Industrial policy never went out of fashion. Economists enamoured of the neo-liberal Washington Consensus may have written it off, but successful economies have always relied on government policies that promote growth by accelerating structural transformation.

China is a case in point. Its phenomenal manufacturing prowess rests in large part on public assistance to new industries. State owned enterprises have acted as incubators for technical skills and managerial talent. Local content requirements have spawned productive supplier industries in automotive and electronics products. Generous export incentives have helped firms break into competitive global markets.

Chile, which is often portrayed as a free-market paradise, is another example.

The government has played a crucial role in developing every significant new export that the country produces. Chilean grapes broke into world markets thanks to publicly financed research and development. Forest products were heavily subsidised by none other than General Augusto Pinochet. And the highly successful salmon industry is the creation of Fundación Chile, a quasi-public venture fund.

But when it comes to industrial policy, it is the United States that takes the cake. This is ironic, because the term "industrial policy" is anathema in American political discourse.  It is used almost exclusively to browbeat political opponents with accusations of Stalinist economic designs.

Yet the US owes much of its innovative prowess to government support. As Josh Lerner, Harvard Business School professor, explains in his book "Boulevard of Broken Dreams," US Department of Defense contracts played a crucial role in accelerating the early growth of Silicon Valley. The Internet, possibly the most significant innovation of our time, grew out of a Defense Department project initiated in 1969.

Nor is America's embrace of industrial policy a matter of historical interest only. Today the US Federal Government is the world's biggest venture capitalist by far. According to The Wall Street Journal, the US Department of Energy (DOE) alone is planning to spend more than 40 billion dollars in loans and grants to encourage private firms to develop green technologies, such as electric cars, new batteries, wind turbines, and solar panels.

During the first three quarters of 2009, private venture capital firms invested less than three billion dollars combined in this sector. The DOE invested 13 billion dollars.

The shift toward embracing industrial policy is therefore a welcome acknowledgement of what sensible analysts of economic growth have always known. Developing new industries often requires a nudge from the government. The nudge can take the form of subsidies, loans, infrastructure, and other kinds of support. But scratch the surface of any new successful industry anywhere, and more likely than not you will find government assistance lurking beneath.

The real question about industrial policy is not whether it should be practiced, but how.

Here are three important principles to keep in mind.

Industrial policy is a state of mind rather than a list of specific policies. Its successful practitioners understand that it is more important to create a climate of collaboration between the government  and the private sector than to provide financial incentives. Through deliberation councils, supplier development forums, investment advisory councils, sectoral roundtables, or private-public venture funds, collaboration aims to elicit information about investment opportunities and bottlenecks.

This requires a government that is "embedded" in the private sector, but not in bed with it.

Industrial policy needs to rely on both “carrots and sticks.” Given its risks and the gap between its social and private benefits, innovation requires rents, returns above what competitive markets provide. That is why all countries have a patent system. But open-ended incentives have their own costs. they can raise consumer prices and bottle up resources in unproductive activities. That is why patents expire.

The same principle needs to apply to all government efforts to spawn new industries. Government incentives need to be temporary and based on performance.

Finally, industrial policy's practitioners need to bear in mind that it aims to serve society at large, not the bureaucrats who administer it or the businesses that receive the incentives. To guard against abuse and capture, industrial policy needs be carried out in a transparent and accountable manner, and its processes must be open to new entrants as well as incumbents.

The standard rap against industrial policy is that governments cannot pick winners. Of course they cannot, but that is largely irrelevant. What determines success in industrial policy is not the ability to pick winners, but the capacity to let the losers go, a much less demanding requirement. Uncertainty ensures that even optimal policies will lead to mistakes. The trick is for governments to recognise those mistakes and withdraw support before they become too costly.

Thomas Watson, the founder of IBM, once said, "If you want to succeed, raise your error rate." A government that makes no mistakes when promoting industry is one that makes the bigger mistake of not trying hard enough.

Industrial policy is a state of mind whose successful practitioners understand that it is more important to create a climate of collaboration between the government and  private sector.