Jobless recoveries and rising income inequality

On one of the listservs to which I belong, one of the members, a professor with a doctorate in literature, threw out the observation that we were entering a post-capitalist era. This claim elicited some interest from other non-econ, non-analytic members. I was mildly exasperated at this example of cognitive pollution. It is a miniscule incident in itself but aggregated over time, you end up with much nonsense and confusion. Its not the camel's nose under the tent that is the issue but rather the eventuality of the camel itself.

There is always something to be harvested though. In trying to imagine the circumstances of what the professor might even think what she might mean by such a claim and how one might address that argument, it forced me to formulate an answer. In doing so, it led me to the hypothesis that there is a causative linkage between capital vs. labor income, TQM/Six Sigma production methods, global competition and income inequality. Let me parse it out.

On the surface, if the claim is to be taken at face value (post-capitalism implying that the importance of capital is reduced or disappears), the claim is refuted by simply looking at the national income accounts. Margaret Jacobson and Filippo Occhino's Labor's Declining Share of Income and Rising Inequality, is a reasonable synopsis. The labor incomes trends are captured in stark simplicity in one graph. Labor's share of national income has declined 3-8% since the eighties. This trend is evident in other OECD countries as well.

The obverse is of course true, capital income is playing an increasing role in national (and personal income). So much for post-capitalism. Growing capitalism is the better description. This is caveated by the acknowledgement that these are trend numbers and they fluctuate with the economic cycle. It is conceivable that the trend might reverse itself, just as it is conceivable that it might accelerate. My money is on the latter for the reasons outlined below.

What are the critical trends in market performance since the 1980s which might have caused a decrease in labor income as a share of national income? Two are obvious and much discussed, 1) globalization and 2) capital labor substitution arising from technology costs. Globalization is reasonably straightforward as an explanatory variable. As the economies of the world have become more integrated (facilitated by improved logistics and supply chains in addition to the more obvious trade treaties), it has allowed for specialization and comparative advantage. Labor intensive activities have moved to low labor cost locations, reducing the amount of labor income in the home country (while also reducing the cost of living). At the same time, Moore's Law has dramatically reduced the cost of technology, accelerating the rate of capital substitution. Simplistically, at some point the cost of a human employee on the assembly line exceeds the cost of a robot.

But that's not all that was happening in the 1980s. Moore's Law was not only about reduced computing costs. It was about miniaturization and digitization as well. Chip technology and digital signals have been displacing electromechanical devices and analog signals. This massive replacement is seen everywhere from the phones we use, the watches we wear and the cars we drive. What we consume in terms of consumer products and services incorporate more complicated (more parts and linear processes) and complex (more parts interacting in non-linear fashions) technologies than ever. We are moving from lower causal density and complexity to higher causal density and complexity.

Which leads to the second major trend from the eighties onwards and which has not been much commented on. The pervasive implementation of Total Quality Management (TQM) and Six Sigma production philosophies and the attendant enabling computer systems, in particular Enterprise Resource Planning (ERP) such as Oracle and SAP.

Our production process are characterized by far greater degrees of causal density and complexity and they are producing goods and services that in themselves are also much more causally dense and complex. When your car won't start, you can't just check the battery, spark plugs and fuel line anymore and expect to solve the problem. You need a $10,000 diagnostic tool that will locate the fault with 99.999666% accuracy.

At six sigma levels of efficiency, production costs really have two components. There is the traditional straight forward direct and allocated costs of production which are reliable, predictable and comparatively low. On top of that, you have to take into account the cost of quality, the cost of not doing it right the first time. When there was low automation and low production tolerances, the cost of fixing something at the end of a production run was relatively low. In the 1950s it used to take 300 man hours to produce and assemble a car (hypothetical number). Today it takes 35 hours. Say you have the same error rate today as in 1950 (implausible) of 1%. For every 100 cars you produce, one needs end-of-assembly-line rectification that takes 3 hours. In 1950, making that repair causes your labor costs to rise by 1%. In 2013, your labor costs increase by 10%.

The consequence is that our production processes have in many respects outstripped human tolerances. Humans don't perform at 99.999666% levels of reliability. It is not so much that humans are more or less expensive than automated processes (though that is a factor) but rather that they are much less reliable. The cost of low reliability with causally dense and complex products with exceptionally high degrees of precision can be catastrophic. So people get engineered out of the production process. That would explain the jobless recoveries of the 2001 and 2008 recessions.

See U.S. Textile Plants Return, With Floors Largely Empty of People by Stephanie Clifford for an example of some of these issues.

How does this relate to the OECD rise in income inequality? As production processes and the products and services they produce become much more causally dense, complex, and precise, they require fewer and fewer employees owing to capital-labor substitution arising in part because of the lower reliability of humans in a Six Sigma environment. The Knowledge, Experience, Skills, Values and Behaviors (KESVB) necessary to succeed in a Six Sigma environment are scarce, very specific, and not easily created through training. See James J. Heckman for the rising importance of non-cognitive skills versus traditional cognitive skills.

We are headed towards a binary economy. One part of the economy is constituted of those local, non-process (i.e. batch manufacturing), non-predictable jobs. Within this part of the economy there are in turn two parts, one of which is high cognitive skills and high non-cognitive skills dependent (think plumbers, electricians, high end restaurant or retail servers) and the other is high non-cognitive skills but low on cognitive skills (hospital attendants, tellers, low end retail). Virtually gone are those jobs with low cognitive and low non-cognitive requirements.

The other part of the economy is high cognitive and high non-cognitive at the far right of the distribution curve: physicians, design engineers, software programming, high end sales, etc. The return to these individuals is exceptional as is characteristic in a highly differentiated, winner takes all economy.

So in a Six Sigma economy, the returns are disproportionate to the most sought after with the scarcest KESVB profiles (a small portion of the population), they are adequate but stagnant for those transitioning away from low cognitive low non-cognitive roles. The returns are worst or non-existent for those low cognitive and low non-cognitive roles partly because there are few if any such roles in a Six Sigma environment and those few that do exist are subject to intense competition.

The upshot is that in an environment of global competition, permeated by Six Sigma production precision, there are fewer and fewer positions owing to capital labor substitution and those roles that do remain are so demanding of very rare and precise combinations of KESVB that the consequence is jobless recoveries and rising income inequality.