The Next Silicon Valley Won't Do Silicon
One of my heroes is Pike Powers, a politician-turned-consultant who helped build Austin, Texas, from a sleepy state capital into one of the world’s premier technology clusters. In an essay for the Federal Reserve Bank of Kansas City, Powers explains how they did it. The key to Austin’s success, he says, was the collaboration between local government, big business and the University of Texas.
These entities agreed that economic development was the ultimate goal, and they drew up long-term strategic plans to lure businesses to the city. The city spent money on schools, roads and water rights. The university built startup incubators and encouraged spin-off companies. Local business organizations campaigned to lure companies to the area. But there was a focus not just on attracting businesses, but on human capital as well -- Austin worked hard to create the kind of city where so-called creative-class types would want to live, which in turn made it more attractive for companies to move there.
It’s hard to know which of these policies were more effective, and which were just for show, but the Austin experiment seemed to work. From 1970 through 2000, Austin’s population growth dramatically outpaced that of Texas as a whole. The consulting firm Savills ranks Austin as the top tech city in the country, and similar lists always include Austin as a strong performer. That growth has provided hundreds of thousands of jobs for the people who moved to the city, and it has produced a host of successful startups that have created valuable technologies for the U.S. and the world.
So can the success of Austin be replicated? This is the question of industrial cluster policy. Instead of traditional development policies, cluster policy aims to make certain cities into hubs for specific industries. The theory is that when you get a bunch of similar companies together, good things happen -- people and companies share more ideas, through casual conversation and workers switching employers. Economists have theorized about these knowledge spillovers since the days of the legendary late-19th-century British economist Alfred Marshall. Strategy researchers have attacked the problem from a different angle but reached similar conclusions. And empirical studies have shown that when a bunch of companies in the same industry cluster near to each other, they tend to be more innovative. So if governments really can create clusters, the benefits could spill over to the whole country and to the world.
Unfortunately, there’s no guarantee that governments can reliably do this. Japan’s Ministry of Economy, Trade and Industry, once famous for its industrial policy, tried in the early 2000s to create a series of industrial clusters. But research productivity went down, not up, and the initiative was widely considered a failure. A slew of attempts to create the next Silicon Valley have failed as well. That raises some uncomfortable questions. Was Austin’s success a one-off? Were the vaunted policy initiatives of Pike Powers and his coalition just a sideshow to a process of natural growth? Or are most governments and industry groups simply lacking in the smarts, the will or the resources that Austin luckily possessed? Are governments simply using the wrong strategies to create successful clusters? And do cluster policies work in developing countries, where technology as a whole is less advanced?
A number of teams of economists have set out to answer these questions in recent years. Aaron Chatterji, Edward Glaeser and William Kerr conducted a study in 2013 in which they evaluated the goals, methods and results of entrepreneurship-focused cluster policies in the U.S. They reach very few solid conclusions:
Some policies do seem to have many upsides and few downsides, such as allowing more skilled immigrants, strengthening education systems, and eliminating unwise regulations. But when we move beyond such simple broad policies towards specific entrepreneurship strategies like clustering, our ignorance becomes obvious. The best path forward involves experimentation and evaluation.
A recent book by a large group of Latin American economists is more hopeful. Alessandro Mafioli found that a number of Latin American cluster projects boosted employment and wages over the years. But because they study whole projects instead of individual policies, he couldn’t draw firm conclusions about which specific policies help and which are useless or worse.
One idea might be to focus on industries other than information technology. There can only be so many Silicon Valley-style clusters in a country -- or in the world -- at one time. Pittsburgh, which, once thrived as America’s steel hub, has made a comeback in recent years as a center for the health-care industry. As in Austin, a university -- in this case, the University of Pittsburgh Medical Center -- formed the core of the new cluster.
So perhaps it would help to have some coordination at the national level, so that each city or region that wants to build a cluster can focus on its strengths. That coordination might be provided by the Cluster Mapping Project, a joint effort of Harvard Business School's Institute for Strategy and Competitiveness, and the Department of Commerce’s U.S. Economic Development Administration. Using this database and its visualization tools, policy makers, business leaders and universities can get a picture of which industries are strong in their region, and which are strong in other regions.
That will let different cities try different cluster policy experiments. Instead of everyone rushing clumsily to build the next Silicon Valley, each city can try its hand at unique regional specialization. A bunch of different policies, in turn, will help economic researchers figure out what works and what doesn’t. When it comes to clusters, we should let a million flowers bloom.
This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
To contact the author of this story:
Noah Smith at email@example.com
To contact the editor responsible for this story:
James Greiff at firstname.lastname@example.org