Christoph Stein: You conclude from your research that supply-side labour market reforms have impaired innovation and labour productivity growth in major OECD countries. Why should labour market institutions influence innovation and productivity?
There are a number of arguments from neighbouring disciplines, such as organizational psychology, that labour market institutions matter when it comes to processes of technological learning and knowledge accumulation. Unfortunately, neoclassical economists have a great talent for ignoring such interdisciplinary insights. And Keynesians don’t perform much better in this respect.
In lots of industries, a business’s technological competencies are not only determined by its contemporary R&D, but also by knowledge and experience from its past projects. For example, the power of ‘Made in Germany’ rests very much upon the capacity to further develop and perfectionize products, processes or systems. Much of the knowledge required for incremental improvements consists of historically accumulated knowledge from people’s experience; part of that knowledge is ‘tacit’. Tacit knowledge tends to be poorly documented or codified. Tacit knowledge can be transferred from person to person in a working process, but it is hard to transfer over geographical distance. In a hire & fire labour market like in the US, it is hard to accumulate such knowledge and to prevent it leaking to competitors. This is one reason why American producers have lost out to competition from German and Japanese suppliers in a whole series of classic industries. The flexible US labour markets that are so much praised by neoclassical economists and/or the much-criticized sticky ones in Germany delineate the difference between Wolfsburg and Detroit; they are one cause of the US Rust Belt.
Which institutions within the firm enhance or impair innovation?
Innovation depends on the trust and loyalty of employees. Monopoly profits from innovation cannot be protected by patents alone. Good protection against imitators also requires secrecy that depends on the loyalty of personnel. Flexible working and high labour turnover rates destroy loyalty. It is therefore not surprising that firms in Anglo-Saxon ‘liberalized’ labour markets maintain substantially larger management bureaucracies for monitoring and control. Another problem is that deregulation on supply-side lines changes the power relations between capital and labour. Easy firing creates a culture of fear that enhances autocratic management practices, as people no longer dare to contradict their bosses. Moreover, people that are easy to fire have motives for hiding information about how their work could be done more efficiently. In the end, management makes poor use of knowledge from the shop floor, while our innovation management textbooks emphasize that, for successful innovation, you need to mobilize knowledge from all layers of the enterprise. Good innovation managers not only tolerate being criticized; they organize sceptical feedback.
What is the role of decentralized wage bargaining in your story?
The demand for decentralization of wage bargaining is a key topic in the reform program of the Troika in Southern Europe and also in President Macron’s recent reform package. From a neoclassical viewpoint, it can easily be defended. Unfortunately, what is ‘good’ for the efficient allocation of scarce resources can be ‘bad’ for innovation that makes resources less scarce. If wage bargaining is done at plant/company level, the innovative market leaders that earn high monopoly rents from innovation will be confronted with higher wage claims. The technological laggards can enforce low wage claims by simply blackmailing their personnel: either you moderate your wage claims and keep your jobs or we go bankrupt. In such a scenario, the Schumpeterian process of ‘creative destruction’ in which better businesses see off the weaker ones no longer works. Too many less talented entrepreneurs will survive. Decentralized wage bargaining at company level is comparable to government imposing a tax on innovation, using fiscal revenues for subsidizing the technological laggards for the sake of rescuing jobs. This implies that part of the labour market reforms of President Macron come down to quite bad industrial policies!
Today’s bargaining at industry level (with government imposing the bargaining outcomes on everybody in the industry) are a terrible labour market rigidity from a neoclassical view. But this labour market rigidity enhances the diffusion of advanced process technologies. It propels technological laggards towards modernizing their equipment and product offerings – or quitting the market.
The economics mainstream still propagates structural reforms of labour markets as a remedy against numerous economic problems …
Such reforms make labour cheaper and this means that investments in modern process technologies become less rewarding. In our panel analysis of 20 countries over 44 years, we estimated: A one-percent lower wage increase will reduce, in the medium term, the growth of value added per labour hour by 0.3 – 0.5 percent. Wherever you had supply-side labour market reforms in the past, you see the same pattern: Growth rates of value-added per labour hour decline. In exchange, labour input rises. Obviously, if you buy fewer robots, you need more hands. Unfortunately, with low growth in value added per working hour, there is, in the end, less National Income to be distributed between capital, labour and the government. A good example is the Netherlands. The Netherlands started in the early 1980s with a policy of very moderate wage claims (‘loonmatiging’). Since 1985, the country’s growth rates of value added per working hour are substantially below the EU average. The political consequences are obvious: there is hardly any room for real wage increases, notably for low qualified workers, and there is permanent pressure for cutting back on public spending and the welfare state. Meanwhile, this turns out to be a fruitful breeding ground for populism, while trade union membership erodes. It also creates hard times for political adherents of the welfare state: they have little to offer their electorate. When public spending is being cut, they are often happy if they can settle for the smaller evil rather than the bigger one.
You have illustrated in your studies that labour productivity growth in Anglo-Saxon countries that deregulated their labour markets is, viewed over decades, substantially lower than in Continental Europe. This is somehow surprising, given the successful firms in Silicon Valley. Aren’t you ignoring an upside of the US model?
The US has a divided economy. There are the successful giants in Silicon Valley; but there is also a broad range of classic industries like automobiles, steel or textiles that have lost out to international competition from German and Japanese suppliers in the field of highly complex products; and they lost out to the Chinese in simpler products. So, Trump is right: the Germans are among those responsible for the US Rust Belt! Only, Trump’s advisors do not tell him that this defeat has to do with highly flexible labour markets in Detroit and fairly rigid labour markets in Wolfsburg.
But Silicon Valley has been a success story. The US Pentagon realized, at a very early stage, that IT is highly relevant for the US Army. In fact, the Pentagon made what we call in Europe ‘industrial policies’: generous sponsoring of IT research and generous procurement contracts for youthful IT firms. Only, in the US, you cannot call that ‘industrial policy’. But if it runs under the heading of ‘National Defence’, it’s fine.
By the way, the recent book by Robert Gordon (‘The Rise and Fall of American Growth’) hints at a problem: The US IT industry is experiencing diminishing returns. Between 1995 and 2005, IT contributed to a substantial increase in labour productivity and total factor productivity in the US. That’s over. The pioneering phase of Silicon Valley is also over. The successful giants of the IT industry increasingly follow an innovation model in which historical accumulation of (‘tacit’) knowledge is becoming more important. If then, in a hire and fire culture, about ten percent of the engineers in Silicon Valley change their job in a year, you get a problem with Pigouvian externalities that make investments in new knowledge less attractive.
In public policy discourse, ‘Innovation’ ranks high. This is in contrast to how little economics has anything to say about innovation. How can one explain that?
Unfortunately, ignorance about innovation holds true for neoclassical and Keynesian economists. Neoclassical theory has at least the theory of endogenous growth. But you are right: even 60 years after the classic article by Robert Solow, knowledge about innovation is still quite thin in the economics mainstream. Anyway, during the last 150 years, it was extremely comfortable for economic modelling to treat technology as exogenous. In fact, Karl Marx was the only modern supply-side economist who took technology somehow seriously. Joseph Schumpeter got his best ideas from him.
Your research is in the Schumpeterian tradition. This blog publishes lots of Keynesian stuff. It seems that the dialogue between Schumpeterians and Keynesians is surprisingly thin. But there must be some overlap!
Right. You have, for example, the ‘demand-pull’ theory of innovation, going back to Jacob Schmookler. From empirical research, it seems that the ‘demand-pull’ hypothesis is realistic. Only, Keynesians hardly know it, let alone include it in their models. Closer to the mainstream, and analogous to Schmooklerian ‘demand-pull’ effects, you have the Verdoorn-Kaldor Effect.
But the dialog between Keynesians and Schumpeterians remains poor. A reason could be that innovation research has somehow a neoliberal image. Perhaps this was one reason why, during my stay at the Böckler Foundation at Düsseldorf, I had a hard time developing cooperation with the many Keynesians there. In any case, Keynesians committed a major blunder by abandoning supply-side economics as an uncontested playground to right-wing economists. As I argued: What the latter have made of it is highly problematic from an innovation perspective. And, as a German, do not forget that, after your Hartz labour market reforms, growth rates in German labour productivity went down by more than half!
In how far does your research influence economic policy in the Netherlands?
I probably have some influence through my contributions in the media. I even received a royal award from our Queen Beatrix for my contributions to national policy discussions. I was disappointed, however, that notably the Dutch social democrats somehow ignored my message. The social-democratic minister of finance (Jeroen Dijsselblom) is still sailing under the doctrine of German Ordoliberalism of the ‘Freiburg School’. In fact, he is the office boy of his Berlin colleague Schäuble. His loyalty to German Ordoliberalism was the reason why he was (re-)nominated as the Head of the Euro Group. By the way, the Dutch social democrats have been heavily punished in the last elections. A disappointed electorate gave them just 9 (out of 150) seats in the parliament. They are now smaller than the Green-Left party and smaller than the Socialist Party.
Couldn’t it be argued that, with the rise of automation technology, there is such a substantial threat to full employment that it is perhaps a blessing in disguise if labour market reforms delay the diffusion of advanced process technology, thus securing jobs?
People should realize that supply-side reforms have not contributed to achieving higher GDP growth. As far as they have favourable effects on labour input, this comes from lower growth in GDP per working hour, leading to a more labour-intensive growth path. Should we bring down labour productivity growth for the sake of keeping jobs? There are more intelligent solutions! In the past, we solved the problem of technological unemployment by negotiating shorter standard working weeks. If you put the total number of hours worked in European countries in 1960 = 100, most countries are now somewhere between 95 and 110. This comes about simply because, over long periods, GDP growth is hardly higher than growth in GDP per working hour. While the number of total hours worked almost stagnated, labour supply increased substantially, notably because of women entering the labour market. Fortunately, the number of working hours per worker went down. For example, an average German worker still worked 2,181 hours per year in 1960, but only 1,371 hours in 2015. There is nothing wrong about robots replacing workers. But if this happens, trade unions should claim some of the productivity gains not as wages but as shorter working hours with wage compensation.
Christoph Stein is a collaborator of the German Economic Journal MAKROSKOP.