Correlated Shortcuts – Industry Window DressingVW Staff
Correlated Shortcuts – Industry Window Dressing
London School of Economics & Political Science (LSE)
Harvard Business School; National Bureau of Economic Research (NBER)
London School of Economics & Political Science (LSE)
February 4, 2016
We explore a new mechanism by which investors take correlated shortcuts, and present evidence that managers undertake actions – in the form of sales management – to take advantage of these shortcuts. Specifically, we exploit a regulatory provision wherein a firm’s primary industry is determined by the highest sales segment. Exploiting this regulation, we provide evidence that investors classify operationally nearly identical firms vastly differently depending on their placement around this sales cut-off. Moreover, managers appear to exploit this by manipulating sales to be just over the cutoff in favorable industries. Further evidence suggests that managers then engage in activities to realize large, tangible benefits from this opportunistic action.
Correlated Shortcuts: Industry Window Dressing – Introduction
Investors are continuously faced with a large number of potential signals that are available to collect and process. Faced with these, investors need to solve the complex resource allocation problem with respect to selecting and processing each potential signal. Indeed, if investors take correlated shortcuts in their investment decisions, then simple pieces of information can remain systematically unreflected in firm prices. Moreover, if firm managers are aware of these shortcuts and their implications, managers may take specific actions to exploit these investment shortcuts.
In this paper, we identify one such shortcut that financial agents take and document how it may affect prices, and further how firm managers may react. Specifically, we examine the primary industry into which each firm is classified. The Securities and Exchange Commission (SEC), in classifying firm operations, designates that each conglomerate firm have a primary industry, determined by the segment with the highest percentage of sales. Using this rule, we exploit situations in which firms tightly surround the discontinuity point of industry classification. For example, a two-segment firm that gets 53% of its sales from technology and 47% of sales from lumber is classified as a technology firm, whereas a firm with nearly identical1 operations but that gets 47% of its sales from technology and 53% of sales from lumber is classified as a lumber firm.
If investors overly rely on this primary industry classification in their investment decisions without fully factoring in firms’ underlying economic operations, they may perceive or treat nearly identical firms around the discontinuity point in different ways. We examine the idea of “naive categorization” by examining both stock return patterns and more directly investor behavior. First, we explore how investors price these conglomerate firms at the cutoff. We find that despite being nearly identical, firms just over the 50% point (in terms of percentage sales from a particular industry) have significantly higher betas with respect to that industry than firms just below the 50% point. So, in the example above, the 53% technology firm’s price moves much more closely with the technology industry than does the 47% technology firm’s price. The difference in industry beta is large both economically and statistically: firms just over the 50% cutoff, on average, have a 60% larger beta (t = 4.91) with respect to the industry in question than those just under the threshold. Importantly, there are no other jumps in industry beta anywhere else in the distribution of firm operations.
Second, corroborating the evidence on industry beta, we find that mutual fund managers exhibit differential investing behavior around the industry classification cutoff. In particular, we focus on mutual funds with a significant sector tilt (based on past holdings). For firms that are nearly identical in their exposures to a particular industry, with the only difference being just above versus below the discontinuity,2 mutual funds that specialize in that industry are significantly more likely to hold firms just above the cutoff than firms just below it. Specifically, the fraction of sector mutual funds investing in a firm is 40% larger (t = 2.55) if the firm is just above the 50% point (in terms of sales from that industry), relative to firms just below the cutoff. Like the beta test, this is the only jump in sector mutual fund holdings throughout the entire distribution of firm operations.
We see the same behavior from sell-side analysts. For each firm, we measure the percentage of sell-side analysts covering the firm from each sector, and find a significant jump in sell-side analyst coverage at the industry classification cutoff. In particular, firms just above the cutoff have significantly more coverage from the classification industry than nearly identical firms just below the cutoff; the former have a nearly 60% (t = 2.27) higher fraction of analysts from the classification industry covering them compared to the latter. Again, we see no similar jumps in coverage percentage anywhere else in the distribution. It is important to note that while our results on industry beta, as well as analyst and mutual fund manager behavior are consistent with financial agents taking correlated shortcuts, they could also be driven by institutional constraints imposed on analysts and investors.
See full PDF below.