New Systems versus Evolution

At a recent talk in London, Chris Blattman asked if a charter city could fail in the same way that public housing failed in Chicago.

After the seminar and in his follow up post, we started a discussion of the deeper issue: What kind of dynamics would be desirable in the space of rules?

To frame the discussion, it helps to have some vocabulary. A rule set is a large collection of rules that are tightly linked. Rule sets can improve through what I will call an evolutionary dynamic based on small, incremental changes or through a new-system dynamic in which an entirely new rule set enters and competes with an existing one.

Chris and I agree that an evolutionary dynamic can lead a vibrant neighborhood but that it can also lead to a dangerous slum with no roads, sewers, safe water, electricity or other utilities.

We also agree that there is a risk associated with new systems. Sometimes they don’t work, as the public housing projects in Chicago demonstrate. Sometimes they work remarkably well. Architecturally similar high-rise buildings in Hong Kong and Singapore provided livable housing for large numbers of working poor in the 1960s and 1970s. (As an aside, Chris and I seem to agree that the key difference between these cases lay not in the hardware or architecture but rather in the supporting rules, particularly those related to crime.)

The challenge that Chris posed to me, one that several others have also suggested, but not as precisely, is how to assess the tradeoff between faster growth and higher risk that the new-system dynamic seems to offer. A related question is who bears that risk. These are important questions. Here are some provisional answers.

Start with the distinction between catch-up growth and frontier growth. Catch-up growth is based on copying existing ideas. Frontier growth involves the discovery and implementation of new ideas.

In catch-up growth, the new-system dynamic can be used to copy existing rule sets. This allows faster growth without the additional risk that comes from using the new-system dynamic at the frontier.

An analogy with firms might be helpful. Amazon.com and Webvan were both new systems that involved innovation at the frontier. Amazon worked. Webvan failed. New systems that are innovative do involve risk.

Contrast this with the process that lead to the spread of modern retailing in the United States. In principle, the ideas and rules developed at Wal-Mart could have been scaled through an evolutionary dynamic that patched and tweaked operations at stores operated by such traditional retailers as Montgomery Wards. In practice, the technologies developed by Wal-Mart spread geographically because new systems – entirely new buildings, with new sets of employees – arose in many different locations. This is an instance of catch-up growth based on the new-system dynamic. Building new Wal-Marts is not as risky as creating a new web-retailing model.

The concept of a charter city does rely on the new-system dynamic, but in the context of economic development, charter cities can adopt a new rule set made up of rules that are known to work well. It can accelerate catch-up growth with little of the risk associated with the development of new systems at the frontier.

A charter city could be also used, perhaps in the United States, to encourage innovation at the frontier of urban living. An entirely new city could introduce changes to transportation systems or energy systems that are improbable along an evolutionary path. This would be like trying to invent Amazon and Webvan. It might be worth doing for the same reasons that it is worth encouraging new start-up firms, to speed up growth at the frontier. But in this case, faster growth would involve more risk.

It is important to recognize that new systems typically need evolutionary improvement after they are introduced. As existing charter cities compete with each other, they will evolve in precisely this sense. Moreover, the new-system approach at one conceptual level (the legal, administrative, and institutional level) does not have to imply a new-system approach at a different level such as neighborhood structure. A charter city could have neighborhoods that evolve according to the kind of evolutionary processes advocated by Jane Jacobs.

A new city could still fail, just as Detroit is failing in the United States. The risk of a failure falls primarily on the owners of the fixed assets that can’t leave, not on the labor income of workers who can leave. This seems like just the right allocation for incentives and risk sharing. If the development authority with administrative responsibility for the city owns the land and collects revenue as rent on long-term leases, it has a strong incentive to get the rules right early on and to help the rules evolve in the right direction thereafter. So too do the diversified private investors in infrastructure and buildings.

An arrangement with many cities that compete with each other is present today in China. Many new or rapidly growing cities compete for residents and investors. Local officials receive a substantial portion of their revenue from gains in the value of land that they own. To my knowledge, none of the new or rapidly growing cities in China has failed to the degree that public housing failed in the United States.

Finally, it is worth noting that roughly 3 billion people will move to cities in the next 50 years. If this growth takes place by an evolutionary dynamic that expands existing cities, many of these cities will fail as badly as the public housing projects did in Chicago. In fact, many of them already do.

The right question to ask is not whether the new-system dynamic, as applied by charter cities, is subject to some risk. Rather, it is whether the risk of failure in charter cities is any greater than along the status quo path. Because charter cities use the new-system dynamic to copy best practice and create strong competitive pressures for rapid evolutionary improvement after they are built, I’m convinced that they offer the prospect of both faster growth and less risk.