-A A +A

Published in the Globe and Mail on April 8, 2013

By Kristian Behrens

Many politicians and business leaders think that creating clusters of economic activity will be a silver bullet for solving a variety of regional or national economic problems. Bold claims that successful economic clusters drive prosperity nationwide require bold evidence – and that evidence is lacking.

The widening gap in Canada-U.S. per capita gross domestic product (GDP) is of concern to both policy makers and ordinary Canadians. Canadian firms’ relative underinvestment in information technology, for example, may explain a large part of the difference in economic performance north and south of the border.

Yet, according to some, the relative underperformance of Canada’s “clustered industries” – the spatial concentration of interrelated industries, specialized services, and customers – may also explain part of that gap.

Canadian manufacturing industries are less geographically concentrated than their counterparts in the U.S. or Britain. Furthermore, there has been a trend of manufacturing sectors becoming less spatially concentrated. Is that trend likely to sustain – or even to increase – the Canada-U.S. productivity gap? And can more geographical concentration help to reduce – or even to close – that gap?

Whereas most cluster studies have focused on specific regions, it is important to think about the broader national consequences of clusters. When some industries cluster in a region, this might come at the expense of other regions or other industries: Gains from clustering in some locations might well be dampened by losses in other locations. Touting the benefits of clusters at the local level might well overstate their overall national effects.

Looking at the evidence this way, it becomes clear that between 2001 and 2009 there has been only a weak link between the geographical concentration of manufacturing industries and their economic performance in Canada. Silicon Valley and Kitchener-Waterloo – the not-so-secret dream of every local policy maker – are exceptions, not the rule.

A doubling in the geographic concentration of firms in an average manufacturing sector would, on average, boost blue-collar wages by about 1 per cent. The effect on white-collar wages in high-tech industries is slightly larger, at about 5 per cent.

While these numbers look exciting at first sight, on a second look they reveal that the productivity and wage effects are rather small. Doubling the geographic concentration of firms in an industry is a herculean task that is likely to be beyond regional – or even national – economic policy. Such a doubling would amount to the reshuffling of large portions of an industry across locations. This could only be achieved at very high cost. Unfortunately, those costs are likely to exceed many times the expected productivity gains from clustering.

If cluster policies – be they local or national – do not provide cost-efficient solutions to the productivity problem, what else can be done? One obvious answer is to expand international trade. There is a much stronger positive relationship between trade and productivity between 2001 and 2009 than between clustering and productivity.

More exports and imports boost both value added per worker and average wages. A doubling of Canadian exports in an average industry increases, on average, blue-collar wages by 6 per cent, an effect that is six times larger than that of a doubling of clustering.

The policy message is that international trade has a stronger positive impact on productivity than the clustering of Canadian manufacturing industries. Looking at trade – and at better tax policy – provides more efficient and cheaper ways to tackle the productivity problem than cluster policy. The issue with trade policy is, of course, the absence of control as to where the positive effects materialize – which is precisely what makes cluster policies so attractive to local policy makers.

In the end, the choice between cluster policies and trade policy boils down to a difficult distributional problem that also has a regional component. Yet, if national economic performance is the key consideration, there is no reason to opt for cluster policies over, say, trade policy.

Kristian Behrens is professor of economics at the University of Quebec at Montreal. He is the author of the C.D. Howe Institute commentary “Strength in Numbers? The Weak Effect of Manufacturing Clusters on Canadian Productivity.”