AEI-Brookings Joint Center Policy Matters 04-10



High-tech Cluster Bombs: Why Successful Biotech Hubs are the Exception, Not the Rule. Scott Wallsten. March 2004.

The first economic-development fad of the twenty-first century has arrived -- Silicon Valley is out, biotechnology is in. The dotcom bust made Internet-led growth less attractive, but dreams of building a vibrant high-tech ‘cluster’ live on. Around the world, politicians wishing to replicate the biotech success stories seen in areas such as San Diego, California, are pouring taxpayer’s money into new initiatives. They imagine that luring the right businesses and research organizations to their backyard will set in motion a virtuous circle, attracting highly educated people to the area who will start more companies, which, in turn, will attract more people who set up even more companies.

There is just one small problem. Despite the promise of thousands of new jobs, cluster-development schemes rarely work. Success stories are the exception, not the rule. San Diego is a real biotech hub, as are Boston’s Route 128 and North Carolina’s Research Triangle Park, where thousands of scientists and engineers busily work on leafy corporate campuses. Yet nobody really knows how to start a successful cluster from scratch. As a result, money spent trying to build one is usually wasted.

Although the desire to set up a biotech cluster is relatively new, industry clusters have been around for some time. In the early nineteenth century, US manufacturing was concentrated in small parts of the northeast and Midwest -- shoes were produced in Massachusetts, and rubber in Akron, Ohio. Carpet producers still congregate in Dalton, Georgia, and furniture makers can be found in High Point, North Carolina.

Why do clusters thrive? In 1920, the economist Alfred Marshall suggested three common-sense reasons: benefits of a pooled labour supply; access to specialized inputs; and information flow between people and institutions. These largely hold true today, although the type of labour and expertise vary by industry.

For example, high-tech regions have lots of venture capital, active universities and productive interactions between academic and industry scientists. Research by economists suggests that knowledge and information will ‘spill over’ from universities to benefit the surrounding communities. But politicians and their consultants have come to see such features as ingredients that, when combined, constitute a sure-fire recipe for high-tech greatness.

It is tempting to think that all you need is a catalyst -- tax breaks, for instance, or a high-profile institute that moves to the area -- to get the process started. After all, Research Triangle Park, the grandfather of research parks established in 1959, became a success after IBM and the National Institute of Environmental Health Sciences set up shop there in the 1960s. Today, more than 38,000 people work in the 131 organizations located in the park.

With success stories such as this in mind, politicians are eager to push their own development ideas. The governor of Florida, Jeb Bush, recently said that giving the Scripps Research Institute in La Jolla, California $310 million in state money and some $200 million in local taxes to build a biomedical research facility in Palm Beach County was “the opportunity of a lifetime”.

Increasingly in the United States, cities, counties and states answer the call and put up taxpayer dollars for these high-tech dreams. Sometimes they think that direct subsidies, such as the Scripps deal, will do the trick. At other times they designate an area as a research park, build some facilities and sweeten the pot with tax incentives to encourage businesses to move in.

In Baltimore, Maryland, officials are hoping that a biotech centre in a run-down section of town will lead to a local renaissance. Likewise, the secretary of commerce for Virginia noted that a new facility being built in Loudon County by the Howard Hughes Medical Institute would be “an extraordinary windfall for Virginia”. State officials encouraged the county to give the institute a break on local property taxes to make the state seem more hospitable to other biotech companies.

Over the years, such activities have seen research parks sprout up like mushrooms around the world. The International Association of Science Parks lists members in more than 80 countries, and the United Nations estimates there are more than 400 parks worldwide. The Association of University Research Parks counts about 180 parks in the United States and has more than 200 members (not all members are parks, many are developers).

Are these initiatives based more on fantasy than fact? Consider a consultant’s report on the predicted economic impact of the Scripps deal in Florida. The report assumes that Scripps’ east-coast growth will mirror its San Diego experience, employing 2,800 people within 15 years and generating more than 44,000 jobs in the state.

Unlike Florida, officials in Baltimore at least seem to acknowledge the risks of their plan to attract biotech companies. And unlike most deals, Baltimore’s plan also includes substantial relocation payments to people being displaced from a poor neighbourhood. So at a minimum (and ignoring the fact that some residents might not wish to move), Baltimore may have found a way to help low-income families seek greener pastures. But predictions of 8,000 biotech-related jobs within ten years will probably turn out to be science fiction.

Consultants’ reports rarely mention failed projects — Florida taxpayers might have been less inclined to hand over half-a-billion dollars if the job-growth predictions had been based on, say, the experience of Prince George’s County in Maryland or San Antonio instead of San Diego. The Maryland Science and Technology Center opened with great fanfare in the late 1980s, promising to employ 12,000 people and generate an additional 25,000 jobs in Prince George’s county. Those dreams never came to fruition, as by the summer of 2003 the park’s developer had built only 240,000 square feet of office space. A local council member recently called the centre a failure, and the city plans to rezone the site to allow residential construction. The state wants a refund on some of the millions of dollars it invested in the site’s infrastructure.

San Antonio broke ground on its Texas Research Park in 1987, with claims that it would lead to 30,000 jobs on the park and another 100,000 spin-off jobs within 30 years. Thirty years have not yet elapsed, but with just 15 companies and about 300 jobs, it does not look too promising.

Both the successes and failures make for fascinating anecdotes, but they cannot answer a fundamental question: do research parks have an effect on regional development? Recently, I set out to investigate this issue empirically.

I assembled data on the number of jobs by industry (from the US Census) and amount of venture capital (from Venture Economics) for each county in the United States covering more than a decade. I then matched counties that established research parks in that time period (the ‘experimental’ counties) with similar counties that did not undertake such a venture (‘control’ counties). This method meant that the experimental group only included counties that established a park in that time period, thus excluding the early success stories, such as Research Triangle Park, as well as older failures. Matching counties was imprecise at best, but I considered counties to be similar if the number of high-tech jobs, venture capital and population were comparable the year before the experimental county set up its park.

I then compared job and venture-capital growth in the experimental and control counties before and after the parks were established. It turns out that the number of jobs or the amount of new venture capital grew no more in the counties that established research parks than in counties that did not. In other words, the data provide no evidence that, on average, setting up a research park made any difference to the region’s economic growth.

I also undertook an econometric analysis that controlled for income, large firms in the county (that could help to attract or spin off smaller firms), research spending by universities, government jobs, a time trend and other factors specific to a given county. This analysis reached the same conclusion. On average, there was no evidence that building a research park made a difference to regional development. If you judge these job-creation schemes not even by their own lofty goals, but by the simple expectation of driving regional growth, there is little evidence that they work.

Some may argue that not every cluster can be a San Diego -- you cannot judge them all by the same criteria. This is exactly right. Almost by definition, a given industry can cluster in only a few places. But those who promote biotech initiatives never say that they want to be a second-rate hub.

How do we explain the exceptions? Some, such as Research Triangle Park, had the advantage of being first. Others, such as Silicon Valley, benefited from a unique set of entrepreneurial and venture capital networks that developed symbiotically with local research institutions like Stanford University. Government policies that encourage entrepreneurs and promote a sound investment climate through, for example, good schools and infrastructure, also help. However, governments -- both local and national -- are not usually good at ‘picking winners’ when it comes to predicting which technology will drive future growth. It is even less likely that a given region will succeed if everyone is chasing the same dream.

The definition of success itself should be closely examined. If a park claims 100 new companies, is that a success? To the park’s promoters it is. But from the region’s perspective it is only a success if those companies did not simply ‘cross the street’ to take advantage of tax credits inside the park boundaries. Even if the businesses are new to the region, it is only a success if the benefits of luring them there outweigh the costs. Viewed from a broader perspective, Floridians will celebrate if companies move from Massachusetts to set up shop near Scripps. But globally, there is no net benefit. It is just a transfer -- of companies from one place to another and of money from taxpayers to mainly wealthy organizations and developers.

The impact on the scientific community is harder to predict. New initiatives presumably mean new job opportunities and potentially higher wages, which could, in turn, attract more people to the field. But if biotech growth is not sustainable then -- as with the dotcom experience -- that influx of people could turn into an oversupply. More generally, failed clusters across the country are unlikely to do much to enhance biotech’s public image.

The obsession with becoming the next biotech hub will fade in time, just as dotcom dreams did. But research parks and similar schemes to promote high-tech development are likely to continue to be irresistible to nearly everyone involved. Businesses -- from high-tech companies to developers -- that get money from these schemes are happy with their windfalls. Politicians are happy to hand out pork while looking like visionaries.

These are tough forces to overcome, but there are things we can do. First, economists and other social scientists should put more effort into understanding regional economies and finding ways to help regions improve on their strengths. Second, citizens should demand honest evaluations of cluster initiatives, instead of glowing reports from the politicians’ hired guns. Some high-tech development schemes may prove to be good investments, but the vast majority should remain in the realm of science fiction, where they belong.

Scott Wallsten is a fellow at the American Enterprise Institute-Brookings Joint Center and a resident scholar at the American Enterprise Institute.

This article appeared in Nature Magazine on March 11, 2004.

See related working paper