It is a stock of past, ploughed-back earnings
corrected
for depreciation (assuming the whole thing is measurable to begin with).
Nitzan Bichler - 2012 - Capital as Power
All it takes is two easy steps.
First, they need to single out on the efficient frontier that particular portfolio (labelled M for convenience) which, when combined with borrowing or lending, yields the highest return
17 Ricciardi (2004) managed to collate a list of no less than 150 unique risk indicators - hardly an indication of unanimity.
? Elementary particles 205
for every level of volatility. And then they make their move. Those who are more risk averse can invest part of their money in M, putting the rest of it into risk-free assets (i. e. lending it to the central bank). And those who are less risk averse can borrow at the risk-free interest rate and invest the extra cash in additional units of M. Life has never been simpler.
CAPM
These guidelines, though, are still limited. Their target is the individual investor who already possesses definite expectations on return and variance and merely awaits instructions on how to diversify. The guidelines are silent, however, on how investors formed these expectations to begin with, and on the market consequences should they all follow the diversification advice. These latter questions were taken up by William Sharpe (1964) and John Lintner (1965) in their capital asset pricing model, or CAPM for short.
On the face of it, the questions seem unanswerable. Since individual inves- tors are assumed to be autonomous, their return/variance expectations are open-ended and can take any value whatsoever. And given that the expecta- tions are unbounded, the consequences of acting on them become unpredict- able. So Sharpe and Linter decided to simplify. What would happen, they asked, if all investors happened to share the same expectations regarding return and variance - and, moreover, if their expectations were the same as the true distribution of outcomes (i. e. if investors knew the stochastic generator of history)?
The scenario is admittedly odd. After all, what can one learn about uncer- tainty by assuming it away? Indeed, what would generate a future variance if all capitalists were the same and if all knew the future variance? Recall, however, that we are dealing here with the articulation of the capitalist nomos. The ultimate task is not to theorize capitalists, but to give them a bureaucratic blueprint. And if you revisit the previous section you will see that, in pursuing this goal, Sharpe and Linter were merely following Keynes' checklist.
And indeed, the answers they gave emerged directly from their assump- tions. If investors all see the world eye to eye, they will all own the same effi- cient portfolio M, and nothing but M. And since they all own only units of M, every owned stock in the market must be part of M. The only portfolio that satisfies these conditions is the market as a whole. So in Sharpe and Lintner's world, investors are fully diversified, each holding onto a propor- tion of the entire market.
Now, recall that diversification reduces volatility because the price move- ments of different assets partly offset each other. Why is the offsetting only partial? The reason is that price volatility is seen as stemming from two distinct sources: one that is unique to the asset itself, and another that is common to the market as a whole. A sufficiently diverse portfolio - which M obviously is - eliminates all unique volatility. Thus, following Sharpe and Lintner, portfolio investors can disregard all price volatility that is out of sync
206 Capitalization
with the market. But no matter how diverse their portfolio, they can never eliminate market volatility, by definition.
In other words, every asset carries in its genes a definite quantum of market volatility, a quantum which in turn is passed on to the portfolio as a whole. 'The risk of a well-diversified portfolio', declare Brealey, Myers and Allen in their Principles of Corporate Finance (2008: 193), 'depends on the market risk of the securities included in the portfolio'. This is not something one argues about: 'Tattoo that statement on your forehead if you can't remember it any other way', they warn. 'It is one of the most important ideas in this book'.
Conveniently, this logic is fully invertible. Since individual assets contri- bute to the market portfolio their own market risk, their contribution can be deduced from the market. Recall that unique risk has been diversified away, so the only thing that makes one asset more or less risky than another is its 'sensitivity' to the overall market. This sensitivity is called beta, and it can easily be measured, or so we are told. The greater the market risk of the asset, the higher beta - and vice versa. 18
Now, if the measured historical beta is equal to the 'true' timeless beta, as the CAPM proclaims, we can calculate the asset's risk premium. By defini- tion, beta expresses the ratio between, on the one hand, the 'excess' return of the asset over and above 'risk-free' assets (asset excess return = r -rrf) and, on the other hand, the 'excess' return of the market over and above 'risk-free' assets (market excess return = rm - rrf):
5. beta = r - rrf rm - rrf
Rearranging Equation (5), the capitalist can obtain the risk premium for the asset:
6. risk premium = r - rrf = beta (rm - rrf)
And, finally, with one more reshuffle, the overall discount rate (r):
7. r = rrf + beta (rm - rrf)
The first component on the right is the 'risk-free' benchmark (rrf), while the second component is the compensation for the asset's risky 'deviations' (beta (rm - rrf)).
18 Beta can be derived in two easy steps. First, assume that all investors consider historical volatility equal to 'true' volatility. Second, run a linear regression with the historical rate of change of the asset price as the dependent variable, and with a constant and the historical rate of change of the market price as the independent variables. Beta is the estimated slope coefficient of this regression.
? ? Elementary particles 207
And that is pretty much it. All that the capitalist now has to do is take the 'risk-free' rate of interest (rrf), the market return (rm) and the estimated beta and plug them into Equation (7). And since all capitalists are assumed to be drones and therefore to do the very same maths with the very same numbers, they all end up with the same discount rate (r). 19 This number is then put into the denominator of the discount formula in Equations (3) and (4) to give the 'true' capitalized value of the asset.
Circularity
CAPM has been a smashing business success. Within a few decades, it has become the basis on which corporate finance courses are structured. Often, it is the only model that MBA students rehearse in some detail. And as the students turn into managers, executives and government officials, they apply what they learnt. In the early 1980s, less than one third of large US firms used the CAPM to compute the cost of equity. By the early 2000s, the proportion was well over 70 per cent, although, to be on the safe side, other formulae are used as well (Gitman and Mercurio 1982; Graham and Harvey 2001).
This success is all the more remarkable given the model's dismal empirical showing. Recall that Sharpe and Lintner assumed that investors know the 'true' variance of returns, and therefore also the 'true' beta - yet that, in prac- tice, they all take a shortcut and use the historical beta instead. No wonder the model crumbles.
First, there is the 'equity premium puzzle': it turns out that, taken as an asset class, equities outperform government bonds by much more than their extra volatility would demand (Mehra and Prescott 1985). And then the puzzle becomes embarrassing. Comparisons of different classes of equities often show returns to be uncorrelated and sometimes negatively correlated with beta values! 'The problems are serious enough', conclude efficient market theorists Fama and French, 'to invalidate most applications of the CAPM' (2004: 43).
Given these difficulties, the CAPM has been refurbished, extended and modified with new and improved techniques and a never-ending flow of fresh data. 20 But the new models share with the old one key feature: circularity. Excess return is the compensation for risk, while risk is measured by excess return. This correspondence holds simply because it should:
[F]ew people quarrel with the idea that investors require some extra return for taking on risk. That is why common stocks have given on
19 Equation (7) produces two special cases. One is for 'risk-free' assets, which, being uncorre- lated with the market, have a beta of 0 and therefore a discount rate of rrf. The other is for the market index itself, whose beta is 1 and whose yield is therefore rm.
20 Two famous extensions/alternatives are Ross' Arbitrage Price Theory (1976) and the augmented CAPM of Fama and French (1995).
? 208 Capitalization
average a higher return than US Treasury bills. Who would want to invest in risky common stocks if they offered only the same expected returns as bills?
(Brealey, Myers, and Allen 2008: 217, first emphasis added)
It is just like Lamarckian evolution. A giraffe grows its neck to reach the high leaves on the tree, and so does the market: it makes prices of volatile assets rise faster in order to give investors a reason to own them.
Risk and power
The issue here is not circularity as such, but the worldview that underlies it. The framework of mainstream financial theories is neoclassical. Its basic units are risk-averse, utility maximizing investors. These individuals are powerless. They are too small to affect the circumstances and hence take them as given. Their only possible course of action is reaction: buying stocks whose return/risk attributes are attractive and selling those that are unattractive (that is, until the market equilibrates their prices to their natural levels). Their focus is price, and only price. The price tells them everything they need to know about return and risk. Whatever lies beneath it is irrelevant. And so the discount formula disappears.
The CAPM reasons the link between return and risk in moral terms: the capitalist 'deserves' higher returns to compensate for higher risk. But if we abandon the fairy tale of perfect competition and efficient markets and return to the real world of organized capitalist power, the capitalization formula comes back into focus and the relationship between risk and return assumes a rather different meaning.
Back on earth, the sequence is as follows: earnings are a matter of power and conflict; the conflict over earnings invites resistance; and resistance breeds volatility and uncertainty. In this way, the capitalist struggle to increase earnings is inextricably bound up with uncertainty regarding their eventual level. The numerator of the capitalization formula becomes inti- mately tied to its denominator.
The degree of confidence
Recall that in capitalism the ownership of an asset is a claim on future earn- ings. The price of the asset, expressed as present value, is merely the capitalist assessment of those earnings. Underlying this assessment are two key con- siderations. One is the level of earnings capitalists expect to receive; the other is the degree of confidence capitalists have in their own predictions. In order to make this degree of confidence explicit, we rewrite Equations (3) and (4), such that:
8. Kt = EE = E * H r rc * ?
? ? 9. Pt = EEPS = EPS * H r rc * ?
In this reformulation, capitalist confidence is expressed in two basic ways. At any given social conjunction, there is a certain benchmark, a rate of return that capitalists feel confident they can get. We denote it accordingly by rc. Finance theory refers to this rate as 'risk-free' - yet explains neither why it is free of risk nor what determines its level. Neoclassicists must resist the temp- tation to equate this rate to the marginal productivity of capital - not only because the latter is logically impossible and empirically invisible, but also because the absence of 'risk' here is secured by . . . the government! As we shall argue in the next part of the book, what enables capitalist governments to set a 'risk-free' rate in the first place - and what makes capitalists view this rate with approval and confidence - is neither 'productivity' nor statist 'distortions', but the overall structure of power in society.
Of course, with the exception of short-term government instruments, capi- talist income is always uncertain (hence the ever-present hype). The conflict that underlies earnings is multifaceted and can develop in many directions. Sometimes capitalist power is sufficiently secure to make capitalists certain of their strategy and the earnings it will generate; at other times, their power is tenuous and future predictions more hesitant. The degree of confidence that emerges from these considerations is expressed, inversely, by the 'risk coeffi- cient' (? ). When capitalists are fully confident, ? is 1. Otherwise, ? is bigger than 1, and it increases as confidence decreases.
Note that this risk coefficient is not the same as the so-called 'risk premium' of finance theory. First, whereas the risk premium pertains to the asset price on the left-hand side of the capitalization equation, the risk coefficient pertains to earnings on the right-hand side. Since the price of the asset involves more than earnings, the two risk concepts cannot be the same.
Second, whereas the 'risk premium' is the designated return for actual volatility, the risk coefficient denotes the confidence capitalists have in their predictions. Of course, volatility and confidence are related, but their corre- spondence is anything but simple. To start with, volatility per se does not generate uncertainty. It is the pattern of volatility that does. The annual earning cycle of ski resorts may be much more volatile - yet far more certain - than the profits of airlines. Insofar as seasonal weather variations prove easier to predict than the vagaries of world travel, capitalists will judge the larger volatility of the former less risky than the smaller volatility of the latter. The other reason is that the past is only a partial guide to the future. This fact has been pointed out by Knight and Keynes, but it takes on a whole new dimen- sion once we bring power into the picture. In the next part of the book we argue that the very purpose of power-driven capitalist accumulation is to reshape society. Capitalists realize that the very nature of their enterprise is to defy prediction, and they therefore take even the most successful forecasting models with a grain of salt.
Elementary particles 209
? ? 210 Capitalization
Lastly, whereas in financial theory a higher risk begets a higher 'risk premium', higher earnings volatility does not imply higher earnings growth, or vice versa. Over the past half-century, the earnings of General Electric rose 10 times faster than those of General Motors - though the volatility of GE's earning growth was far smaller than GM's (Bichler and Nitzan 2006). Contrary to Lamarckian finance, earnings volatility does not entail a 'premium'. And the reason, again, is at least partly associated with power. Bernoulli's capitalists are risk averse because their goal is hedonic consump- tion: their next yacht is assumed to be slightly less enjoyable than the previous one. But power-driven capitalists are very different. As we shall see, their goal is not more income, but a larger distributive share of income. Redistribution, however, does not obey the laws of diminishing marginal utility, so there is no longer reason to assume that capitalists are risk averse.
Most of the world's leading capitalists, including some of the biggest so- called portfolio investors, are not very diversified. In fact, many are highly focused. The reason they have 'made it' is that, unlike the passive individuals that populate the CAPM, they are active. 'Speculating and playing with power is more exciting than playing roulette', writes Traven in The White Rose: 'At roulette influence cannot be exercised' (1929: 96). Big capitalists do not take the odds as given; they try to change them. They struggle to increase their earnings and surrounding hype, and they similarly try to tame volatility. They are not only risk takers, they are also risk shapers. And as they embrace risk rather than shy away from it, the moral link between earnings volatility and earnings growth breaks down.
For the large capitalists, reducing earning volatility is a major obsession. But the reason is not hedonic payoff, but predictability and control. Knight and Keynes identified the problem: the unbridgeable gap between uncertainty and risk. Organized capitalist power is an attempt to defy the problem, if only in appearance: by shaping society, capitalist power 'translates' undefined uncertainty into seemingly quantitative risk. Capitalism is uncertain partly because the conflictual power logic of accumulation makes it so. But power also means ordering, and from the standpoint of capitalists this ordering is the degree to which they can contain their own uncertainty. Partly objective, partly inter-subjective, this degree is captured inversely by the 'risk coeffi- cient'.
Toward a political economy of risk
Clearly, there is an urgent need for a critical political economy of risk. Yet both orthodox economics and heterodox Marxism are unable to develop such a political economy from within their own frameworks.
Neoclassical theorists are ambivalent about the subject. On the one hand, risk taking is one of the key justifications for riches and wealth. On the other hand, to recognize risk and uncertainty is to accept that foresight is flawed and information inherently incomplete, if not absent. This ambivalence may
Elementary particles 211
explain why in their textbook, Samuelson Inc. devote two pages out 932 to the notion of risk - enough to pay tribute to the issue without undermining every- thing else (Samuelson and Nordhaus 1992: 673-74).
The situation is not much better with Marxist theory and radical institu- tionalism. Although earnings and risk are two sides of the same power process, Marxists have had plenty to say on the former and almost nothing on the latter. Marx's own work emphasizes the 'iron laws' of history and is indifferent to its uncertainties. The concept of risk is neither covered nor indexed in Bottomore's Dictionary of Marxist Thought (1991). There is no entry for risk in The Elgar Companion to Radical Political Economy (Arestis and Sawyer 1994). Risk is indexed twice - in reference to 'decision making' and to 'utility' - both subjects of minor interest to radical political econo- mists. It is similarly absent from The Elgar Companion to Institutional and Evolutionary Economics (Hodgson, Samuels, and Tool 1994). 21
In both the neoclassical and Marxist cases, the neglect is rooted in the materialistic notion of accumulation. Capital in these frameworks is a material entity, denominated in productive/hedonic units, and so risk, by its very 'immaterial' nature, must be external to that entity. Risk can certainly influence accumulation; but it can never be integral to it. It is only when we conceive of capital as power that risk can be made inherent to accumulation.
Summing up
In the three chapters of this part of the book, we have argued that political economists have been barking up the wrong tree. Capital is neither a hedonic entity nor a social amalgam of abstract labour, but a capitalization of expected future earnings. The process of capitalization, having emerged in Italy during the fourteenth century, has since expanded and developed to become the most adhered-to convention of the capitalist nomos. It encom- passes more and more aspects of human life - from the individual, to social relations, to society's broader interactions with the environment.
The process of capitalization differs from the material amassment of 'capital goods' not only conceptually, but also empirically. We have seen that, historically, the growth trajectories of pecuniary capitalization and the 'capital stock' have oscillated in opposite directions. Both mainstream and Marxian theories explain this mismatch either as a mismeasurement, or simply as the distortion of 'real' by 'fictitious' capital. But the fact of the matter is that there is nothing to distort or mismeasure. The two concepts are fundamentally different, so there is no reason why they should be equal in magnitude or move in the same direction to start with.
21 Farjoun and Machover's Laws of Chaos (1983), which we mentioned in Chapter 6, offers a notable stochastic exception to Marxist determinism. But their account focuses on the ex post dispersion of prices and profit rates and does not extend to the ex ante notions of uncertainty and risk.
? 212 Capitalization
'Material capital' is backward looking.
It is a stock of past, ploughed-back earnings corrected for depreciation (assuming the whole thing is measurable to begin with). Capitalization is forward looking. It discounts the earnings of the future. Moreover, whereas 'material capital' is one dimensional, based on earnings only, capitalization is multidimensional. It consists of four elemen- tary particles - actual future earnings, hype, a confident rate of return and a risk coefficient. Finally, material capital is seen either as distorted by power (in the neoclassical case) or supported by power (in the Marxian case) - yet in both cases, power is external to capital itself. By contrast, the elementary particles of capitalization are all about power.
These broad contours of capitalization are the sketch from which one can begin to theorize and research the architecture of capitalist power. Of course, a full account of this architecture cannot be attempted in a single volume. Therefore, in the remainder of the book, we focus on what seems to be the most salient dimension: the power underpinnings and implications of capi- talist earnings. Hopefully, this analysis will provide insight and raise ques- tions to encourage further explorations into the subject.
Appendix to Chapter 11: strategists' estimates of S&P earnings per share
Figure 11. 2 presents earnings-related data for the S&P 500 group of compa- nies over the period between 1988 and 2006. The chart contrasts the actual level of earnings per share (EPS) with the consensus estimates of strategists made one to two years earlier. The figure also plots a hype index, calculated as the ratio of estimated to actual earnings.
The thin line in the upper part of the figure plots the actual EPS. The data pertain to annual earnings, which usually are reported during the first quarter of the following year. For example, the annual EPS data for 1999 would be reported during the first three months of 2000.
The thicker broken line in the upper part of the chart shows the corre- sponding consensus estimates. The data points on this line are monthly read- ings, denoting the consensus forecasts made during the previous year. For instance, the data point for January 1999 is the estimate, made in January 1998, for the 1999 annual earnings; the data point for February 1999 is the estimate, made in February 1998, for the 1999 annual earnings; and so on. The last estimate for each year is made in December of the previous year. This setup gives us up to 11 observations for each year's earnings (some months do not have reported estimates). The first estimate is made 23 months prior to the last month of the projected year (for instance, in January 1999 for the year ending December 2000), and the last estimate is made 12 months before that year's end (in December 1999 for the year ending December 2000). The series appears 'broken' since every December the forecast shifts to next years' earnings.
The thicker line at the bottom of the figure computes a monthly hype index
70 ? ? ? ? 4. 0 3. 8
Elementary particles 213
? ? ? index
ratio
? ? ? ? IBES consensus EPS estimate made in the previous year (left)
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? Actual EPS
(left)
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? Hype Index
(right)
? ? ? ? ? ? www. bnarchives. net
? ? ? ? 60
50
40
30
20
10
0 -10
3. 6 3. 4 3. 2 3. 0 2. 8 2. 6 2. 4 2. 2 2. 0 1. 8 1. 6 1. 4 1. 2 1. 0 0. 8 0. 6 0. 4 0. 2
-20 ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? 0. 0 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010
Figure 11. 2 S&P 500: earnings, earning estimates and hype
Note: EPS denotes earnings per share. The Hype Index is the ratio between the consensus EPS
estimate and the actual EPS. Source: IBES through WRDS.
based on the two earning series. Each observation measures the ratio between the consensus forecast made in the same month during the previous year and the actual EPS for the entire year. For example, the data point for January 1999 would express the ratio between the forecast made in January 1998 for 1999 EPS and the actual EPS in 1999.
The figure depicts several clear patterns:
1 Despite the short forecast period, the forecast errors are very large. According to the hype index, the range of errors is 69 per cent - from a maximum overshooting of 43 per cent (hype reading of 1. 43 in May 1991), to a maximum undershooting of 26 per cent (hype reading of 0. 74 in May 1995). The standard deviation of the errors is 17 per cent - three times larger than the average EPS growth rate of 5. 8 per cent.
2 Over time, the errors tend to 'average out'. For the period 1987-2006, the mean value of the hype index is 1. 02 - only marginally highly than a
214 Capitalization
neutral value of 1. In the long run, strategist expectations oscillate around
actual earnings.
17 Ricciardi (2004) managed to collate a list of no less than 150 unique risk indicators - hardly an indication of unanimity.
? Elementary particles 205
for every level of volatility. And then they make their move. Those who are more risk averse can invest part of their money in M, putting the rest of it into risk-free assets (i. e. lending it to the central bank). And those who are less risk averse can borrow at the risk-free interest rate and invest the extra cash in additional units of M. Life has never been simpler.
CAPM
These guidelines, though, are still limited. Their target is the individual investor who already possesses definite expectations on return and variance and merely awaits instructions on how to diversify. The guidelines are silent, however, on how investors formed these expectations to begin with, and on the market consequences should they all follow the diversification advice. These latter questions were taken up by William Sharpe (1964) and John Lintner (1965) in their capital asset pricing model, or CAPM for short.
On the face of it, the questions seem unanswerable. Since individual inves- tors are assumed to be autonomous, their return/variance expectations are open-ended and can take any value whatsoever. And given that the expecta- tions are unbounded, the consequences of acting on them become unpredict- able. So Sharpe and Linter decided to simplify. What would happen, they asked, if all investors happened to share the same expectations regarding return and variance - and, moreover, if their expectations were the same as the true distribution of outcomes (i. e. if investors knew the stochastic generator of history)?
The scenario is admittedly odd. After all, what can one learn about uncer- tainty by assuming it away? Indeed, what would generate a future variance if all capitalists were the same and if all knew the future variance? Recall, however, that we are dealing here with the articulation of the capitalist nomos. The ultimate task is not to theorize capitalists, but to give them a bureaucratic blueprint. And if you revisit the previous section you will see that, in pursuing this goal, Sharpe and Linter were merely following Keynes' checklist.
And indeed, the answers they gave emerged directly from their assump- tions. If investors all see the world eye to eye, they will all own the same effi- cient portfolio M, and nothing but M. And since they all own only units of M, every owned stock in the market must be part of M. The only portfolio that satisfies these conditions is the market as a whole. So in Sharpe and Lintner's world, investors are fully diversified, each holding onto a propor- tion of the entire market.
Now, recall that diversification reduces volatility because the price move- ments of different assets partly offset each other. Why is the offsetting only partial? The reason is that price volatility is seen as stemming from two distinct sources: one that is unique to the asset itself, and another that is common to the market as a whole. A sufficiently diverse portfolio - which M obviously is - eliminates all unique volatility. Thus, following Sharpe and Lintner, portfolio investors can disregard all price volatility that is out of sync
206 Capitalization
with the market. But no matter how diverse their portfolio, they can never eliminate market volatility, by definition.
In other words, every asset carries in its genes a definite quantum of market volatility, a quantum which in turn is passed on to the portfolio as a whole. 'The risk of a well-diversified portfolio', declare Brealey, Myers and Allen in their Principles of Corporate Finance (2008: 193), 'depends on the market risk of the securities included in the portfolio'. This is not something one argues about: 'Tattoo that statement on your forehead if you can't remember it any other way', they warn. 'It is one of the most important ideas in this book'.
Conveniently, this logic is fully invertible. Since individual assets contri- bute to the market portfolio their own market risk, their contribution can be deduced from the market. Recall that unique risk has been diversified away, so the only thing that makes one asset more or less risky than another is its 'sensitivity' to the overall market. This sensitivity is called beta, and it can easily be measured, or so we are told. The greater the market risk of the asset, the higher beta - and vice versa. 18
Now, if the measured historical beta is equal to the 'true' timeless beta, as the CAPM proclaims, we can calculate the asset's risk premium. By defini- tion, beta expresses the ratio between, on the one hand, the 'excess' return of the asset over and above 'risk-free' assets (asset excess return = r -rrf) and, on the other hand, the 'excess' return of the market over and above 'risk-free' assets (market excess return = rm - rrf):
5. beta = r - rrf rm - rrf
Rearranging Equation (5), the capitalist can obtain the risk premium for the asset:
6. risk premium = r - rrf = beta (rm - rrf)
And, finally, with one more reshuffle, the overall discount rate (r):
7. r = rrf + beta (rm - rrf)
The first component on the right is the 'risk-free' benchmark (rrf), while the second component is the compensation for the asset's risky 'deviations' (beta (rm - rrf)).
18 Beta can be derived in two easy steps. First, assume that all investors consider historical volatility equal to 'true' volatility. Second, run a linear regression with the historical rate of change of the asset price as the dependent variable, and with a constant and the historical rate of change of the market price as the independent variables. Beta is the estimated slope coefficient of this regression.
? ? Elementary particles 207
And that is pretty much it. All that the capitalist now has to do is take the 'risk-free' rate of interest (rrf), the market return (rm) and the estimated beta and plug them into Equation (7). And since all capitalists are assumed to be drones and therefore to do the very same maths with the very same numbers, they all end up with the same discount rate (r). 19 This number is then put into the denominator of the discount formula in Equations (3) and (4) to give the 'true' capitalized value of the asset.
Circularity
CAPM has been a smashing business success. Within a few decades, it has become the basis on which corporate finance courses are structured. Often, it is the only model that MBA students rehearse in some detail. And as the students turn into managers, executives and government officials, they apply what they learnt. In the early 1980s, less than one third of large US firms used the CAPM to compute the cost of equity. By the early 2000s, the proportion was well over 70 per cent, although, to be on the safe side, other formulae are used as well (Gitman and Mercurio 1982; Graham and Harvey 2001).
This success is all the more remarkable given the model's dismal empirical showing. Recall that Sharpe and Lintner assumed that investors know the 'true' variance of returns, and therefore also the 'true' beta - yet that, in prac- tice, they all take a shortcut and use the historical beta instead. No wonder the model crumbles.
First, there is the 'equity premium puzzle': it turns out that, taken as an asset class, equities outperform government bonds by much more than their extra volatility would demand (Mehra and Prescott 1985). And then the puzzle becomes embarrassing. Comparisons of different classes of equities often show returns to be uncorrelated and sometimes negatively correlated with beta values! 'The problems are serious enough', conclude efficient market theorists Fama and French, 'to invalidate most applications of the CAPM' (2004: 43).
Given these difficulties, the CAPM has been refurbished, extended and modified with new and improved techniques and a never-ending flow of fresh data. 20 But the new models share with the old one key feature: circularity. Excess return is the compensation for risk, while risk is measured by excess return. This correspondence holds simply because it should:
[F]ew people quarrel with the idea that investors require some extra return for taking on risk. That is why common stocks have given on
19 Equation (7) produces two special cases. One is for 'risk-free' assets, which, being uncorre- lated with the market, have a beta of 0 and therefore a discount rate of rrf. The other is for the market index itself, whose beta is 1 and whose yield is therefore rm.
20 Two famous extensions/alternatives are Ross' Arbitrage Price Theory (1976) and the augmented CAPM of Fama and French (1995).
? 208 Capitalization
average a higher return than US Treasury bills. Who would want to invest in risky common stocks if they offered only the same expected returns as bills?
(Brealey, Myers, and Allen 2008: 217, first emphasis added)
It is just like Lamarckian evolution. A giraffe grows its neck to reach the high leaves on the tree, and so does the market: it makes prices of volatile assets rise faster in order to give investors a reason to own them.
Risk and power
The issue here is not circularity as such, but the worldview that underlies it. The framework of mainstream financial theories is neoclassical. Its basic units are risk-averse, utility maximizing investors. These individuals are powerless. They are too small to affect the circumstances and hence take them as given. Their only possible course of action is reaction: buying stocks whose return/risk attributes are attractive and selling those that are unattractive (that is, until the market equilibrates their prices to their natural levels). Their focus is price, and only price. The price tells them everything they need to know about return and risk. Whatever lies beneath it is irrelevant. And so the discount formula disappears.
The CAPM reasons the link between return and risk in moral terms: the capitalist 'deserves' higher returns to compensate for higher risk. But if we abandon the fairy tale of perfect competition and efficient markets and return to the real world of organized capitalist power, the capitalization formula comes back into focus and the relationship between risk and return assumes a rather different meaning.
Back on earth, the sequence is as follows: earnings are a matter of power and conflict; the conflict over earnings invites resistance; and resistance breeds volatility and uncertainty. In this way, the capitalist struggle to increase earnings is inextricably bound up with uncertainty regarding their eventual level. The numerator of the capitalization formula becomes inti- mately tied to its denominator.
The degree of confidence
Recall that in capitalism the ownership of an asset is a claim on future earn- ings. The price of the asset, expressed as present value, is merely the capitalist assessment of those earnings. Underlying this assessment are two key con- siderations. One is the level of earnings capitalists expect to receive; the other is the degree of confidence capitalists have in their own predictions. In order to make this degree of confidence explicit, we rewrite Equations (3) and (4), such that:
8. Kt = EE = E * H r rc * ?
? ? 9. Pt = EEPS = EPS * H r rc * ?
In this reformulation, capitalist confidence is expressed in two basic ways. At any given social conjunction, there is a certain benchmark, a rate of return that capitalists feel confident they can get. We denote it accordingly by rc. Finance theory refers to this rate as 'risk-free' - yet explains neither why it is free of risk nor what determines its level. Neoclassicists must resist the temp- tation to equate this rate to the marginal productivity of capital - not only because the latter is logically impossible and empirically invisible, but also because the absence of 'risk' here is secured by . . . the government! As we shall argue in the next part of the book, what enables capitalist governments to set a 'risk-free' rate in the first place - and what makes capitalists view this rate with approval and confidence - is neither 'productivity' nor statist 'distortions', but the overall structure of power in society.
Of course, with the exception of short-term government instruments, capi- talist income is always uncertain (hence the ever-present hype). The conflict that underlies earnings is multifaceted and can develop in many directions. Sometimes capitalist power is sufficiently secure to make capitalists certain of their strategy and the earnings it will generate; at other times, their power is tenuous and future predictions more hesitant. The degree of confidence that emerges from these considerations is expressed, inversely, by the 'risk coeffi- cient' (? ). When capitalists are fully confident, ? is 1. Otherwise, ? is bigger than 1, and it increases as confidence decreases.
Note that this risk coefficient is not the same as the so-called 'risk premium' of finance theory. First, whereas the risk premium pertains to the asset price on the left-hand side of the capitalization equation, the risk coefficient pertains to earnings on the right-hand side. Since the price of the asset involves more than earnings, the two risk concepts cannot be the same.
Second, whereas the 'risk premium' is the designated return for actual volatility, the risk coefficient denotes the confidence capitalists have in their predictions. Of course, volatility and confidence are related, but their corre- spondence is anything but simple. To start with, volatility per se does not generate uncertainty. It is the pattern of volatility that does. The annual earning cycle of ski resorts may be much more volatile - yet far more certain - than the profits of airlines. Insofar as seasonal weather variations prove easier to predict than the vagaries of world travel, capitalists will judge the larger volatility of the former less risky than the smaller volatility of the latter. The other reason is that the past is only a partial guide to the future. This fact has been pointed out by Knight and Keynes, but it takes on a whole new dimen- sion once we bring power into the picture. In the next part of the book we argue that the very purpose of power-driven capitalist accumulation is to reshape society. Capitalists realize that the very nature of their enterprise is to defy prediction, and they therefore take even the most successful forecasting models with a grain of salt.
Elementary particles 209
? ? 210 Capitalization
Lastly, whereas in financial theory a higher risk begets a higher 'risk premium', higher earnings volatility does not imply higher earnings growth, or vice versa. Over the past half-century, the earnings of General Electric rose 10 times faster than those of General Motors - though the volatility of GE's earning growth was far smaller than GM's (Bichler and Nitzan 2006). Contrary to Lamarckian finance, earnings volatility does not entail a 'premium'. And the reason, again, is at least partly associated with power. Bernoulli's capitalists are risk averse because their goal is hedonic consump- tion: their next yacht is assumed to be slightly less enjoyable than the previous one. But power-driven capitalists are very different. As we shall see, their goal is not more income, but a larger distributive share of income. Redistribution, however, does not obey the laws of diminishing marginal utility, so there is no longer reason to assume that capitalists are risk averse.
Most of the world's leading capitalists, including some of the biggest so- called portfolio investors, are not very diversified. In fact, many are highly focused. The reason they have 'made it' is that, unlike the passive individuals that populate the CAPM, they are active. 'Speculating and playing with power is more exciting than playing roulette', writes Traven in The White Rose: 'At roulette influence cannot be exercised' (1929: 96). Big capitalists do not take the odds as given; they try to change them. They struggle to increase their earnings and surrounding hype, and they similarly try to tame volatility. They are not only risk takers, they are also risk shapers. And as they embrace risk rather than shy away from it, the moral link between earnings volatility and earnings growth breaks down.
For the large capitalists, reducing earning volatility is a major obsession. But the reason is not hedonic payoff, but predictability and control. Knight and Keynes identified the problem: the unbridgeable gap between uncertainty and risk. Organized capitalist power is an attempt to defy the problem, if only in appearance: by shaping society, capitalist power 'translates' undefined uncertainty into seemingly quantitative risk. Capitalism is uncertain partly because the conflictual power logic of accumulation makes it so. But power also means ordering, and from the standpoint of capitalists this ordering is the degree to which they can contain their own uncertainty. Partly objective, partly inter-subjective, this degree is captured inversely by the 'risk coeffi- cient'.
Toward a political economy of risk
Clearly, there is an urgent need for a critical political economy of risk. Yet both orthodox economics and heterodox Marxism are unable to develop such a political economy from within their own frameworks.
Neoclassical theorists are ambivalent about the subject. On the one hand, risk taking is one of the key justifications for riches and wealth. On the other hand, to recognize risk and uncertainty is to accept that foresight is flawed and information inherently incomplete, if not absent. This ambivalence may
Elementary particles 211
explain why in their textbook, Samuelson Inc. devote two pages out 932 to the notion of risk - enough to pay tribute to the issue without undermining every- thing else (Samuelson and Nordhaus 1992: 673-74).
The situation is not much better with Marxist theory and radical institu- tionalism. Although earnings and risk are two sides of the same power process, Marxists have had plenty to say on the former and almost nothing on the latter. Marx's own work emphasizes the 'iron laws' of history and is indifferent to its uncertainties. The concept of risk is neither covered nor indexed in Bottomore's Dictionary of Marxist Thought (1991). There is no entry for risk in The Elgar Companion to Radical Political Economy (Arestis and Sawyer 1994). Risk is indexed twice - in reference to 'decision making' and to 'utility' - both subjects of minor interest to radical political econo- mists. It is similarly absent from The Elgar Companion to Institutional and Evolutionary Economics (Hodgson, Samuels, and Tool 1994). 21
In both the neoclassical and Marxist cases, the neglect is rooted in the materialistic notion of accumulation. Capital in these frameworks is a material entity, denominated in productive/hedonic units, and so risk, by its very 'immaterial' nature, must be external to that entity. Risk can certainly influence accumulation; but it can never be integral to it. It is only when we conceive of capital as power that risk can be made inherent to accumulation.
Summing up
In the three chapters of this part of the book, we have argued that political economists have been barking up the wrong tree. Capital is neither a hedonic entity nor a social amalgam of abstract labour, but a capitalization of expected future earnings. The process of capitalization, having emerged in Italy during the fourteenth century, has since expanded and developed to become the most adhered-to convention of the capitalist nomos. It encom- passes more and more aspects of human life - from the individual, to social relations, to society's broader interactions with the environment.
The process of capitalization differs from the material amassment of 'capital goods' not only conceptually, but also empirically. We have seen that, historically, the growth trajectories of pecuniary capitalization and the 'capital stock' have oscillated in opposite directions. Both mainstream and Marxian theories explain this mismatch either as a mismeasurement, or simply as the distortion of 'real' by 'fictitious' capital. But the fact of the matter is that there is nothing to distort or mismeasure. The two concepts are fundamentally different, so there is no reason why they should be equal in magnitude or move in the same direction to start with.
21 Farjoun and Machover's Laws of Chaos (1983), which we mentioned in Chapter 6, offers a notable stochastic exception to Marxist determinism. But their account focuses on the ex post dispersion of prices and profit rates and does not extend to the ex ante notions of uncertainty and risk.
? 212 Capitalization
'Material capital' is backward looking.
It is a stock of past, ploughed-back earnings corrected for depreciation (assuming the whole thing is measurable to begin with). Capitalization is forward looking. It discounts the earnings of the future. Moreover, whereas 'material capital' is one dimensional, based on earnings only, capitalization is multidimensional. It consists of four elemen- tary particles - actual future earnings, hype, a confident rate of return and a risk coefficient. Finally, material capital is seen either as distorted by power (in the neoclassical case) or supported by power (in the Marxian case) - yet in both cases, power is external to capital itself. By contrast, the elementary particles of capitalization are all about power.
These broad contours of capitalization are the sketch from which one can begin to theorize and research the architecture of capitalist power. Of course, a full account of this architecture cannot be attempted in a single volume. Therefore, in the remainder of the book, we focus on what seems to be the most salient dimension: the power underpinnings and implications of capi- talist earnings. Hopefully, this analysis will provide insight and raise ques- tions to encourage further explorations into the subject.
Appendix to Chapter 11: strategists' estimates of S&P earnings per share
Figure 11. 2 presents earnings-related data for the S&P 500 group of compa- nies over the period between 1988 and 2006. The chart contrasts the actual level of earnings per share (EPS) with the consensus estimates of strategists made one to two years earlier. The figure also plots a hype index, calculated as the ratio of estimated to actual earnings.
The thin line in the upper part of the figure plots the actual EPS. The data pertain to annual earnings, which usually are reported during the first quarter of the following year. For example, the annual EPS data for 1999 would be reported during the first three months of 2000.
The thicker broken line in the upper part of the chart shows the corre- sponding consensus estimates. The data points on this line are monthly read- ings, denoting the consensus forecasts made during the previous year. For instance, the data point for January 1999 is the estimate, made in January 1998, for the 1999 annual earnings; the data point for February 1999 is the estimate, made in February 1998, for the 1999 annual earnings; and so on. The last estimate for each year is made in December of the previous year. This setup gives us up to 11 observations for each year's earnings (some months do not have reported estimates). The first estimate is made 23 months prior to the last month of the projected year (for instance, in January 1999 for the year ending December 2000), and the last estimate is made 12 months before that year's end (in December 1999 for the year ending December 2000). The series appears 'broken' since every December the forecast shifts to next years' earnings.
The thicker line at the bottom of the figure computes a monthly hype index
70 ? ? ? ? 4. 0 3. 8
Elementary particles 213
? ? ? index
ratio
? ? ? ? IBES consensus EPS estimate made in the previous year (left)
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? Actual EPS
(left)
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? Hype Index
(right)
? ? ? ? ? ? www. bnarchives. net
? ? ? ? 60
50
40
30
20
10
0 -10
3. 6 3. 4 3. 2 3. 0 2. 8 2. 6 2. 4 2. 2 2. 0 1. 8 1. 6 1. 4 1. 2 1. 0 0. 8 0. 6 0. 4 0. 2
-20 ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? 0. 0 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010
Figure 11. 2 S&P 500: earnings, earning estimates and hype
Note: EPS denotes earnings per share. The Hype Index is the ratio between the consensus EPS
estimate and the actual EPS. Source: IBES through WRDS.
based on the two earning series. Each observation measures the ratio between the consensus forecast made in the same month during the previous year and the actual EPS for the entire year. For example, the data point for January 1999 would express the ratio between the forecast made in January 1998 for 1999 EPS and the actual EPS in 1999.
The figure depicts several clear patterns:
1 Despite the short forecast period, the forecast errors are very large. According to the hype index, the range of errors is 69 per cent - from a maximum overshooting of 43 per cent (hype reading of 1. 43 in May 1991), to a maximum undershooting of 26 per cent (hype reading of 0. 74 in May 1995). The standard deviation of the errors is 17 per cent - three times larger than the average EPS growth rate of 5. 8 per cent.
2 Over time, the errors tend to 'average out'. For the period 1987-2006, the mean value of the hype index is 1. 02 - only marginally highly than a
214 Capitalization
neutral value of 1. In the long run, strategist expectations oscillate around
actual earnings.
