The large structural changes of our time require ever-changing work tasks, and hence, skills. But to identify these new skills, let alone deliver them, we need an understanding of the specific process of skill formation. Bob Hancké argues that this needs to be a messy, political process rather than a technocratic exercise. Only this way can labour market actors build a common set of rules and understandings, which are embodied in institutions.
Skills are the future. Training and education are the best safeguards against unemployment and low wages. Skills improve not just labour productivity, but also allow the use of more sophisticated tools and machines. Companies, industry associations, consultancies, and strategy units in governments all agree: The industries of the future will be built on very different skill sets than the ones we know today. Governments make it the centrepiece of their education, industrial, and regional development policies. It is one of the few areas where market-oriented and more interventionist policymakers and observers agree.
Yet skills embody a paradox: They are a necessary ingredient of modern advanced capitalist economies, but actually very difficult to get right. Producing skills often requires the simultaneous orchestration and coordination of many actors, the need for thick institutions to contain the actors, and ways to monitor the skill production. And that all happens against the background of often at least partially diverging interests among these actors. Coordinating all those actions is hard but necessary.
The labour market problem today, as Nick Barr explains, is exacerbated by a series of new developments that originate in the combination of new products and services (from electric vehicles which require different skill sets to an economy based on gigs); automation (poised to supplant most of the jobs that have provided a stable livelihood for the bulk of the workforce in the OECD); and emerging new models of work organization (hybrid work and WFH, project-based self-employment, etc, often related to the Covid shock). All this suggests new skills and more training. But that is where the problem starts.
In essence the issue of skills has three dimensions: Do we – always – need more skills? Do we need new specific skills or, instead, more general skills? And, probably most importantly, how would we know what we need? After a short analytical introduction to the problem of skills, these three questions will guide the rest of this paper. Foreshadowing the analysis here: Rather than being a technical exercise (or even a survey-based inventory), producing skills – defining them and building mechanisms through which they can be acquired – is a messy, political process, in which parties with partly competing visions (actors in government, industry, and labour) build a common set of rules and understandings, which are embodied in institutions. Those jointly managed institutions are needed to have a better chance at identifying the skills needed; and an understanding of the specific process of skill formation is needed to arrive at these newly needed skills.
Why is skill formation a problem?
A short excursus on skill formation – often an elusive political economy problem – will allow us to get a measure of the problem. For modern advanced capitalist societies skills are a crucial ingredient of their economic success, because these economies have little else: few raw materials, agricultural monopolies or other natural endowments that produce wealth. Yes, some have wine and gas or oil, but the first is a relatively small part of GDP, even in France and Italy, and the latter is in many cases as much as curse as a blessing – such cheap money is usually not wisely spent.
While very important, skills are not simple to produce. Even if we know (or think we know) what we need, we have to design a system that will successfully generate them. Start from the basics: With very few exceptions, most skills are useful in different organisations – in other companies in the same sector (eg. skilled machine tool builders, doctors or accountants), in similar occupations in different companies and sectors (eg. ICT or mechanics), or in many different professions in many different sectors (eg. economists). As a result, a company may well train an employee, but another company can employ her without incurring the training costs for a slightly higher wage than the former intended to pay. The former thus stops training lest they subsidise the competition, and the skills that benefited everyone, at least in principle – the employee, the industry, the companies, and society as a whole – are not there.
Skills are underprovided as a result of the collective action problem that ensues, unless one or more of three conditions are met: the benefits accrue, entirely or as good as, to those who fund the training (company or individual); the government funds a large part of the training and makes it accessible to all; or all companies in a sector (or a region) contribute more or less equally to the funding of a training. The latter requires, of course, that companies are incentivised to do so – in this simple prisoners’ dilemma situation a confluence of interests among actors is not enough for a sophisticated system for training to develop. That is the bare bones political economy problem.
So, do we need new skills in the future?
The generic answer to the question of the need for new skills is almost invariably yes. But sometimes old skills, which rely on tacit knowledge make more sense. If something can be automated, it will be. But tacit or practical knowledge is hard to codify and therefore automate. If machines take over the ‘easy’, repetitive tasks, we may need to develop the ‘difficult’ (from a machine’s point of view), unique tasks. This does not have to be esoteric stuff. A cleaning robot, for example, would probably throw away a Rolex watch left in a hotel bathroom as rubbish, while the human cleaner would recognise it in a fraction of a second and alert the reception. We are simply better equipped to deal with this eventuality than the robot – who may well be a much better cleaner in every other way.
Something similar happens with hybrid working and work from home (WFH). While we need the stimulus that comes with simultaneous physical presence, many, and probably most, jobs could improve tremendously, for both the employee and the company, if they were treated as a series of project-like task bundles. But doing so likely requires a portfolio of old-fashioned ‘people skills’, workflow management skills, and possibly even ways to handle low-level logistics problems. These perhaps redundant skills may be useful even for staff that is not likely to use them since they could step in during emergencies or give their advice on how to improve processes.
The upshot is that we need to take a careful look at the work that can be usefully farmed out to machines, and upgrade the remaining jobs. Importantly, this is not a static problem, whereby we identify tasks that exist now: the redesign of jobs will generate new task profiles and, thus, new skills, or at the very least old skills repurposed for these new working arrangements.
Different specific skills or more general skills?
A second problem is how to think about the nature of future skills. Broadly speaking, we distinguish between two types of skills: those that belong to a particular occupation and can only or almost solely be used in that occupation. We call those ‘specific’ skills, of which nurses, machine tool builders, doctors, lawyers or plumbers are paradigmatic examples; or those skills that rely on insights and analytical knowledge that can be used in many different occupations and settings – so-called ‘general’ skills, for instance, engineers or economists.
The difference between the two is not the level of abstraction or the depth of knowledge but the restrictive nature of the setting within which the skills can be used. A lawyer, doctor or plumber know a lot of things, including sophisticated abstract principles, that civilians do not know – try installing a heating system in a new house if you don’t think so; and then do it again in the neighbouring, also new but differently designed house. That is why these jobs very often have dedicated training institutions (medical, nursing or law schools, or long apprenticeships) and highly policed labour markets (numerus clausus, public exams, certificates, etc.).
Now, the issue with regard to future skills is whether we need (more but) different specific skills, relying on specialised knowledge that can only be used in a limited number of situations, or more general skills instead. Ignore the examples above: the world will always need nurses, doctors, plumbers and lawyers. But will we need machine tool builders in a world of electric vehicles – and as many as we have now? Will we need steel workers in a circular economy? Will we generally need many manual workers in the future, or do we actually need trained engineers to monitor and control the computers that perform most of the manual work? The argument for emphasising general skills in future is strong: if the future is this uncertain, societies hedge by diversifying their skills portfolio – but that implies, in this instance, also hedging the individual skill portfolio and concentrating on highly transferable general skills, at the cost of the stable jobs, careers, beautiful products, and politics that came with a labour market built around specific skills. And if we lose the labour market as a coordinating institution altogether, in a dystopian generalised gig economy, the risk of job loss has become endemic and only general skills can save you.
How would we know?
But probably this extreme scenario is a little far-fetched, and the choice between new specific and more general skills is one of balance – a bit of both, in other words. That raises the third problem with regards to skills. How would we know? What is the mechanism for aggregating the information on future skill needs and how do we organise the information clearing so skills that are necessary find employers and employees willing to invest in them, individually, collectively, privately, or publicly?
This is an underestimated dimension of the skills problem. One issue is breadth. Even if a company knows what it wants, it might set the parameters for training too narrowly, so that skills can only be used in that one place; this may result in life-time employment, but not necessarily of the type compatible with open labour markets. A worker in such a situation would never have a chance to leave for another job, and aggregated over the population at large, we would end up with stagnant, closed, quasi-feudal societies. No worker would willingly do this, of course, and training falls through. But the company may also be simply wrong: it thinks it needs more skilled machine tool builders, trains them, and then discovers that the technological and organisational model that it based its decision on is obsolete. Plenty of lives can be ruined in this way, as the US Midwest’s deindustrialisation drive over the last three decades demonstrates.
While the knowledge problem is a generic one, there are no easy answers. Hayek and his followers would emphasise the ability of a decentralised market economy to do the work: millions of more or less informed decisions will aggregate up to an optimal solution. Ignore, as this argument does, the existence of market failures and concentrate on the paradox that what may be socially optimal could be an individual catastrophe. What if everyone simultaneously makes the same mistake (as happened during the dotcom boom and the financial crisis)? Imagine, for example, that all car companies (and their unions) think that the turn to electric vehicles implies that we need better trained skilled workers in the area of mechatronics (combining mechanics and electronics), but one company decides to have a decentralised system, where they name and market a brand, hire Amazon for the logistics and Tandy (Radio Shack) for the generic parts of an electric vehicle; they would make a very cheap, customised car in a local garage in a week or less. This may sound like social science fiction, but so was reliable communication on a tablet computer via Skype or Zoom only a generation ago.
Yet the state, the immediate mirror solution, does not necessarily work much better. You don’t need to think of the excesses of centralised planning in the Soviet Union (with its 5000 right shoes instead of 2500 pairs) to see the problem. Even strong, democratic states in open societies and market economies have failed. France has worked hard on its version of a German-style dual training system since the mid-1970s and failed. The UK destroyed functioning apprenticeships since its forced deindustrialisation and has struggled to rebuild them ever since.
Hayek was right about the problem here – the overly centralised nature of decision-making – but wrong about the market as the solution. The German training system (which also exists in a similar form in Austria, Belgium, Switzerland and elsewhere) works well because it is a private yet non-market, and collective yet decentralised information and decision-making mechanism. It includes business associations and chambers of commerce, employers and trade unions, and works councils and HR departments, all balanced in a long-term, mutually beneficial arrangement in which all actors hold soft or hard veto positions.
But those organisations, in this or a similar form, with the strategic vision to disentangle short-term costs and long-term gains, are not equally strong or may not even exist everywhere. One of the reasons why the introduction of training systems in France and the UK has been a failure is the relative lack of underpinning institutions that would govern the process of building and running a functioning training system: unions and employers are too weak, chambers of commerce are absent as local players, and links between schools, universities and companies are either too close or too tenuous.
The political economy problem at the core of skills
The vexed question of skills is, in essence, a political economy problem. It is very difficult to think of the future of skills without a history of collective organisations that produce frameworks allowing all stakeholders in the skills debate to have their say. Not only does such a set-up balance the different claims; it will also produce genuinely better outcomes because all opinions can be challenged and need to be explained. Once those arrangements are in place, the contents of skills and the concomitant training system can be addressed.
The generic issues with regard to skills are simple, in other words: you need a mechanism to elicit and aggregate relevant information; a space where ideas can be voiced, evaluated, modified or withdrawn; underpinning arrangements that would make decisions binding yet revisable without undermining trust; and a monitoring and policing mechanism that all parties deem legitimate. And all these prior arrangements have to be up for grabs at (almost) any time, since you cannot innovate the product (skills) without knowing (and innovating) the process of building the product.
Balancing those needs is a highwire act – which helps us understand why modern France and the UK, while achieving most formidable political accomplishments in the history of humankind by most standards, have hit a wall so many times. For the technocrats in the Elysée and Number 10 defining the new skills was the first, but also the last step. They ignored the need for the institutional infrastructure that addressed the different political dimensions of the problem, including the possibilities of constructive disagreement. Instead of concentrating on the contents of skills, they would have been better advised, perhaps, to think about the less glorious but more substantive job of incentivising socio-economic actors to develop the arenas necessary for such negotiations.
Conclusion
Skills are, with capital, the economic life blood of modern advanced capitalist democracies. But they are also a high-stakes gamble for individual employees and companies. Getting them wrong has a high cost – and the more uncertain the environment, the higher the chances for failure, dragging along entire communities and industries. Skills are therefore not just an economic but also a political issue, on which different parties in the economy project their life chances.
Political problems require political solutions. In Germany and its satellites, this has taken the form of a complex dual training system that involves all relevant actors – companies, unions, governments, and industry associations – who negotiate a new future of the economy, of companies, and of skills. Not everywhere are the institutional endowments present to address the skills problem in this way. But, in a basic sense, all have to think about the central political questions. The central issue is, therefore, how to translate these crucial conditions for skill development into institutions that strengthen those bequeathed on countries and sectors by history, and how to overcome the structural constraints that history imposes. Exploring the art of the possible, in other words – or politics.
© PEACS OG October 2020