web server statistics
American Accounting Association

CONTACT: Ben Haimowitz (212-233-6170)

CEO retirement is big loser for shareholders, when chief's pension is based on company's late-stage performance, new study finds

Earnings manipulation in pre-retirement period leads to subsequent stock decline

Does the approach of a CEO's retirement present hazards to investors? Certainly, chiefs that amass extravagant riches in departing the firm provoke shareholder outrage; yet, past research has yielded little evidence that, in general, voluntary CEO departures lead to abnormal negative returns for their firm's stock.

But now a new study in The Accounting Review, one of the journals published by the American Accounting Association, finds that investors have more cause for vigilance about forthcoming CEO retirements than previous research has suggested.

The problem is nothing as obvious as golden parachutes or outlandish retirement bonuses. What the new study finds is that a relatively obscure but widespread form of executive pay called SERP (supplementary executive retirement plan) can lead to negative abnormal stock returns of as much as eight to nine percent over the three years following a CEO's retirement, if the SERP is based on company performance over the final years of the chief's tenure.

In the words of the study, authored by Paul Kalyta of McGill University, "Results provide evidence of income-increasing earnings management in the pre-retirement period in firms with performance-contingent CEO SERPS...Retirements of CEOs with no SERP and CEOs whose SERP benefits are not contingent on firm performance do not lead to abnormal market returns around CEO retirement announcements. In contrast, the post-retirement period in firms with performance-contingent CEO SERPs is characterized by negative long-term abnormal returns."

CEO earnings manipulations in this latter group of firms, the study finds, arouses the market's concern only after the announcement of the chief's forthcoming retirement. In effect, the announcement acts like a wake-up call, ending "an ex ante asymmetry between shareholders and the CEO regarding the timing of CEO retirement," writes Prof. Kalyta. He adds, "Although the market may have expectations about management turnover, the CEO is in a better position to predict the timing of his/her retirement than any other individual...Hence, any increases in discretionary accruals in the final years prior to CEO retirement may not be unraveled by the market until after the retirement announcement, causing a temporary overpricing of firm equity [during those years]."

The magnitude of the market's belated negative response, the professor finds, depends on "the size of discretionary accruals in the pre-retirement periods." Thus, while estimated post-retirement market returns over three years average about minus five percent for companies with performance-contingent SERPS, they average more than minus eight percent for those companies in which discretionary accruals  (and therefore earnings manipulations) are above the median for the group as a whole.

Comments Prof. Kalyta: "Corporate accounting provides many opportunities, all perfectly legal, to manipulate earnings by accelerating recognition of revenue and deferring recognition of expenses. There is significant room for discretion in accounting, and CEOs can use this for a variety of purposes. To use it for essentially selfish reasons, even when legal, certainly raises ethical questions, particularly when, as this research reveals, doing so destroys a considerable amount of shareholder value."

In the U.S. and most developed countries SERPs exist to circumvent regulatory limitations on retirement income under regular pension plans. As the study explains, "The Internal Revenue Service caps the annual compensation used for determining pension benefits...at $235,000. Consider the CEO who retires with a $2 million pre-retirement base salary, 35 years of pensionable service, and a pension plan that calls for two percent of the final base salary multiplied by the number of years of pensionable service to be paid to the CEO annually upon retirement. In this case, the CEO's pension is limited to $164,500...a modest fraction of pre-retirement income for many executives."

About three fourths of the largest corporations have SERPs  to increase executives' post-retirement income beyond the regular pension limit, with some plans determining pensionable earnings from base salary only and others determining it from salary plus performance bonus and other performance-based compensation. The difference between the two plans can be considerable. Prof. Kalyta explains that a single performance bonus of $1 million can increase a CEO's annual pension by as much as 100 percent, and a $5-million bonus can produce a yearly increase of 500 percent. "In other words," the professor writes, "CEOs with performance-contingent SERPs have strong incentives to make accounting choices that increase firm short-term income in determination years and, therefore, amplify the value of their pensions." 

To find what effect these incentives might have on a firm's accounting and the subsequent effect on the company's stock, Prof. Kalyta collected data on Fortune 1000 firms whose chief executives retired during the ten years 1997 through 2006. The final sample consisted of 388 CEOs, of whom about 30 percent had no SERPs. Among the 273 chiefs with SERPs, 63 percent had plans that were contingent on firm performance during the period that determined the size of the exec's pension, a period that averaged 3.22 years.

In the case of CEOs with performance-contingent SERPs -- and only in those instances -- a clear pattern emerged of above-average discretionary accruals in the four years prior to CEOs' retirement followed by below-average discretionary accruals in the four years following retirement. The greatest proportion of discretionary accruals occurred in the second and third years before retirement then dwindled in the final year. Why this decline? "The CEO may have limited means to increase firm short-term earnings after several years of abnormally high discretionary accruals," Prof. Kalyta surmises. "In addition, boards and stakeholders may become more vigilant as the CEO gets closer to retirement."

The professor doesn't favor a regulatory fix for the problem his research has uncovered. "Compensation contracts should be determined by the market," he says. "The best antidote is increased vigilance on the part of boards and investors. Requisite information on pension plans of top executives is already disclosed in proxy statements. With awareness of the pitfalls of performance-contingent SERPs, investors have the means today to be wary of this benefit in firms with CEOs who may be nearing retirement.

"In addition, boards should exercise greater care than in the past when designing SERPs. People tend to think that performance-contingent compensation is always desirable, but that is not necessarily the case. It just does not make much economic sense for performance in one or a few years to determine the amount of benefits that executives or their surviving spouses will take from the shareholders' pockets for many years to come."

The study, entitled "Accounting Discretion, Horizon Problem, and CEO Retirement Benefits," is in the September/October issue of The Accounting Review, published every other month by the American Accounting Association, a worldwide organization devoted to excellence in accounting education, research, and practice. Other journals published by the AAA or its specialty sections include Accounting Horizons, Issues in Accounting Education, AUDITING: A Journal of Practice and Theory, Behavioral Research in Accounting, The Journal of the American Taxation Association, and The Journal of Management Accounting Research.



© 2010 American Accounting Association All rights reserved.