When a state raises its income tax rate, evidence indicates that some taxpayers do leave the state in that concurrent year. Even though it is reasonable to presume that it takes time for household to respond to an income tax rate change as migration entails many costs and friction, things do appear to change quickly for some taxpayers. Tax laws are not changed in the dark. These are public events and high wealth citizens are usually well informed. The proposed tax rate changes would be reported in local newspapers before they were voted on by state legislatures. Some households probably prepare for such a turn of events while changes to the tax law are being debated. However, transplanting oneself to Florida or other low tax rate state goes beyond getting a new driver license. According to the Wall Street Journal (January 8, 2020) the state of New York conducted 15,122 residency audits form 2013 to 2017. Of those audits 52% of New Yorkers lost their case costing them an average $144,000 in additional taxes and penalties. Having said that, how many taxpaying households are we talking about leaving when a state raises its marginal personal income tax rate on the highest income earners? We look at the historical experience of three states and also four pieces of academic research. The academic research is important because econometric techniques allow the researcher to hold everything else constant.
We start first with IRS data for three states. The IRS has kept net-outmigration data by state for seven income cohorts since 2012. Our data on net-outmigration comes from this IRS state migration database for the years 2012-2018.
We are especially concerned about households with incomes greater than $200K because they will be impacted the most by an income tax rate change. The IRS further subsets their state migration data by age cohorts. We have restricted our work to ages less than 65 years for this $200K+ income bracket. We are not interested in retirees. Essentially, we are trying to understand net-outmigration rate for the very highest income earners in a state during their prime working ages during the year concurrent to the tax increase.
How many households leave in response solely to an income tax rate increase? In this initial experiment using actual state data, we are not able to hold other reasons constant (like the relative strengths of other state economies, weather, and crime). The data does not for allow for us to isolate these factors. We have to turn to more formal academic research, which we do below, to isolate the impact. Here we just note what happened in the year of the income tax rate increase to the number of high-income taxpayers ages 45-64. In addition to the reasons cited above, some high-income individuals work for a company and get transferred to another city. This migration would show up in our net-outmigration data, but obviously is not driven by direct economic forces. Also, here we do not consider cases when a state lowered its income tax rate because evidence shows that the response is much more muted in that initial year.
Table 1 show the three states that raised their income tax rate during the period 2012-2018. We have just these three cases when a state has raised its income tax rate in the time period that overlaps the IRS data on net-outmigration. Table 1 shows our three cases. Column three of Table 1 shows the income tax rate change. Column four shows the stock of high-income-taxpayers for all age cohorts in the state in the year prior to tax rate increase passage. Column five shows the response (the change) in the number of high-income taxpayers in year concurrent to the high tax rate. Column 6 is the percentage change relative to the stock. Column 7 shows the percentage change for a given tax rate increase – the relative response (an elasticity). This column translates the percentage changes in Column 6 into a uniform relative response rate given a 1% tax rate increase. The table shows a large range of relative responsiveness. Taxpayers in the states of Maryland and Connecticut exhibited a larger response than did taxpayers in Minnesota. A reasonable initial hypothesis is that taxpayers in those two states had more options about where to move than similar taxpayers in Minnesota.
A secondary (but equally important issue) is how long did the impact last?
In Chart 1, we show elasticities for Maryland five years after the income tax rate increase. By 2018, the 25 basis point income tax rate increase that Maryland had instituted in 2013 had been in effect for 6 years. The first bar in Chart 1 is the elasticity of 1.11 which we showed in the table above. Table 1 also shows the data behind the 1.11. An additional 395 high-income households left the state in 2013. This 395 was on top of the 501 high-income households which had left the state in the year before the income tax rate increases (the year 2012). The total number of high-income net-outmigrants in 2013 had thus jumped to 896. In 2012, the year before the income tax rate increase, the total number of high-income taxpayers in the state was 142,970. We use this stock of high-income taxpayers as the denomination for all six periods. The 501 divided by the 142,970 relative to a 25 bps increase in the income tax rate yields the elasticity of 1.11. Again, people leave a state for all kinds of reasons and we make no attempt here to keep other things constant. So the elasticity of 1.11 applies to people leaving for all reasons, but this change did occur after the state’s income tax rate went up by 25 basis points. This large jump is noteworthy. Some group of high-income taxpayers must have been making plans to leave as legislation was being enacted. We do see large jumps in the initial year for other states.
There is an additional, equally important, piece of information in Chart 1: The impact of the income tax increase lasted into the second year. The elasticity of 0.86 is pretty close to the 1.11 of year 0. We see that after one full year of the tax rate increase being in effect, the net-outmigration pick-up over year t-1 is almost as high as it was in the year the tax increase was implemented. However, in the second full year after the tax increase, net-outmigration fell back down close to the 2012 amount of 501 households (again, 2012 is year = t-1 for Maryland). But then in 2016, the third period after the income tax rate increase, net-outmigration jumps back close to the level of year 0 with an elasticity of 0.92. Burnout in the sense that net-outmigration is lower in a given year t than in year 0 did occurred in year 2, but then net-outmigration picked up again. It bounced around, but stayed relatively high constant amount in 2016. In 2017 and 2018 net-outmigration in Maryland nearly triples. The 2017 elasticity is 2.77. It could be that it took households three years to respond because of the complexities of leaving. The takeaway being that in this one example, the net-outmigration bump from an income tax rate increase does slow to a burnout stage but then accelerates after 4 years from time 0.
In 2012, Connecticut continued with its series of income tax rate increases by raising its marginal income tax rate on the highest income tax bracket by 20 basis points to 6.70 percent. So it would not be true that all of the net-outmigration in the years following 2011 would have come from that 20 basis point increase. Nonetheless, net-outmigration jumped by 15% to 583 high-income taxpayers in 2012 from 440 in year t-1 (a 0.71 elasticity). The following year, net-outmigration rose to 646 high-income taxpayers. So the impact was felt even stronger the year after the tax increase. Using the same denominator and the same income tax rate changes as we did for year 0 yields an elasticity of 1.02 in year 1 (close to Maryland’s year 0 elasticity of 1.11). With the income tax rate still at 6.70 percent, in the second full year after the tax increase, net-outmigration count rose to 669. So for Connecticut we are seeing the elasticity rising again in the second years the tax rate was in place. Finally, in the third full year after the tax rate increase, net-outmigration was only 515. This is still higher than the year before the tax increase. The elasticity in year 3 is 0.37. The important point here is that the impact of an
income tax rate increase grew for two additional years after the tax rate change. Burnout in the sense that net-outmigration is lower in year t than year 0 occurred in year 3. In 2016 (year 4 above), Connecticut raised its marginal income tax rate on high-income earners another 29 bps. In that same year net-outmigration jumped again to 788 taxpayers. This is an elasticity of 0.70. In year 5, after both income tax rate increases (totaling a combined 49 bps) net-outmigration spikes with the elasticity reaching 1.67.
It is interesting to note that over the seven years in Chart 2, the state lost about 2.4% of its highest income earners (using the 2011 count of high-income earners as the denominator). It is true that the state continued to mint new high-income taxpayers, so losing some of its highest income earners may not have presented the state with a tax revenue problem. Nonetheless, a horse race between how fast the state can mint new high-income taxpayers and fast it loses them is being played out in the state every year. In fact, the year-over-year (FY19/FY18) change in personal income tax revenue collections for Connecticut were negative in FY19 indicating that these high personal income tax rates and the 2017 Tax Cuts and Jobs Act have pushed Connecticut past this tipping point.
In 2013 Minnesota raised its marginal income tax rate by 2 percent. Net-outmigration of
high-income tax payers jumped by 241 taxpayer. The elasticity was a muted 0.12 and remained at that level for the six years up to and including 2018 (Chart 2). So the impact stayed roughly constant – it did not burn itself out, nor accelerate. The response in Minnesota was about 1/10 the response in Maryland and Connecticut. The weak net-migration response maybe related to Minneapolis, the state’s major city, being far removed from good alternative places to live.
A problem with looking at the actual IRS state migration data is that although we know that the income tax rate changed in a given year, other things changed also. Here other things are not held constant.
Several academic research studies attempt to establish the relationship between income tax rate increases and net-outmigration using data on other non U.S. geographies for very high income earners. All of the studies calculate elasticities with respect to a 1 percent income tax rate change, so we can compare their results to the IRS data. Akcigit et al. (2015) find elasticities for the number of domestic and foreign inventors with respect to personal income tax rate equal to 0.03 and 1. Moretti and Wilson (2017) find large, stable, and precisely estimated effects of personal taxes on “star” scientists’ migration patterns. They calculate a flow elasticity whereby a 1% increase in income due to a personal income tax rate cut increases net-outmigration by 0.4% per year, each year. This is an elasticity of 0.40. More recently using U.S. data, Giroud and Rauh (2019) answer the same question (do state income tax rates impact location choice) by looking at how state tax rates impact counts of S corporations. They find that a 1% increase (decrease) in the statutory personal income tax rate corresponds to a 0.12% decrease (increase) in the
number of employees belonging to pass-through firms, an elasticity of 0.12. Lastly, Haidorfer and Sussman (2020) using the same IRS data as in Charts 1 to 3 on taxpayers with incomes greater than $200K for ages less than 65 find that a 1% higher state income tax rate increases yearly net-outmigration by 110 high-income earners per 100,000 high-income taxpayers in the first year. This is an elasticity of 0.11.
These academic studies use cross sectional data and differing tax rates at a given point in time for several different geographies. They do not focus on an individual state to see how net-migration changes over time as we do in Charts 1 through 3. These results are in Chart 4 (All papers are on the CHRR site).
There are several points to be made here:
Current research show that a statistically significant number of high-income taxpayers vote with their feet -- taxes do impact where people choose to work and live.
Each states has its own elasticity. The precise amount varies quite a bit among states and seems to depend on how many good alternative geographical locations are available to taxpayers. The more geographic alternatives that taxpayers have for moving, the higher will be the state’s elasticity.
A paucity of data does not allow for a direct calculation of the impact of a rate change over time. Does net outmigration accelerate, stay constant or burn itself out? The IRS data presented above suggests that the impact of a tax rate increase on net-outmigration holds steady, or rises slowly, for 3 years, burns itself out, but then picks up again.
Econometric analysis for 50 states and Washington, DC indicate that elasticities from research are independent of the initial income tax rate that the state starts from. In the parlance of economics, the specification is linear – the response of taxpayers to an income tax rate increase of 1% is the same if the state has a very low income tax rate, or a very high income tax rate. This result is not very intuitive.