Where have all the banks gone? IAMECON wins another federal research grant
October 12, 2019
Where did the PPP Loans Go?
May 5, 2020
I have been a diligent advocate for dense cities in my Urban Economics lectures at UT Austin, since 2015. Finally, I am proven wrong by COVID-19.
Our team collected county level statistics (N=1103) from John Hopkins University along with other Census sources, in order to statistically assess the factors determining the growth rate of daily cases from one county to another. Our preliminary findings indicate that (not surprisingly but statistically significantly):
– Spread rate is increasing with density, at a decreasing rate
– Spread rate is lower in more rural areas
– Spread rate is increasing with the population percentage of 15 to 39 year olds
– Spread rate is decreasing with duration of exposure, at an increasing rate
We defined the spread rate as the compound average daily growth of cases since the first reported case.
A state level comparison of average growth rates (average of county growth rates for each state) shows the nature of the relationship with population density: “increasing at a decreasing rate”
As one would recognize, one needs to jointly analyze the effects of these factors on the corona spread rates in order to decompose the relationship. In order to do so, we have ran a cross-sectional regression at the county level data, and our main findings are as follows and the dataset and the code used for estimation is available here:
County Level Regression results – Dependent variable Daily Average GRowth Rate of Covid-19 Cases
| (1) Spread Rate Exposure > 14 days | (2) Spread Rate Exposure between 14 and 31 days | (3) Spread Rate Exposure > 14 days &Density < 1000 (pop /square mile) | |
|---|---|---|---|
| Density | 0.001*** (0.000) | 0.003*** (0.001) | 0.035*** (0.004) |
| Density^2 | -0.000*** (0.000) | -0.000*** (0.000) | -0.000*** (0.000) |
| RUCC_2013 | -1.643*** (0.154) | -1.529*** (0.151) | -0.719*** (0.149) |
| Days since first reported case | -0.351*** (0.098) | -0.067 (0.457) | -0.809*** (0.093) |
| Days since first reported case ^2 | 0.000 (0.001) | -0.008 (0.011) | 0.005*** (0.001) |
| Pop (ages 15 to 39) % | 0.197* (0.055) | 0.172* (0.060) | 0.118** (0.057) |
| Constant | 24.012* (3.210) | 22.675* (5.743) | 28.216*** (3.009) |
| Observations. | 1103 | 1065 | 981 |
| R-squared | 0.479 | 0.490 | 0.535 |
| Standard errors are in parenthesis *** p<0.01, ** p<0.05, * p<0.1 |
To conclude, “rurality”, “population density” and “people of ages 15 – 39” are the three major determinants of how fast COVID-19 has been spreading in the community. Of course, we currently lack a very important component, the prevalence of “social distancing” in each county, which we currently control for using the population percentage of people ages 15 to 39.
We are in the process of creating a custom measurement system for community enforcement of social distancing rules, and will update soon.
