(1) Some places are good at generating wealth due to a combination of natural amenities and other fixed assets, e.g., talent agglomerations rooted in economic decisions made centuries ago. Amenity-rich state like New Jersey or California really can function as “stationary bandits,” at least for a while: affluent citizens won’t make the decision to leave such states lightly. The trouble is that amenities rooted in things other than climate and natural beauty can erode over time.
New Jersey, for example, fares poorly on the Kauffman Index of Entrepreneurial Activity by State. This could be true for many reasons, including the high cost of living. There is no clear correlation between tax levels and the level of entrepreneurial activity, e.g., California has a steeply progressive tax system, yet it has a fairly high level of entrepreneurial activity. Does this mean that other states can and should adopt California’s tax policies? The danger there is that California can “afford” its tax policies only because of very unusual, hard-to-replicate factors, like the self-reinforcing concentration of VCs and talent in Silicon Valley. It’s not clear that California’s tax policies are building on this advantage rather than undermining it.
Which leads us back to Young and Varner:
The millionaire tax is a major policy experiment both in the context of the New Jersey tax system, and in the context of what other states were doing during this time (which, for the most part, were cutting rather than raising top income taxes). Another factor that makes New Jersey an ideal case study is its unique location as a core part of the New York metropolitan area, which includes 21 million residents across four states: New York, New Jersey, Connecticut, and Pennsylvania. Figure 3 shows the geography of the tax systems in this area. On one hand, being situated across the river from Manhattan — with its cultural draws and high-paying employers — might seem to support New Jersey’s ability to tax the rich. Yet, the wealthy from New Jersey have three other state tax systems they can arbitrage without leaving the New York metropolitan area. For example, high earners living in Bergen County, New Jersey, can move about 30 miles to Fairﬁeld County, Connecticut, and watch their marginal tax rate fall from 8.97 percent to 5 percent. Few other places in the country make it easier to move to a different state without leaving one’s city or completely separating from the social ties of friends and family. This jurisdictional proximity would suggest unusually intense regional tax competition. Nevertheless, large differences remain in top marginal tax rates. [Emphasis added]
This raises an interesting question: what has happened to housing prices in Fairfield County, Connecticut as opposed to Bergen County, New Jersey over this period of time? Have we seen greater creation of new millionaires in southwestern Connecticut than in northern New Jersey? Indeed, if Young and Varner were looking for a natural experiment, one wonders why they didn’t consider comparing high-earners in New Jersey to their counterparts in Connecticut rather than rely on less-affluent households as their control group. There are many confounding variables in both cases. Yet I think it’s safe to say that if the creation of new millionaires proceeded at a markedly faster pace in Connecticut, which saw a smaller tax increase over the 2000-2007 period, we’d have an interesting result.
Stamford has emerged as an important secondary financial center in the New York metropolitan area over the last decade, and one wonders if Connecticut has widened its lead as the managers of hedge funds have decided that Connecticut would be a more desirable environment for upwardly-mobile for financial professionals. Might New Jersey have proven more competitive? Again, the question isn’t just about retaining current residents, but about attracting footloose residents from other states, e.g., affluent New York city residents who decide to move to the suburbs.
The story of Stamford reminds me of a recent post by Tino Sanandaji:
When you put a lot of skilled people together in knowledge industries and entice them to work hard, productivity appears to go up exponentially. This is what we observe in cities such as New York, and in elite organizations such as Mckinsey, Google or Harvard. These high-skilled individuals and organizations create a great deal of so-called innovation spillovers. The people who developed the microprocessor thus only captured a small fraction of the societal value created as private wealth. This is one reason I believe the standard estimates underestimate the negative effects of taxes on the economy, since they neither include innovation and knowledge spillovers or the effect skilled people have on the productivity of other skilled people.
Could it be that Connecticut’s tax policies facilitate the rise of a talent agglomeration that has yielded significant economic dividends for the state? The rise of Stamford has presumably generated employment for workers with complementary skills, among other things.
(2) New Jersey hasn’t developed a financial center comparable to Stamford. Rather, it continues to rely heavily on commuters who work in New York city, a fact that has important tax implications:
For New Jersey residents who work in New York, the new millionaire tax would in many cases have little effect on their income tax bill. Although New Jersey added a higher top rate (8.97 percent), effective rates remained lower than New York State effective rates for many top earners. This is because New York rates are higher than New Jersey rates in the lower income brackets and because the New York “claw back” feature applies the top rate (which ranged from 7.7 to 6.85 percent during the 2004–2007 period) to all income for sufﬁ ciently high earners, as the beneﬁ ts of lower bracket rates are phased out. Thus, New York State’s effective tax rate was higher than New Jersey’s on joint ﬁ lers’ taxable income between $0 and $1.34 million (2004–2005) and between $0 and $804,000 (2006–2007).9 We cannot observe in our data who works in New York speciﬁ cally, but we can identify people who have employment earnings from outside New Jersey. This allows us to estimate the behavioral response among people who deﬁ nitively pay all their tax in New Jersey.
Finally, it is important to note that earnings from investments (which make up about 30 percent of income in our data set) are taxed only where people live. For a substantial portion of the income of wealthy taxpayers, location of employment is not a constraint on tax competition. [Emphasis added.]
This is very interesting and important. It suggests that what we want to pay careful attention to the impact of the millionaire’s tax on migration for households that work (and pay tax) entirely in-state and people who rely heavily on investment income. Let’s keep this in mind.
New Jersey experienced net out-migration of millionaires in every year from 2000–2007, averaging a net outﬂ ow of 459 per year, or 1.2 percent of the state’s millionaires. Nevertheless, the stock of millionaires increased substantially over this period, rising 43 percent (from about 33,000 in 2002 to about 47,000 in 2006). New Jersey is a producer of millionaires, not an importer. At current trends, migration is not an important determinant of the stock of millionaires in the state. Changes in net migration represent only 3.3 percent of the typical year-to-year ﬂuctuation (standard deviation) in the number of millionaires in New Jersey. Income dynamics, rather than migration trends, are the central force that determines the tax base for the millionaire bracket. Migration, nonetheless, has certainly increased since the new tax was introduced in 2004. In the pre-tax period (2000–2003), net out-migration averaged 9.3 per thousand millionaires. Since the new tax was imposed, net out-migration has risen to 14.5 per thousand, an increase of 56 percent. Yet, as the baseline migration rates are so low, the impact on the stock of millionaires is very small. The increase represents a total loss of 5.2 households per thousand.
This initial analysis offers two basic conclusions: (1) migration has a small impact on the millionaire tax base, accounting for only 3 percent of the variation in the number of millionaires each year; and (2) there is an observable increase in migration associated with the introduction of the new millionaire tax bracket[Emphasis added.]
Again, wouldn’t it be helpful to know about Connecticut’s performance on this score relative to New Jersey, particularly as it relates to residents that work (and pay tax) entirely in-state? The 2000-2007 period saw a sharp increase in the stock of millionaires across the region.
Then there is following, which merits close attention from policymakers interest in the lessons for state governments:
Model 5 focuses on people who earn all of their wages in New Jersey, and thus paid all of their state income tax in-state. This group represents 62 percent of the overall sample (those earning more than $200,000), meaning that 38 percent earned some of their wages out of state.15 Model 5 shows that the DiD estimate for millionaires is 2.5 (semi-elasticity of 0.28) but does not achieve statistical signiﬁ cance. For the top 0.1 percent, the estimate is (a non-signiﬁ cant) 9.8 with an implied semi-elasticity of 0.42. While not statistically signiﬁcant, the ﬁndings have economic signiﬁcance (McClosky and Ziliak, 1996) and are clearly suggestive: in states where fewer people work (and pay income tax) out of state, other things being equal, the effects of a millionaire tax may be larger than in New Jersey. [Emphasis added]
It is possible that a millionaire’s tax will do less harm in states like Maryland and Virginia, where large numbers of high-earners work in a neighboring high-tax jurisdiction, than in other states.
Young and Varner draw a similar conclusion on investment income:
Model 6, in contrast, focuses on people who earn all their income from investments. This group should be particularly sensitive to the tax rate for two reasons: (1) they are not tied to the state by an employer; and (2) like the previous group, all of their income is taxed in New Jersey. The estimates indicate the new tax raised migration by 6.4 per thousand among millionaires and by 27.4 per thousand among the top 0.1 percent of earners. The semi-elasticities are 0.64 and 1.17 respectively, although only the effects for the top 0.1 percent are statistically signiﬁ cant. While these individuals represent a small portion of the overall sample (9 percent), they seem more sensitive to the tax rate, particularly when their incomes are extremely high.
Though the effect is small, it does seem worthy of note.
(3) A few (even more) miscellaneous thoughts: The authors go on to discuss the impact of the housing boom on increasing the amount of out-migration across high-earners above and below the $500,000 mark:
New Jersey’s overheated housing market provides a plausible explanation of why out-migration has increased across the top of the income distribution, regardless of the new tax bracket.
Yet another reason to harp on the evils of the mortgage interest deduction and building restrictions. At the tail end of the paper, the authors mention many confounding variables, e.g., the expectations of workers in their control group, etc. I found this passage particularly interesting:
If the population that earns more than $500,000 in a given year is a continuously changing group, then the millionaire tax targets transitory spikes rather than permanent incomes. In aggregate, it is a tax on the extreme right-tail of the income distribution, but because this group is constantly changing, the tax has only a small effect on any individual’s “permanent” tax rate (i.e., the effective tax rate on their permanent income). This, in turn, means that while high incomes are more highly taxed, there are few speciﬁ c individuals with an incentive to use migration as a way to avoid the tax. In short, the transitory nature of very high incomes improves the efﬁ ciency of millionaire taxes, raising substantial revenues while creating few adverse incentives. Taxes on transitory income spikes are certainly less distortionary (and less migration-inducing) than taxes on permanent incomes.
This is an excellent point. It also reminds us of the danger of relying heavily on a volatile tax base, an issue that Josh Barro has raised.
What is less convincing is the following:
Finally, a possible concern is that our period of analysis — four years before the tax, and four years after the tax — is insufﬁcient to capture the potentially very long term effects on migration of raising the tax rate. We believe that potential migration effects should happen fairly quickly. When a new tax is imposed, people have an incentive to move as quickly as possible. Immediate migration allows movers a longer period over which to amortize the ﬁ xed costs of moving and gives more years to accumulate the beneﬁ ts of a lower tax regime.
As the story of Stamford suggests, the question is whether raising the tax rate has an impact on coordination across high-skilled individuals within firms, as suggested by recent research by Raj Chetty, John Friedman, Tore Olsen, and Luigi Pistaferri. From their abstract:
We show that the effects of taxes on labor supply are shaped by interactions between adjustment costs for workers and hours constraints set by firms. We develop a model in which firms post job offers characterized by an hours requirement and workers pay search costs to find jobs. We present evidence supporting three predictions of this model by analyzing bunching at kinks using Danish tax records. First, larger kinks generate larger taxable income elasticities. Second, kinks that apply to a larger group of workers generate larger elasticities. Third, the distribution of job offers is tailored to match workers aggregate tax preferences in equilibrium. Our results suggest that macro elasticities may be substantially larger than the estimates obtained using standard microeconometric methods. [Emphasis added]
As a policy matter, we should be indifferent to individual high-earning households. We should care a great deal about whether we are undermining the talent spillovers that represent an investment in our collective ability to generate wealth.
Young and Varner have written a very interesting paper. I’m sorry to say that most casual readers of Robert Frank’s post will leave without understanding its implications and the many things the authors miss, due to the limitations of the data.