Month: May 2018

My talk at MIT Sloan School on simulations and science

Here is the video, a bit short of an hour, you may recall this was my abstract:

What happens when a simulated system becomes more real than the system itself?  Will the internet become “more real” than the world of ideas it is mirroring? Do we academics live in a simulacra?  If the “alt right” exists mainly on the internet, does that make it more or less powerful?  Do all innovations improve system quality, and if so why is a lot of food worse than before and home design was better in 1910-1930?  How does the world of ideas fit into this picture?

By the end I considered whether we might make science better by first making it “worse.”  I also covered The Phantom Tyler Cowen, and whether attempted refutation is the best way to approach a new idea.

Authoritarian gridlock

Legislative gridlock is often viewed as a uniquely democratic phenomenon. The institutional checks and balances that produce gridlock are absent from authoritarian systems, leading many observers to romanticize “authoritarian efficiency” and policy dynamism. A unique data set from the Chinese case demonstrates that authoritarian regimes can have trouble passing laws and changing policies—48% of laws are not passed within the period specified in legislative plans, and about 12% of laws take more than 10 years to pass. This article develops a theory that relates variation in legislative outcomes to the absence of division within the ruling coalition and citizen attention shocks. Qualitative analysis of China’s Food Safety Law, coupled with shadow case studies of two other laws, illustrates the plausibility of the theoretical mechanisms. Division and public opinion play decisive roles in authoritarian legislative processes.

That is from Rory Truex, via the excellent Kevin Lewis.

What will it take to reduce Bay Area housing costs?

2016 academic analysis by David Albouy, Gabriel Ehrlich and Yingyi Liu estimated that, in general, rents decrease by 3 percent for each 2 percent increase in the housing stock. (This estimate is close to the estimate of a lengthy blog post analysis at Experimental Geography, done two years ago, looking specifically at San Francisco’s history over the last six decades.)

If our goal is to reduce the average market-rate apartment rent to 27.5 percent of median household income (the midpoint between the 25-30 percent range that is normal), that means reducing the rent from $43,200 to $24,895, a 42.4 percent reduction. Using our ratio of a 2 percent housing stock increase leading to a 3 percent decrease in rents, that means, keeping all else equal, the Bay Area would theoretically need to increase the number of housing units overnight by 28.3 percent. (Let’s round up to 30 percent to make the subsequent calculations more intuitive).

…For example, if it takes 20 years to make up our housing deficit, and underlying trend growth for the U.S. population is 0.7 percent per year (15 percent over 20 years), and the average household size remains 2.3 persons, then the Bay Area will need to grow households 30 percent more than the amount of households needed to accommodate trend U.S. population growth (i.e. 30 percent more than the underlying 15 percent population growth), for a total growth of housing stock of approximately 50 percent over 20 years.

Let’s state it plainly: The Bay Area must increase its total housing stock by 50 percent over the next 20 years to bring affordability down to a reasonable level.

That is from the excellent Patrick Wolff.