The Price of Delay

And why a recent paper likely overstates the harms

I avoid writing about zoning and housing regulation because I really don’t have anything interesting to say about it. Obviously, it is a very big deal. Deregulating it is the most obvious way we could risklessly grow the economy. Hsieh and Moretti estimate the gains from SF and NYC alone hypothetically deregulating housing, after correcting some errors, at a 36% increase in GDP. It is the primary reason for the non-convergence of US regional income, according to Ganong and Shoag. It used to be that people could move from poor regions to rich regions, and earn more. Now their earnings are eaten away by housing costs. And on, and on, and on.

I have read this all before. Doubtless you have read at least some of this before. It gives the feeling of preaching an old sermon, worn out years before. Nevertheless, it does deserve some repeating — and I have a new paper to discuss. 

I say this all to preface talking about new research. I am quite skeptical of their estimations, and believe them to be substantial overestimates. I have absolute faith that the permitting process is reducing the number of units available, but cannot believe the magnitudes. These criticisms do not undermine the thrust of the paper, and are simply quibbles over estimate size.

The cost of housing regulation is not only from explicit bans and bars on housing. It also comes from uncertainty. When we make it so housing is default illegal, and you need permission from someone to build, you cannot effectively plan for the future. Your investments are in danger of being totally wasted. A new paper quantifies these harms and finds that they are actually extremely high. 

“Development Approval Timelines, Approval Uncertainty, and New Housing Supply: Evidence from Los Angeles”, by Stuart Gabriel and Edward Kunga, tracks a large administrative dataset of all new housing started in Los Angeles between 2010 and 2022. The average product took four years to complete — of which 37% was waiting for regulatory approval. Risk-averse developers cannot predict how long the regulatory process will go on, discouraging investment. They estimate a 25% reduction in the length and volatility of permitting would increase housing completions by 20.9%. This regulatory approval is, damagingly, not even entirely concentrated at the beginning of the process. You will need to obtain additional permissions as the project goes along, tying up resources in half-completed projects. They estimate that speeding up approval times by 25% the rate of housing construction can be sped up by 11.9%, simply by freeing up resources. 

This is a result I don’t really believe, by the way. It rests on the resources used in construction not being transferable to other projects. Is this really plausible? A work crew which can’t get permission to proceed can be employed on other projects. A piece of construction equipment can be sent to another work site. We cannot simply assume that increasing the speed of construction projects doesn’t affect the completion of other projects. To take an extremely simplified example, suppose a developer has one tractor and work crew, which they use on one project at a time, and two running projects. Any non-simultaneous delay could allow them to work on the other project instead, and thus removing the obstacle of permitting would speed up some projects at the expense of another.

I checked with my papa, who ran a construction company for several decades. He felt that companies always wanted backup projects. Delays are common, and you want to be able to put your capital to work doing something. He worked in laying large sanitation and water pipes around Phoenix in the 70s and 80s, where the regulatory burden was not so intense. Nevertheless, simple concern about “will the pipes be here on time” is sufficient to encourage multiple projects being doable.

The results from reducing uncertainty are on sturdier ground, although they are constrained in how confident they can really be. The estimates derive from comparing differences in approvals between districts, and controlling for observables. There is no natural experiment here to try and get exogenous variation, but that’s ok. The estimates are still large and substantial. There really is no plausible story in which faster approval times make developers less certain and less willing to invest — all there is to be argued about is its effect size. This isn’t like the minimum wage, where you can construct *a* model with counterintuitive effects. The only cost is some externalities to neighbors, and no one would seriously argue that these externalities are large enough to outweigh the benefits of more housing construction.

The lesson is ultimately one we can take from Coase’s “The Problem of Social Cost”. Were there no transaction costs, we would have no uncertainty. It would not matter whether the right to build were default legal or illegal. With transaction costs, the default matters more than anything. A country like England, where all land use is required to go through government approval, is considerably harmed compared to the US, where land use, provided it matches what the land is zoned for, is default legal. I am an institutions skeptic — I worry that they are endogenous to other factors. This strikes me as a very plausible example of the deep institutions actually mattering, and not being merely cover for other things. I think these studies may not have been done yet. The La Porta, Lopez-de-Silanes, and Shleifer paper on the impact of legal origins does not appear to consider its impact on housing construction times, nor does googling turn up any papers. Anyone who knows a paper is strongly encouraged to let me know! And even better, they should dunk on me on twitter, for being so silly as to not know the paper!