Notes on Works in Progress Issue 15
Works in Progress is a beautiful little booklet, published each quarter by a division of Stripe. Loosely paraphrasing from one of its editors – Sam Bowman – “our thinking is the social benefits of GDP growth are appreciated, but remain under-rated”.
“You are all potential writers for Works in Progress”, Sam posited to the audience at the latest edition’s launch in Dublin. “Our focus is on policy ideas that are tractable“.
~~~
I recommend the magazine (free online to read here). Now for a few notes on the seven sections in the latest edition.
~~~
1. On the Falling Rates of Drink Driving
Nick Cowen presents the drastic decline in deaths over the past decade in the UK, attributing this to stronger deterrence via a) stronger policing, b) harsher penalties and c) an increase in social disapproval of drinking and driving.
I can attest to to a similar pattern in Ireland. We were hammered with television ads of car crashes and lost loved ones throughout 2000-2010, and I was breathalyzed twice from age 17 to 21. I’d often hear how “even one drink would put you over the limit”.
When I moved to Boston in 2010, Americans appeared cavalier – having a drink or even two seemed the norm. Fast forward to 2020; I’m back in Ireland, and drink driving norms appear to have dropped back a little – “just the one and you’ll be alright”.
2. Getting materials out of the lab
Ben Reinhardt takes on the question of why it can take decades for new materials like carbon fibre to successfully be adopted in commercial products. Reinhardt points out how academia incentivises the discovery of materials that are novel, over those that are scalable and useful. In earlier decades, organisations like Bell Labs avoided this disconnect. Bell Labs did so by co-locating teams doing fundamental research with those taking products to market. Today, that approach is largely spurned by bigger companies and the responsibility rests on startups. Unfortunately their funding structure is more amenable to software or pharmaceuticals than new materials.
Reinhardt’s diagnosis of the challenge in getting new materials to market is as follows:
The ultimate use-case for new materials is often unknown for large periods of time. This was the case for carbon fibre – before finding a place as a lightweight but strong structural element in aircrafts and elsewhere.
For startups – who compete with software and pharmaceutical ideas for capital – the time horizon for finding a use case do not justify the the costs of development for a new material.
Academic publishing norms incentivise novel materials/papers over materials that combine novelty with a pathway for reaching commercial scale and adoption.
Meanwhile, corporations – for reasons not entirely explained in the piece – nowadays leave early stage research to startups and acquisitions.
In sum, there is no complete and well funded pipeline to develop new materials and take them to market.
I add to this analysis with two further points:
The high costs of trial and error disfavour materials versus software startups. In materials startups, as in hardware startups, the costs of trial and error are higher than in software. This is a distinct and separate to the argument that capital costs are higher in hardware than software. The payoff from trial and error is fat-tailed. As such, what matters is a) being able to try as many things as possible and b) lowering the costs of trials. Ideas that are conducive to cheap trial and error (i.e. software) will appear to display higher margins and thus be the preferred destination for venture capital. Therefore, in a world where entrepreneurial talent is limited, hardware and materials businesses receive less talent and capital than software businesses.
Materials Startups often operate in Monopsony Markets. A material is just one portion of a final product. The production of physical products requires significant capital and benefits from economies of scale. A startup selling to such a producer may often have weak pricing power – as there are few such producers. Alternatively, if a startup decides to vertically integrate and manufacture themselves, this greatly reduces the percentage annual return on capital to investors (because the rate of return will now be heavily weighted towards manufacturing, which is a low rate of return business).
Solving (speculatively) for the equilibrium:
The Bell Labs of today are the software companies (Google, Meta) because software generates better returns than hardware for reasons of network effects AND lower costs of trial and error.
The startups of today also are software companies.
But what about the manufacturing companies today? Why don’t they do fundamental materials research?
The talent for such research is doing software (or finance or consulting)?
The CEOs/board of manufacturing companies are stock option earners rather than equity owners/founders, so they don’t think long-term like founders/owners, which is required to justify investing in materials research.
I don’t find my explanations entirely satisfying.
I commend Reinhardt for taking on the important topic.
~~~
3. Why prediction markets aren’t popular
J Zachary Mazlish and Nick Whitaker address the question of why political and (non sporting) event prediction markets are not popular. The core arguments are:
Unlike stock investing, where stocks as a whole typically go up, betting on political events is zero-sum. Taken together, the bettors on each side do not gain anything. In fact, allowing for time and operational costs, betting markets are negative sum (ignoring that the prices themselves provide valuable information – a significant assumption?).
Unlike sports gambling, many political and event prediction markets are for events further in the future, making them less interesting to recreational gamblers.
To the extent political bets could be useful for companies hedging against certain political events, such hedges might just as easily (or more easily) be achieved through futures markets (e.g. taking a position on the future price of oil).
While betting is restricted in the US, the Irish and British have long been able to bet on a very wide range of events. Empirically, the amount of money betted is heavily weighed towards near-term sporting events. Non-sports betting, such as on upcoming elections, appears to be a much smaller portion of a bookie’s business. As such, it appears true that there is little market demand for event and political betting markets.
For me, the (speculative) use case for prediction markets might be to help with decision making – perhaps as an (imperfect and incomplete) guide to voters or to policymakers on what might work or be implemented. One might argue that – to the extent this is useful – such markets might be offered by bookmakers already, but if the value of the information is primarily a public good, one could not expect such markets to naturally be funded. Perhaps then there is a case for subsidising such markets. Although, I’m skeptical of such subsidies working well, as the question of what market types to subsidise itself would become political. Rather, I prefer an informational/awareness approach, i.e. perhaps by setting up new markets at a small scale (even on just an altruistic basis) – the value of this information can be discovered. Alternatively, perhaps the creation of prediction markets (e.g. on Manifold Markets) can be used as a tool for political parties that themselves decide to subsidise or just set up such polls, i.e. a market of political parties type argument.
I wouldn’t be too harsh on prediction markets at this point for use in public policy. Surfacing information is hard and achieving accountability is hard in political systems, so we shouldn’t set the bar too high for ideas that might be of help.
~~~
4. Congestion Pricing in New York
Ben Adler traces the history of trying to introduce congestion pricing in New York City over the last decades up until the point today where – finally – it may be introduced (although uncertainty remains).
Two things come to mind in reading the piece:
It strikes me that allocating the proceeds of congestion pricing to public transport is a bad idea, not because it is bad to fund public transport, but because such a guaranteed flow of funds is not aligned with using that money efficiently. On the other hand, fees and taxes that are not earmarked for specific spending purposes get lost in a master budget where there can be similar problems of accountability. Either way, it strikes me that public transport is only partially an issue of funding, but mostly an issue of improving incentives for designing and operating the infrastructure efficiently.
After decades of political effort trying to get congestion charges, I wonder whether that political effort could have better been spent elsewhere – for example on improving zoning regulations.
~~~
5. Gentrification as a Housing Problem
Anya Martin reviews evidence of how direct efforts to slow down gentrification (e.g. through rent controls and policies against housing development) tend to be ineffective in reducing the displacement of residents.
To me, it is obvious that – when demand for housing in a certain area increases – the only way to slow rising prices and displacement is to increase the supply of housing – both in that area and around that area. But this isn’t a widely held view! So, I’m missing something. What is it? Is it just a natural human desire for more of the same – i.e. to freeze things as they are? Or, is it because there are real and concentrated costs to allowing development in your neighbourhood. I can’t stop thinking of how a friend of mine is (rightly?) worried about on street parking being a problem for him if there is more development locally.
I recently got a pamphlet from a local politician suggesting that we work to reduce local property taxes. If anything, perhaps we need to increase property taxes, so that there is a locally and communally felt benefit to increased property prices and development. I’m not convinced that’s enough though. People view housing prices as volatile and random, so I fear that reductions in things like public car parking spaces – albeit of negligible value compared to housing appreciation – is something tangible and outweighs the volatile benefits of house value appreciation.
~~~
6. A history of Britain’s housing (financial) crises in the 1900s
In one of the longer pieces of this Works in Progress issue, Samuel Watling charts the history of volatile housing prices in the UK. There are two phases of the housing price cycle. One phase is where policy artificially boosts the supply of mortgages – through low interest rates and tax breaks for home ownership – resulting in soaring property prices. The second phase is where building societies are mandated to reduce mortgages, resulting in a housing crash. What both phases have in common is that the supply of housing in the UK does not respond well to changes in prices, because it is extremely difficult to get permission to build.
I was surprised – although not entirely shocked – at how low the elasticity of supply for housing is in the UK – about 0.3, compared to 1.6 in the US. Interestingly, the US, Canada and also Denmark, Finland and Sweden have quite high elasticity – meaning that housing supply does adapt when prices go up and down. That said, I suspect the elasticity in US supply is not uniform – and – perhaps in cities like Boston or San Fran is similarly inelastic to the UK as a whole (which is perhaps dominated by London? (is that the cause of the skewed result?).
~~~
7. Why are buildings today not so ornamented?
This piece by Samuel Hughes was my favourite and covers theories as to why architecture has so much ornamentation before 1900, while most buildings after that are minimalist.
One theory is that labour has become relatively expensive, so we can’t afford ornamentation anymore. However, the counterpoints to that are a) that automation has made decoration a lot cheaper and b) actually housing at the budget end of the scale DOES have ornamentation (check out the article for some nice photos – “budget” houses built today in the UK are not so different from 100 years ago). So, most likely the decision to go minimalist is more to do with the cultural/artistic influences of Picasso and Stravinsky (and Joyce apparently?).
I’m a fan of bringing ornamentation back. It’s something that should be easy to do in web development – I just need to put aside a little extra time on my designs and put in those few swirls and flourishes.