Last week David Cameron launched ‘East London Tech City‘, which he hopes will become ‘one of the world’s great technology centres’. Could it work? Up to a point, according to research from LSE. Science parks can pay off, although it’s unlikely they’ll create the next Silicon Valley.
Theoretically, we shouldn’t need science parks at all. In spatial economics models, firms sort across space to optimal locations. In practice, this doesn’t always happen. Planning restrictions limit space; businesses may lack the funds to move; and some firms will head to prestigious addresses, rather than the most productive.
So policies that try to cluster firms together might be a good idea. Theory and evidence suggest businesses benefit from co-location. Big labour markets, a rich mix of input-output linkages and knowledge spillovers help firms become more innovative – and more productive. In turn, this helps explain why cities form and grow.
So, can science parks can replicate these dynamics? Christian Helmers, a research economist at SERC, is trying to find out. There are now at least 85 parks in the UK, with over 76,000 workers on site. Helmers looks at two: Cambridge Science Park, the UK’s first and most prestigious, and the St John’s Innovation Centre next door.
Christian is interested in whether science parks drive up firm’s innovative activity. So he tests whether co-location raises patenting rates. He finds it does. He also looks at what types of firms tend to gain. Controlling for various other factors, he finds that inside the park, firms of the same industry tend to patent more. Science parks can pay off.
What does this mean for policy? Christian suggests policymakers should promote specialised science parks, dedicated to single industries. I think there are some wider lessons here too – not all of which are good news for East London.
1) Lifecycle – Helmers finds no significant effect of firm age on patenting – that is, start-ups don’t particularly benefit from being in science parks. This echoes other work which suggests that economically diverse cities act as ‘nurseries’ for young firms – basically offering a big pool of ideas, suppliers and people. In turn, it suggests that unless they’re in big cities, incubators might not be effective.
2) Mix and scale – Christian’s research basically tells us that at small scale, similarity matters – firms gain from having others like them around. (In the jargon, science parks exhibit ‘Marshall-Arrow-Romer’ externalities.) But at city scale, the opposite seems to be the case. Economic diversity matters. Jane Jacobs first suggested knowledge spillovers across industries – recent work by Duranton and Puga and a team at SERC empirically confirms this . So we shouldn’t expect science parks to drive cities.
3) Expectations – Silicon Valley is not a valley – it’s a city-region. By contrast, science parks are tiny. Taken together, Cambridge Science Park and the St John’s Centre cover around 2m square feet. At a conservative estimate, Silicon Valley covers 1300 square miles. To put point 2) another way, no science park (or Silicon Roundabout) is going to be the next South Bay.
4) Surroundings – some of Christian’s results have to be driven by what’s outside the science park – in this case, Cambridge University and the Cambridge high-tech cluster. All the firms in the park benefit from high quality research and a big pool of skilled workers. The East London park will have to draw on the whole of London’s innovation system if it’s to benefit its tenants. The London location will clearly help. But as Vivek points out here and here, high-tech growth is fundamentally about people and culture – not property.