Analysis of CFRA Earnings Score Model Performance

Transcription

1 Phone: CFRA SCORE PERFORMANCE Analysis of CFRA Earnings Model Performance By Suraj Srinivasan, PhD September 4, 2013 CONTACT Dan Mahoney, CFA, CPA Phone: (646) Michael Chupeco Phone: (646) This report investigates the returns prediction performance of the Accruals model score (CFRA Earnings ) for the recent time period as well as the performance of the model leveraging industry scoring. The analysis below was performed by Suraj Srinivasan, Associate Professor in the Accounting and Management area at Harvard Business School, on behalf of CFRA. For a full background of Professor Srinivasan, please see Appendix 2. Executive Summary The results presented below are for the universe of firms in Compustat and CRSP for the time period Results using both annual and quarterly data are provided. Analysis was also conducted on the original sample in the Richardson et al paper 1 and the original results were verified for quarterly data also. Table A below presents a summary of hedge returns for all options considered. 1. Accrual model in a more recent time period performs similar to original Richardson et al results. Magnitude of returns is similar to the original paper. The CFRA quarterly adjusted model performs better than the annual model with raw and sizeadjusted returns of 13.98% and 13.13%, respectively. 2. These results validate the use of the model with recent data. 3. Within industry decile formation does not improve the performance of the Accrual model. Industry is defined using 2 digit SIC codes. Table A: Summary of hedge returns for different portfolio formation methods All returns are in percent Annual Quarterly Raw Returns Size adjusted Returns Raw Returns Size adjusted Returns Accruals Table Accruals (industry level) Table Richardson, Sloan, R., S., Soliman, M., and Tuna, I., Accrual Reliability, Earnings Persistence and Stock Prices.

2 Table 1. Sample Description The table presents the number of firms in the sample for each fiscal year. The sample consists of 145,581 firm-quarter and 36,321 firm-year observations from fiscal year 2001 to Panel A. Sample Distribution by Year Annual Quarterly Fiscal Year Freq. Percent Freq. Percent , , , , , , , , , , , , , , , , , , , , , , Total 36, , Panel B. Descriptive Statistics Adjusted CFRA was calculated using the CFRA performance adjusted methodology for both annual and quarterly data. In all tables returns and size adjusted returns are in percent. Annual Data Variable Obs Mean Std. Dev. Median P5 P95 Adj. CFRA 36, Return 36, Size_Adj Return 36, All rights reserved. This document may not be reproduced or re-disseminated in whole or in part without prior written permission from CFRA. 2

3 Quarterly Data Variable Obs Mean Std. Dev. Median P5 P95 Adj. CFRA 145, Return 145, Size_Adj Return 145, Table 2. Future Annual Mean Raw and Size-Adjusted Returns Portfolios are formed based on Adjusted CFRA Accrual. Firm-quarter (or firmyear) observations are ranked quarterly (or yearly) and assigned in equal numbers to decile portfolios based on the total accrual score. Hedge returns represents the net return generated by taking a long position in the Low portfolio (Portfolio 1) and an equal sized short position in the High portfolio (Portfolio 10). T-statistic tests whether the hedge return is statistically different from zero. The portfolio returns are equalweighted mean annual or quarterly buy-hold return. The size-adjusted return is calculated by deducting the value-weighted average return for all firms in the same size-matched decile, where size is measured as market capitalization at the beginning of the return cumulation period acquired from CRSP. The return cumulation period begins four months after the end of the fiscal year and continues for 12 months. Panel A. Annual The table presents future annual mean size-adjusted returns for decile portfolios formed on Adjusted CFRA Accrual. The sample covers 36,321 firm-year observations for the period Portfolio N Adj. CFRA Return Size_Adj Return 1 3, , , , , , , , , , T-Stat All rights reserved. This document may not be reproduced or re-disseminated in whole or in part without prior written permission from CFRA. 3

4 Panel B. Quarterly The table presents future quarterly mean returns and size-adjusted returns for decile portfolios formed on CFRA Accrual. The sample covers 145,581 firm-quarter observations for the period Quarterly returns are cumulative buy and hold returns for three months starting four months after the end of the fiscal quarter. Portfolio N Adj. CFRA Return Size_Adj Return 1 14, , , , , , , , , , Annualized T-Stat Table 3. Industry-Level Decile Formation Industry-level decile portfolios are formed based on CFRA Accrual within the twodigit SIC industry code. Firm-quarter (or firm-year) observations are ranked quarterly (or yearly) and assigned in equal numbers to decile portfolios by industry based on the total accrual score. Hedge represents the net return generated by taking a long position in the Low portfolio (Portfolio 1) and an equal sized short position in the High portfolio (Portfolio 10). T-statistic tests whether the hedge return is statistically different from zero. The portfolio returns are equal-weighted mean annual or quarterly buy-hold return. The size-adjusted return is calculated by deducting the value-weighted average return for all firms in the same size-matched decile, where size is measured as market capitalization at the beginning of the return cumulation period acquired from CRSP. The return cumulation period begins four months after the end of the fiscal year All rights reserved. This document may not be reproduced or re-disseminated in whole or in part without prior written permission from CFRA. 4

5 Panel A. Annual The table presents future annual mean size-adjusted returns for decile portfolios formed on CFRA Accrual by 2-digit SIC code. The sample covers 36,321 firm-year observations for the period The hedge portfolio size-adjusted return for Adjusted CFRA score is 11.2%. Portfolio N Adj. CFRA Return Size_Adj Return 1 3, , , , , , , , , , T-Stat Panel B. Quarterly The table presents future quarterly mean size-adjusted returns for decile portfolios formed on CFRA Accrual by 2-digit SIC code. Sample covers 145,581 firm-quarter observations for the period Quarterly returns are cumulative buy and hold returns for three months starting four months after the end of the fiscal quarter. Portfolio N Adj. CFRA Return Size_Adj Return 1 15, , , , , , , , , , Annualized T-Stat All rights reserved. This document may not be reproduced or re-disseminated in whole or in part without prior written permission from CFRA. 5

6 APPENDIX CFRA Earnings Model Overview CFRA s earnings quality model provides a proprietary quantitative earnings quality ranking for over 14,000 publicly traded firms globally. The model universe includes over 9,000 in North America and more than 5,500 companies internationally. Originally developed with three Wharton accounting professors, the back tested methodology measures the more discretionary, non-cash component of a company's earnings, which tends to be less reliable and less persistent than earnings supported by cash flows. Companies at higher risk levels are more likely to report disappointing results, restate earnings, and face regulatory enforcement action and class action law suits. At the other end of the spectrum, the model effectively highlights companies with understated earnings potential that are more likely to outperform peers and the broad market, and represent lower restatement and litigation risk. The CFRA model, combined with CFRA's best-in-class research offerings and global team of analysts provides clients with the industry's broadest and most in-depth forensic financial statement research and analysis on the market. CFRA Methodology Earnings quality means different things to different people. At CFRA, we view a high quality accounting system as one where the reported accounting rate of return provides a reasonable approximation of the firm's internal rate of return (IRR), and accounting book value provides a good measure of invested capital. Our focus then is to exploit information in the financial statements to measure the difference between a firm's reported accounting rate of return and its IRR. To do this, we measure the distortions embedded in the financial reporting process by decomposing reported earnings into an accrual and a cash flow component. The key difference between the accrual and cash flow components of earnings is that the accrual component involves a greater degree of management subjectivity. The accrual component of earnings typically incorporates estimates of future cash flows, deferrals of past cash flows, allocations and valuations, all of which rely on estimates, which introduces a significantly higher degree of subjectivity than simply measuring periodic cash flows. Thus, when the accrual component of earnings is unusually high, overall earnings will be less persistent. CFRA s quantitative approach to measuring earnings quality is to provide a comprehensive definition and categorization of accruals. Early academic research focused on a narrow definition of current accruals, ignoring many long-term operating accruals (such as capitalized expenditures and other intangible assets) which exhibit very similar issues related to the risk of management subjectivity All rights reserved. This document may not be reproduced or re-disseminated in whole or in part without prior written permission from CFRA. 6

7 CFRA s comprehensive definition of accruals captures all accruals and deferrals embedded in the financial statements that relate to both operating and investing activities of the firm. In our model, total accruals are measured as the change in net operating assets and are comprised of a current (operating) and non-current (investing) component as follows: Accruals = Change in Net Operating Assets Accruals = Current (Operating)Accruals + Noncurrent (Investing)Accruals. Current (operating) accruals capture the change in current assets, net of cash and shortterm investments, less the change in current liabilities, net of short-term debt. These accruals form the core of traditional measures of accruals used in earlier academic research. The major operating accrual accounts include trade accounts receivable, inventory and accrued liabilities, as well as other current assets and liabilities. These accounts tend to best reflect the subjectivity in revenue and expense recognition, and the extent to which a company tends to apply conservative or aggressive accounting practices. According to an Earnings Quality study authored by University of Michigan Accounting Professor Patricia Dechow and listed in a 2004 CFA Institute publication, the majority of SEC Accounting and Auditing Enforcement Releases in recent years relate to misstatements in accounts receivable and inventory accounts. Non-current (investing) accruals capture the change in non-current assets, net of longterm non-equity investments and advances, less the change in non-current liabilities, net of long-term debt. The most significant investing accrual components on the asset side are typically property, plant and equipment (PP&E) and intangibles. There is considerable subjectivity involved in the initial decision of which costs to capitalize for both PP&E and recognizable internally generated intangibles (such as capitalized software development costs), and subsequent decisions related to the allocation of these depreciable costs over the assets' estimated useful life. These non-current assets are also subject to periodic write-downs when they are determined to have been impaired, and the determination of such a permanent impairment involves considerable managerial discretion. On the liability side there are a variety of accounts such as longterm payables, deferred taxes and postretirement benefits which also tend to reflect subjective estimates. Growth and Performance Adjustment A proprietary aspect of the CFRA model is the adjustment that we make in our calculations to account for differences in growth and performance. Accruals naturally occur for firms that are growing because they have growing receivables, inventories, and other working capital, all of which are non-cash items. For these firms, a high level of accruals may not be a major concern. Conversely, poorly performing firms are more likely to liquidate inventories and reduce working capital. Thus, we expect to see a All rights reserved. This document may not be reproduced or re-disseminated in whole or in part without prior written permission from CFRA. 7

8 relatively low level of accruals. If instead, the magnitude of accruals is relatively high for such a firm, this is a warning sign. Our proprietary adjustments help to parse out cases where high accruals are less of a concern and cases where earnings manipulations are more likely to be the reason for the high level of accruals as well as cases where high accruals are a stronger indicator that forward earnings growth rates are unlikely to persist. The proprietary adjustment is embedded in our final accrual decile ranking All rights reserved. This document may not be reproduced or re-disseminated in whole or in part without prior written permission from CFRA. 8

9 APPENDIX 2 About Suraj Srinivasan Suraj Srinivasan is an Associate Professor in the Accounting and Management area at Harvard Business School. He teaches the second year MBA elective Business Analysis and Valuation Using Financial Statements and to executives, Strategic Financial Analysis for Business Evaluation. He also teaches in corporate governance programs Making Corporate Boards More Effective, Audit Committees in a New Era of Governance and Compensation Committees: New Challenges, New Solutions. Prior to joining HBS, Professor Srinivasan was an Assistant Professor of Accounting at the University of Chicago Graduate School of Business from where he received the Ernest R. Wish Accounting Research prize in Professor Srinivasan s research examines corporate governance in the U.S. and internationally. He has studied issues such as the impact of globalization on corporate disclosure practices and compensation arrangements in international companies, the effect of securities regulation on incentives of companies to cross list in the U.S., incentives of audit firms to provide high quality audits, and reputational consequences for corporate directors when companies experience financial reporting problems. Professor Srinivasan earned a bachelor's degree with honors in electrical and electronics engineering and a master's degree in physics with honors from Birla Institute of Technology and Sciences in India prior to earning an MBA from the Indian Institute of Management Calcutta. He also received a doctorate degree in business administration from Harvard Business School in 2004 where he received the George S. Dively Award for outstanding thesis research All rights reserved. This document may not be reproduced or re-disseminated in whole or in part without prior written permission from CFRA. 9

10 The content of this report and the opinions expressed within are those of CFRA. This analysis has not been submitted to, nor received approval from, the United States Securities and Exchange Commission or any other regulatory body. While CFRA exercised due care in compiling this analysis, CFRA AND ALL RELATED ENTITIES SPECIFICALLY DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, regarding the accuracy, completeness or usefulness of this information. and assumes no liability with respect to the consequences of relying on this information for investment or other purposes. In particular, the research provided is not intended to constitute an offer, solicitation or advice to buy or sell securities. CFRA s financial data provider for financial companies is SNL FINANCIAL LC. CONTAINS COPYRIGHTED AND TRADE SECRET MATERIAL DISTRIBUTED UNDER LICENSE FROM SNL. FOR RECIPIENT S INTERNAL USE ONLY CFRA, CFRA Accounting Lens, CFRA Legal Edge, CFRA, and all other CFRA product names are the trademarks, registered trademarks, or service marks of CFRA or its affiliates in the United States and other jurisdictions. CFRA may be protected by U.S. Patent No. 7,974,894 and/or other patents. If you have any comments or questions, please contact cservices@cfraresearch.com. CFRA Contact Information Phone + 1 (212) Web cservices@cfraresearch.com All rights reserved. This document may not be reproduced or re-disseminated in whole or in part without prior written permission from CFRA. 10