Measuring Urban Government Productivity Using Public Big Data

In this Big Data age, what do we now know about urban government productivity?  Whether the city is Baltimore or Chicago or San Diego, when a city spends an extra dollar on public goods such as street safety or public transit, how much local public goods are produced?  This is a surprisingly difficult question to answer and in the next series of blog posts, I will argue that this “opacity” benefits several interest groups including the Mayor, the urban government and the public sector unions.

Back in 2017, I co-authored a paper that was published in the Journal of Public Economics that investigated the productivity of the public transit bus sector in over 100 cities over a 20 year period.

A more productive public sector has a lower average cost per bus mile.   There is a simple Leontief production function producing bus miles.  The inputs are;  a bus, some gasoline to drive the mile and a driver. Given that the bus and gasoline are sold on world markets, there is a single price for those inputs and the only source of variation in productivity is the price of labor.  I believe that quality variation is small here. A bus driver is a bus driver.  We document that where public sector unions are more powerful that average costs of service provision are higher.    We document that when cities replace a progressive mayor with a Republican mayor that these costs decline.  Urban politics matters in determining the price of local public goods.  Progressive cities face higher prices for local public goods.

The open question here is what is the extra quality and quantity of local public goods produced when an extra $ is spent?  Does the answer differ across cities with weaker and stronger local unions? Does the answer differ depending on the city’s demographics?  For street safety and fire protection and public education, what is the marginal product per dollar spent?  Rick Hanushek’s work offers one perspective and recent work by Bo Jackson and co-authors  offers an alternative perspective.

Today, cities are publishing data on teacher salaries and school level test scores.  Cities are publishing data on total pay (salaries and overtime) for each fireman in each year.     Read our 2019 report studying public pay in Baltimore, Boston and NYC using public Big Data.  

From such data, can economists estimate production functions of local public goods to infer the marginal product and average product of these public employees per dollar spent?  (the bang per buck condition familiar from intermediate microeconomics).  I claim that we cannot and that this creates “opacity” as economists cannot judge whether taxpayers are getting a good deal when the local government devotes tax $ to greater public employment.

In the next series of blog posts, I will sketch out additional data that urban governments should post to help to build confidence that urban governments are delivering quality service and to be held accountable to skeptical tax payers (and suburbanites considering moving to the cities).

The theme of these blog posts will focus on how to build trust in urban government through transparency and accountability.   Analytics are being used in pro-sports (Money Ball) in benchmarking academics (Google Scholar) and in business (Amazon is a trillion $ company because of its use of analytics).  I will propose a research endeavor to slightly disrupt business as usual in the urban public sector. 

For PHD economists, the research agenda here is to measure a structural local public goods production function in the presence of a heterogeneous production function and the possibility of x-inefficiency such that the equilibrium L/K ratio lies inside the production possibility frontier.