“If your experiment needs statistics, you ought to have done a better experiment.” Ernest Rutherford
Sometimes you need to get into the fundamentals to understand if your belief system is sound. In corporate governance literature of the last two decades, there is no more fundamental concept than Tobin’s Q, which legions of law professors have used as a proxy for firm value. Based on regression analyses examining variations in Tobin’s Q, they have made definitive pronouncements about any number of corporate governance topics, from staggered boards to the value of activism. Yet tracing the evolution of Tobin’s Q to its current state—a state completely alien to the original conception—reveals a twisted tale, proceeding like an epidemiological disaster in which Tobin’s Q transforms from an innocent and useful organism in macroeconomics to an unrecognizably mutated and widespread disease in corporate governance literature, infecting policies and practices throughout the corporate governance world.
The way has been lit by Professors Robert Bartlett and Frank Partnoy (“B&P”) in their wonderfully clear article “The Misuse of Tobin’s Q.”
B&P introduce their story by noting that Tobin’s Q has been used as a proxy for firm value in over 300 law review articles on corporate governance. However, Tobin did not design Q as a proxy for firm value, but rather as a macroeconomic indicator of investment opportunities. How we got from Tobin’s Q to the current misused and mutated Q (what we will call “Mutant Q”) does not reflect particularly well on academia or the hypo-critical consumers of its work product at Institutional Shareholder Services (ISS) and Glass Lewis. B&P lay out an aggregation over decades of errors, illogical leaps, over-simplifications and over-generalizations that created Mutant Q and allowed it to flourish in academia, and it becomes clear that the generalized use of Mutant Q both made possible a generation’s worth of literature promoting the virtues of a shareholder dominated model of corporate governance and at the same time ensured that the foundation of this literature would be unsound. It is one of those odd cases, that you sometimes find in tracing back case law precedent, where the complications and long, tortured history of a doctrine’s evolution somehow garner more acceptance for the doctrine than common sense ever could.
The Healthy Organism. Tobin’s Q started out its life in the late 1960s as the invention of Yale economics professors Tobin and Brainard. These decorated practitioners of the dismal science reduced to a formula the insight that, on a macroeconomic level, “the market valuation of equities, relative to the replacement cost of the physical assets they represent, is a major determinant of investment.” The idea, once stated, seems straightforward enough—if the market value of the equities representing capital goods is greater than the replacement cost of those capital goods (i.e., if the quotient Q—resulting from dividing market value of equities by the replacement value of the underlying capital goods—is greater than 1), then investing in more of those capital goods is likely a good idea. This idea was useful in explaining, among other things, why an easy monetary policy spurred investment—by lowering the replacement costs of capital goods, Q would increase and, if Q increased to greater than 1, investment should increase as well.
What does this have to do with firm value? Nothing. That is the point. This is the beginning, before the mutations.
Jumping Species. For over a decade, Tobin’s Q stayed within the confines of macroeconomics. Q jumped unexpectedly to microeconomics in a 1981 study by Messrs. Lindenberg and Ross (“L&R”). For the first time, instead of using Q as an indicator of macroeconomic investment opportunities, L&R measured Q for individual firms, associating a high Q with the ability to charge monopoly rents. However, L&R’s study was not focused in any way on firm value—they were looking at “the dispersion of market power across firms and industries.” In other words, there was no expectation by L&R that one could move from a macroeconomic aggregation across the economy to a firm-by-firm measurement of Q and find that those individual measurements had any particular significance. While they did imply that high Q firms might be more profitable, the applicability of this implication to any particular firm was brought into question by L&R explicitly, as they noted that “beyond the usual caveats concerning data availability and quality” a high Q could simply reflect lower costs that were not reflected on the books.
This is a critical point and one that seems to have been lost on the generations that came after L&R. Strikingly, L&R’s study found the highest Q values for early tech companies like Xerox, 3M, IBM and Polaroid, drug companies like Schering-Plough, Eli Lilly, and Smith Kline, and strong consumer brands like Coca Cola, Gillette, Johnson & Johnson, Avon and Kellogg—all companies where the denominator for Q (the replacement value of physical assets) did not take into account potentially huge sources of value, i.e. intangible assets like goodwill, patents, and other IP. (At this point in its transformation, the denominator of Q did not include any intangible assets.) L&R do not suggest that Avon (with a Q of 8.53), or Coca Cola (with a Q of 4.21) were somehow “higher value” or more profitable than Mobil (with a Q of 1.20). The only valid inference would be that Avon was, relative to its physical assets, more highly valued than Mobil—which really has nothing to do with the actual relative values of Mobil and Avon, much less the relative virtues of their managers or board structures. Scaling market values against the replacement cost of physical assets does make it easier to compare firms of different sizes and valuations, but in a world where intangible assets are valuable, the Q that you are comparing doesn’t of itself have much meaning. As if to drive home the point, L&R explicitly stated that firms with higher Qs “are often those with relatively unique products, unique factors of production, and so forth.”
Given their singular focus on market power, L&R were presumably far from dreaming that Q—through the magic of regression analysis—would eventually be deemed to be indicative of the desirability of a particular board structure, a particular level of management ownership, limitations on shareholder bylaw amendments, supermajority voting provisions, poison pills or golden parachutes. This seems especially true when one considers that their own study seems to attribute a high Q to high value intangible assets, rather than to any super-attenuated feature of a firm’s governance structure. Nonetheless, subsequent governance studies placed huge pressure on the accuracy of Q as a proxy for firm value—a pressure that Mutant Q cannot begin to bear.
Patient Zero. By the early 1990s, a 1988 study by Professors Morck, Shleifer and Vishny (“MSV”), together with the L&R article, was commonly cited “as the justification for using q as a proxy for firm value.” It is worth noting that in MSV’s study Q was still very far from its final mutation. Most significantly, the denominator for MSV’s Q consisted of estimates of the actual replacement value of a firm’s physical assets, rather than the book value of all assets. More important than their definition of Q, however, was their use of Q.
MSV asserted, non-controversially, that “Tobin’s Q is high when the firm has valuable intangible assets in addition to physical capital, such as monopoly power” and—citing L&R— “goodwill, a stock of patents, or good managers.” In other words, if the firm has valuable intangible assets, and that value is recognized in its market value (the numerator of Q), but not in the denominator of Q, then Q is higher than for comparably valued firms without intangible assets. Reducing the denominator increases the quotient Q. Straightforward math.
What MSV did next, however, was revolutionary, and perhaps the reason that their article has been cited over 9,600 times. MSV decided to use Q as their “proxy for market value of the firm’s assets.” In effect, MSV decided that Q could stand in as a scaled value to compare firms of different sizes—firms with higher Qs being deemed to be relatively more valuable, as they were apparently more effective in converting physical assets into market value. Then MSV went further—creating a cottage industry for finance and law professors—by deciding that differences in this scaled value could be made attributable solely to firm management. More specifically, they hypothesized that by looking at the association between the percentage of a firm owned by management and the Q of such firm, they could determine the ideal level of management ownership.
First, let’s examine the idea that Q reflected a firm’s scaled relative value, reflecting its effectiveness in converting physical assets into market value. Remember that MSV’s mutation of Q reflected only physical assets in the denominator. Accordingly, Q could be viewed as a testament to management’s ability to wring value from physical assets, but only if the firm (and all the firms it was compared to) had no other intangible assets. MSV themselves acknowledge the existence of other intangible assets, such as “monopoly power, goodwill, a stock of patents, or good managers.” So Q is necessarily not just a measure of the firm’s effectiveness at converting physical assets into market value, but, as much as anything else, a measure of how many valuable intangible assets a firm had that were not reflected on its balance sheet. One of these valuable intangibles could of course be a good management team, but to measure the management team’s effectiveness, you would have to control for all of the other possible intangible assets affecting a firm’s Q. MSV rather blithely did attempt to do so, but, as discussed below, it was a remarkably ineffective attempt.
Second, the idea that deviations in Q could be traced to the effectiveness of management deserves a few observations.
- To step back a minute, if you somehow could design a Q that represented the market value of a firm’s equities over the replacement value of all its assets, all you would have, in Tobin’s world, would be a Q that indicated whether more investment might be a good idea. In this world, a high Q would simply mean that you might have a firm with valuable investment opportunities and a management team that was missing out on those investment opportunities – not a more valuable firm.
- In MSV’s Q, where intangibles were not accounted for at all, a high Q could represent, among many other things, “monopoly power, goodwill, a stock of patents, or good managers.” MSV acknowledge that Q “is a very noisy signal of management performance,” but it was the signal they chose to listen to. MSV did, to their credit, make an attempt to control for the presence of two intangible assets—valuable IP and a valuable brand—by controlling for R&D spend and advertising expenses. However, their attempt was singularly inapposite. A firm with a high R&D spend could be indicative of a need to catch up to a competitor who had already accumulated valuable IP, or indicative of the fact that the firm valued IP more than its competitors and so regularly invested in R&D. R&D, in short, is not a good proxy for the value of a firm’s “stock of patents.” Similarly, a firm’s high advertising budget could indicate that a firm is spending to protect its mighty brand, or could indicate that the firm is spending mightily to develop a brand. Not to mention that not all firms yield the same returns from their investment in R&D and advertising.
Finally, the MSV study was also notable for its attribution of all management success or failure to one aspect of management: the percentage of management ownership. MSV did not attempt to control for the myriad other variables that could affect management performance, such as tenure at the company, years of industry experience, education levels, reporting structures, budget to effect change, firm culture, or (to cite later favorites) entrenchment, board structure, and compensation structure. This approach of assuming away variables too numerous to contemplate is a feature of many later studies, and another weakness of the academic approach that should worry anyone charged with making actual decisions about the corporate governance structures at a particular firm.
In short, neither the L&R or MSV studies, which launched a thousand paper ships, sets a proper foundation for using Q as a proxy for firm value. While there were attempts to design a Q that better served this purpose, as discussed below, those attempts were doomed to failure precisely because the radical transformations that led to “Tobin’s Q” being used as a proxy for firm value were not thoughtfully made, or properly justified after the fact. Yet having embarked on this tempting path, where so many ideas on governance could easily be “tested” statistically across firms with this handy tool and some drudgery on the part of research assistants, an academic publishing onslaught was predictable. This onslaught only accelerated as Q further mutated to make its use even simpler, though the results were unfortunately no more reliable.
Mutant Q Under the Microscope.
The modern formulation of Mutant Q is perhaps even less comforting than earlier versions for those hoping to rely on an assumption of academic rigor. While Q has evolved, it has evolved primarily to make its calculation simpler for purposes of scholarly manipulation, not to make it more meaningful.
It took a fair amount of work to determine the replacement value for physical assets in the older versions of Q, and these older versions suffered from the complete exclusion of intangible assets from the denominator of Q. Both problems were “solved” by the expedient of simply using the book value of total assets (rather than the replacement value of physical assets) as the denominator of Mutant Q. Book value of total assets is of course far less work-intensive to determine than the replacement value of physical assets—an obvious attraction to researchers eager to test their theories. Mutant Q as a result is now quite simple to determine. This simplicity may be why B&P refer to Mutant Q, somewhat less provocatively, as “Simple q,” while noting, somewhat more provocatively, that “corporate finance scholars routinely and sanguinely use this market-to-book version of Simple q largely without question, perhaps because it is wrapped up in the lore of “Tobin’s q,” which might mask the fact that it is merely a “market-to-book”’ ratio.” They find its acceptance “interesting and surprising, given how many scholars warned, two decades ago, about its potential problems.”
There are a number of questions B&P would like those corporate finance scholars to consider, but perhaps the most important turn on the comparability of the book value of assets of different firms. In particular:
- Physical Assets: The basics of plant, property and equipment are initially reflected on the books at historical cost and subject to widely varying depreciation schedules, which depreciation schedules are not consistent across firms and have nothing to do with deterioration in economic value. How could those book values be said to accurately reflect replacement value, or even be proportionately related to replacement value? This problem alone introduces a large and random measurement error into the calculation of Mutant Q.
- Intangible Assets: A huge part of the value of many modern corporations is reflected on their balance sheets only in the most random manner. If an acquiror buys a target company, the acquiror’s goodwill will largely be written up to reflect the purchase price of target, effectively capturing the value of target’s intangibles in the book value of acquiror. If, on the other hand, the acquiror developed intangibles of the same value in-house, through, e.g. R&D expenditures, those intangibles would not be reflected as an asset on the balance sheet. B&P calculate Microsoft’s Q by adding an estimate of the market value of its intangibles (which was $54 billion higher than the book value of those assets) to the denominator of Q. This results in a Q of 1.77. If the market value of the intangibles were not included in the denominator of Q, Microsoft’s Q would be 3.25. Accordingly, if Microsoft were acquired tomorrow, its “value”—measured by Q—would be cut in half. It is hard to believe that Q is a good proxy for Microsoft’s value if so much turns on this accounting distinction. It is equally hard to believe that corporate directors, shareholders and proxy advisory firms make decisions about governance based on studies premised on this distinction—but they do.
As B&P conclude: “Despite their common use, market-to-book proxies for Tobin’s q…are unreliable measures of firm value…likely to produce biased estimates due both to omitted assets (e.g., intangibles) and firm specific details that can systematically alter Simple q (e.g., the level of current assets, depreciation, etc.). Scholars should view with suspicion any assertions that firm characteristics [e.g., board structure] affect firm value because they affect Simple q.”
Autopsy of a Super-spreader
To put their criticisms to the test, B&P re-ran the numbers on a particularly influential corporate governance study, substituting their best guess at the true replacement value of assets (including intangible assets) for the book value denominator used in the original study. Strikingly, the results of the study— Professors Bebchuk, Cohen and Ferrell’s “What Matters in Corporate Governance,” published in 2009—do not hold when the denominator is altered by this nod to reality. This study—which has been cited over 550 times since it was published—uses Mutant Q to determine the importance to shareholders of 6 corporate governance features that (in its far from neutral parlance) “entrench” management: staggered boards, limits to shareholder bylaw amendments, supermajority voting provisions for mergers and charter amendments, poison pills and golden parachutes. The study purported to show, in a qualified manner, that elimination of these features correlates with, and may even cause, increased firm value—as measured by Mutant Q. However, there is no correlation when using B&P’s replacement value Q. One can only guess whether the 550 studies that came after “What Matters in Corporate Governance,” would be similarly rendered meaningless by adoption of a more realistic Q.
Why does this matter? Primarily because it means literally hundreds of well-regarded studies on corporate governance matters, studies that have been the justification for spectacularly aggressive changes to corporate governance practices, derive their conclusions from a fundamentally unworkable conception of firm value. These studies have influenced a generation of thought on corporate governance, and in spite of the randomness of the errors introduced by Mutant Q, often seem to drive in the same ideological direction— towards less protection for directors, and more immediate control by shareholders. These studies are still cited by ISS and Glass Lewis as the basis for opposition to staggered boards and other structural corporate defenses critical to allowing directors to pursue long term strategies. Stripped of these traditional defenses, companies are significantly more vulnerable to “investment limiting” campaigns by activists, to pressure from activists and others to sell, spin off, distribute more cash, or otherwise take actions that pump up the stock price in the near term but may have a negative effect on long term value. The net effect is to leave the management of public companies not to directors but to the loudest and most pressing shareholders. B&P’s article demonstrates convincingly that the assumption of academic rigor granted to these foundational studies is unwarranted. Deference to ISS and Glass Lewis policies that are based on these studies is similarly unjustifiable.
 See Robert P. Bartlett and Frank Partnoy, The Misuse of Tobin’s Q (SSRN, June 28, 2018), available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3118020.
 To cut to the chase, Mutant Q is essentially a market-to-book value ratio. It is speculation, but one has to wonder if a market-to-book value had been introduced 30 years ago as a proxy for firm value, whether anyone in academia would have bitten.
 Bartlett and Partnoy, The Misuse of Tobin’s Q at 9 (cited in note 1).
 See generally Eric B. Lindenberg & Stephen A. Ross, Tobin’s q Ratio and Industrial Organization, 54 Journal of Business 1 (1981).
 Id at 30.
 Id at 30.
 Id at 29.
 See generally Randall Morck, Andrei Shleifer & Robert W. Vishny, Management Ownership and Market Valuation: An Empirical Analysis, 20 Journal of Financial Economics 293 (1988).
 Bartlett and Partnoy, The Misuse of Tobin’s Q at 15 (cited in note 1),
 Morck et al., Management Ownership and Market Valuation at 296 (cited in note 8), generally citing Lindenberg and Ross, Tobin’s q Ratio and Industrial Organization (cited in note 4).
 Id at 294.
 Id at 296.
 Id at 296.
 Philip H. Dybvig and Mitch Warachka, Tobin’s q Does Not Measure Firm Performance: Theory, Empirics, and Alternatives (SSRN, March 5, 2015), available at https://ssrn.com/abstract=1562444 or http://dx.doi.org/10.2139/ssrn.1562444.
 For those who would like a bit more precision, Simple q is formulated as q=MVE+BVD-DT/BVE+BVD, where MVE=market value of equity, BVD=book value of debt, DT=deferred taxes, and BVE=book value of equity. Since BVE+BVD= TA, where TA= the book value of total assets, this can also be written as q=MVE+BVD-DT/TA. Some versions ignore the enhancement of DT, writing the formula as q=MVE+BVD/TA.
 Bartlett and Partnoy, The Misuse of Tobin’s Q at 16 (cited in note 1).
 Id at 31-32.
 Id at 50.
 Lucian A. Bebchuk, Alma Cohen and Allen Ferrell, What Matters in Corporate Governance? 22 Review of Financial Studies 783 (2009).