Analysis D: Child heat full distributional exposure change over time

[FanWangEcon]

We study the distribution of temperature exposure for children, and examine how this shifts over time. We focus on children with several alternative definitions for age ranges. We compare temperature and demographic data between 1990 and 2020.

Key Statistics

Based on daily or hourly temperature in a particular calendar year, we compute the share of children exposed to $>x$ degree of temperature. We do this for a grid of temperature from very low (e.g., -40) to very high (e.g. +40).

Note that this is under the assumption that children do not move. We know the number of children in county $z$ in year $y$, and we know the hourly temperature from ERA5 for various grid-cells within that county during the course of the year. It would be nice if we knew also the number of children in each grid-cell, but we don't. To match population data to temperature data, we:

1. Find the grid cells that match with each county in year $y$. Note that the county boundaries shift over time, and the number of counties differ each year, but that does not impact our analysis because we are not trying to compare county across time, but using the county as discrete elements for our discrete probability mass distribution of children.
2. For each hour (or alternatively for each day), we take the average temperature across cells matched to each county
3. For each county, now we have an average hourly or daily temperature. We then compute the share of hours (or days) in a year temperature is above degree x.
4. At very low temperature, the share of hours (or days) in a year above the low temperature level, say $-40$, is approximately 100 percent. As we move to higher temperature, the share of hours (or days) in a year below a very high temperature level say $40$ is approximately 0 percent. As we shift from low to high temperature, we could trace out a CDF, the distribution of temperature exposure for each city across hours (days).
5. Now suppose we have two cities, one has 1 child, the other has 2 children. Suppose 20 percent and 50 percent of hours in each city is over $20$ degrees in a particular year. We can construct a distribution for the number of children exposed to different shares.