I am teaching population analysis to the undergraduates this year which is a first for me and them. We talked about the simple calculation of growth rates and how population change is actually more complicated and subtle than an overall population growth rate. This got us to the idea of turnover, that is just measuring the births, deaths, inflows, and outflows that go into the changing population. Basically just seeing how much is changing compared to the overall population level. So I thought I would have some fun and apply this to North Dakota. This could be especially important given that some North Dakota counties experienced high degrees of population change due to economic circumstances.
The data come from the county-to-county migration estimates for the ACS 5 year 2011 to 2015 data as well as the POPEST data series for 2017, which includes population by county for 2011 to 2015, and the births and deaths data by county as well. I calculated county growth from 2011 to 2015 from the annual population estimates in the traditional fashion. The one difference here is that I expressed that growth as a change per 1,000 people since that is how I express the turnover rate. The standard growth rate map looks like:
Recall the end point here is 2015, so the timing is right at the end of the oil boom when the Bakken region was the driver of many aspects of the state including population. As can be seen no county outside the Bakken area was anywhere near the level of growth seen by that part of the state.
The 5-year ACS flow estimates are for the full five years for each county so I totaled births and deaths for the same five years to make them comparable. Adding all four components gave me an estimate of the total turnover in the population. The resulting map is not really too surprising.
If you were to guess the four counties with the highest turnover it would be Fargo, Grand Forks, Burleigh, and Ward. And you would be correct. The Bakken area had the most growth but that is due in many ways to the smaller size it started from in 2011. In order to control for scale I take these turnover numbers and take them out of 1,000 of population:
As you can see this created a great deal more variability than you might expect but that is really, really exciting. Recall that to get the rate I then divided by the total population over the five years and multiplied by 1,000 to get the turnover per 1,000 population. The churning in the Bakken area is not too much a surprise with the anecdotes of short timers coming in, working a time, and then leaving. Adams and Hettinger counties showing up in the high turnover rate category was a bit of a surprise though. Notice too that Ward and Grand Forks counties are higher in turnover rate than Fargo and Burleigh counties. This corresponds with some of the anecdotes about the difficulties these areas have keeping people over the last several years. Emmons, Kidder, and Slope counties ending up on the low end of the turnover may or may not be a surprise. I will need to think about that a bit more.
When thinking about growth in population, which most areas in North Dakota think about quite a bit, it is really important to realize that growth rates are important and tell part of the story, but not the full story. There are inflows and outflows of population, births and death, such that the composition of the population changes as a result and you need to think carefully about these types of issues when discussing policies. These are the same types of issues and discussion we should be having at the national level right now but are not. Hopefully we can do this better here.
Update (9/5/2018): I changed the title to reflect better the topic of the post.