Americans think the French don’t work hard. Thomas Piketty’s Capital in the XXIst Century has 970 pages in the French edition before me. That’s with merely a table of contents. Following the annoying and backward French habit, there’s no index. The scholarly apparatus – bibliography, data tables, and more – is shifted to a website. You can download the lot in a zip file of a mere 13 MB.
It matters that Piketty has written a doorstopper not an article. The book is very accessible to the general reader with stamina; in fact it’s too leisurely and repetitive for my taste, more appropriate to the series of lectures on which it is no doubt based. But it is massively learned. My heart warms to a writer who takes Jane Austen and Honoré de Balzac as serious and acute witnesses to the economic life of their times. His interest in social history is not that of a dilettante like me. He has done primary research in the Paris city archives on the inheritance patterns of the bourgeoisie of the Belle Epoque. His research style owes a lot, explicitly, to the Annales school of historians like Febvre, Braudel and Bloch, who were not afraid of the challenge of understanding entire complex societies over long sweeps of time.
Freshwater economists will not be able to challenge this historical depth simply by fanning their vapours with elegant DGSE models. Jane Austen is outside their professional frame of reference. Piketty has restored economic history to its rightful place as the test of large-scale theory, as Milton Friedman accepted. Graduate students in economics are I hope heading for the history departments to learn about the methods. With luck, Piketty may shift the paradigm of economic science in a good direction, with more humility and a willingness to use all the evidence.
Beyond simple praise – read the book! – there are three ways to criticise a magnum opus like this. Niggles: e.g, in his long time-series charts, he often shifts the horizontal scale half-way without flagging it. This sort of thing would be useful in quantity, but requires real work. Frontal attack: the theory is wrong because. Piketty is out of my league and anyway it looks pretty convincing to me. Finally, there’s complaining that the author should have written a different book. That’s my line. It has the additional advantage that I don’t need first to finish the one he wrote.
The book he didn’t write is about innovation. There’s no index, but using the informative chapter headings I ran down a very few pages on technology and technical progress, none specifically on innovation. Piketty follows economic tradition since Solow in treating technical progress as a uniform and exogenous force. To the very limited extent he goes further, it’s treated as the spread of knowledge, presumably largely by education.
I suggest that is inadequate, and that innovation specifically is a key part of the mechanisms of capital accumulation and inequality. If that’s so, then we can look for different ways of interrupting the process of rising inequality of income and wealth that is threatening our democracies.
A toy model of innovation might look like this. We have exogenous scientific progress, which is knowledge generated largely in universities. It is spread through the economy via scientific and technical education, and other formal and informal social interactions (apprenticeships, water coolers, conferences, blogs). The knowledge has to be embodied in skilled workers and in published resources of various sorts that they read and refer to. This yin-type disseminated knowledge is the precondition and enabler of innovations, which are yang-type embodiments of new ideas not in the minds of workers but in tradeable products and processes.
What do we know about innovation? There’s no grand theory, but a few stylised facts are plain:
- Innovation always has a local origin. This may be one man in a small workshop, like Gutenberg or Harrison. It may be a small team, like Steve Jobs and the designers of the first iPhone. It’s never spontaneously generated over a wide area.
- Innovations diffuse through the economy gradually, taken up selectively by a minority of the skilled workers (update) and entrepreneurs (endupdate) capable of applying them. The timescale may be months, years, or decades. It’s never instantaneous.
- Innovations form a cascade. One innovation enables and potentiates another. A single fundamental discovery like the transistor or the printing press can lead to decades or centuries of change and millions of new products.
I infer from these facts that it does not make sense to study innovation in an equilibrium. Since growth (in the sense of rising GDP per head) depends absolutely on innovation, any growing advanced economy is in a state of perpetual technical disequilibrium at any timescale of interest. Poor countries can in theory grow largely by capital accumulation, but in practice they also grow by technological emulation, and are similarly always in disequilibrium.
The two processes of innovation, diffusion and cascade, feed into capital accumulation, profits, and inequality. Successful innovators like Apple earn high profits, expand the production frontier, and accumulate capital. At the same time they destroy the capital of backward competitors and producers of goods that consumers substitute away from, increasing capital destruction above the base rate of depreciation. The net rates of profit and accumulation are the sum of the gains of innovators and the losses of the backward. Growth is not a rising tide lifting all boats; it’s a complex Darwinian struggle with winners, losers and bystanders.
Imagine a warm bath with a hot tap and a drain sensitive to pressure. The volume of water is the quantity of capital, the temperature the rate of profit. Hot water coming in raises the level of the bath and its temperature; the drain outflow lowers the volume. The drain doesn’t cool the bath, so the analogy is imperfect. However, the metaphor has the bonus feature that left to itself the bath cools down slowly to ambient temperature, corresponding to a gradual decline in the no-innovation rate of profit towards the steady state, as in the classics.
Meanwhile, what is happening to wages? These are determined by the average productivity of labour applied to the whole capital stock. Profitable employers need not pay more, unprofitable ones cannot pay less. The labour market allows the former to earn superprofits and accumulate capital rapidly. These gains to capital outweigh the losses and bankruptcies of the failing businesses.
This stylised picture or template – it’s not a proper model, but could be turned into one – puts innovation at the centre of capital accumulation and therefore of inequalities of wealth. Innovation combines with inheritance through financial mechanisms that allow inherited wealth to be steered preferentially into high risk but high-yield investments, including those in innovations.
My Schumpeterian churning mechanism is intuitively plausible. How much does it explain? Are superprofits to innovators, and capital destruction for the losers from change, significant or marginal? How much of the national income estimates for depreciation (a not trivial share of GDP: 11%-19% in the 8 richest countries 1970-2010, dixit Piketty, table 5.3) reflect physical wearing-out, and how much premature obsolescence driven by innovation? We can point to innovations that destroyed a lot of prior capital, like the motor car or the supermarket, and others like Portland cement that a priori had little effect – a few makers of lime for mortar lost out, but any Victorian builder could easily switch to cement and speed up his work considerably. (It would be interesting to look at the effect on the viability of monasteries from the loss of the book market to the new printers around 1500, just before the Reformation.) We can point in our own day to innovative companies with high returns on capital like Apple (27% ROCI) and Google (15% ROCI), and many others like solar panel makers that barely make a living. It would take a proper research programme, using Piketty’s storehouse of data to begin with, to answer the question.
As a blogger I’m allowed to jump to conclusions. If my picture has any truth to it, and I feel it has quite a lot, we can think of tackling inequality through socialising the mechanisms of innovation. It’s a weakness – the great and admitted weakness – of Piketty’s book that his proposed remedies, a return to confiscatory progressive taxes on high incomes and inherited wealth, look politically impossible. Keynes proposed an intellectual revolution, but the policies he put forward were positive-sum games for both capitalists and wage-earners, though not to rentiers. My idea is in between. Even a drastic change in incentives for innovation does not hurt existing vested interests very much, and they are what counts in politics. The proposal is challenging but not I think fanciful.
What would an alternative system of innovation look like? Fortunately we have a great deal of historical experience with state-sponsored innovation. From the British Longitude Prize of 1714 to the Internet protocols, governments have quite successfully promoted innovations that fitted their goals. From the start of the arms buildup before WWII (say 1935) to the end of the Cold War in 1991, governments dominated new technology. Radar, jet planes, computers, rockets, nuclear weapons and reactors, high-speed trains and the Internet all arrived through public not private funding and purpose.
It would therefore be technically possible to abolish patents and replace them with other well-tried incentives and mechanisms that eliminate or greatly reduce the private appropriation of benefits – many of them derived anyway from publicly-financed scientific research. We can have government-run or university laboratories, large prizes, or contracts with profitmaking firms; and place the resulting intellectual property in the public domain. (See my proposal for a Qatent Office, a public-domain counterpart to the Patent Office.)
The scheme would largely remove the superprofits of first movers. It would look rather like the ARM ecosystem, in which new processor architectures and designs are licensed to any electronics company that is ready to pay. The licensees compete between themselves in the new products using the processors (cascade innovation). These markets are generally competitive and normal-profit, with the striking exception of Apple, whose monopolistic profits result more from superior ergonomy and style than its in-house SOC designs. At the other end of the chain, the destruction of the profits and capital of those left behind would continue unabated. The net effect would be to lower returns to capital and net accumulation.
Besides the political feasibility of such a system, we need to address the question whether it would be an adequate replacement for the one we have. Would we get new products like the iPad? Would we get its lower-cost imitators (diffusion) and rivals (cascade)? The Soviet Union notoriously had an effective system for developing new military hardware, and a failed one for civilian products. However, that was an extreme case. Civilian products always had low priority, and the absence of hard budget constraints and informative prices removed the incentive on the manufacturers of televisions and washing machines to make ones that worked properly, let alone improve them. Lockheed and Thiokol, dependent on Pentagon and NASA contracts, are not models of efficiency, but they retain an incentive to perform.
The incentive for diffusion is still present. Since we lose the deadweight of patent rents, diffusion could even be faster, as in software today. That leaves the generation and the cascade of new ideas. It would be prudent to work for now on the presumption that there are significant innovations like the iPad that depend on expectations of high rents. However, as in science, the motivation of innovators is fame as well as – and probably more than – money. This aspect would be unchanged, so the loss would be partial. Finally, the finance sector today is a hotbed of perverse negative-sum innovation. Governments will not pay to replace this, which is all to the good. However, these innovations are not protected by intellectual property privileges, and in any case need much more severe regulation – not to encourage but to contain.
We seem to have a choice here. We can socialise the generation of useful innovations entirely, and accept some loss in the pace of change, perhaps considerable, especially in consumer products. Or we can go half-way, weakening rather than abolishing patents, and expanding government finance for innovation rather than expecting it to take over completely. The choice depends on the impact on accumulation and inequality. If we can generalise the situation of the German Mittelstand, with many rich but not super-rich owners, that would solve our political problem. It may not be enough.
To get back to Piketty. I plead with him to write this book, or get a team of his brightest students to do so. The study of the mechanisms of innovation lends itself perfectly to the eclectic methods of the Paris school: indeed it’s hard to see how anything else can work. Innovations are incommensurable; each has a unique life-story, like a living creature. But there is a massive amount of data available, from biographies, case studies, patent offices and tribunals, and for the losers, from bankruptcies. We have for instance a detailed chronology of the spread of printing through Europe from Mainz in the 25 years after 1454. Tracking innovation cascades requires genuine technical understanding – again readily available in the South Bank. The mathematics of spatial and branch diffusion is not that of equilibrium equations but (I surmise) of Markov chains, graphs and cellular automata, which can outflank the DGSE guys with a better class of inaccessibility. My thesis that innovation is coupled to superprofits and accumulation is testable from the data sources he already has. Allez-y!