|
Page 1 of 4
Various criticisms have recently been made with respect to ubiquitous style boxes, including the contention that they cost investors absolute and risk-adjusted performance during bear markets. While some of this criticism is thought provoking, blaming the size/style framework for unsatisfactory performance is misplaced. Size and style explain a large part of the cross section of equity returns, and the framework has been an invaluable resource for facilitating performance measurement and passive investment. However, style boxes should not be interpreted as a cage. It may very well pay to invest with managers that fall “outside the box,” but surely it also pays to know that fact. Conversely, it is surely beneficial not to overpay for exposure within style boxes. While there are no universally accepted definitions of growth and value, there is a body of topical literature in which factors consistently appear as helping to define growth or value, or both. Ideally, style benchmarks would reflect theory as it evolves through time.
First-generation style benchmarks were built around single factors such as book to price. Since growth and value were defined by a single variable, value consisted of stocks with high book-to-price values, and growth consisted of those with low book-to-price values. However, relying on one measure to differentiate growth from value was justifiably criticized. Pope, Rakvin and Platt (2003)1 point out some of the limitations of book to price and state that while it does seem to have explanatory power, “there logically must be … additional information in other factors.” As a solution, next-generation style benchmarks assessed growth and value along separate dimensions by identifying distinct growth and value factors.
While multifactor style indexes were an important innovation in style benchmarking, they also suffer shortcomings. Over time, developments in financial research, changes in accounting standards and shifts in investment sentiment jointly contribute to variations in growth and value factor sensitivities. Given the inherently factor-based nature of style indexes, good benchmarks would ideally incorporate some level of dynamism. On the other hand, it is important to avoid too much change because benchmarks provide a standard for performance measurement. By periodically updating factors every five to 10 years, style benchmarks can capture an aspect of market representation that may be ignored in static formulations. For example, much work has been done in recent years on styles factors, so the universe of potential factors is of a considerably different makeup than it was 10 or 15 years ago. Style benchmarks should take these developments into account. Less directly but perhaps just as significantly, changes in accounting standards may affect factor efficacy in growth and value differentiation. Periodic factor updates would capture changing dynamics on both fronts without over-fitting to recent history.
The Need For Factor Revision
With the 2009 rebalance of its U.S. style benchmarks, S&P Indices revised the factors it uses to categorize growth and value stocks and create its style indexes. S&P embarked on this update with the goal of addressing the need for increased dynamism in style benchmarking. Measures of the need to review and potentially revise style factors include the frequency, breadth and depth of changes in accounting standards over time. Since the original deployment of S&P’s current factors, many changes have occurred in accounting standards and also, perhaps, in the factors that best reflect the investment decisions of growth and value managers. Figure 1 shows the number of new standards issued by the Financial Accounting Standards Board (FASB) during 2005-2009.
Some of these changes in standards have had far-reaching consequences. One example is the statement of financial accounting standards (SFAS) No. 158, released in September 2006, which changed the rules under which public companies report pension-related assets and liabilities on the balance sheet. Mulford, Quinn and Swanson (2008)2 show the effects of SFAS No. 158 on total assets, total liabilities and shareholder equity, as well as on related profitability measures such as return on assets (ROA) and return on equity (ROE). Of the 24 companies studied (Dow Jones Industrial Average components with defined benefit plans affected by SFAS No. 158 in 2006), the change in ROE after adjusting for the new standard ranged from 0.5 percent to 173.9 percent. To analyze ROE through time as a determinant of returns, we ranked the S&P 500 universe and measured ROE quartile portfolio returns for seven nonoverlapping periods covering January 1988 through August 2009. Figure 2 shows the results; it seems clear that ROE performance is highly time-varying.
|
Previous Page
