Industrial policy is back. Vince Cable, Britain's Secretary of State for Business, Innovation and Skills is calling for a "proper industrial policy" to create "a different type of economy".
In the past week, I've been invited to two thought-provoking events on the subject.
The first to be organised by Chuka Umunna, Vince Cable's Labour counterpart, featuring the distinguished innovation scholars Carlota Perez and Mariana Mazzucato;
The other is a collaboration between three left-leaning think-tanks: the IPPR, the Resolution Foundation and the New Economics Foundation featuring a distinguished panel of wonks and Vince himself.
The debate in the UK has parallels in the US, where the related issue of how (and whether) to save American manufacturing is a hot topic.
A lot of the debate so far has focused on the question of whether industrial policy is a good idea from an economic point of view.
This is an interesting question but one unlikely to be resolved definitively in the near future.
Making the case for industrial policy
While we are waiting for an answer, a more relevant question from a policymaker's point of view is "what arguments do you need to win to put a credible industrial policy in place"?
Imagine yourself in the shoes of a policymaker thinking about the future of the economy.
If you like, you can be the parallel-universe Secretary of State for Business, or Shadow Secretary of State.
You're concerned at Britain's economic prospects after the financial crash and worried about how we're going to stimulate growth across the country.
Could industrial policy do the trick?
Now for industrial policy to be a runner for you politically - to be something that you can credibly advocate, design, and campaign on - you have to answer three questions.
Not only whether it "works" in the abstract but also, whether you can build effective policies around it, and whether you can sell it to the electorate.
So far, too much of the debate has focused in a very general way on the first question and not enough on the last two.
Let me explain, starting with the first question.
1. Does industrial policy work?
The first question you need to answer if you want to "do" industrial policy is a basic one: does it work?
There is a strong vein in British (and American) political discourse that holds that it simply doesn't: industrial policy means subsiding British Leyland and "picking winners" that turn out to be anything but.
Since 2008, this view has been challenged.
This is the argument we've seen played out in newspaper columns and think-tank meetings in recent months. The debate typically moves quickly to the analysis of history.
The most intuitive way to address the question "does industrial policy work?" is to ask "has industrial policy ever worked?"
Here, advocates of industrial policy have made a good showing, quoting lots of instances of economic growth that appear to have followed on from government action.
The most obvious examples are innovative economies with so-called "developmental states", such as Finland or Taiwan. (Here is Esko Aho, Finland's former prime minister, talking about the role of government innovation policy in unlocking Finland's economic turnaround and rapid growth in the 1990s.)
Effectiveness in Anglo-Saxon economies
It can be argued this works in Anglo-Saxon economies too.
One of the main points of Will Hutton and Philippe Schneider's The Failure of Market Failure or Mariana Mazzucato's The Entrepreneurial State is that most, perhaps all, innovation success stories in the US and UK have involved state action.
From the public R&D grants behind the technology used by Google or the biotech industry to the British government's gift of the 400MHZ spectrum to Vodafone, the hand of the state looms large.
Fred Block's Swimming Against the Current highlighted the importance of a "hidden developmental state" in underpinning the apparently free-market innovation seen in the US economy.
The Keynesian economist Ha Joon Chang argues (for example in this video, half-way down the page) that the early industrialisation of the UK relied on industrial policy, a case first made by Friedrich List and Alexander Hamilton in the nineteenth century.
Of course, not all aspects of this argument are uncontroversial.
The question of whether specific innovations resulted from industrial policy can be disputed. Readers of the American debate on industrial policy may have been amused to see hydraulic fracturing ("fracking") described as the result of industrial policy in an article by the think-tanker Michael Lind only a week after it was highlighted by former White House chief economic adviser Michael Boskin as an innovation that developed without government help.
But despite these quibbles, our hypothetical policymaker has some decent ammunition if she chooses to argue that industrial policy can work.
There's still more work to do. Policymaking isn't just about economic analysis: the specific design of policy and political communication matter too.
These two areas have been considerably less well analysed in the debate so far.
2. Can you design specific policies likely to work in the future?
It's not enough to argue that some industrial policies have worked in the past.
Our hypothetic Secretary of State also needs to get specific: to outline what a modern-day industrial policy would look like in practice.
Here she faces two problems: firstly, it's an area where the devil is in the detail, and secondly, it's an area where the required policy expertise is in relatively short supply.
Let's start with the point about detail.
Knowing that, for example, Vodafone benefitted in its early days from free spectrum and relaxed planning rules for its base stations, going on to become a telecoms titan is interesting, but doesn't really help a policymaker trying to design an industrial policy to support nanotech.
One of the valuable insights from five decades of innovation studies is that the peculiarities of different technologies and industries matter and that effective policy reflects this. The specific mechanisms of industrial policy matter too.
The recent decision of the EPSRC to exercise a "shaping capability" to concentrate physical science research looks like exactly the kind of thing an industrial activist might want to do more of. But rightly or wrongly, it seems to have attracted the opprobrium of several distinguished scientists.
To be credible, an active industrial policy needs to manifest itself as specific suggestions about policy: how should we fund research and development? What intermediate institutions should we have, and how should we fund them? What role should government play in the finance of new ventures?
The details matter.
Finding helpful advice
This is complicated by the second problem: the kind of expertise needed to make these decisions is in relatively short supply.
Doing industrial policy well requires knowledge of businesses and technologies, not just economics and policy. These are relatively expensive skills for politicians to access: it's not economical for an Opposition to employ advisers with this experience, and even in government, the civil service's capability to provide this kind of advice has to some extent been hollowed out by decades of government antipathy to industrial policy.
Of course, the private sector stands ready to advise the government on this kind of question but taking advice from companies on how government should regulate and subsidise them raises obvious questions of its own.
Before our would-be industrial policymaker lets out a frustrated cry of "I can't get the staff!", it's worth entertaining a more troubling possibility.
Some of the analysis required to make good industrial policy may not be possible at all, or at the very least may be extremely difficult, even for experts.
Professor Mazzucato uses the phrase "the entrepreneurial state", to describe a government running a good industrial policy.
One definition of entrepreneurship is making decisions, and committing resources, under conditions of uncertainty.
Cognitive psychology suggests that these are the kind of decisions that experts make badly.
In the 1950s, Paul Meehl showed how experts, in his case clinicians, underperformed simple algorithms in predicting the long-term prognoses of their patients; Philip Tetlock demonstrated the disastrous failure of political experts to make accurate predictions about their fields of expertise; and a wide range of psychologists and other researchers have replicated their findings in other areas.
Even with the right experts, it is difficult to make effective industrial policy to the extent that involves making bets about future technologies, demand and capabilities. At the very least, it argues for an approach that takes into account the fallibility of experts, and embraces multiple methods and approaches.
So politician trying to push industrial policy must not only make the broad intellectual case for it, they also need to make specific policies, of a kind that require business and technological expertise that is likely to be in short supply among the people they can readily turn to for advice, and that even then may let them down.
3. How do you describe what you are doing to the electorate?
Having won the conceptual argument for industrial policy and designed a set of specific and practical policy proposals, there is one final question for our would-be industrial policymaker to address: how to sell the package to the public.
We've already established that industrial policy is, rightly or wrongly, a hard sell in Anglo-Saxon political systems.
In fact, it is noticeable that there are broadly speaking two types of countries that have enacted effective, explicit policies based around innovation: small, highly democratic countries with high levels of political consensus, like Finland and Sweden in the 1990s; and smallish, less democratic countries with high levels of political consensus, like Singapore, Taiwan or Korea in the 1980s.
It's unclear that the same political tactics that worked in these countries would work in the UK today: at the very least, it is a political risk.
There is one other option: the US model, Fred Block's "hidden developmental state", in which the extent of industrial policy not a subject of political debate.
This too has its pros and cons: keeping innovation off the political radar gives it protection from political vicissitudes, but it makes it hard to justify large investments. The US to some extent squared this circle by justifying investment on mission-related grounds: in particular national defence and federally funded health research.
Some have argued that combating climate change could play the same roll in a future UK industrial policy. But again, this is not a political slam dunk. In particular, it makes it difficult to engage the public as partners in innovation policy, meaning it is likely to remain a technocratic exercise.
With more and more of the population educated, interested in science, design, or the creative industries, and active in innovation in their own right, failing to tap the "geek factor" and co-opt it may result in industrial policy that is not only less popular, but also worse.
Not a level playing field
So as we've seen, if our hypothetical policymaker wants to put in place an active industrial policy, she needs to tackle three issues: the conceptual case for industrial policy (where most of the debate has focused); the specifics of policy design (which is difficult because it is detailed and the skills to advise on it are relatively scarce); and how to communicate it to the public.
No wonder it is daunting.
But there is one further difficulty: the opponents of industrial policy don't have to win all three battles. Perhaps unfairly, they only have to win one.
Either by showing that industrial policy generally doesn't work, or by arguing that it is not possible to design effective policies to make it work in the future, or by designing more convincing messages, they can in practice undermine an attempt to implement an industrial policy.
Unfair perhaps. But forewarned is forearmed.