0.55189 .0001 interact + / - -0.02937 .0295 [[alpha].sub.0] + - -0.00481 .1534 [[alpha].sub.1] [[alpha].sub.2] + + 0.52252 .0001 [[alpha].sub.3] F-statistic .0001 as p-value [R.sub.2] .1142 Adjusted [R.sub.2] .0923 Sample Size 2904