Brussels’ push for centralisation is based on an outdated view of how societies work
In complex systems, top-down decision making reveals its limitations.
Lesen Sie die deutsche Version hier.
During the opening months of 2021, the Covid crisis has focused attention In Europe on an important political principle. We see the European Commission, motivated as ever by the drive to an «ever closer union», attempting to assert control over not just the purchase and distribution of vaccines to member states but over the pace at which it should be carried out. A uniform speed is the ideal of the central planners in Brussels.
Against this, many individual countries have suddenly become enthusiasts for a more devolved, less centralised system of decision-making. When Hungary broke ranks and began importing vaccines in its own right from Russia, it seemed to fit the stereotype most people have of that country being an outlier within the EU. Now, however, both France and Germany are contemplating the same policy.
These tensions merely reflect existing pressures within Europe. Germany and France back moves by both the Commission and the Central Bank towards a more centralised federal state. Yet within a number of countries, such as Spain and Belgium, there is strong support in some regions for much more local autonomy, or even outright independence. Britain of course took the ultimate step of exiting from the direction of travel to a centralised European state. But within the United Kingdom itself there are separatist pressures, principally in Scotland but also in Northern Ireland and Wales. All three countries, particularly the latter two, depend heavily on subsidies from England to sustain their living standards.
Fast vaccination, poor cancer treatment
Generations of economic policymakers have been raised in the mechanistic, centrally planned view of the world: to achieve a particular set of aims, draw up a list of policies, and simply tick them off. It is a comforting environment in which to live, being seemingly dependable, predictable and controllable.
But the limits to this philosophy were exposed in the major natural experiment which took place in the second half of the 20th century between centrally planned systems and the market oriented economies. Whether it was the United States and the Soviet Union, East and West Germany, North and South Korea, or even China as a planned economy and China with a capitalist face, the decentralised systems were the decisive victors.
Sixty years ago, however, the result seemed to be in doubt. The Soviet system appeared to be superior, capable of making technological advances much more rapidly than the West. In a few short years, it had caught up with America in terms of nuclear weapons. In 1957, it had launched the first human into space.
The Soviets wanted to develop a missile capable of launching a satellite into orbit. A very challenging task, but one with a very clear goal. In this they were successful. But their centralised system was completely unable to cope with the different challenges of a consumer-oriented economy.
In such an economy, instead of decisions being made at the top and then cascading down in an ordered manner, the demands flow in the opposite direction. Millions of individuals make their own decisions, in ways which may be hard to anticipate by the companies which serve any particular market.
The strengths and weaknesses of a centralised system of decision-making can also be seen in the example of the British health system. In 2021, this has far outstripped the performance of other European countries in vaccinating the population.
A key reason for this is that the health system in the UK is highly centralised, far more so than any other Western health service. It was set up in the aftermath of the Second World War, when central planning seemed normal.
The clear aim has been to vaccinate tens of millions of people in the space of just a few months. The target was set, and the command percolated down the system for people to deliver it.
More generally, however, the British centralised health service performs poorly in comparison to those of other European countries. There are many different, and indeed competing, targets to meet. What are the priorities in terms of what should be treated, who should get it, how should it be delivered?
In terms of outcomes such as cancer survival rates, the British record is poor. It is poor, too, at preventative medicine, at preventing health issues from arising. An important reason why death rates from Covid have been so high in the UK is that the level of obesity is the highest of any major European country, a problem which the centralised health service has not been able to solve.
«Racism» on a chessboard
In certain circumstances, centralised systems, whether political or economic, can indeed deliver good results. When there is an unequivocal target, a structured, hierarchical system along which commands are passed can work well. But modern economies and societies more generally are not constructed around this machine-like structure. Instead, they are more like biological systems, dependent on subtle interactions between their component parts. In the scientific jargon, they are complex systems.
In everyday English, the words «complex» and «complicated» are often used as synonyms. But here the word «complex» has a very specific meaning. A machine can be very complicated, but its workings can be understood completely and therefore can be readily controlled.
The rules which describe the behaviour of a complex system can be very simple. But the fact that the component parts interact, that the behaviour of each one can be altered by these interactions, means that they are much less susceptible to precise control.
An example is the model of residential segregation in cities developed by Thomas Schelling, an American polymath and Nobel Laureate in economics, some fifty years ago. In American cities, there are high levels of racial segregation. In general, most neighbourhoods are dominated by one particular ethnic group or other. The obvious conclusion to draw from this clear pattern observed at the aggregate level is that there are substantial levels of racial prejudice amongst individuals. Why else would we observe such high levels of residential segregation?
Schelling showed that this observation could easily arise in a situation in which people had only a very slight preference for living amongst people of their own ethnic group. They did not have to exhibit any strong prejudice at all.
He imagined a very large chessboard in which equal numbers of pieces of two colours are scattered at random. A small number of squares are left vacant. A piece is selected at random to decide whether it wants to move to an empty square, also selected at random. The decision rule is very simple. Each piece is surrounded by eight squares. Schelling defined these, plus the square the piece itself occupied, as the neighbourhood of the piece. The piece is content if there are 5 pieces of its own colour and 4 of the other. So people are happy to live in mixed neighbourhoods. The other «colour» is readily accepted. It will only move if it is in a minority in the neighbourhood. Pieces are selected at random in the model to decide whether or not to move.
Quite rapidly, a surprising result emerges. Despite the tolerance shown at the individual level, strong patterns of segregation are seen at the aggregate level. And each time the game is played, any given area of the board can be dominated by either colour. It is not possible to predict at the outset exactly what will actually happen as the game unfolds.
A key feature of the Schelling model is the interactions of individuals. A piece may be content with its location. But one of its neighbours, which itself has a slightly different neighbourhood, may decide to move. Its place may be filled in a way which then induces the first piece to decide to move.
Complex systems are fundamentally different from mechanical ones. Accurate prediction is a very challenging task. Working out the implications of any given intervention is difficult. In short, they are not easy to control.
An important piece of practical evidence is the very poor record of economic forecasts. For example, the Survey of Professional Forecasters in the US does exactly what its title suggests. It surveys a wide range of forecasts of the American economy, and has done since 1968. In terms of GDP growth, the average forecast for one year ahead over this period has a zero correlation with what actually happened.
It is not that economists are stupid. In a system in which millions of firms and consumers make individual decisions which interact with each other, accurate prediction is a challenging task and, by implication, so too is control.
How can these inherent limits to knowledge in complex systems be best handled in terms of political structures? A strong argument for devolving decision-making within an overall federal structure is that it enables the effects of different ways of approaching a problem to be observed.
I referred above to the very significant «natural» experiment which was carried out between the centrally planned economies and the more free-market oriented ones of the West. It was not set up by deliberate design in a laboratory. It was the consequence of decisions made for completely different reasons than those of scientific enquiry. The more evidence of this kind which can be gathered, the better informed policy making can be. And decentralised systems offer more opportunities for natural experiments to occur.
Of course, what works well in one particular region or locality might not be directly transferable to another. But developing a practical knowledge base of the impact of different approaches to a problem – by observing what actually happens when a particular policy is carried out – is important.
In complex systems, outcomes are very often different to those which are intended, no matter how carefully thought through the planning might be. To emphasise again, in complex systems there are inherent limits to knowledge.
The advantages of decentralisation are made apparent in Hayek’s seminal 1945 paper «The Use of Knowledge in Society». He argued that information is dispersed across society, and a central planner can never hope to know it all. Hayek had this insight decades before the mathematical formalisation of complex systems began. A related concept is that of tacit knowledge, developed by Michael Polanyi in the late 1950s. People have a lot of skills, ideas and knowledge which is not easy to codify.
A practical current example is the Covid vaccination programme which is underway. In the UK, the uptake in general is very high, but there is scepticism in a small number of ethnic groups. Of course, there is a role for a centralised approach to encourage people to get «jabbed». Government advertising, using celebrities who particularly appeal to such groups, is useful.
But the most effective persuasion is being carried out at a very local level. Local elected councillors and religious leaders have a detailed knowledge of the communities they serve. They are well placed to deliver messages which are specific right down to the level of the individual household. All the above arguments point in the direction of decentralised decision-making.
The capacity of centralised bureaucracies to resist any such pressures should never be underestimated. We might usefully recall the fact that in 1919 Lenin became concerned that his revolution was becoming swallowed up by centralised red tape and bureaucracy. He established a commission to investigate the problem. By the end of 1920, this commission itself employed 100’000 people.
The problem of transactional costs
But clearly some degree of centralised decision-making is needed. The UK, for example, is a nuclear power. But it would scarcely be desirable to issue each citizen with his or her own nuclear device, even if, stretching imagination to the limit, it were either technologically possible or affordable.
We might once again turn to economics for some inspiration on how to think about the question. Ronald Coase was awarded the Nobel prize in economics in 1991 in part for a paper he wrote in the 1930s whilst still an undergraduate student. There is in fact a growing academic literature around the potential application of Coase’s insights to political systems, inspired by a 2003 paper by economist Daron Acemoglu.1
Coase asked the question: why do firms exist? In principle, every aspect of the processes of production and distribution could be carried out using bilateral trading contracts between agents. His answer was that firms exist because of the existence of a large number of costs in individual market transactions. The cost of obtaining a product is not just the price. There are, for example, costs in both searching for and processing the information about alternatives. There are often bargaining costs, the time and effort taken in negotiating the price. The creation of a firm, in which individuals collaborate, reduces such costs substantially.
Essentially, we are faced with a trade-off. There are costs, discussed above, associated with centralised systems, in particular, their inability to gather valuable information specific to individual localities and communities limits their effectiveness.
But there are what we can describe as transactional costs associated with devolved structures. Some of the really enthusiastic supporters of cryptocurrencies, for example, imagine a future in which every individual can create his or her own currency. But even equipped with massive computing power and the internet, it is hard to see how such a system could possibly be more efficient than one in which a national central bank holds a monopoly on the creation and maintenance of a shared currency. It is true that a central bank could make a very serious error and, for example, create hyperinflation. But in developed economies there are very few such cases, they are very rare events. In contrast, the transaction costs involved in a system with personal currencies are constantly present.
The ideology of centralisation
In an important sense, Western societies have gone through an evolutionary process which has resulted in a reasonable balance being struck between these two sets of costs. In economic life, it does seem that societies with structures at either extreme of this spectrum are much less successful than those in which some sort of balance is struck between them.
However, centralised bureaucracies are generally biased towards centralised solutions, because these look very efficient when viewed from the centre. There is therefore a constant pressure towards more rather than less centralisation.
We see the massive centralising forces exhibited by the European Commission. It is based on an outdated view of how societies actually work. But it was supported by the prevailing ideology of the leading member states, driven by historical contingency to set up a structure which would minimise the risk of a further catastrophic European war.
Perhaps the shambles which it has created over the Covid vaccination process will cause many of the member states to pause and consider putting a halt to these overly centralising tendencies.
Daron Acemoglu, 2003. Why not a political Coase theorem? Social conflict, commitment, and politics. Journal of Comparative Economics, 31(4), pp.620-652. ↩