Hyun Song Shin wrote in 2010,
John Geanakoplos deserves credit for having blazed the trail on the analysis of leverage and collateral and their crucial impact on the workings of the financial system.… In developing these ideas, he has been ahead of the pack. Indeed, until recently, there was no pack.
In their paper titled “Holding idle capacity to deter entry”, Geanakoplos et al. (1985) extended the work of Dixit (1980) in showing that strategic interaction could support equilibria with firms holding idle capacity to deter entry. In other words, firms may want to build moats to protect their cash cows. It is noteworthy that this paper appears during the great post-1982 merger wave in which more than one in four Fortune 500 manufacturers faced a hostile take-over bid, and the stock market finally became ‘a market for corporate control’, as Manne had seen coming two decades earlier. It was also in 1985 that Jensen and Ruback showed that the market for corporate control serves the shareholder interest, the organizing principle of neoliberal market society then being articulated.
The Chandlerian firms did not submit willingly. Indeed, they had to be dragged kicking and screaming to submit to the new disciplinary regime where they were expected to report to Wall Street analysts every quarter. They much preferred the good old days when the bankers had waited on them. Now, a great industrial firm’s strategic vision was to be subjected to the second opinion of a 28-year-old bank analyst from Harvard Business School. Worse still, they watched their own capital stock transformed into collateral for debt to finance the very barbarians knocking at their gates. How did American industrial capitalism go from occupying the commanding heights of the economy to being at the beck and call of Wall Street?
The fortunes of the great manufacturers had begun to turn in the late-1960s, although it was a while before anyone noticed. In Capital Resurgent: Roots of the Neoliberal Revolution, 2004, Duménil and Lévy documented the onset of structural crisis — the secular decline in profit rates that can be said to have been evident as early as 1966. The structural crisis morphed into a crisis of macroeconomic stability as the Keynesian mechanism, that had hitherto worked like clockwork, refused to work in the face of the simultaneous presence of high inflation and high unemployment. The long-running crisis of the Seventies undermined not only the credibility of Keynesian macroeconomic stabilization but also the Treaty of Detroit on which the corporatist-liberal synthesis of midcentury American capitalism had rested.
Global finance had punctured the ‘quasi-public international financial order’ even before Bretton Woods came into force in 1958. Midland Bank accepted the first eurodollar deposit in 1955. The Bank of England, Midland’s regulator, called the executives in for tea. Thinking through the implications for London’s place in global finance, they let it fly. In 1962, Kennedy became the first president to receive regular updates on the eurodollar market. By the end of the decade eurodollar deposits were approaching the scale of the world’s central bank holdings combined. The power of unregulated capital flows was soon rediscovered. The pound was bagged in 1967, the Deutschmark in 1969, and the biggest fish in the tank, the US dollar, in 1971. Center country policymakers were forced to relinquish any control over capital flows as the crisis worsened in the Seventies. (See Burn, The Re-Emergence of Global Finance, 2006).
The crisis came to a head in November 1979. Speculative attacks on the dollar caught a President in reelection mode. After panicked conversations with Wall Street bankers and Treasury officials over the weekend at the Crawford retreat, the President gave in to their solution. Carter appointed the inflation hawk par excellence, Paul Volcker, to chair the Federal Reserve. The Fed had played second fiddle to elected officials in the management of the macroeconomy in the midcentury synthesis. All that was about to change almost overnight as the reins of the world economy were seized by Volcker.
Volcker immediately set about delivering the bitter medicine. He would eventually raise the policy rate to nearly 20 percent per annum. The sky high rates were attended by an ultra-strong dollar as globally-mobile financial capital flooded to dollar assets. The economy plunged into the deepest recession of the second half of the twentieth century. Industrial firms were hit by the triple whammy of a severe cyclical downturn, unprecedented cost of borrowed capital, and a superstrong dollar that made them uncompetitive against Japanese and German competition.
American Business focused its fury on what they saw as Japanese currency manipulation. Caterpillar Inc, the icon of American industrial capitalism, led the charge. They wanted Treasury to force the Japanese to liberalize their financial system, in particular by removing capital controls. A young Paul Krugman, then at the President’s Council of Economic Advisors, immediately pointed out that the super strong dollar was the consequence of the sky-high rates. And that deregulating Japanese capital flows would increase the financial demand for dollar assets and thereby further strengthen the dollar. Treasury saw it coming but, wink-wink, went on the trip anyway with the Business lobby. The Japanese signed on. The effect was exactly as Krugman had predicted. (Krippner, Capitalizing on Crisis, p. 98.)
Even as it devastated American industrial capitalism, the Volcker coup proved to be a veritable bonanza for Wall Street. The flow of tribute to the Street in the form of interest payments turned into flood with the Volcker hikes just as the strength of the dollar sharply increased their net worth. Soon this would be joined by the confiscation of firms’ internal surpluses in the form of dividends and share buybacks.
But redistribution of the surplus was a nonsolution to the loss of underlying dynamism. How was dynamism to be restored in the heart of America’s industrial economy? What was required, it was felt in forward-looking circles and would soon become received wisdom, was a thorough-going reconstruction of the rules of the game. The diagnosis explained the loss of dynamism by the ossification of American industrial capitalism. If the United States was going to emerge from rust-belt pathos, the Chandlerian firms would have to become more competitive. Put simply, dynamism was to be restored by going back to the basics. Chandlerian autonomy had to give way to the logic of discipline. (Roberts, The Logic of Discipline: Global Capitalism and the Architecture of Government, 2011.)
As as late 1980, the Chandlerian firms had called for ‘The Re-industrialization of America’, as a special issue of BussinessWeek, the semi-official mouthpiece of American industry, demanded. The problem is identified as the ‘stunted growth in productivity’ in US manufacturing. The goal was to ‘reverse the decline of the manufacturing sector’ by, astonishing as it seems in retrospect, emulating ‘the most successful industrial nation in the world’ and following an aggressive and interventionist ‘industrial policy’. Was this Japanese counterfactual plausible?
In the event, as their fortunes deteriorated further, American industry was forced into a fighting retreat across all fronts. They finally managed to obtain relief from the super strong dollar with the Plaza accord of 1985. But by then they had already capitulated to the new rules of the game. Yet, as we shall see, the new rules proved more effective in restoring upper class revenues and power than in reversing the loss of underlying dynamism.
The principal problem of contemporary Western political economy is the vanishing of broad-based growth. Growth theory is of no help of course. The dominant term that explains growth in modern growth theory is the unexplained residual called total factor productivity, ie that portion of income growth which cannot be explained by growth of capital or labor inputs. It reflects technical change and gains in knowhow by situated communities of skilled practice. We lack even a toy model of this all important phenomena. Beyond, of course, a general suspicion of a classical liberal bias.
Robert Brenner has argued that overcapacity in global manufacturing is the cause of secular stagnation:
Since 1973, the economies of the advanced capitalist countries have performed ever more poorly. The growth of GDP, investment, productivity, employment, real wages, and real consumption have all experienced a historic deceleration, which has proceeded without interruption, decade by decade, business cycle by business cycle, to the present day.
The source of this loss of dynamism has been the deep fall, and failure to recover, of the economy-wide rate of profit, a process that took place mainly from the late 1960s to the early 1980s and derived largely from the relentless buildup of overcapacity across the global manufacturing sector.
Overcapacity can certainly erode profitability. The logic of Geanakopolas’s deterrence game offers a plausible mechanism for why overcapacity may persist. But what does the data tell us? In an earlier attempt, I tried to test Brenner’s hypothesis by looking at the performance of profit rates in the US economy. I found that the economy-wide rate of profit has been substantially restored. Brenner pointed out in a response that I was miscalculating the economy-wide rate of profit. Instead of nonresidential fixed assets, for the financial sector the denominator ought to be financial capital. That was an important corrective.
Subsequently, I looked at the share of corporate profits in national income. This doesn’t answer the question: what’s the rate of profit for the average firm? Instead what it answers is: what portion of national income is cornered by US corporations? By this measure, corporate profitability has revived fully. Indeed, in the postcrisis era, the share has been fully restored to the level of the Sixties’ boom. How can the economy-wide profitability of firms have revived without a restoration of the rate of profit? Aren’t we missing something?
What we are missing is a model of how the burden of the loss of dynamism was distributed in American society. Even absolute gains by the corporate sector together with absolute decline in everyday living standards are consistent with the same rate of growth. Figure 3 is suspect because it mixes distributional questions with questions of underlying growth. I agree with Brenner that the central puzzle concerns the latter. What is going on beyond the chimera of financial expansions and contractions? Is Brenner right? How might we test the Brenner Hypothesis?
In trying to answer Brenner, I have come ever closer to his position. I spent a great deal of time thinking about how to incorporate financial sector profits into an overall assessment. The obvious thing to try, of course, is market equity. But that ends up measuring valuation cycles rather than underlying dynamism. Even if we index as a proportion of GDP, we end up picking up cycles of polarization and de-polarization. It’s clear that the share of corporate profits in the national pie was restored by the growth of profits in the financial intermediary sector without a restoration in the core of the industrial economy.
Then I had a realization that was an echo of a eureka moment I had when I was working on my dissertation on black hole geometry. I realized then that the global geometry of the black hole was locally measurable. That’s exactly the move we shall make here. We don’t need a representation of economy-wide profits to subject the Brenner Hypothesis to empirical interrogation. We may use a micro-local representation instead.
More recently, I have realized that I did not actually test a crucial working part of Brenner’s thesis in the first pass. Namely, does greater overcapacity actually explain declining profit rates in US manufacturing? A clear prediction of the Brenner Hypothesis is that subsectors of US manufacturing with greater overcapacity should have seen a steeper decline in profitability. Put another way, the decline in profit rates ought to track capacity not just over time (see next figure) but also in the cross-section.
We obtain corporate profit data from Table 6.17, “Corporate Profits Before Tax by Industry” and capital stock from Table 4.1, “Current-Cost Net Stock of Private Nonresidential Fixed Assets by Industry Group and Legal Form of Organization” of National Data published by the Bureau of Economic Analysis (BEA) of the US Department of Commerce. We restrict the sample to manufacturing subsectors in order to test Brenner’s main hypothesis, and to the period 1972-2015 due to data availability. For each subsector and each year in the panel, we define the profit rate to be the ratio of pretax profit and net stock of nonresidential fixed assets. We obtain capacity utilization from the Federal Reserve. We match the two data sets by year and NAICS code. Our sample consists of N=836 observations covering 44 years and 19 manufacturing sectors.
Let’s begin by examining the broad diachronic pattern by looking at the averages of the two variables. We can see that there hasn’t been much of a revival in manufacturing profitability. Fluctuations in the two series seem correlated. We shall get to the bottom of this covariation presently.
First, what does the overall pattern look it? We plot the distribution of profit rates achieved since 1972. We know that manufacturers’ profits were even higher in the 1960s than in the 1970s. Yet, in this picture, the seventies look pretty dandy compared to later. The whole distribution climbs down to a trend-less cyclical pattern after the Volcker coup. An alternate explanation for the climb down pattern we see here is that double-digit inflation somehow inflated profit rates (despite the fact that we are using current cost estimates and not historical cost — the latter inflates profits on firms’ books because fixed assets acquired earlier seem cheaper compared to current period profits). But as we shall see, the Brenner Hypothesis is not at the mercy of the climb-down at all.
In order to test the Brenner Hypothesis, we propose to apply asset pricing theory to the cross-section of underlying profit rates in US manufacturing. The empirical implications of the Brenner Hypothesis are clear. If overcapacity is driving profit rates in US manufacturing then this information should be recoverable from the diachronic and synchronic pattern of manufacturing profits. But, first, what is the null?
We propose a null model of profit rates informed by the decision-making problem of the marginal entrant making entry or exit decisions; specifically, by the parameters of the response function of the marginal entrant whose marginal value of wealth prices the projects. In equilibrium, the profit rates from potential projects ought to be such that the marginal entrant is indifferent between them.
The marginal value of wealth to the marginal investor encodes how much value she attaches to a dollar in different states of the world. In the Capital Asset Pricing Model, the systematic factor is the market rate of return, ie the marginal value of wealth to an investor holding the market portfolio. Here we take US manufacturing sectors as the asset universe with the understanding that the marginal “investor” making entry and exit decisions is a great industrial firm. Of course, only a container of a situated community of skilled practice, with the relevant skill-set, can credibly threaten entry.
Following the Capital Asset Pricing Model, we posit that the marginal entrant holds the “market portfolio” in manufacturing. The systematic factor in this frame of reference is thus the overall rate of profit in US manufacturing, that we shall call the macro factor. The prediction is that sectors that do less badly in a general downturn in manufacturing profits can get away with a lower hurdle rate of return. On the other hand, sectors that get killed when you are already getting killed all around should sport high profit rates.
Profit rates and capacity utilization are far from stationary. Even de-trending and an AR(1) correction for the left-hand side variable failed to rule out autocorrelation in a Durbin-Watson test of the OLS residuals. However, first-differencing works. We choose the first-difference of the profit rate displayed in Figure 1 as the macro factor. It captures the cyclical component of manufacturing profits that is the dominant pattern of the graph displayed in Figure 5. This single factor model does an excellent job in explaining the variation in manufacturing profit rates, as the next figure shows.
Let’s go back now to Brenner’s alternative. We ground Brenner’s hypothesis micro-theoretically in deterrence theory. Industrial firms face relentless competition from global rivals. They should redeploy to sectors that sport excess profits until such profits are bid away. How then can excess profits persist?
Put simply, idle capacity is being used to deter entry and protect moats. In other words, the build-up of idle capacity is the flip-side of industrial concentration in the tradable sector. We shall test whether the pattern of excess profits and idle capacity is the result of the ‘rationalization’ of product markets. This is the process by which monopolistic sectors of industrial economy were oligopolized through mergers and the activism of private equity. This oligopolization required the buildup of massive idle capacity to deter entry and enforce equilibrium. Otherwise it could not survive for very long at all. Manufacturing, situated entirely in the tradable sector, is like the balance of power in Central Europe for centuries — peace could not obtain for a second if force was not opposed by counterforce.
What private equity and the bankers accomplished was to push US manufacturing to a Pareto sub-optimal separating equilibrium with nonzero idle capacity on the equilibrium path. The massive excess capacity we documented earlier are great moats meant to protect cash cows, as Warren Buffett has long practiced in his investment philosophy. This can be seen a counter-movement from capital in the face of the neoliberal intensification of the global condition. Might this not be a good way to link the neoliberal counterrevolution, complete with the barbarians, to Brenner’s diagnosis; all of it grounded in the systemic logic of Geanakoplos’s family of deterrence games? But what can the evidence bear?
Figure 3 displays the linear relationship between changes in profit rates and capacity utilization in the forgetful representation (ie forgetting the spacetime structure of the panel data). The gradient is very significant.
How well do the two models do compared to each other? The next figure displays the fixed effects in an apples-to-apples comparison. We can see that Brenner’s model survives the introduction of our systematic factor. Although the macro factor, which as we have seen is nothing other than the cyclical factor in Figure 5, is more important than shocks to capacity utilization, idle capacity is, as we shall see, low-balled by this measure.
We go deeper and subject the hypothesis to a more stringent test. A strong version of the Brenner Hypothesis would imply that sectors with higher capacity beta would sport higher rates of return. The interpretation would be that our marginal entrant, a great industrial firm, strategically chooses to divest from subsectors with greater excess capacity. This response function of the marginal investor should leave a clear imprint on the cross-section of expected profit rates in the form of a positive linear relationship between factor betas and mean profit rates.
We test this prediction of the Brenner Hypothesis against our data set using standard 2-pass regressions. Specifically, we project profit rates [r(i,t)] for each sector onto our risk factor — shocks to capacity utilization and/or shocks to mean profit rate in manufacturing — in the time-series [t=1,…,T] to obtain factor betas. In the second pass, we project mean profit rates of the subsectors onto the betas in the cross-section [i=1,..,n] to obtain the price of risk (lambda). We exclude the petroleum products sector, the most volatile of our sectors which obfuscates the underlying structural relationships.
Testing Brenner’s model, we find that the price of shocks to capacity is significant at the 1 percent level. Shocks to capacity utilization are thus priced into the cross-section of profit rates in US manufacturing. Asset pricing theory’s price of risk is in our frame the shadow price of costly signaling required to sustain a separating equilibrium. The foundations of oligopoly and overcapacity are the same.
What about our macro factor? Or, what have we found out about the behavioral pattern of American manufacturers beyond that predicted by Brenner and Geanakoplos? We find that fluctuations in the aggregate profitability of US manufacturing are also priced into the cross-section of manufacturing profits. That is, expected profit rates are proportional to the degree of covariation of sectoral profit rates to fluctuations in the profit rate of US manufacturing as a whole. The issue is covariation with the pricing kernel (the marginal value of wealth to the marginal investor) not return volatility per se. This is as true of the marginal entrant in the market for manufactured goods as a marginal investor in asset markets. The pattern is consistent with the null model of profit rates.
Brenner’s secret supremacy is revealed when we run a horse race in the 2-pass regressions. We find that sensitivity to capacity utilization trumps sensitivity to the cyclical factor. Although the macro factor becomes insignificant in the cross-section once we include capacity, that is probably due to the narrowness of our cross-section. What is hard to contest is the supremacy of shocks to capacity utilization in explaining the cross-section of profit rates in US manufacturing. Whereas the gradient of our macro factor explains 32 percent of the variation in expected profit rates, Brenner’s gradient explains 43 percent. To wit, if the behavior we are positing to our marginal manufacturer is true, and that is what the data is telling us, industrial capitalism cares more about avoiding crowded sectors than sectors which do less badly during downturns.
We have documented this behavioral pattern of US manufacturers. This is the explanandum, brute empirical fact that we have established. The question is what is the explanation of this pattern?
We have suggested a theoretically-grounded lens for understanding the evidence we have marshaled. The appearance of Dixit (1980) and Geanakoplos (1985) in the early-1980s is no coincidence. It announces the arrival of a new logic of discipline to American manufacturing. Put crudely, Wall Street arranged the oligopolization of US manufacturing to new equilibria with bigger moats. It’s the merger wave that prompts Geanakoplos to think of the problem and offer a closed-form solution.
Geanakoplos nailed it. The logic of discipline of a capital market-centered system is clear. Greater idle capacity may be the recipe for a healthy stock price. Given the strategic dilemma, the culminating point of capacity expansion is reached at suboptimally high investment in capacity in order to deter entry. For in this manner stable oligopolies could replace monopolistic competition in sector after sector, niche after niche. Wall Street’s schemes of rationalization of sectors automatically promoted greater concentration through the valuation effect. This parlor trick restored Wall Street profits, and upper class investment incomes, without restoring underlying dynamism. And perhaps making it worse.
What we have uncovered in profit rates is a behavioral pattern of American manufacturers. They have been caught up in a much more intense and intensifying competition with each other since the container revolution of the late-1960s and the complete annihilation of distance. The intensifying global condition plunged US manufacturing into a profound crises of profitability from which it failed to emerge despite extended surgery by Wall Street.
We have shown that the Brenner Hypothesis is priced into the cross-section of profit rates in US manufacturing. We have suggested a historical conjuncture to situate the diachronic logic of the Copernican revolution whereby Chandlerian firms reoriented to face capital markets. What followed can be read off the cross-section of profit rates, as we have shown. The logic of discipline of the global condition in the neoliberal era was lived by great industrial firms fending off barbarians at the gates. This involved more than the rise to primacy of Wall Street. And more than the general discursive shift towards neoliberalism. It required a loss of self-confidence by the great industrial firms. A loss that can hardly be restored without the restoration of underlying profitability.
We have related Brenner’s diagnosis to the industrial structure of American manufacturing. The logic of deterrence theory suggests a novel solution to the central problem of Western political economy. For if Brenner is right that the loss of dynamism is due to ‘the relentless buildup of overcapacity across the global manufacturing sector,’ and we have failed to falsify it, then we have in effect isolated a channel from industrial structure to idle capacity and thence to underlying dynamism. In light of Brenner’s compelling diagnosis of the Western predicament, an aggressive pursuit of industrial deconcentration as a policy can then be seen not only as an attack on vertical income and wealth polarization, but also as an effort to restore dynamism through the logic clarified by Geanakoplos.
The data set and technical details about our methods are available upon request.