CAKE WITHOUT FLOUR
Some growth economists might regard the considerations we have just reviewed as rather quaintly anachronistic in putting so much emphasis on the material nature of economic production. Well-established patterns of economic growth show that as incomes rise, the proportion of output as measured by such indexes as real GDP consisting of material goods steadily declines. The major sources of growth in incomes (and, given the way we measure GDP, in indexes of output) shift to the tertiary sector, particularly services. The chief input to services is human intelligence, and at least in some accounts, intelligence is an unlimited resource. So why couldn't real GDP, measured to include the use-value of services, continue to grow without limit?
There are some immediate problems with this conception. Strictly speaking the production of almost all services does require material and energy inputs, as the gigantic server farms required for information technology are a concrete reminder. Maintaining the human capital to provide a glittering array of intellectual services requires material and energy inputs, and these very likely increase as the quality of intellectual output rises.
But this vision of endless growth without material or energy inputs requires some re-examination of just what it is that we regard as output and try to measure in indexes like real GDP. Some rapidly growing service industries, such as finance, seem to be able to produce increasing measured output without much input increase, even of human employment, at all [Basu and Foley 2011; Foley 2011]. An examination of the issues raised by the growing significance of service industries, which have no measurable output, raises some deep questions about the conception of economics.
The paradigmatic economic interaction for economic theories rooted in the marginalist revolution, such as neoclassical economics and its various descendants, is a transaction in which one good moves from the possession of an agent who subjectively values it less to the possession of another agent who values it more, in exchange for another good (in many transactions money). As the familiar Edgeworth-Bowley box construction illustrates, this type of transaction puts both agents on a higher (or at least no lower) indifference curve, and thus achieves a Pareto-improvement in the allocation of existing resources. Many financial transactions are of this type, for example, initial public offerings to take companies public, real estate brokerage, insurance contracts, and other more exotic forms of financial arbitrage. It is important to remember, however, that the transfer of existing goods or assets in these transactions is not production. When financial intermediaries appropriate some part of the economic surpluses generated in these transactions as revenue, however, economic statisticians have felt compelled to regard the resulting incomes as part of national income, and to invent an imaginary product, financial services, to put on the product side of the accounts as a counterpart.
It is hard to imagine limits to the magnitude of subjective economic surpluses that could be realized through transactions of this type. If, for example, policies or the historical evolution of the division of labor increase economic insecurity by eroding the institutions of traditional societies, one can easily imagine an unlimited expansion of insurance transactions as a result. But from the point of view of classical political economy, it is the increase in material productivity of labor, not the increase in economic insecurity associated with the expansion of the division of labor, which is the source of improvements in economic welfare. This point of view is deeply embedded in the methods of national income accounting, for example, in the fundamental rule that transactions involving the transfer of existing assets do not constitute production of goods and services, no matter how much economic surplus they may represent.
The classical political economists and Marx addressed these issues through the concepts of "productive" and "unproductive" labor. In the version of this distinction, Marx distilled from his critical review of Adam Smith, productive labor (whether it produces material goods or services, since providing haircuts is hard to distinguish from making hats) returns the costs of production with a profit, while the cost of unproductive labor is paid out of revenues without any recovery or return. This classical-Marxian line of thinking puts the origin of the incomes from the production of "services," such as finance, in a different perspective.
This perspective is perhaps most clearly articulated in Marx's analysis of wage labor and the origins of surplus value. Productive labor is responsible for the whole value added in production, but receives only a fraction of the value added in the form of the wage. The resulting surplus value constitutes a pool of potential revenue for which capitalist producers, landowners, intellectual property owners, financial firms controlling money capital, and the state compete. The implications of this analysis, which, unfortunately, is for the most part systematically excluded from the modern economics curriculum, are far-reaching. No particular capitalist firm, no matter how large in revenue and employment, can have much direct effect in increasing the pool of surplus value. Thus "money-making" in capitalist society is proximately based on taking surplus value away from others. In an economy where resources and intellectual property command enormous rents, there may be a vanishingly small connection between the revenue of any entity and its actual contribution to production of useful output.
Many people today are dazzled by the apparently magical ability of innovators to appropriate enormous revenues on the basis of ideas and their manipulation alone. This phenomenon has understandably spawned theories of a "new" economy, supposedly based on new principles of the creation of value. Classical-Marxist political economy, in contrast, locates incomes to innovation not in new principles of the creation of value, but in new (or newly important, since most of these "business models" have actually been around for a long time) modes of appropriation of surplus value. As Slavoj Žižek vividly points out, increasing returns in the appropriation of rents for intellectual property simultaneously obscure the origin of the resulting enormous incomes in the pool of surplus value appropriated from productive labor and mystify the factors behind the increasing inequality in the distribution of these revenues [Žižek 2012]. The origin of the rent of a particularly exploitable resource like a waterfall or a petroleum deposit is hard enough to understand, but at least the owner of a waterfall cannot allow an unlimited number of cotton mills to exploit the resulting usable energy. By contrast, the owner of the rights to distribute a piece of software that, due to network externalities, becomes a technical standard, can allow an effectively unlimited number of users to install the software and charge each of them a fee.
It would be, however, a peculiar political economy that convinced itself that the increasing returns in the rents to artificially created assets, such as systems software, were a remedy for thermodynamically imposed decreasing returns to resource use in material production.