[ I came across Lyman Stone on Twitter, where he has interesting insights to share on migration. Stone, who is an economist for a US government agency, in his personal time blogs on migration issues at Medium and is well worth a read. To give you a flavor of what you’ll find, he graciously allowed me to repost the piece below – Aaron. ]
What Changed in the New IRS SOI Data?
I’m just going to put up one chart today. The IRS recently began publishing its Statistics of Income migration data again. However, the new SOI is much changed, and much improved. It includes a bunch of new migrant files that weren’t included in the old version, due to greatly improved methods of assembling the data. The result is a slightly higher estimated migration rate (more in line with ACS). The biggest changes occur for high-income groups. That’s neither here nor there for my purposes, but there are a lot of academic papers out there that use IRS SOI that may need to be recalculated now that the number of high-earners in the data just took a big step upwards. The point is, the new and old data can’t really be linked together in a very meaningful way.
But that’s no big deal, because the new data is far and away better than the old data. Aside from being more complete as I mentioned above, the new IRS SOI data actually gives us much more complete information than the data from 1990–2011. For 2011–2012 and 2012–2013, we now have inflows and outflows for every state by age and income bracket of the head of household. By the way, that’s age crossed by income. So we can estimate migration rates for a given age group, for a given income group, or for a given age-income group. Very fancy. This is the only place you can find this detailed of migration flows crossed by two variables without either delving into microdata or similarly labor-intensive processes. Want to know how many people earning between $50,000 and $75,000 over the course of the year between the ages of 45 and 55 moved to Alabama from 2012 to 2013? Easy: 898. But they claimed 1,934 exemptions, for an average tax-household size of 2.15 people. Oh, and the average AGI per tax returnin this group was $61,476. Pretty neat stuff; much more than we could do with the old SOI data.
But there’s another useful trait to the IRS data. ACS data cuts out at either $75,000 or $100,000 in income per year: very much in the middle class. We can’t use ACS to say very much about high-earners. But IRS SOI data offers a breakout for $200,000 and over in annual income. In other words, its wider range of income brackets allows us to make more interesting conclusions about money and migration.
Exploring IRS Migration Data: The Young, the Old, and the Rich
Click to enlarge:
The above chart shows one set of columns for each age range and, within that, one column for each income group within that age range. So, for example, the tallest bar is about 6.7%, for individuals under 26 filing tax returns with AGI over $200,000. These are the most mobile people. The next most mobile are 26–35 year olds with AGI under $10,000. For every group other than those under 26 and those over 65, the <$10k bracket has the highest migration.
Age, or the migration life cycle, matters more than income.
What we can see quite clearly is that, while very high incomes and very low incomes do impact migration rates, incomes from $10,000 to $199,000 have very similar migration rates for most age groups. The outliers are essentially the superstars and the unemployed.
Within each age group, variation tends to be about a percentage point; for the high-migration groups it can be more. But migration varies by about 3 percentage points across the age groups. In other words, likelihood of migration has more to do with stage of life than with income.< Now, that said, money does seem to matter, especially for younger migrants, and especially if you’re rich or poor. Anywhere in the middle and you’re fine. For low-income people migration becomes more probable as unemployment is likely and the rewards to leaving are much more likely to outweigh the risks. An active job search may also make it more likely a person hears about employment opportunities somewhere else. On the other hand, a person who loses their job may migrate somewhat less aspirationally, moving in elsewhere with friends or family. Across age groups, migration is lowest for middle-class earners.
For the rich, migration may be more likely for several reasons. First, tax filers making over $200,000 per year may be more likely to derive a substantial share of their income from financial assets, allowing flexibility of residence. Financial security may also make them more able to finance the costs of migration without much concern, making the risk-reward calculation of migration more favorable.
I’ll continue exploring the new IRS data as more of it becomes available; they should be releasing the 2013/2014 data within the next month or so. Improved IRS SOI data is cause for celebration: this provides another baseline, alongside CPS and ACS, of now much-improved quality. While historical comparisons to IRS data may not be entirely appropriate, the new series offers robust new information not previously available to most users. And this new information reveals at least one interesting trend: migration falls with age, and is lowest at all ages for middle-income people. The control for age and income allows a specificity not available in other readily available data products. Migration is higher for the very low-income, and for high-income individuals, and these effects are most pronounced for young people.
This post originally appeared at In a State of Migration on August 24, 2015.