(Excerpt from Project Proposal) Separating Facts and Forecasts in Business Performance Measurements: Intertemporal Financial Statements and SEC’s Safe Harbor Rules on Forecasts by Jonathan Glover, Yuji Ijiri, Carolyn B. Levine, and Pierre Jinghong Liang Carnegie Mellon University In Consultation with James L. Murdy, CEO, Allegheny Technologies, Inc.
Research Problem: Recent reform movements in accounting and corporate governance point to a significant weakness in existing financial statements in which forecasts (e.g., pension liabilities, impairments, depreciation, etc.) are mixed with facts and presented as a single number. Our project focuses on separating facts and forecasts and reintegrating as a single number with two parts, just like complex numbers which have the “real” part and the “imaginary” part while the two together constitute a single number. With forecasts clearly identified as such, investors are forewarned and the SEC’s safe harbor rules on forecasts would apply which explicitly recognize the inherent subjectivity and inaccuracies of forecasts. Our project will explore the resulting “intertemporal” financial statements and extend the outcome, using analytical models, to other business performance measurements. Research objective: To improve the trust people place on business performance measurements, net income being one of the most widely used business performance measurements, by means of clearly separating facts and forecasts.
Research Methods: We will work on general formulations of the problem dealing with positive as well as negative effects of separating facts and forecasts in business performance measurements. Specific applications to enhance financial statements will be developed building on the large scale business-game simulation models that have been used at Carnegie Mellon University as the required core of the MBA program continuously over the past 45 years with periodic updates and overhauls (now involving global operations and simultaneous participation from Japanese and Mexican universities). The school enjoys annual participations of over 300 “mock” board members (for 50 companies run by 6 students each) from downtown Pittsburgh ranging from corporate executives, accounting practitioners, and entrepreneurs. The research team will be assisted by a number of such people as project consultants, in particular James L. Murdy, CEO of Allegheny Technologies, Inc. (and a former CFO and Executive Vice President of Gulf Oil Corporation in Pittsburgh until it was bought out by Chevron), who have assisted our school over the past three decades. Description of proposed research activities: Stimulated by the passage of the Sarbanes-Oxley Act in July 2002, our team started working on the research problem last summer and completed a working paper, “CEO/CFO Certification and Emerging Needs to Separate Facts and Forecasts: Exploring ‘Intertemporal Financial Statements’ with Two Time-Phases,” by Jonathan Glover, Yuji Ijiri, Carolyn B. Levine, and Pierre Jinghong Liang, December 2002 (see Social Science Research Network at http://www.ssrn.com). We wish to take liberty of submitting this paper as an attachment. It represents a first step in what we see as a much larger research program--we were not able to fully describe many of the interesting problems and possible solutions in this first paper, both
2
on financial statement issues and agency theory issues. We wish to build on this paper and extend it to a monograph on facts and forecasts in business performance measurements with the funding assistance from the Business Measurement Research Program. Indication of the novelty of the topic: Separating facts and forecasts in a comprehensive manner on financial statements is a novel idea which, to our knowledge, has never been explored before. While component breakdowns of an account by means of tables and footnotes have been widely practiced, our attempt here is to construct comprehensive breakdowns of all accounts simultaneously (analogous to general equilibrium models, at least in spirit) so that each column of all financial statements balances and articulates on its own and, when added across columns, the sums agree with the traditional financial statements. Likewise, our analytical formulation of the use of forecasts showed an unexpected discontinuity. The analytical model itself is novel in its focus on robust contracts. Both the analytical model and financial statement applications stem from our team’s perspective that facts and forecasts should be separated for two reasons. First, people react differently to facts and forecasts and second, financial statements are required to serve two different purposes simultaneously, namely the need to evaluate the past and the need to predict the future. Financial statements have been decomposed along the “time” dimension (e.g., quarterly statements) and the “space” dimension (e.g., segment reporting) where the primary benefit is the decomposition of data from consolidated financial information. Separating facts and forecasts represents an analogous decomposition along the “legal” dimension. The fact-forecast separation allows users to better understand the extent to
3
which financial statements contain estimates and forecasts and provide forecasts the appropriate legal status under safe harbor rules. The emerging legal characteristic of accounting is, we believe, fundamental to our understanding of accounting's role in the economy and society at large. A full treatment of the subject requires developing conceptual constructs as well as practical measurement procedures. A successful development of these constructs and procedures would open up an array of new research questions on business measurement, which would improve both the theory and practice of accounting. Expected contribution and benefit of the proposed research project: In our paper mentioned earlier, we comment on the benefit of separating the two in view of the Sarbanes-Oxley Act. CEOs/CFOs can mitigate the unfair exposure to the risk of forecasts inevitably going wrong, as long as their forecasts are “prepared in good faith and with a reasonable basis” under the SEC’s safe harbor rules on forecasts. Thus corporations have incentives to disclose facts and forecasts separately. Investors are fully warned of the risk of forecast errors and they will welcome the additional column because it allows the risk assessment of forecasts. The regulators will welcome the new disclosure because it has always welcomed any reconciliations that make financial statements more transparent. Significance for the field and for the researchers’ development: People treat facts and forecasts differently because a) forecast measurements are inherently less accurate than fact measurements or b) forecast measurements are more manipulable than fact measurements. CEO/CFO certification was introduced to prevent the latter but CEOs and CFOs may suffer from the former, the so-called alpha risk, the risk of the innocent get-
4
ting punished because of the forecasts’ inherent lack of accuracy. Analytical models we may use, such as agency theory which is the analysis of the basic business relationship in which managers act on behalf of owners, highlight the important role of business measurement in evaluating managerial and firm performance. This framework provides an excellent vehicle by which to formulate and analyze this fundamental dilemma in using forecasts in business performance measurements, to which our study is expected to make contributions as well as to benefit our research careers. Standard agency theory models assume that the principal (contract designer) perfectly knows the environment that applies. This assumption leads to extreme valued contracts that do not work well if the environment is different than expected. The part of this project focusing on managerial incentives will attempt to develop an understanding of how facts and forecasts can be used in robust contracts, i.e., contracts that are fairly insensitive to slight changes in the description of the modeled environment. Our hope is that allowing for some uncertainty about the contractual environment will generate contracts that more closely resemble those we observe or provide suggestions on improving existing contracts. Early analysis points to a number of promising conjectures about the ways in which robustness requirements will limit the use of forecasts in contracts (for example, that robust contracts will use forecasts in production and other investment decisions, but not solely to improve risk sharing). Analytical models are only a part of the project, and we expect that our contributions will be both theoretical and applied, using a variety of research methods.
5