“You can see the computer age everywhere but in the productivity statistics” | Robert Solow[1]
In a recent lecture, Adair Turner takes on Solow’s paradox: why, given that computers give us such massive productivity gains, are the national productivity statistics so disappointing? Turner answers the paradox by explaining what else is going on in the economy. Given the advance of technology over the 30 years since Solow’s observation the implications are now even more important.
Our mental model comes from history
Our mental model of productivity growth is likely to be based on the historic transition from agriculture to manufacturing. Strong productivity growth on the farms meant fewer labourers were required to produce the same amount of food, so the displaced workers headed to the city and found work in a factory. The production line was mechanised somewhat, so the factory workers became more productive alongside the famers (who continued to experience strong productivity growth). Consequently the productivity of the overall economy was very strong.
Consider an alternative history, where there was no city and no factory to absorb the displaced farm labourers. Instead the farmers hire the former labourers as domestic servants. As the role of domestic servant cannot be automated, the productivity gains will be very small – it will take about the same time to prepare a meal or iron a shirt in 5 years’ time as it does now. In this case, the strong productivity growth within agriculture will be diluted by the slow growth in the other sector, and so the economy will show muted productivity growth. Worse than that, the passage of time means that the share of labour in high productivity farming will fall (fewer labourers are required each successive year) while that in the static productivity sector rises, meaning that the economy’s overall productivity will progressively decline.
This has implications for our current economy. It seems clear that the digital revolution and rise of the robot will bring very strong productivity growth to certain sectors. But the productivity of the overall economy will depend on what the displaced workers end up doing. The worry is that the new jobs will be in services or the gig economy which are hard to automate.
The growth of zero-sum activities
The second strand of Turner’s argument is that there will be an increase in zero-sum activities. These are activities ‘in which different people compete against one another for a share of the economic cake, but where all of their activity adds not at all to the sum total of goods and services’. To be fair, he doesn’t label this as unambiguously bad, partly because he sees himself within the zero-sum category. The point, however, is that these activities do not raise aggregate human wellbeing.
To illustrate, assume that the displaced farm workers above do not have the option of becoming domestic servants. Their choice now is either to become a criminal, or a police officer. The same amount of food is being produced, so there is no increase in human welfare, but the farmers now need to pay the police to protect them against the criminals. Yes, the police are valuable – but only because the criminals destroy value. We are now in the territory of distributing value rather than creating value.
Applying this to the modern economy, Turner supplies a long but not exhaustive list of zero-sum activities including cyber criminals / defenders, tax accountants and lawyers, marketing / advertising / communications consultants, lobbyists, “much of financial trading and some of asset management”. Some of the zero-sum activities may be admirable (eg campaigning for cause A) but all of them distribute rather than create value. In this context, it is striking how large a proportion of the zero-sum activities are well-paid, which means that a large component of an economy’s high quality talent is competing over the distribution of total income, rather than creating new income. And, as a bonus, the zero-sum activities look less vulnerable to automation – making them an even more sensible career choice.
The value of nil- or low-cost products is mismeasured in GDP
Turner discusses a third effect of technological progress, namely that accurately incorporating the true benefits of free technology services in GDP is “impossibly difficult”[2]. Combining the three effects can fully explain Solow’s paradox – it is perfectly possible to have strong productivity gains in parts of the economy, while the overall economy has low, and possibly falling, productivity growth rates in its official figures.
Why is this important?
There is no better way to express the importance of productivity growth than to quote Paul Krugman[3]:
“Productivity isn't everything, but, in the long run, it is almost everything. A country's ability to improve its standard of living over time depends almost entirely on its ability to raise its output per worker.”
At its best, the investment industry can fund economic growth and development that raises living standards (wealth and wellbeing) for everyone. Given the logic above this would involve allocating new capital to positive-sum, value-creating businesses, and depriving zero-sum businesses of new capital[4]. It also supports engagement and acting as an owner. However, is this a values-based position, like impact investing, or a finance-based position that can pass the fiduciary test? It ought to be finance-based, as over long periods price should reflect fundamentals – suggesting that the market beta reflects the aggregate value creation of the mix of businesses. But the price of a share in a relatively efficient market should create roughly equal expected returns for all businesses. If true, then a bias to positive-sum businesses should not need to come at a cost to returns (unless we believed they had become systematically overvalued).
For me, the key idea to come from these considerations is whether the investment industry should (or will) aim to increase its societal influence by being an active shaper of an evolving investment opportunity set, or whether it will be a passive taker of the set of securities available for portfolios. This shaping could be at the security level, at the sector level (exercising biases towards value creating sectors and/or against sectors that are at risk of stranding), at the asset class level (in preferences between public and private markets) or in real asset choices (for example in favouring real estate choices that have certain sustainability characteristics).