The Technological Externality of Scaling
“Carbon taxes, even were they politically feasible, accomplish some decarbonization via incentivization of green energy scaling, and some via degrowth. The technological externality of scaling — which, by the way, is global, not just localized in the U.S. — means that we want to do more via the former and less via the latter. Thus, industrial policy like the IRA should be our go-to policy for fighting climate change.”
A classical liberal’s instinct is to dismiss this. Scaling isn’t an externality. Its value is captured by those who achieve economies of scale.
But we can assume Smith knows that. He probably means something else.
Google is no help. The phrase appears to be unique. But it hints at a concept commonly deployed to justify subsidies to help immature technologies scale up. Failing his own explanation, let us assume that is what he is referring to.
This is the concept that industrial policy (i.e. picking winners) can help drive immature technologies down the “learning curve” not just through Research & Development, but also simply by encouraging economies of scale. The reduction in their unit cost extends beyond the assisted projects and benefits society.
But economies of scale benefit society whether they are delivered by market forces or government subsidies. The benefit alone does not justify the subsidy. The supposed externality is needed to justify an intervention.
The externality is presumably that the cost reductions are available to other manufacturers besides the first movers. If the first movers could capture the value, they could plan to recoup their early investment from widespread deployment at the reduced costs, in the normal way. But if competitors can free-ride on the cost-reductions achieved by first movers, then first movers are deterred from making the investments to achieve the necessary scale, and society at large misses out. Industrial policy corrects that “market failure”.
What are the limits of this argument? If government has sufficient expertise to correct such market failure, why wouldn’t they do it every time? Why would they be more expert in one area than another?
Smith’s answer appears to be that there are circumstances where intervention is more justified. He cites two: climate change and security. But why only two? Other circumstances may be less existential. But if they provide an economic benefit, why would we forsake that? The more widely governments deploy their expertise to address scaling externalities and deliver economies of scale, the greater the economic benefit and the more prosperous the society.
And yet the opposite relationship is found in reality. The more widely governments try to direct an economy, the worse the economy does. Communism was the climax of the planned-economy fetish, and it shot blanks wherever it was tried.
Why would this be? Surely governments have the best expertise available to them? Many observers of the perverse outcomes of government programmes reach for cynical explanations: stupidity or malintent. That is unnecessary and (in my opinion) mostly untrue. Government advisers are not generally stupid people and they mostly have good intentions. And yet systems that rely on them to direct the economic process will fail.
Innovation is key. An economy treading water is falling behind. It might not seem too bad if future generations enjoyed no better than our technology and standard of living. But imagine if our grandparents had thought the same. Many more people would be hungry or dying prematurely, and most of what we take for granted for our information technology and leisure would not be feasible. Despite the many things wrong with our world, technology has improved it hugely. Cuba is what an economy that treads water looks like.
Central planners (and their academic boosters) love to think they are good at spotting the innovations that will succeed and backing them, but they underestimate at least two related factors.
One factor is luck. People like to think success is the result of genius and effort. They certainly help, and it is in our interests to incentivise them. But they do not guarantee success. Supposed geniuses are often wrong (especially about what non-geniuses want), and effort won’t make them right. To maximise the chance that some good ideas will win through, you want as many competing ideas as possible. Picking winners acts like a filter that aims to remove the duff ideas but in practice removes as high a proportion of the good ideas as the duff ones.
That is because of the second factor: non-conformity. Almost by definition, deep innovation tends to spring from non-conformity. But government winner-picking rewards conformity with bien pensant beliefs. It could not be otherwise. Could you imagine a minister announcing that “all the experts tell us this won’t work, but we’re going to back it anyway”?
These factors make winner-picking an ineffective driver of economic progress even if we assume that those implementing it are intelligent and well-meaning. In my experience, they are no less so than the rest of the population, though not as superior as they imagine.
You can add to this, if you choose, more cynical explanations such as libertarian characterisations of public choice theory. I would advise not to, for reasons I will explain in another piece.
If winner-picking is harmful to economic progress, then Smith’s calculation is reversed. The more existential the threat (such as climate change or national security), the more important is the diversity of the market and the more harmful is the conformity of industrial policy.
Without the protection of government support, what happens to the ideas that are never developed because investors could not see a way to protect their innovations from later competition sufficiently to justify the risk of early-stage investment? There are several answers:
- This is what the imperfect intellectual property (IP) protection systems (patent and copyright) are for. The imperfections mean this is not a strong argument.
- Perhaps more importantly, even without IP protection, first movers gain know-how, which they can hope to leverage against new entrants.
- In some circumstances, network effects are a powerful prize that first movers are aiming to grab.
- This fear misreads the psychology of entrepreneurs. They will happily take any protection that is offered, but many will not be deterred by its absence. They generally believe that their business nous will continue to find clever ways to stay ahead of the chasing pack.
- They have to persuade (naturally more cautious) investors of that, but the best entrepreneurs should have capital from previous successes that allows some degree of funding R&D from the balance sheet. One success does not guarantee others, but it is not a negative indicator, and probably is a positive indicator to a degree. Winner picking suppresses that signal, because the winners are typically “de-risked” and therefore able to attract reduced-cost funding regardless of their track record.
- They do not need to capture the market, just take a sufficient share for a long enough period to recoup the investment. Some people are satisfied with enough (and the stimulation of the innovation process regardless of the scale of the reward) and do not need a pot of gold to get out of bed. Of course, they will take the gold if it materialises.
- The alternative is not a world without risk. Activist government creates regime uncertainty and encourages conformity (as nicely demonstrated in Roger Koppl’s Big Players and the Economic Theory of Expectations).
- Some ideas will not be developed. So be it. Even fewer ideas are developed under the winner-picking scenario, deliberately. The objective of industrial policy is to focus resources on the pre-judged winners.
- Those that are developed may not be scaled as fast as could have been delivered through winner-picking. That is a negative if we know winner-picking will pick winners more successfully than the trial-and-error of the market. But if the opposite is true (as I contend), this is an avoided negative, because industrial policy scales rapidly (at taxpayer cost, with the concomitant economic drag) the losers as well as the winners it chooses.
The free market isn’t and doesn’t have to be perfect, just better than the alternatives. It is.
Advocates of industrial policy assume their desire to replace the messy market process with a more scientific approach has a bearing on the efficacy of that option. Reality works in the opposite direction. The inefficacy of winner picking makes it undesirable, regardless of what one might wish were the case.