Illinois, Again, and Why Nets Aren’t Everything
Yesterday I got involved in a conversation with Daniel Kay Hertz and Michael Lucci about migration in Illinois. It started with this post by Daniel, writing for the left-leaning Center for Tax and Budget Accountability.
His key graph is this one:
So let’s parse what this is saying. For every 10,000 residents in NW Indiana, 71 of them moved to Cook County per year. And for every 10,000 residents in Cook County, 18 of them moved to NW Indiana.
Daniel’s interpretation is that this suggests that folks in NW Indiana have a stronger preference for Cook County than folks in Cook County have for NW Indiana. This, despite the fact that when you take the raw net migration, Cook County loses out to NW Indiana. Because Cook County has ~9 times the population of NW Indiana, even its lower relative rate of bilateral outflows means it has negative net flows.
Michael Lucci, of the right-leaning Illinois Policy Institute, also writes a lot of stuff about Illinois migration. His take on Daniel’s post was:
And, in addition, he suggested:
So here’s the problem. We have two very different ways of looking at this data. The question is, who’s right?
Well, as Obi-Wan Kenobi would say, from a certain point of view, both!
Net Migration Matters
Let’s start with Michael’s point. He correctly asserts that Cook County, Chicago, the Chicago MSA, and Illinois generally experience negative net migration. The result of this negative net migration is that these areas experience substantially slower population growth, indeed most recently actual population decline. Because other areas are experiencing growth, these regions have lower “market share,” i.e. represent a shrinking share of national/regional population.
This is all true! And if you want to figure out why a region’s population is shrinking or growing, the net migration balance is an essential piece of that story. Furthermore, negative net migration does usually indicative a relative lack of… something. It’s hard to say what. It may be a relative lack of good amenities vs. other places, or a relative lack of jobs, or a relative lack of good governance, or an excess of taxes. In some sense, net migration does provide a broad indicator of what places people perceive to be falling short. Thus, Michael’s most basic point, that negative net migration suggests Chicago/Cook County/Illinois is somehow falling short, is correct.
But Net Migration Isn’t Enough
Let’s imagine a prison in Cook County is closed and the prisoners are transferred to southern Illinois. The data we have will show this as outflows from Cook County. But surely nobody in their right mind would suggest this is because of tax policy, or bad standard of living, or excessive housing costs, or regulation, or poorly funded schools, or whatever. We’d say it’s because of, well, whatever led to a prison being closed!
Or consider if a large number of non-religious Chicagoans converted to a new religion, and began recruiting spiritual leaders for their congregation from outside the region. How would we model this? No economic model of migration of which I’m aware includes a religious component. There wouldn’t be an observable income shock. This would just be a pure change in consumption preferences among Chicagoans resulting in a sudden demand for people trained at the Non-Chicago Religious Seminary.
Or consider the person who moves to Chicago for a job offer without doing much research then, upon arrival, discovers that holy cow sales taxes are really high and their paycheck doesn’t go as far as they thought. Time to pick up stakes and move on! This person would have a tax-motivated migration… without any change in tax policy! Residence in Chicago allowed them to acquire better information about local tax policy!
In all these cases, net migration balances would wrongly inform us about what’s happening in the region in question. Less-than-voluntary migratory flows (prisoners, but also school-age children, some religious orders, military personnel, etc) certainly tell us little or nothing about amenities and costs in an area. Shocks to migration driven by idiosyncratic preference changes are also hard to see as being driven by specific local costs or amenities. And of course, different migrants have different sets of information available to them when they assess Chicago’s fitness for their preferences.
The way we try to get at this as economists is by assessing “revealed preferences.” So we try to control for things that we don’t consider core regional preferences; that is, things people will almost always prefer or disprefer. So, more jobs should attract people. More crime should repel people. Better schools should attract people. Expensive cost of living should repel people. High wages should attract people. High taxes should repel people. When we control for these, we can get an unexplained residual; idiosyncratic migratory factors for a region.
Some researchers call this the amenity value of a region. But that’s wrong too. Consider a place with a university. It will have high 18 year-old inflows and high 22-year-old outflows. Depending on graduating class sizes, you may get substantial changes in the net migration across these age groups despite literally zero change in the underlying amenities. And since this group has a very high migration rate, this tail can wag the net migration dog.
Some things are amenities to some people but costs to others. Some things are amenities of shifting value over the life cycle. These are complicated effects, but by controlling for migratory sub-groups, we can usually sort this out and get a subgroup-specific residual, so we can do a kind of subgroup-weighted-regional-amenity-score. In practice nobody does this because it’s a pain to calculate, but it’s what researchers should do if they want to be precise.
But even that isn’t quite right. Economists today generally suggest that economic agents respond within certain information parameters. That is, people don’t respond to signals they do not receive. Obvious, but important.
The signals received by an outflow vs. an inflow can vary widely. A person who leaves has already collected large amounts of residency-based information than a person who is just arriving. That is, outflows reflect better-informed preferences than inflows. This is a really nifty quirk of information-based models: inflows are a less reliable indicator of local amenities than outflows! Inflows don’t have as much information about local conditions as outflows, so they can’t be seen as being as reliable of indicators.
This model is now widely used in the economics of migration. Studies of repeat migration, failed migration, iterative migration, etc are now the workhorses of the discipline, and they include explicit or implicit models about information acquisition and preference formation. For a fantastic example of these information-based models of migration which I’ve covered before which also happens to be the best recent paper on why migration is falling, see here.
There’s no really good way to test how much information migrants have. But what we do know is that repeat- and return-migration has fallen in the last two decades more than migration generally, suggesting inflows have been obtaining information which better enables them to avoid dis-preferred locations. Now, this information may also be causing potential migrants to miss out on potentially preferable migrations… but the bigger effect is fewer people moving to places they’re likely to dislike.
The point of this is simple.
If you want to assess what places people “prefer” or not; places they are “voting” for or against via migratory revealed preferences, net migration isn’t actually enough.
Net migration is a non-negotiable starting point. But it is not the ending point. It is a necessary component for explaining revealed preferences, but not a sufficient one.
Gross Rates Matter As Much As Net Rates
Full disclosure: I’ve chatted with Daniel in the past about estimators similar to the one he used, and have advised their use in an Illinois context in the past. I use very similar metrics myself. For example, I showed the net rate as a % of the partner-county vs Fayette County in my recent post about Lexington, KY. I’m a fan of looking at these partner-specific rates as better first-approximations of revealed preferences than just pure net rates.
However, I worry about comparing the two sides of a bilateral flow between two very differently sized places. Imagine I compare the migration rates between NYC and Pike County, KY. It’s almost impossible that a larger share of NYC moves to Pike County than vice versa, because NYC is a much larger share of Pike County’s possible migratory universe than Pike County is of NYC’s possible migratory universe. Basic mathematics imply that NYC will always look more preferable than Pike County no matter what the relative empirics actually suggest.
What we really want to do is pick two locations of about the same size, but say on opposite sides of the IL/IN border. Ideally both should be equidistant from Cook County, have similar commuter flows, similar crime, similar economies, etc.
Daniel suggests that Will County, Illinois may be a good candidate, as it’s a similar size and distance from Cook County as the NW IN counties. And he finds a rate of 77/10k vs 18 for NW IN, suggesting people in Indiana are indeed slightly less likely to move to Cook County. Meanwhile, Will County gets 21/10k movers from Cook, suggesting Cook County folks maybe like Will County a bit better than they like NW IN.
I don’t know if Will County is really comparable to NW Indiana. But this type of comparison is probably a better indicator than raw rates between two very dissimilar places. Or, you could try to account for the potential migratory universe and make adjustments… but that’s a lot of work (interestingly, this is similar to an instrumental variable occasionally used to predict shocks in migration, but I digress).
Can’t We All Just Get Along
There is nowhere in the US with as much politically-charged domestic migration commentary as Illinois. This is probably because it’s a big state drifting into decline, so the impacts are being felt and the stakes are high, and local interests are big enough to fund robust local media and commentary. In such an environment, local politics look more national. Now, I’ve got no dog in this fight, but it does strike me that the major groups looking at migration in Illinois really need to figure out what argument they’re actually having. Are they arguing about how to boost population? How to boost population share? How to boost migration, regardless of total population impact? Does population total matter at all, or do we only care about revealed preferences? What is the question we want to figure out?
There are people in Illinois doing good, serious work on this question, Michael and Daniel not least among them. But it does seem like the partisan lines as they’re drawn mean one side refuses to consider anything besides net migration, and the other is committed to insisting there’s no problem. My worry for Illinois is that these two positions may be durable even as decline continues, as seems likely.
For Cook County, a plausible range of specifications for birth, death, and migration rates puts 2020 population somewhere between 5.08 million and 5.24 million. The midpoint estimate is 5.16 million, lower than the 2010 Census, and reaching that upper end would require some very significant improvements in all components of population change. In other words, continued decline is probable for Cook County and for Chicago, thus any liabilities will be more and more difficult to finance.
For Illinois on the whole, the high estimate for 2020 I come up with is around 12.87 million people, with a low estimate of 12.64, and a midpoint around 12.7. That is, even with a robust turn and recovery beginning in 2017, Illinois will barely eke out population above its 2010 Census population, while the most plausible estimate given current trends and population is probably continuing, if somewhat slower, decline.
Here’s a thought for Illinois. CBTA cannot solve the state’s migration problems. IPI cannot solve the state’s migration problems. And if the strategy for trying to do so involves promoting acrimony and bitter division and adopting an intellectual culture of mutual suspicion and hostility, the result will be more outflows. Migration responds not only to economics, but also to social and cultural norms. Want people to stick around? Maintain a non-toxic public forum. Hold hands and sing kumbaya. Really, advice for people working on these very technical, data-intensive issues in every state: set up a standing lunch date with your regular opponent so you can at least ensure you’re gonna get along personally, avoid ad hominems, and, hey, you may even find there are useful and mutually beneficial exchanges of information.
I know that sounds ridiculous, but it turns out that nobody, not even in Illinois, cares about migration in its own right. It’s a downstream product used as a stick to beat the other side over the head. The problem, however, is that migration is the problem! It’s very negative and stripping Illinois of people! Illinois needs people who see taxes, healthcare, education, regulation, as being downstream from migration, and thus are willing to work on the issue directly, rather than as a proxy war for the budget.
I know I’m sounding all lecturey here but seriously, if there’s any state that should be laser-focused on figuring out migration, it’s Illinois. The state still has more births than deaths, but that will change before you know it. And when it does, the jig is up and if you haven’t solved the problem by then, it’s time to think about which counties to merge and which cities to abandon in their totality. Have fun with that.
Check out my Podcast about the history of American migration.
If you like this post and want to see more research like it, I’d love for you to share it on Twitter or Facebook. Or, just as valuable for me, you can click the recommend button at the bottom of the page. Thanks!
Follow me on Twitter to keep up with what I’m writing and reading. Follow my Medium Collection at In a State of Migration if you want updates when I write new posts. And if you’re writing about migration too, feel free to submit a post to the collection!
I’m a native of Wilmore, Kentucky, a graduate of Transylvania University, and also the George Washington University’s Elliott School. My real job is as an economist at USDA’s Foreign Agricultural Service, where I analyze and forecast cotton market conditions. I’m married to a kickass Kentucky woman named Ruth.
My posts are not endorsed by and do not in any way represent the opinions of the United States government or any branch, department, agency, or division of it. My writing represents exclusively my own opinions. I did not receive any financial support or remuneration from any party for this research.
