I estimate that ongoing state tax revenue losses from billionaire departures will run between $3.53 billion and $4.49 billion per year, mostly from forgone individual income tax collections, with smaller effects from sales taxes and economic spillovers. Under standard discount rate assumptions, the net present value of these recurring losses outstrips the one-time revenue the initiative’s drafters project from the wealth tax itself, and runs to several multiples of the revenue estimated by a competing analysis.

You can read the full paper here. Downloads of the paper are appreciated—mostly because I think (hope) it’s worth reading, but also because downloads improve the paper’s visibility for others looking for information on the California wealth tax.

My calculations are specific to California, but a similar effect should be anticipated in any other state that contemplates a wealth tax.

There’s a robust debate about whether billionaires will be able to avoid some or all of the wealth tax by relocating out of state. This is true both for departures before the January 1, 2026 “snapshot” residency date, since the California Franchise Tax Board might conclude that domicile has not shifted based on a totality of factors, and also for departures after that date, as courts could very well revise or strike the residency date in the initiative. In a forthcoming Tax Foundation analysis, I plan to explore the many unresolved questions about California residency for wealth tax purposes.

For ongoing liability under other taxes, however, none of this matters. It only matters that billionaires leave, regardless of whether the move is successful in avoiding wealth tax liability. (Of course, they’re leaving because they justifiably believe that this may eliminate their liability.) Many are unlikely to come back—especially if they believe future wealth taxation is on the horizon—and at some point, they will certainly be domiciled elsewhere, depriving California of tax revenue.

That’s what my analysis explores.

In my paper, I divide California’s 212 billionaires into four groups based on their primary source of wealth: public founder equity, private operating businesses, financial fund management, and diversified holdings. These groups differ on three relevant dimensions: mobility, California-source income retention after departure, and employment and investment spillovers from their departures.

Public company founders are highly mobile and retain almost no California-source income after relocating, but their departures produce little employment spillover, since a publicly traded company will not follow its founder. Private operating business owners are less mobile but generate much larger economic spillovers when they do leave, since their employees, suppliers, and operations sometimes move with them. Fund managers fall in between, as do those with diversified holdings.

Notably, with how heavily concentrated billionaire wealth is at the top, departures by those the most motivated to leave can have an outsized impact.

Most of the empirical literature on wealth taxation comes from Europe. This evidence is valuable but will almost certainly understate behavioral responses at the U.S. state level. Crossing state lines is far easier than changing countries, so migration elasticities ought to be considerably higher. Additionally, European wealth taxes also tend to exempt business assets, meaning that the measured response is substantially that of successful small business owners and professionals rather than the founders of large publicly traded companies. My analysis thus considers a range of migration elasticities, regarding the European experience as the floor.

I invite you to read the full paper.

Separately, I also have an op-ed on the philanthropic implications of billionaire outmigration and a Tax Foundation piece explaining why valuation by founders’ voting interests remains a valid concern.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

In tax year 2025, federal, state, and local tax collections totaled $7.58 trillion, which is $22,182 per capita and $56,247 per household. Twenty-nine percent of U.S. personal income goes to taxes, meaning that, collectively, it takes Americans 106 days to earn enough to remit taxes at all levels. Coincidentally, that date is just around the corner, too: April 16th, one day after Tax Day.

Let’s break that down:

• Individual income taxes raised $3.34 trillion in 2025: $2.74 trillion at the federal level and $592 billion at the state and local level. That’s just under 47 days’ worth of personal income, taking us to February 16th.

• Federal social insurance and retirement taxes come next at $1.77 trillion. That’s a little under 25 days’ worth, which brings us to March 12th.

• Property taxes, which are almost exclusively local (but include taxes on businesses’ machinery and equipment, not just real property), brought in $827 billion (11.5 days), and now we’re to March 24th.

• Sales, excise, and miscellaneous taxes generated $785 billion, with about $600 billion of that coming from state and local sales taxes, adding 11 days and bringing us to April 4th.

• Corporate income taxes, at $603 billion ($424 billion at the federal level), add another 8.5 days and take us through April 12th.

• Tariffs (customs duties) generated $264 billion, up from $79 billion in 2024, adding a little under 4 days and taking us through April 16th, the day when the country will have collectively earned enough to pay all of 2025’s federal, state, and local taxes.

These charts are embeddable for sharing in blog posts or in other online content:

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

• They allow lawmakers to meaningfully commit to future rate relief, providing greater predictability for businesses and individuals engaged in economic decision-making.

• They statutorily prioritize rate relief, making tax cuts, rather than spending growth, the default response to greater fiscal capacity.

• They allow rate reductions to be phased in responsibly, making further reductions contingent on revenues exceeding established benchmarks and giving legislators time to evaluate the sustainability of the ultimate target rate, rather than adopting a large rate cut one fell swoop.

But the appeal and logic of tax triggers can also incentivize lawmakers to wrap largely revenue-insensitive phased tax reductions in “trigger” language, presenting them as revenue-dependent when they really aren’t. Sometimes states can afford a tax cut without triggers, and sometimes lawmakers may genuinely want to cut the size of government along with the tax reduction. But if that’s the case, lawmakers need to be honest with the public, and just as importantly, with themselves. Just because a tax trigger has been created doesn’t mean it’s effective.

Some triggers are ineffective because lawmakers were overly restrictive when designing its conditions, meaning that tax cuts don’t actually trigger even in periods of sustained revenue growth. Others are the opposite: superficially, they seem to make future rate reductions conditional, but in practice, they just act as stepped reductions, or trigger at inopportune times.

This is often inadvertent, as tax trigger design can be confusing. But if policymakers adopt an approach that purports to only reduce rates subject to revenue availability, and then a reduction gets triggered during an economic downturn, that’s bad for everyone: for agencies facing unexpected spending cuts, for lawmakers who might have to vote for a tax increase, and for an electorate that believes, with justification, that it was promised something else.

This is an argument for designing tax triggers correctly, and also for calling out triggers that aren’t working as intended. To that end, I have a new piece on North Carolina’s current revenue triggers, demonstrating how they fail to achieve their objectives, with the potential to trigger rate reductions even in the midst of a recession.

Lawmakers who believe in the pro-growth tax reforms North Carolina has adopted over the past 13 years, which include significant income tax cuts, should want to preserve those gains by guarding against an accidental misstep that could undermine everything they have accomplished.

You can read the piece here.

(I do offer one caveat: I wrote this piece several months ago, but it was only published yesterday, a few days after the release of a revised state revenue forecast. The new numbers wouldn’t materially change the analysis. Had the piece been written using the new data, every argument would remain and the scenarios outlined would be quite similar. But I did want to acknowledge that the piece does not use the latest data, something I may try to address in an update.)

If lawmakers use triggers, they should genuinely be triggers. And if policymakers want to ensure that the progress they make on tax reform and tax relief remains in place, they should want to guard against an overextension that could sour their colleagues on tax reform as a whole.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

The reasons for rising gasoline prices aren’t a secret. The world is experiencing a supply shock. Tinkering with gas taxes won’t reopen the Strait of Hormuz, and it won’t deliver anything close to dollar-for-dollar savings to motorists.

Crude oil supply has been disrupted, and gasoline has become scarcer relative to demand. Prices have risen to ration the available supply and bring the market into equilibrium. The quantity of gasoline demanded at a lower price would outstrip demand.

We know what happens when policymakers set price ceilings during an energy crisis, and thankfully, most lawmakers seem to have learned the lessons of the 1970s. No one wants to revisit that era’s gas shortages.

Suspending a gas tax isn’t an attempt to cap prices, but instead an effort to reduce a price input. But while this lowers one component of the retail price, it does nothing to reduce the underlying scarcity. If the market-clearing price is higher than it was before the conflict with Iran—and it clearly is—then some of the tax reduction will be offset by higher pre-tax prices.

Recent state-level price changes provide evidence of this. The average cost of fuel varies across states for many reasons, including gas taxes, environmental regulations, transportation costs, labor and operating costs, and market competition. During a supply shock, we would expect to see relative price differences decline, with lower-price states experiencing larger percentage increases in price at the pump than states that already had high prices. Market forces drive prices everywhere toward a new equilibrium, regardless of their starting point. Differences will remain, but we’d expect them to be somewhat smaller.

And they are. Using AAA’s daily average gas price data for February 6th (pre-supply shock) and March 16th, I compared initial gas prices with the percentage increase in prices post-supply shock. The effect is robust and in line with expectations.

To get technical for a moment, the coefficient is -0.074, meaning that for every $1 that a state’s gasoline price was above the national average on February 6th, the post-shock price increase was 7.4 percentage points smaller. States below the national average experienced larger percentage increases.

We can measure this convergence in other ways as well. Post-supply shock, relative price gaps shrank. The coefficient of variation fell from 15.3 to 13.2 percent, meaning that interstate price differences narrowed relative to the national price level. This is consistent with the regression results.

Throwing a gas tax holiday into the mix—whether federally or at the state level—won’t yield dollar-for-dollar savings, because marginal price changes are being driven by increased scarcity.

That doesn’t mean there wouldn’t be any savings. But during a supply disruption, the pass-through won’t be anywhere near complete. If the lower (or eliminated) tax induces additional consumption, the increased demand will undercut the intended relief, absorbing some of it into industry profits rather than passing it along to drivers. That’s not because oil producers, refiners, or service stations are being greedy; it’s just the intersection of supply and demand curves.

Prices are an allocation mechanism. During a supply shortage, efforts to hold down prices without addressing supply are unlikely to work as well as hoped. Motorists won’t get anywhere near as much relief as they expect, while transportation funds are deprived of revenue for road maintenance. It’s no surprise that gas tax holidays are proposed during periods of greater oil scarcity, but the problem that prompts the policy fix is also the reason the policy fix will underdeliver.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

Holmes’s line about taxes as the price of civilization is often quoted, but many are unaware of the context, from a case that hinged on whether an imposition was a tax or a penalty. “Taxes are what we pay for civilized society … A penalty, on the other hand, is intended altogether to prevent the thing punished.”1

All policy involves tradeoffs, and all things being equal, any tax will tend to produce less of whatever it is levied upon. (“Any tax is a discouragement,” as Justice Holmes wrote in another case.2) For most major taxes, however, this is an unavoidable consequence, not the policy objective. Or at least, that’s largely been true historically.

The left appears increasingly divided on the primary purpose of progressive taxation—whether redistribution is primarily about floors or ceilings.

For one camp, which has dominated real-world progressive policymaking, the basic logic of a progressive, redistributive system is that there is a compelling need for social insurance, housing, education, and health care policies that ensure a sufficient standard of living for all, and that wealthier households have a greater ability to pay for these programs and can thus be expected to contribute more. In this view, wealth is not bad in itself (indeed, it’s necessary for the system to work), it just creates additional obligations to support the welfare of those less well-off.

For another camp, wealth itself is a problem to be solved. New taxes are proposed to address income inequality, with less focus on raising the absolute position of the least fortunate and more focus on closing the relative gap. Taxes are conceived less as the price we all pay for government—even if some are expected to pay much more than others—and more as a penalty on something bad (particularly high incomes or net worth). This doesn’t mean, of course, that those in the latter camp don’t care about the programs that additional revenue funds, but the difference in focus matters.

In one sense, this divide is nothing new. It’s Rawls vs. Dworkin. It’s the difference between instrumental and intrinsic egalitarianism. These debates about equality, egalitarianism, and distributive justice can be found in any textbook. But until recently, a case can be made that most left-leaning lawmakers saw progressive taxes as a means rather than an end: they were, in their view, the fairest way to pay for programs they supported, many of which benefitted lower-income households. They were not intended as punishment of wealth and did not suggest disapprobation.

By contrast, consider the proposed California wealth tax, which is to be imposed on “the activity of sustaining excessive accumulations of wealth” (emphasis mine). Or look at Washington, where lawmakers are considering an income tax that only taxes millionaires, or proposals at the federal level that would wipe out income tax liability for many households (eliminated under $92,000 for married couples under Sen. Van Hollen’s proposal, and under $75,000 with Sen. Booker’s), funded by higher rates on high earners.

Politicians on the right are playing the same game, whether it’s “no tax on tips” or state-level proposals to eliminate income taxes for the majority of households. There’s an acknowledgment embedded in each of these proposals, sometimes implicitly and sometimes explicitly, that taxes are viewed as penalties, and that “deserving” people shouldn’t pay taxes.

It’s hard to imagine how this is supposed to work in the long run. There are economic objections: extraordinarily high rates on marginal income and capital accumulation are uniquely harmful. There are stability objections: over-indexing on high-net-worth individuals introduces greater volatility (capital gains realizations are far more volatile than wage income, for instance), which is poorly calibrated to cover recurring, essential expenditures. But there’s also a policy incoherence that those on both the left and right may eventually rue.

For those on the right, pursuing policies that exempt many voters from formerly broad-based taxes will make it harder to keep rates on remaining taxpayers in check. It’s surprising that more conservatives aren’t wary of this, and perhaps less surprising that many progressives are okay with it.

But for the left, convincing voters that middle-class families shouldn’t pay taxes is playing with fire. There’s no realistic level of tax on the wealthiest households that can fund their preferred size of government—or indeed our current size of government—for very long if taxes are reimagined as a penalty on the rich. That’s especially true if the actual stated goal of some (not all) of these proposals is to curb “extreme” or “excess” wealth. The traditional progressive focus on wealthy taxpayers disproportionately funding government is premised on the idea that we will continue to have—and be delighted to have—wealthy taxpayers.

At the risk of oversimplification, the Scandinavian model of progressivity involves broad-based taxes with highly progressive expenditures, while the American model accomplishes more redistribution through taxes themselves, with much more progressive taxes but somewhat less progressive spending.

A Tax Foundation analysis found that the bottom quintile receives $6.17 in transfers per dollar paid in federal, state, and local taxes, while the top quintile receives $0.11. We can also look at this as “effective fiscal incidence,” essentially the net of tax and transfers. If a household paid 10% of their income in taxes but received redistributive transfers worth 30% of their pretax income, they net +20%. The Tax Foundation calculates that the bottom quintile nets +127%, the third quintile nets -2%, and the top quintile nets -31%.

But notably, while this means that about half of households are net beneficiaries under the current tax-and-transfer system, it doesn’t mean that the average household isn’t paying taxes. The Tax Policy Center estimates that the average total federal tax rate for the middle quintile is 7.5%, even though that quintile’s effective individual income tax rate was negative.

In the short run, some progressives might view the adoption of a Van Hollen-style proposal as a victory. Likewise, in the short run, some conservatives might be excited to carve up state income taxes by progressively zeroing out liability for many households. But in the long run, none of this seems sustainable.

Sen. Russell Long, who chaired the Senate Finance Committee for fifteen years, summed up the basic philosophy behind this populist turn for both the left and the right: “Don’t tax you, don’t tax me, tax that fellow behind the tree.” But public finance will be far more precarious if policymakers of all stripes reconceive taxes as penalties and convince the public that taxes are something only other people should pay.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

It is widely acknowledged that those subject to estate taxes take steps to reduce or avoid liability, including relocating to states without an estate tax in the years prior to their death. This, of course, deprives their former state of revenue under income and other taxes for the remainder of their lives, which represents a substantial offset to the revenue generated by the estate tax on those who remain.

While there’s existing literature addressing this question in much more sophisticated ways, one advantage of a Substack is that it provides the opportunity to explore questions through new lenses and to see if anything comes into view, even if the vision is a bit hazy.

We know which states federal estate tax collections come from. We also have details about federal income tax liability by state, broken down by income class. All else being equal, we would expect to see a reasonable state-by-state correspondence between (a) federal income tax liability for households with $1 million or more in adjusted gross income and (b) federal estate tax liability in those states.

We wouldn’t expect the ratios to be completely consistent across states, since the class of million-dollar-a-year earners is not coterminous with the class of people who will eventually owe federal estate taxes. Nevertheless, this is still a decent proxy that allows us to test whether high earners are fleeing to avoid estate taxes in their final years, which would show up in abnormally low ratios of federal estate tax liability to federal income tax liability for high earners in those states.

As you’ll see, this analysis comes with a lot of caveats and shortcomings, some of which could be addressed with a more rigorous study design and some of which are inherent in the data. That’s why I think Substack is the perfect place to explore this: we can look at some suggestive data that gives us directionally meaningful information, while acknowledging a lack of precision.

Even with the limitations, moreover, we get a couple of useful natural experiments: the 2001 Economic Growth Tax Relief and Reconciliation Act, which phased out a federal credit for state-level estate taxes between 2002 and 2006 (it fully offset liability prior to that), and the COVID-19 pandemic, which increased mortality in ways that presumably limited the ability to plan for estate tax avoidance.

An initial glance at the data is highly suggestive: there are consistently higher federal estate tax payments from states without their own estate taxes than we would expect based on federal income tax liability by state, which could be evidence of wealthy residents relocating in their final years to avoid state taxation of their estate.

The Tax Cuts and Jobs Act (TCJA) roughly doubled estate tax exemptions, which explains why—with a lag based on filing dates—estate tax collections fell so dramatically in the later years of this series.

Though not definitive, it is interesting that the lowest ratios between states with and without an estate tax are in 2020 and 2021. Predicting the timing of one’s death is always imprecise, but the pandemic resulted in substantial excess mortality. It took the lives of many people who might not have anticipated being in their final years and would not have been able to plan accordingly. We would therefore expect to see that state-level estate taxes had less of an effect on where people’s estates were taxed in those years—and indeed, that seems to be the case. (It is slightly surprising that this much of the effect shows up in 2020 itself, and other factors could be at play as well.)

There are many complicating factors. Especially in smaller states, the death of one billionaire can skew a single year’s collections dramatically, though we largely solve for this by aggregating state data. For small states where one or two payers are the predominant sources of federal estate tax revenue in a given year, IRS data is sometimes suppressed for privacy reasons, meaning that some states must be excluded from the analysis in certain years (excluding their income and estate tax revenue), which is another shortcoming. This data issue is greater post-TCJA, since the higher exemption means far fewer taxpayers.

Another issue: Delaware and New Jersey repealed estate taxes during the period in question. (New Jersey retained an inheritance tax.) Here, I took the conservative option of classifying them as non-estate tax states in all years subsequent to repeal, though we would expect some lag if people seek leave estate-taxing states several years prior to their passing. (In this instance, Delaware’s data are suppressed in some years anyway, so the issue is primarily about New Jersey.) We might also expect to see that, with the passing of years, states that repeal their estate taxes show a higher federal estate-tax-to-income-tax ratio after repeal, and we do. This is more gratifying than seeing the opposite result, but we can’t be terribly definitive based on a few years’ data, from a notoriously volatile tax, in a single state.

Then there’s a more fundamental question: if we’re measuring late-in-life tax responses, what is the driver: these states’ estate taxes, their overall tax burden, or both? Many of the states with estate taxes also feature above-average individual income taxes, which retirees—no longer attached to a place of work—might move to avoid. Retirees may also land in states without an estate tax for largely unrelated reasons, perhaps seeking sun and golfing more than a particular tax climate.

In a more sophisticated analysis, we could seek to control for these differences. For today’s simple tour through some basic data, I’ll just provide a baseline by comparing the states that currently have an estate tax to those that currently lack one, from both (1) 1997-2000, before any discussion or implementation of the federal reform that eliminated the credit for state estate taxes, and (2) 2015-2018, the four most recent successive years without significant exogenous factors (TCJA or pandemic effects).

Back when every state had a fully offset estate tax, the states that later repealed them had 38.7 percent higher federal estate tax collections as a percentage of federal income tax receipts from the highest-earning households. Now their collections are 70.3 percent higher. Something seems to have changed.

I could add many additional caveats, including the fact that $1 million in AGI in the late 1990s is not the same as $1 million in the late 2010s. If someday I choose to return to these data, I will investigate whether there are ways to firm up the calculations and isolate the effect. For now, I offer the figures as a potentially interesting observation, using this Substack as a scratchpad. Take it for what you will.

Certainly, the findings are in line with expectations.

A 2023 study by Enrico Moretti and Daniel Wilson concluded that 35 percent of local billionaires leave states with an estate tax, and that states lose 69 percent of estate tax revenue to reduced individual income tax collections. Recalculating just using states with individual income taxes, those states lose about 86 percent to reduced income tax collections, to say nothing of losses under other taxes. Some states, they find, actually lose revenue due to the estate tax, chiefly if they have high-rate income taxes.

Much earlier, Jon Bakija and Joel Slemrod explored the broader phenomenon of the rich fleeing high-tax states through the lens of federal estate tax returns. The period they analyzed was before the end of the federal credit, and with an eye toward that new reality. They calculated that if people moved five years before death in response to state estate taxes, the total revenue loss to the state would be 1.73 times as large as the revenue loss from the estate tax alone.

Both these papers offer far more scholarly insights into the question I sought to address here. But sometimes a simple look at the data can be instructive, too, despite the many limitations.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

The answer matters as policymakers evaluate measures for property tax relief, but it’s not straightforward. Tax collections have risen dramatically in recent years, but this did not happen in a vacuum: the population rose, total floor space increased, and people increasingly opted for more expensive homes. It’s not just that all home values appreciated—though most did, substantially—but also that there are more households, and that homeowners increasingly demanded larger, more expensive homes in more desirable neighborhoods.

Property tax critics often default to nominal property tax increases when reporting rising tax burdens. They sometimes argue that inflation adjustments are unnecessary or inapt because owners’ income (including retirement income) isn’t guaranteed to rise with inflation. But governments’ costs of providing services rise over time, and ignoring inflation dramatically overstates the increase in property taxes. When critics do concede the necessity of inflation adjustments, they rarely go any further.

Any long-term evaluation of property tax burdens must be inflation-adjusted. But a question then arises: which measure of inflation is best? The “standard” measure of inflation is typically the Consumer Price Index for All Urban Consumers (CPI-U), based on a U.S. city average, though sometimes versions specific to particular metropolitan statistical areas are used for state- or region-specific calculations, e.g., CPI-U for Cincinnati-Hamilton OH-KY-IN.

Increasingly, the federal government uses Chained CPI-U, particularly for longer time horizons. Whereas traditional CPI uses a fixed basket of goods to measure price changes in time, Chained CPI accounts for a greater level of substitution: if beef prices go up disproportionate to other meats, consumers might buy more chicken. This is a good measure for some things, but probably not for property or for local government services.

Finally, state and local government officials often prefer the Implicit Price Deflator for State and Local Government Purchases (IPDSL), which yields the highest inflation rates and thus the lowest rate of real growth in property tax collections.

The goods and services that governments purchase differ in composition from the typical basket of consumer goods, and many undergo greater cost increases. To the extent that this effect derives from the mix of goods (e.g., all construction costs are rising faster than CPI-U and construction costs are a greater share of governmental than household budgets), accounting for that through a governmental measure is reasonable. But to the extent that governments are less efficient than the private sector and operate in markets with less competition, using an inflation measure that captures rising governmental costs is probably a bad way to contain those costs.

Additionally, the Implicit Price Deflator combines state and local government expenditures, and thus includes the rapidly rising costs of health care and social welfare services largely delivered by state governments, which are less pertinent for local governments and thus less relevant to the property tax debate.

Because of this, my preference is to use the Consumer Price Index rather than the Implicit Price Deflator, but—where revenue constraints are imposed, like with property tax levy limits or other tax and expenditure limitations—to build in some fixed growth factor above CPI to account for its limitations. Just using the Implicit Price Deflator gives too much away.

With a standard inflation adjustment, property taxes rose 70 percent in real terms between 1995 and 2023.

Of course, inflation adjustments are not the whole story, though many analyses treat them as such. The population is also growing, while families are shrinking. A larger population means more housing, and at least theoretically, smaller families mean more distinct housing units proportional to population.

Interestingly, however, I see little evidence that declining household size was significant in aggregate. Part of this seems to owe to an example of Simpson’s paradox: household sizes are declining across most demographics, but the fastest-growing populations have larger (if still declining) household sizes. Immigration and increased ethnic diversity appears to be stemming the tide on the average household size. Later marriages and fewer children, moreover, may be partially offset in the data by more 20- and 30-somethings living with roommates.

Household size may not matter as much as expected, but population growth certainly does. Simply inflation-adjusting property tax collections over time cannot capture this, and thus dramatically overstates the increase in property tax burdens. After adjusting for population growth, the real increase in property tax burdens was 34 percent between 1995 and 2023.

Even that may not tell the whole story, because the composition of that property has changed. Houses are getting larger even though families are getting smaller. Floor space has increased by almost two-thirds since 1995, with large increases in both residential and commercial property. (Industrial square footage has fluctuated, but current square footage is similar to 1995 levels.) And that’s just one possible metric, just one way that properties have become more valuable.

So should we adjust for floor space or other ways that properties have improved over the decades? That’s a tougher question than the population adjustment, because while it makes sense to tax each property on its market value relative to other properties, it’s not necessarily the case that property taxes need to rise because real property get more expensive in aggregate—even when the greater expense is due to genuine improvements, like larger or more up-to-date homes.

Certain government services may cost more to provide to nicer homes, and people in wealthier neighborhoods may demand more or better government services. Still, the cost of schools, roads, and courts doesn’t change just because people’s houses got larger.

If you use floor space as a proxy for qualitative improvements in real property, the real increase in property taxes between 1995 and 2023 drops to a mere 5 percent. I think this is far too low for the reasons stated above, but it’s still a potentially valuable data point that might prompt us to apply some discount to the 34 percent real growth indicated by population and inflation growth alone.

Another way to think about property tax burdens is to express them as a share of income. As people become wealthier, they may purchase more expensive housing, which will cause aggregate property tax burdens to rise. But if income rises faster than housing prices, property taxes as a share of income can still fall.

During periods of robust economic growth, we would expect to see property tax burdens decline as a share of income, because even though demand for pricier housing might increase, there are upper limits for most purchasers: someone with a net worth of $10 million probably lives in a much more expensive house than someone with a net worth of $500,000, but it’s likely not twenty times as expensive.

Indeed, this is what we see in the data: property tax collections rose as a share of income during the housing bubble of the 2000s but have been declining ever since. That doesn’t tell us that property taxes are at the right levels, but it does offer some good news: overall, ability to pay is improving despite rising property tax bills, because income is rising even faster.

None of this tells us the “right” level of property taxation, and by any reasonable measure, property tax burdens have risen considerably faster than inflation. But how we measure the increase matters. The data tells a story, but selective or inadequately considered use of data can give the wrong impression.

Rising property tax burdens are a real issue, and one that demands real solutions. But our sense of what policy solutions are appropriate will be influenced by our sense of the breadth of the problem—meaning that it matters how we measure.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

Assuming ten duty days in California, which is typical for the Super Bowl, Seahawks quarterback Sam Darnold will owe an estimated $192,189 for playing today—more than the $113,000 he’ll make if his team loses, and even slightly more than the $188,000 he gets if he wins.

Jock taxes are just a special instance of nonresident income taxation. Many states, including California, require nonresidents to remit tax even if they only spend one day in the state.

I’ve created an embeddable calculator that allows you to select any player, set certain parameters (including adjusting the number of duty days), and see how much they’ll owe to California for playing in the Super Bowl. Click here or on the image below to go to the calculator, where you can use the “Embed” link to put this on your own site if you wish.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

That’s the subject of my latest blog post for the Tax Foundation, which you can find here.

Whereas most states used to be in the broad middle, the new distribution is increasingly bimodal. So-called millionaires’ tax proposals in Michigan, Rhode Island, Virginia, and Washington would further widen the gap, as would—in the opposite direction—additional proposals for rate reductions in a number of states.

I typically use this newsletter either to write about something that doesn’t quite fit for publication through my normal channels or to expand on something I’ve written elsewhere, but today’s email is an extremely short one with nothing more than the goal of sharing my Tax Foundation analysis, which you can find here. I believe this trend is important enough to amplify in an email, as it has significant implications for the future of state tax competition.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

Where the CEA went wrong is in attempting to calculate the rates at which a sales tax on a broad base of personal consumption could replace existing collections from states’ current individual income, corporate income, and sales taxes. The CEA estimated that an average rate of 6.23% would suffice. I ran the numbers for a Tax Foundation analysis and came up with 17.51% as a national average.

My goal in today’s newsletter is not to pick on the CEA. Instead, I want to explain one of the ways their calculations went awry, because they weren’t the first to make this mistake and they won’t be the last. Perhaps this explanation can spare others some embarrassment and provide useful insight into how to calculate the potential revenue effects of sales tax base broadening.

Let’s use Connecticut as an example. The state currently imposes a 6.35% sales tax, a corporate income tax with an 8.25% top rate, and an individual income tax with a top rate of 6.99%. (As an aside, the state’s income tax features a highly unusual recapture provision, which phases out the benefit of lower rates. In other words, high earners pay 6.99% on all taxable income, not just marginal income.)

The White House believes that these taxes can all be replaced with an 8.77% broad-based sales tax. I’m going to show you why the real rate is at least 17.22%.

Here’s what the relevant taxes raised in Connecticut in FY 2025. The pass-through entity (PTE) tax is an extension of the personal income tax on pass-through businesses, with PTE tax payments credited against ordinary individual income tax liability.

• Individual Income Tax: $12.96 billion

• Corporate Income Tax: $1.40 billion

• Pass-Through Entity Tax: $2.37 billion

• Sales and Use Tax: $6.46 billion

• Total: $23.21 billion

That’s the number to replace: $23.21 billion. Now let’s figure out how to get there.

To their credit, the CEA attempted to design a tax that only applies to final consumption. (About 41% of current sales taxes fall on business inputs.) I will follow their lead on this.

Adjusting to match fiscal year data, Connecticut’s total FY 2025 personal consumption expenditures ran about $251.5 billion. If sales tax were imposed on every last dollar of personal consumption in Connecticut, the replacement rate would be 9.23%, already slightly above the White House’s estimate of 8.77%. We should also recognize that no tax gets perfect compliance. Sales tax revenue estimates typically assume 85–⁠89 percent compliance. At 89%, the revenue-neutral rate is 10.37%.

But stopping there would be fantastically wrong. Even the CEA’s faulty calculations recognized this.

The Bureau of Economic Analysis (BEA) publishes an annual series called Personal Consumption Expenditures. It’s a measure of the economic value of personal consumption, not of transactions. Even if you’ve paid off your mortgage, for instance, you “consume” housing services by living in your own home, which the BEA estimates with an imputed rental price of owner-occupied housing. The sales tax couldn’t apply here even if policymakers wanted it to (and they don’t), since there’s no actual transaction.

The CEA claims it accounted for three exceptions—housing plus two preferential policies:

• The sales tax would not apply to rent or housing more generally.

• The sales tax would not apply to groceries.

• The sales tax would not apply to any category of good that is already taxed under an excise tax or other selective tax (gasoline, alcohol, tobacco, etc.).

Adjust for those (I only included the selective taxes they mentioned, plus insurance, which is subject to insurance premium taxes), and now we’re at 14.13%. And we’re still not done.

As I noted in my Tax Foundation analysis, the CEA’s base includes, among other things:

• All healthcare expenditures, including those covered by private insurance (funded by premiums subject to separate excise taxes) and those paid for by government through Medicare, Medicaid, and other programs (which cannot be taxed as a matter of law).

• Consumption that does not involve a financial transaction, including the full imputed value of banking services that are largely funded by banks’ reinvestment of depositor funds, not by direct fees from depositors.

• The value of services provided by nonprofits at free or subsidized rates, including scholarships and endowment subsidies that reduce college tuition, free or subsidized medical care, aid provided by charitable organizations, and the full operating expenses of houses of worship (none of which are transactions and none of which could be subject to sales tax).

• Other purchases that are not legally taxable, including internet access and purchases from the US Postal Service.

Excluding anything else that doesn’t involve a transaction (uncharged banking services, growing produce on your own land, etc.) and anything that can’t be legally taxed (Medicare, Medicaid, USPS deliveries, internet access, etc.), brings the rate to 20.10%.

But the CEA did offer some preferential exemptions, for groceries and items subject to separate excise taxes. There are arguments against these exemptions, and they certainly aren’t legally or practically necessary, though they’re certainly popular. Adding those back to the base yields the minimum possible legal rate, if lawmakers are willing to adopt a truly broad-based sales tax that falls exclusively on final consumption (avoiding any tax on business inputs).

That rate is 17.22%.

Finally, if policymakers couldn’t stomach taxing health care (beyond government-provided services), private education, financial services, and professional services (accounting, legal, etc.), and also adopted the exemptions the CEA incorporated in its plan, the replacement rate would skyrocket to 30.40%.

Incautious use of the PCE series has led many people awry, not just the economists at the CEA. Perhaps others can learn from their mistakes.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

A new paper from Rachel Moore, Brandon Pecoraro, and David Splinter at the Joint Committee on Taxation provides compelling evidence that (1) the revenue-maximizing rate is much lower than many have believed and (2) Laffer Curves, in their words, “are flat”—that is, there’s a large range of top rates where further rate increases yield very little change in revenue collections.

The red line in the above chart is the one to pay attention to, and you can see that federal taxes raise about the same amount of revenue across a range of more than ten percentage points on the rate. The authors describe this by saying that the curve is flat along the top. I prefer to think of it as a mesa.

Unlike most prior research, which used simplifying rate assumptions (e.g., considering only labor income and ignoring lower rates on capital gains income, or disregarding taxes other than the federal income tax), the new JCT paper employs more realistic tax assumptions, yielding better-calibrated results. The JCT economists model the stylized tax bases used in previous analyses, some of which are too narrow and others too broad, while contributing their own “true base” (the red line in the above chart).

They find that overall federal tax revenues only increase by 0.5% when the top federal income tax rate increases from 37% to 40%, and that it plateaus after that before eventually yielding consistent revenue reductions on the other side of the curve.

The authors identify an equity-efficiency trade-off in which GDP falls while tax progressivity increases, noting that over the flat region, “increasing the top rate raises relatively little revenue,” whereas “raising top rates primarily trades off between progressivity and growth.”

Don’t miss the significance of this combined with that flat region: across that region, higher marginal rates increase progressivity by reducing the after-tax income of high earners, not by increasing the income of low earners.

When higher rates yield higher revenues, governments can allocate some of that additional revenue to programs that benefit lower-income households. When raising top rates yields flat-lining revenue, the additional progressivity owes entirely to economic drag that reduces the income of those subject to the higher rates. The rich get less rich (and the economy gets smaller), but lower-income households don’t benefit. In fact, they stand to lose as well: lower economic growth is bad news across the income spectrum.

The JCT economists also calculate the revenue-maximizing top rate for the federal income tax when taking all taxes at all levels (federal, state, and local) into account, and find that we’re already on the mesa:

This suggests that there’s little aggregate revenue-raising potential of raising state or federal rates, though that doesn’t mean that state-level rates are themselves approaching the wrong side of the Laffer curve. Raising state rates can simultaneously shrink the pie but grow an individual state’s slice of that pie.

States that raise taxes should expect reductions on both the extensive (outmigration and capital flight) and intensive (reduced labor, investment, etc.) margins. This reduces the yield of those tax increases, and the JCT paper suggests that the systemwide effect may not involve much, if any, new revenue, but the tax-hiking state itself would still expect some level of revenue increase.

Of course, revenue maximization should not be the goal of tax policy in the first place. If 40% is the revenue-maximizing federal income tax rate, but a rate hike from 37% only yields extremely modest revenue gains while doing much more significant economic harm, that’s a strong argument against targeting revenue maximization even if you’d otherwise prefer the government to collect additional revenue. The slope of the curve before it goes flat is highly relevant here.

Moore, Pecoraro, and Splinter have substantially improved our understanding of revenue-maximizing tax rates. This paper has a lasting real-world impact.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

Last week, I explained how this provision and several others combine to massively overvalue net worth for wealth tax purposes. Valuation based on super-voting shares has been flagged as a major issue by many in California’s tech community as well. Founders and early investors often hold super-voting shares that could lead to valuations that far outstrip their actual stake in their company, potentially forcing them to sell off a significant portion of their shares and send stock prices plummeting.

That argument received pushback from four professors associated with the wealth tax initiative, who claim that the founders of publicly traded businesses would not have their holdings valued by voting or direct control rights. This provision, they note, does not apply to publicly traded assets—and therefore, they claim, it won’t apply to super-voting shares in publicly traded companies.

Their argument collapses two categories. The initiative says the voting-share valuation rule does not apply to publicly traded assets, not to shares in publicly traded entities. The memo assumes that these are identical, i.e., that all shares in a publicly traded company are publicly traded assets, even if the share class does not trade. This is not the case.

Alphabet (Google), Meta, and Airbnb, for instance, are publicly traded companies. But their founders’ Class B super-voting shares cannot be listed on an exchange. They lack a secondary market, and their value is tethered to specific founders. Those shares are not publicly traded, even though they’re shares in publicly traded companies.

The professors also argue that any improper valuation could be resolved through a dispute mechanism under which default valuations can be contested. The option of challenging a default valuation is never a sufficient solution to overly aggressive valuation rules, but is particularly inadequate here, as the initiative imposes a chilling effect on certified appraisals through severe penalties on both taxpayers and appraisers who submit valuations that California’s Franchise Tax Board (FTB) ultimately rejects.

If the FTB determines that a taxpayers’ super-voting shares should have been valued based on controlling interests, or at least valued at some premium over other company stock, the taxpayer would face a penalty of up to 40% and their appraiser could get hit with a penalty of up to 4% of the underpayment—which is potentially ruinous for appraisers, since they are tax professionals, not billionaires.

I have written a memo explaining in greater detail why, despite wealth tax supporters’ downplaying of the issue, aggressive valuation rules create the very real possibility of taxing controlling interests. The memo has been posted to SSRN and can be downloaded here.

Click Here to Download the Memo

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

A fault line is emerging between the majority of states that have cut individual income taxes in pursuit of greater tax competitiveness and a minority of states that are doubling down on high taxes on high earners. We are headed toward a new reality in which there is no such thing as a “typical” rate—just low rates and high rates.

Today I want to share four I’ve written recently on these proposed taxes on high earners:

• A recent paper exploring six provisions of the proposed California wealth tax that dramatically overstate taxpayers’ net worth.

• A blog post explaining how the proposed millionaires’ tax in Washington would undercut small business owners, startups, and tech employees.

• A new paper out today, co-authored with my Tax Foundation colleagues Janelle Fritts and Nicole Fox, estimating the effects of the proposed Michigan high earners’ tax on job creation, migration, gross state product, and innovation.

• An op-ed, reprinted below, on the state economic harms of wealth taxation.

Although these pieces are specific to California, Michigan, and Washington, much of the analysis should be valuable in the context of proposals in other states as well. (If you’re interested in analysis for your state, please get in touch.)

Proposed Wealth Tax Creates Long-Term Budget Risks for California

This op-ed originally ran in the Orange County Register on January 14th.

California’s proposed wealth tax has reportedly already driven at least six billionaires out of state, and proponents haven’t even started collecting signatures yet. The preemptive exodus, and the prospect of significantly more departures later, illustrates the initiative’s fiscal risk: not only revenue under the new tax, but also sustained collections from California’s existing taxes, hinge on the decisions of a small number of highly mobile individuals who are being incentivized to leave.

As the nonpartisan Legislative Analyst’s Office noted in its evaluation of the initiative, the new tax would yield a temporary state revenue increase from the wealth tax but a “likely ongoing decrease in state income tax revenues.” A half-dozen billionaires have already left. Many more might depart if the initiative appears to have a good shot at passage, and still more would leave if it’s ratified by voters. Even with passage, moreover, the tax’s prospects are uncertain given the inevitable legal challenges. California could lose a significant share of its billionaires – and many other high earners with them – over a tax that might never generate a dime.

California’s tax system is uniquely susceptible to this risk because it relies so heavily on high earners. If billionaires leave and if they take other high earners with them, California’s top-heavy tax system would take a substantial hit. Filers with more than $1 million in annual income were responsible for around 40 percent of California’s personal income tax collections by 2019, and those with $5 million or more in income – fewer than 10,500 filers – paid 20 percent of the income tax. At a rough estimate, the state’s 200 wealthiest households, those potentially subject to the wealth tax, likely remit $5 billion or more each year in income taxes on their own.

If billionaires leave California to avoid the wealth tax, they won’t just uproot themselves. For each billionaire departure, many other high-paying jobs will follow, as cities like Austin, Phoenix, and Tampa Bay become magnets for departing tech executives, employees, and investors.

High earners tend to cluster around growing businesses and venture capitalists. If senior executives and major investors depart, they will take offices, enterprises, and, over time, high-paid employees with them. And in a state where the loss of 10,500 people would wipe out $25 billion just in personal income tax collections (based on current revenue estimates), to say nothing of revenue from other taxes, it wouldn’t take that many departures to create an ongoing budget crisis. Losing even a fraction of the state’s high earners creates a budget hole that the wealth tax cannot fill, and isn’t even designed to fill.

The proposed initiative is a one-time tax with earmarked revenue. Even if it were not tied up in litigation, its collections would not offset reduced revenue for education and other core government services. And whereas the initiative creates a one-time wealth tax, an exodus of jobs and high earners (including not just billionaires but also many other highly paid employees, particularly in the tech industry) yields lower revenue year after year. That puts pressure on school budgets (since about 40 percent of personal income tax revenue goes to education) and all other state government functions.

Under the initiative, wealth tax liability is based on residency as of January 1, 2026. However, no one is confident that this retroactive and non-proportional “snapshot” residency requirement will withstand legal scrutiny. That includes the initiative’s drafters, who included language asking courts to substitute later dates if legally necessary to preserve the tax. Billionaires could move later in the year – even after Election Day – in the expectation that the courts would revise the residency requirement.

Under the wealth tax proposal, California is at risk of losing something it can’t easily recover: not just a specific set of billionaires who may never return, but the state’s seeming inevitability as the hub for new tech investment and innovation.

Other cities and states have long chipped away at Silicon Valley’s edge, but California’s fortunes rose in tandem with the rise of artificial intelligence. If some of the tech industry’s leading figures depart to avoid a one-time wealth tax that may never even go into effect, other states’ technology hubs – the country’s Silicon Hills, Slopes, and Prairies – will be the ballot measure’s true winners, and California’s economy and state budget its losers.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

Levy limits cap the annual increase in property tax collections across a jurisdiction, rolling back rates (mill levies) to keep collections in check as values rise. It is generally agreed that levy limits should roll back rates based on the increase in value of existing properties, whereas new property should be outside the cap, since new businesses and a larger population create new costs for localities. Population or business growth shouldn’t reduce per capita collections. But while, with one partial exception, all states with levy limits concur with that reasoning, they don’t agree on how to implement it—and the differences matter.

Boise, Brownsville, and Boston each represent a structural approach to new construction under levy limit regimes. Boise stands for Idaho’s unique design, under which revenues from new construction can count toward the cap (though in real-world Idaho, only under unusual circumstances). Brownsville stands for a Texas system under which new construction is excluded from the allowable growth calculation. And Boston stands for Massachusetts’ system under Proposition 2 ½, under which the value of new construction is added to allowable growth.

In each case, the type only refers to the broad mechanism, not the details. Idaho has two caps, and the one that includes the value of new construction has an 8% growth factor, making the restriction far less stringent than it appears at first glance. The strict design of Texas’s cap is diluted by a generous 3.5% growth cap. And Massachusetts’ limitation, though “leaky” in design, has a 2.5% cap with no inflation adjustment. Reality is nuanced. But a typological use of these jurisdictions helps explain the three basic ways in which levy limits can handle new construction.

Imagine a community with $100 million in housing stock and a tax rate of 10 mills (1%), generating $1 million in property tax revenue in the first year. With a 2% levy limit, if there were no new construction and the value of existing properties rose 5% to $105 million, the levy limit would cap the increase in tax liability at $1.02 million by rolling back the mill levy to 9.71 mills.

Now imagine there’s also $5 million in new construction, and consider how levy limits work under the Boise, Brownsville, and Boston systems.

If new construction is included in the cap (the “Boise system”), then all $110 million in housing can only raise $1.02 million in revenue, forcing the mill levy to 9.27. The existing properties, which paid $1 million in taxes while valued at $100 million, pay $974,000 in tax while valued at $105 million. Total tax collections are $1,020,000, the same as if there had been no new construction.

If new construction is excluded from the cap (the “Brownsville system”), then the rollback is the same as if there were no new construction: to 9.71 percent. The new construction is disregarded for purposes of calculating the levy limit, and the new rate is applied to both old and new property. The existing properties, which paid $1 million in taxes while valued at $100 million, pay $1,020,000 when valued at $105 million. Total tax collections on old and new property are $1,068,571.

Finally, if new construction is added to the cap (the “Boston system”), then the allowable increase in collections is the sum of (a) last year’s collections plus 2% and (b) the value of new construction times last year’s rate. Under a 2% growth cap with an initial rate of 10 mills, therefore, subsequent year collections can be $1,020,000 (existing property + 2%) + $50,000 ($5 million in new property times last year’s rate) = $1,070,000. The rollback is calculated against this amount, yielding a new rate of about 9.73 mills. The existing properties, which paid $1 million in taxes while valued at $100 million, pay $1,021,364—an increase of almost 2.14% despite a 2% cap.

Here’s what that looks like:

This design choice is only one of many important considerations in designing levy limits. Policymakers must also decide on the allowable growth rate (and whether to incorporate an inflation adjustment, ignored in the example above for simplicity’s sake) and determine what spending obligations, if any, are outside the cap, among other considerations. But the distinction highlighted here is important yet little-noted. It seems quite probable that lawmakers in states with what I’ve called the Boston system (new property added to the cap) are unaware of the Brownsville system (excluded from the cap), and vice versa.

As policymakers consider the adoption of levy limits in new states, they should (at least for this) think Brownsville, not Boston—and certainly not Boise.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

But first, please permit me a digression for a career update and a quick request.

With the dawn of 2026, my new business, Walczak Policy Consulting, is up and running. In addition to my continued association with the Tax Foundation, where I am now a Senior Fellow, I am working with the Platte Institute as a Senior Tax Policy Advisor and will be writing for the Mackinac Center as an Adjunct Scholar. I will be announcing additional affiliations soon, and publishing with other organizations on a project-specific basis.

My request: would you consider sharing my LinkedIn announcement with your network?

R&E Expensing Restoration Costs Are Mostly Timing Effects. But What Does That Mean?

Decisions about conformity to the One Big Beautiful Bill Act (OBBBA) are front-of-mind for many lawmakers as state legislatures reconvene for 2026. That’s understandable, but hesitation about restoring first-year research and experimentation (R&E) deductions is troubling. As I have written here and elsewhere, allowing an immediate deduction for research and development costs is sound tax policy—and, until very recently, there was near-universal agreement on this point. The federal government adopted the provision in 1954 and every state with a corporate income tax followed suit.

I won’t rehash the details of how a gimmicky reconciliation provision shifted the R&E deduction from first-year to five-year amortization in 2022, or how the OBBBA restores long-standing policy. You can read more about that in my prior coverage of the issue.

What I’d like to focus on: what it looks like for a state to transition back to first-year R&E expensing after shifting to amortization. I and others discussing the issue often note that, for states, amortization—and its reversal—is mainly a timing effect. But what exactly does that mean? My goal here is to demonstrate the timing effect with tables and charts.

When businesses are required to deduct their expenses over multiple years, they still get to deduct the full cost of that expense, but the delay comes at a cost in terms of inflation and the time value of money. Additionally, some businesses may have cash flow issues, particularly in their early years. They may have significant research expenses but not be profitable. More to the point, they may appear profitable—and be taxed as if they are profitable—if their revenues are counted in the current year, but the bulk of their R&E and other capital expenditures are shifted to later years. Amortizing R&E expenses can tax research-heavy firms at the wrong time, when they have limited ability to pay.

States, of course, generate additional revenue through amortization, since it erodes the present value of the deductions. But this additional revenue through inflation and other timing penalties is not particularly large, and states don’t need it. Lawmakers surely didn’t think they needed it between 1954 and 2021, when first-year expensing was permitted.

The problem is that now that states have shifted to amortization, restoring immediate expensing pulls more deductions into the 2025 tax year. We’ll walk through how this works, starting with this observation: the timing benefit of the shift to amortization was short-lived. All R&E is deducted eventually, so if a business can only deduct 20% of the cost in a given year, that may be a “savings” of 80% for the state in that year, but the other 80% of the deduction simply shows up as an obligation in future years.

Just like “buy now, pay later” schemes don’t actually reduce how much you pay, five-year amortization doesn’t reduce the size of the deduction states ultimately have to offer. And since new investments are made every year, it doesn’t take long until equilibrium is restored.

Imagine that every year, corporations invest 100 units in R&E. (The value doesn’t matter. That could be $100 million, $10 billion, whatever—we’ll keep this simple and just use consistent units.) Under first-year expensing, they’d collectively deduct 100 units from taxable income each year. Under five-year amortization, they spread it out—over six years, confusingly enough, due to a “split-year” convention where you actually deduct 10% in the first and last tax year.

It looks something like this. (We’ll get to the box in a minute.)

Note that in 2026, a state using amortization is offering deductions for investments made in 2022, 2023, 2024, 2025, and 2026. By next year, enough time will have passed since the implementation of 5-year amortization that, neglecting inflation and investment growth, states would be offering just as much in annual deductions as they were prior to amortization. All the savings were concentrated in a few years, as can be seen in the column on the right showing the total deduction value by year.

Now imagine a state follows the OBBBA in restoring first-year expensing. All recent expenses still being amortized can “catch up,” while all expenses made in 2025 and beyond are again deductible in full in the first year. That looks something like this:

You may have guessed that the units in the yellow box on the first table sum to 310, the deduction value for 2025 in this second table. This is a timing shift, though a dramatic one. All the deferred deductions from the past few years “catch up” in one year. Basically, states just unwind their prior policy, reversing their temporary windfall of recent years, but there’s certainly a short-term cost. (It’s important to note that corporate income taxes only account for about 2 percent of state revenue, and that R&E policy is a small slice of that small pie.) For tax year 2027 and onward, the annual deductions are the same as under amortization.

Delaware is the first state to choose a different option: requiring R&D investments already made under amortization to continue under that convention, but new investments obtain the benefit of first-year expensing. This isn’t entirely fair to prior investment, but it’s economically efficient inasmuch as retroactively restoring the deduction can’t change any of the investment decisions already made. (Congress had to deal with a different calculus: had it only restored the provision prospectively, companies may have delayed R&E investments in 2025, creating substantial economic disruption.)

The Delaware approach spreads out the “cost” of clawing back the state’s previous windfall over several years, making it more manageable for states with tight margins. That looks something like this:

Here’s what this looks like in cumulative deductions between 2020 and 2031. Assuming, for now, that firms make the same amount of R&E investment each year, there’s a convergence of both OBBBA conformity scenarios with pre-amortization policy. If, by contrast, states stick with 5-year amortization, every future year features the same amount of deductions (accumulated across multiple years), but the 2022-2026 shift is never fully “caught up.”

And here’s how that plays out year-by-year:

These are, of course, stylized examples. Research expenses aren’t identical from year-to-year. In fact, they’re growing, recently at a rate of about 5% per year. With consistent growth rates, the math doesn’t change much. With more variation, you get a bit more noise and a slight further shift in timing effects, but the basic story is the same.

States received an unexpected windfall when R&E deductions were postponed starting in 2022. There’s a cost to unwinding that policy. From one perspective, of course, it’s not a cost at all: it’s simply returning the windfall. But to the extent that this shows up now, states can choose to absorb the timing effect in one year or to follow Delaware’s model and spread out the reversal of the previous timing shift. What they shouldn’t do is hold so firmly to their one-time windfall that they permit their tax codes to penalize R&D in perpetuity.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services, both project-specific and on retainer (or in visiting fellow-style roles). If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

There are sound economic reasons for cities and states to attract data centers. Meanwhile, evidence is mounting that, while these operations can be a source of significant new tax revenue, the investments will go elsewhere if taxes are too onerous. I’ll explore this below, expanding upon a paper I recently published with the Tax Foundation.

SALT News and Updates

• At a rehearing, Colorado’s title board denied title setting for a proposed graduated rate income tax ballot measure, determining that the proposed initiative violated the state’s single subject rule. The initiative will not appear on the 2026 ballot.

• Several Alabama cities are challenging the state’s Simplified Sellers Use Tax regime, designed to provide a single point of tax collections and administration for remote sellers. A key point of dispute is the ability of large multistate corporations to use the system even if they also have physical presence (distribution centers, stores, etc.) in the state. Alabama is one of only three states without central sales tax administration, and obligating remote sellers to remit to each jurisdiction separately raises “undue burden” arguments under the U.S. Constitution.

• The 5% one-time wealth tax ballot measure in California has been given a title (“Imposes One-Time Tax on Certain Individuals and Trusts”) and can now be circulated for signatures.

State and Local Taxation of Data Centers

In a new Tax Foundation paper, I calculate tax liability for a $1 billion model data center in twelve representative jurisdictions across the country. This edition of The SALT Road provides some additional analysis and expands on some arguments advanced in that paper, which I encourage you to read for a more comprehensive treatment of the subject. (The paper, while not terribly long, runs 14 pages. This newsletter, mercifully, does not.)

California has a lot of data centers: 326 by one count, or 7.6% of all data centers nationwide. That’s not terrible, but it’s a surprisingly mediocre performance for the state that lays claim to Silicon Valley and is at the epicenter of AI development. California represents 14.0% of GDP but, despite being a tech-oriented state, only hosts the aforementioned 7.6% of data centers, ranking 38th on the ratio of share of data centers to share of GDP, sandwiched between Alabama and Kansas.

California imposes the nation’s highest tax burden on data centers, largely because its high-rate sales tax applies to purchases of servers, chips, and other data center business inputs.

Virginia, the historic “home of the internet,” offers considerably better tax treatment. It leads the nation with the most data centers (670) and the highest ratio of data center share to GDP share (15.5% of data centers, 2.6% of GDP). But even Virginia’s historic advantages have not been enough to stem the tide of new data center developments in lower-tax systems.

Texas does quite well, with 429 data centers. But states that really punch above their weight (in addition to Virginia) include Oregon, Iowa, Montana, Nevada, North Dakota, Arizona, Wyoming, and Ohio. Each state has its own advantages.

Oregon imposes lower tax liability than other West Coast states while still offering proximity to Silicon Valley and Seattle. Many of these states have low utility costs, an important consideration for data centers, and the climates of northern states reduce cooling costs. But there’s another reason these states stand out: the data center industry is heavily taxed, making tax burdens a high-salience issue, and these states offer better tax treatment than many of their peers.

The largest data center investments, unsurprisingly, are going to lower-tax jurisdictions. The OpenAI/Oracle “Stargate” project, which could ultimately entail $500 billion in capital investment, is being developed in Texas, and Google recently announced $40 billion in new investment across three Texas campuses, while the “Frontier” data center facility will bring a $25 billion investment to a single Texas location.

AWS, meanwhile, plans a $15 billion data center expansion in Indiana and $10 billion investments in both Mississippi and North Carolina. A multi-tenant $10 billion campus is also under construction in Mississippi, while Meta is making a $10 billion single-site investment in Louisiana. In the next tier down, Google has large investments planned in Iowa and Arkansas, and a variety of companies are opening new facilities in Ohio, Pennsylvania, Wisconsin, and elsewhere.

Nowhere to be found: California or other high-tax states. Even historically competitive Virginia isn’t attracting new data centers at the same rate, though several significant projects are still underway.

It’s one thing for Texas to outcompete California, but when Iowa, Louisiana, and Mississippi are emerging as better locations for the largest data centers while historic tech-heavy states aren’t even on the map, something notable is happening.

In my recent paper, I estimate that a $1 billion data center faces an average effective tax rate of 48.3% on its profits before targeted incentives, with three-quarters of that tax burden being location-dependent. Data centers are highly capital-intensive operations that require frequent upgrades to servers and chipsets. They face significant taxes on their real property, and, depending on state-level policy choices, can also be exposed to extremely high sales and business tangible personal property tax burdens.

In many states, sales tax exemptions for data centers are structured as an incentive, available only if investment and job creation targets are met. This leads many to regard the exemptions as preferential treatment, but it's actually the opposite: business inputs are supposed to be exempt from the sales tax. Making that exemption conditional is worse than ideal treatment, not better.

Some states provide temporary abatements of tangible personal property (TPP) taxes on data centers’ machinery and equipment (servers, chipsets, chillers, etc.). These exemptions are somewhat more preferential, though they’re generally available on the same terms provided to other businesses with substantial capital investment. Ideally, moreover, property tax would only be imposed on real, not tangible, property. While it’s unfortunately quite common, taxing capital investment is poor policy.

If data centers didn’t get the benefit of sales tax exemptions and didn’t receive their current levels of TPP tax relief, I estimate their all-in effective tax rate would hit a ludicrous 90.1%.

If states choose to subject capital-intensive businesses to sales and TPP taxes on all of their equipment, these taxes—which are not tied to ability-to-pay—can wipe out virtually all profits. That’s a good way to ensure that investments go elsewhere.

For more on data center taxation, you can read my recent paper here.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services. If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

But let’s be honest: it’s almost Christmas. I don’t want to write about tax policy and you don’t want to read about it. Today I’m taking a detour to answer a question I’ve often wondered about while watching football: just how accurate are those in-game win probabilities, anyway?

Throughout most of the fourth quarter of Thursday night’s Rams-Seahawks thriller, ESPN and others gave the Rams overwhelming odds, peaking at 98.8% for ESPN. The Seahawks won in overtime, producing this dizzying real-time win probabilities chart:

Sam Darnold indisputably engineered a highly improbable comeback. But with 10 minutes left in regulation, were the Seahawks’ odds of victory really just 1.2%? And every other time we’ve seen these win probability charts go wild: are the probabilities miscalibrated, or do we only remember the games that go against the odds? (Even a 1.2% chance should hit once every 83 times, after all.)

My guess was that these win probabilities were good in aggregate, but perhaps overly sensitive to swings in the action. But I didn’t know, so I decided to find out.

I started by pulling win probabilities for each game of the 2025 season as of the end of the 1st, 2nd, and 3rd quarters, and comparing them to the actual results.1 Because some odds were extremely rare (not many teams have an 80% win probability at the end of the first quarter), smoothing created too many data anomalies, so I settled for “bins” of 10 percentage points.

I asked the question: how often does a team with, say, 30-40% odds as of the end of the first (or second or third) quarter go on to win the game? If the probabilities are reliable, the result should be in the middle of the range.

Turns out, they are:

Interestingly, probabilities were noisiest at the end of the third quarter. Perhaps the models get more conservative toward the end, understating the odds of victory for the team with the lead and creating the dip visible between 40 and 70% win probability. But overall, these results suggest that in-game odds of victory are quite accurate.

Hold on, though: selecting the end of each quarter might have a smoothing effect that washes out an oversensitivity to big plays and lead changes. How often have we seen a 4th quarter score send win probabilities soaring, only to see the lead change yet again?

How often? I honestly had no idea. It seems like it happens with some regularity, but am I only remembering the outliers? I decided to test this as well.

This time, I pulled win probabilities after every single play all season long to identify any time in a game that a team’s win probability exceeded a certain threshold. If a team’s win probability ever eclipses, say, 70%, then they should only lose 30% of the time. But does that hold, or do we see teams’ probabilities spiking (or plummeting) and then recalibrating?

It holds remarkably well:

If anything, win probabilities again prove slightly conservative. If at any point in the game, a team’s win probabilities hit 70%, that team wins 72% of the time. At 80%, their actual probability is 82%, and at a 90% projection, they win 93% of the time.

I suppose I shouldn’t be surprised that win probabilities are well-calibrated—which does not, of course, make them terribly useful. The outcome of most games remains reasonably uncertain well into the action, which is why win probabilities so often hover between 40 and 60% until late in the game.

Statistically, moreover, with a 272-game regular season, we should expect to see about 14 teams blow 95% victory odds each season, with 11 such collapses anticipated by this point in the year. To date, nine teams have posted such spectacular losses, with one other game just missing the mark. I’m using nflverse data, which gave the aforementioned Rams 99.2% odds of victory at peak, compared to ESPN’s 98.8%.

The Chicago Bears Threaten to Leave Over Taxes and Subsidies

I promised I’d get to taxes eventually. Here goes.

Last week, Chicago Bears team president Kevin Warren floated the possibility of moving the Bears to Northwest Indiana amid ongoing negotiations around a new home and a new stadium for the team.

The Bears have two asks. They want $855 million in public funding for new infrastructure (roads, utilities, commuter line adjustments) to support a stadium in the suburbs, along with legislation exempting construction materials from the sales tax for, and freezing property tax assessments on, megaprojects like their proposed stadium. The assessment freeze would be based on the land value assessed before major construction begins.

I discuss this at greater length in a blog post for the Tax Foundation, so I’ll cut to the chase. Illinois lawmakers are right to balk at the team’s demands for special treatment, but the Bears front office has a point, too: Illinois, and specifically Cook County, taxes are very high. I estimate that unabated property tax burdens for a new stadium complex at the team’s proposed new home in Arlington Heights, Illinois would run about $210 million per year, compared to $50-75 million per year if the team relocated to northwest Indiana.

In other words, operating in Illinois is expensive, whether you’re the Bears or any other business. Chicago and the state of Illinois have long sought to patch over an uncompetitive tax code with targeted incentives, but when those aren’t available, staying put is costly—even for a team as embedded in the community as Da Bears.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services. If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

This edition of The SALT Road isn’t representative of its typical content, but if you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

SALT News and Updates

• Hawaii Gov. Josh Green has floated the possibility of partially reversing the state’s recent income tax cuts, retaining cuts to lower rates but restoring higher rates for the top brackets, in response to the state’s budget challenges.

• Illinois has decoupled from full expensing for machinery and equipment and updated its statutes to transition from taxing GILTI to the much more expansive NCTI.

Worth Reading

• Significant tax changes will go into effect in many states on January 1st. My Tax Foundation colleagues Nicole Fox and Jacob Macumber-Rosin have a round-up.

State Tax Competition Under Single Sales Factor Apportionment

New York policymakers appear to believe that, in an era of single sales factor (SSF) apportionment, corporate rates don’t matter. They’re mistaken—and understanding why is important for lawmakers everywhere.

New York Gov. Kathy Hochul has been guarded in her comments on the possibility of raising the state’s corporate income tax, but in private conversations with NYC Mayor-Elect Zohran Mamdani, her office reportedly takes the idea of raising the state’s corporate income tax rate to 11.5 percent seriously.

Mandani characterizes this as matching New Jersey’s top corporate rate, currently the highest in the country, but the comparison is misleading. The New York City metropolitan area has three layers of corporate taxation: the standard state corporate income tax rate (currently 7.25%), a 30 percent MTA surcharge (which functionally adds 2.175% to the rate), and New York City’s 8.85% corporate tax, for a combined rate of 18.275%. If the state’s rate increased to 11.5%, the MTA surcharge would be worth 3.45% for a combined increase of 5.525% and an all-in rate of 23.8% on corporate income apportioned to New York City.

To put this in context, just the proposed increase is about in line with the total rate in many states. At 23.8%, state and city corporate taxes would exceed the federal burden (21%) and roughly match the average burden of OECD countries.

Lawmakers sometimes convince themselves that corporate rates don’t matter since, under single sales factor apportionment (where the taxable share of a multistate corporation’s profits is assigned based on the share of sales into that state), companies can only avoid burdens by restricting their markets. Moving payroll or property doesn’t make a difference.

There is truth to this, but it ignores two important points: one, the tax discourages businesses from expanding physical operations to increase their sales footprint in local markets, and two, it has the potential to drive up prices for local consumers. These effects are largely hidden from residents, who won’t connect higher prices or reduced consumer choice to high corporate rates, but that doesn’t make them any less real.

Corporate tax incidence and its economic consequences is hotly disputed. A popular public finance textbook observes that “[t]he incidence and long-run economic effects of corporate income taxes is one of the most unresolved and controversial topics of public finance. The special aspects of state government use of corporate income taxes, especially formula allocation of the tax base, complicate matters still further.”1 Even while acknowledging competing estimates, however, we’re not left entirely in the dark. There are some things we can say with considerable confidence.

The first is that corporate income taxes are uniquely harmful to the economy. Studies routinely report employment, wage, and establishment elasticities of -0.4 or more, meaning that jobs, wages, and the number of corporate establishments each decline by more than 0.4% for each one percentage point increase in the corporate tax rate.2 These are economists’ results even with the prevalence of sales factor apportionment, which should attenuate some of the localized effects of corporate income taxation.

The second is that the burden of corporate income taxes falls on three groups of people: owners/investors, in the form of lower return on their capital investment; workers, in the form of lower wages; and consumers, in the form of higher prices. Estimates of incidence vary and can differ based on industry, state tax structures, how highly capitalized a market is, and more. But the incidence on owners and investors is far lower than most people likely believe.3

And the third is that apportionment matters—though not always in the ways policymakers anticipate.

Under formulary apportionment, the corporate income tax is essentially a factor tax: while its base is net income (profits), its channels are payroll, property, and (increasingly) sales. The economist Charles McLure long called the corporate income tax a set of three taxes on sales, wages, and property. Now, in most states, it is primarily a tax on sales. The upshot is that a substantial share of the tax’s incidence will come in the form of an implicit sales tax, raising prices.

Research bears this out. One study finds that every percentage point increase in a state’s corporate tax rate leads to a 0.24 percentage point increase in retail product prices. The researchers find that, when effects on prices are taken into account, 52% of the corporate income tax’s incidence falls on consumers, 28% falls on workers, and only 20% falls on shareholders.4

If anything, the 0.24 percentage-point increase in retail prices is a lower bound estimate, given the study’s design. The researchers employ an identification strategy that relies on comparing the prices of products sold in one state that are produced by firms with significant operations in different states, exploiting rate increases in the production states (which are exogenous to demand conditions in the destination state) and ascertaining whether prices rose in the destination states.

Given how many of the states in which production took place have single sales factor apportionment, and given that the study measures a rate increase in the production rather than sales market, an elasticity of 0.24 is notable. It’s indicative of a highly robust sales effect for corporate income tax increases, if it would show up even under these circumstances. In more concentrated markets (where there’s less competition), the researchers find an even higher price pass-through, with an elasticity of 0.31.

The study could not explore the impact of destination-state SSF corporate income tax increases due to its design. But if tax increases in other states raise prices, we should expect an even greater effect within the state that raised taxes.

Not everyone expects this, however. The economic development argument for single sales factor apportionment is that it diffuses costs and concentrates benefits. Because the tax falls on sales rather than in-state payroll or property, the tax falls on capital investment nationwide, but the taxing state gets the full benefit of the revenue.

There’s something to this. Sales factor apportionment doesn’t have the same disincentives for locating payroll and property in a given state that traditional three-factor apportionment did. But consistent economic findings that higher corporate rates hurt state economies, even under SSF apportionment, should tell us it’s far from the whole story. McClure and others provide a clue to what’s going on.

Even in today’s global economy, many C corporations are regional, and they experience the full brunt of a state’s high corporate income tax. And even many national and multinational corporations have business models that require them to enter specific geographic markets. Large brick-and-mortar retailers are an obvious example, though far from the only one. Many service businesses also need a meaningful local presence to operate in a given market. If selling into New York City yields a 23.8% tax on net income sourced to the city, that may well deter a large company from opening new stores there. The additional sales generate diminishing after-tax returns.

This is especially true due to how apportionment works. New York City is wealthy, but it’s also expensive. Some businesses might enjoy uniquely high profit margins in the City, but for many, margins will be lower there even if they remain profitable (and at high volume). Imagine a company has $1 billion in sales nationwide with a 5% profit margin ($50 million), and that 10% of its sales are in New York City ($100 million), but its profit margin there is only 3% ($3 million). Apportionment would expose $5 million in profits to the City, not $3 million, and tax them at high city and state rates.

That will affect businesses’ decisions to expand their footprint in a high-tax jurisdiction, as investments elsewhere could yield a far higher after-tax return. High corporate income taxes deter local investment even under single sales factor apportionment, because they reduce the after-tax gains from new sales in that market.

But what about online retailers and other remote sellers? Are they fonts of free cash for states with high CIT rates, since they are hardly likely to redline their markets, and the economic effects of the tax will be diffused nationwide? Not really.

Again, it’s important to remember here that online retailers and online service providers—however prominent when we think of big business—are a far cry from being “the economy.” More to the point, however, high state-level corporate taxes on online retailers affect that state’s economy, in the form of higher consumer prices.

We’ve known for decades that online retailers use variable pricing for many products, based on user location, interests, income, and a host of other factors. The precise details of algorithmic pricing are typically opaque. Still, given what we know about how corporate income taxes raise prices, algorithmic price discrimination is an obvious channel for adjusting prices to account for higher corporate taxes in some destination markets. When states and cities raise corporate income taxes under single sales factor apportionment, they are, to a substantial degree, taxing their own consumers.

Elected officials in New York and elsewhere might think that higher corporate taxes diffuse their adverse impact across the entire country, while concentrating the benefits of the revenue raised from those taxes. Research, however, aligns with theory: even with most states adopting single sales factor apportionment, high corporate rates hurt the taxing state’s economy, reducing gross state product, personal income, innovation, and purchasing power.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services. If you are in the market for tax policy research or know someone who is, please let me know.

Please Share this Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

This is wrong—very wrong.

Secretary Bessent doesn’t say that states’ policies could reverse the benefit of the new federal provisions, but I’ve now had multiple people tell me they’ve heard that message, so let’s be abundantly clear: whether or not a state incorporates these deductions into their own tax codes, they’re not doing anything to deprive their taxpayers of the full benefit of the federal deductions.

What Bessent does say is that a coterie of elected officials in New York, Colorado, Illinois, and the District of Columbia are thwarting state-level tax relief. It’s an odd list because it’s so capricious and incomplete. Only eight states are currently in line to bring in some or all of these new provisions: seven because the state’s tax code begins with federal taxable income rather than adjusted gross income, and one (Michigan) through legislation expressly conforming to the new deductions for tips and overtime, and the new senior bonus. To Bessent’s list, you could add every state with a Republican governor except those in the five Republican-led states that use federal taxable income as their income starting point.

In other words, New York Gov. Kathy Hochul, Illinois Gov. JB Pritzker, and D.C. Mayor Muriel Bowser haven’t done anything at all. They simply haven’t championed adopting these deductions voluntarily, just like most of their gubernatorial peers, including most Republican governors.

Colorado, meanwhile, is one of the seven states that begin their tax calculations with federal taxable income. While lawmakers made a few tax changes in a recent special session, they left the deductions for tips and vehicle loan interest alone, along with the new enhanced senior deduction. (Earlier, Colorado preemptively decoupled from the overtime pay deduction.) Governor Jared Polis, therefore, is the rare governor presiding over a state that does offer most of these deductions, yet he inexplicably makes Bessent’s naughty list.

These deductions are temporary, arbitrary, costly, and confer little economic benefit. There’s no compelling reason why a low- or middle-income tipped employee should pay less income tax than a non-tipped worker making the same amount. Similarly, it’s not obvious why overtime pay should receive better treatment than ordinary wage income, or why the tax code should put a thumb on the scale in favor of taking out larger car loans.

Conversely, the restoration of full expensing of machinery and equipment under § 168(k) and of research and experimentation costs under § 174 is extremely high-impact tax policy, yet the Secretary has trained his ire on (select) states that don’t proactively adopt these non-neutral personal deductions rather than on the states that actively decouple from neutral cost recovery.

As I’ve written elsewhere, the OBBBA gets business expensing right, and states should follow suit. You’d think these far more economically salient policies would be the ones worthy of the Treasury Secretary’s attention.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services. If you are in the market for tax policy research or know someone who is, please let me know.

A Request: Please Share This Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested, either by forwarding this to them or using the “share” button below. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak

SALT News and Updates

• California’s Attorney General is expected to issue a title for a proposed one-time 5% wealth tax ballot measure very soon, which would allow proponents to begin circulating petitions to place it on the November 2026 ballot.

Worth Reading

• My Tax Foundation colleagues Manish Bhatt and Abir Mandal have a new paper on P.L. 86-272, the 66-year-old federal law that limits state tax nexus for companies with only minimal contacts in a state. In recent years, some states have sought to erode the law’s protections by using, e.g., “cookie nexus” for physical presence. Manish and Abir argue against efforts to make the law a dead letter and make recommendations for strengthening and modernizing the law.

• There’s a good article in Tax Notes State on evolving deference standards post-Loper Bright, with a particular focus on California and Texas.

Some States May Accidentally Tax International Income Through Unintended NCTI Conformity

Lawmakers in many states voted to decouple from the tax on global intangible low-taxed income (GILTI)—but in some cases, those actions will do nothing to prevent their states from conforming automatically to its successor, the tax on Net CFC-Tested Income (NCTI). And if policymakers objected to taxing GILTI at the state level, they’ll really hate how NCTI functions in the states.

At the federal level, NCTI operates as a minimum tax, imposing compensatory U.S. tax on income that is only minimally taxed abroad. U.S. parent companies can claim tax credits for foreign taxes paid by their foreign subsidiaries (controlled foreign corporations, or CFCs). If the foreign subsidiaries are meaningfully taxed abroad, they don’t owe anything in U.S. federal tax. But states don’t offer foreign tax credits, so state-level NCTI taxation applies to an apportioned share of all income of foreign subsidiaries, regardless of where the income was generated or how much tax was paid on it abroad.

I have a Tax Foundation blog post highlighting how some states are at risk of accidentally taxing NCTI, but I want to use this venue to explain precisely how this happens.

Consider three different approaches to excluding the Tax Cuts and Jobs Act’s tax on global intangible low-taxed income (GILTI) under IRC § 951A from state tax codes.

• Iowa: “Subtract, to the extent included, global intangible low-taxed income under section 951A of the Internal Revenue Code.” Iowa Code Ann. § 422.35(12).

• North Carolina: “[Deduct] Any amount included in federal taxable income under section 78, 951, 951A, or 965 of the Code, net of related expenses.” N.C. Gen. Stat. § 105-130.5(b)(3b).

• Louisiana: “… there shall be allowed for each taxable year a deduction equal to the amount of dividends that would otherwise be included in gross income.” La. Rev. Stat. Ann. § 47:287.738(F)(1).

Did you catch the differences?

Iowa—along with Kansas, New Hampshire, and Tennessee—statutorily decouples from a particular provision called GILTI, which no longer exists. The new tax on NCTI is its successor and was written to the same code section, but NCTI is not GILTI, and the deductions those four states have for GILTI should not be assumed to capture NCTI.

North Carolina, by contrast, is one of fourteen states that exclude GILTI by statutory reference to § 951A or to a broader segment of the Internal Revenue Code that includes § 951A. In these states, subtractions or deductions for GILTI will provide the same exclusion for NCTI, since it is drawn to the same statute.

Louisiana, meanwhile, is one of twelve states where GILTI was excluded from the tax base through a determination that it constituted dividend income or Subpart F income and was thus eligible for deductions related to dividend or Subpart F income.

The same logic that underlies tax administrators' treatment of GILTI as dividend or Subpart F income should also apply to NCTI, though that is not guaranteed. (Arizona is of particular note because the statute explicitly identifies “global intangible low-taxed income” as constituting dividend income. That reference will not apply to NCTI, though arguably their general rule should have applied to GILTI—and would therefore apply to NCTI—even without that explicit language.)

GILTI and NCTI are not, strictly speaking, Subpart F income, which is an income class for income from a CFC that U.S. investors must include on their own income tax returns. However, federal law stipulates that GILTI and NCTI income “shall be treated in the same manner as” Subpart F income, and the two states that exempt GILTI (and now, presumably, NTCI) as Subpart F income both pick up that federal requirement.

Similarly, GILTI and NCTI are not necessarily dividends under the definition in IRC § 316. Importantly, however, they comport with the definition of deemed dividends as developed in court decisions and administrative rules. All states agreed that GILTI should be treated as dividend income. In states that offer a dividends-received deduction, GILTI received the benefit of that deduction. That’s true even in states that tax GILTI: some states only deduct a given percentage of dividend income, which kept some GILTI in the tax base.

States that offered a deduction for GILTI by classifying it as dividend income should continue to do so under NCTI. Still, it wouldn’t hurt for policymakers to gain assurances on this point.

Finally, even states that already taxed GILTI and now tax NCTI might want to scrutinize the way they conform. The ideal policy is to decouple, but even where NCTI is intentionally taxed, anachronistic references to GILTI can complicate matters. In Oregon, for instance, both the addback of the § 250 deduction and the provision of a partial dividends received deduction make specific reference to GILTI.

Of course, if you’re not particularly familiar with GILTI or NCTI, all of this might be pretty opaque. Let me close by quoting at length from one of my prior Tax Foundation blog posts discussing the state implications of a transition from GILTI to NCTI:

GILTI and the States

Prior to the Tax Cuts and Jobs Act (TCJA) of 2017, the US taxed the worldwide income of US corporations and their affiliates, including controlled foreign corporations (CFCs) based abroad, while allowing such companies to take credits against their US liability for foreign taxes paid. Under the TCJA’s territorial tax system, foreign income is not taxed by default, but Congress wanted to avoid profit-shifting activity in response. GILTI was intended as a minimum tax on certain foreign earnings, undermining the potential tax benefit of such profit shifting. The new NCTI regime arguably provides a better calibration at the federal level, but a far worse one for states incorporating the provision into their own tax codes.

Under GILTI, federal law sought to distinguish between routine and “supernormal” returns, with lawmakers postulating that a CFC’s returns above 10 percent of the value of its tangible assets very likely constituted income from intangibles (e.g., patents, trademarks, copyrights, and other forms of intellectual property from which royalties are derived). This initial 10 percent was excluded under the qualified business asset investment (QBAI) exclusion, which represented a rough-and-ready deemed return on physical capital. Any returns above 10 percent were subject to GILTI.

The remaining income, subject to GILTI, received both a deduction (initially 50 percent) and an offset worth 80 percent of foreign taxes paid. The deduction meant that the US tax rate on GILTI was lower than the rate on US income, reflecting the fact that it was earned abroad and is not an ordinary part of the tax base. The 50 percent deduction under § 250, therefore, functionally turned the 21 percent corporate income tax rate into a 10.5 percent rate on GILTI. (The deduction was scheduled to decline to 37.5 percent in 2026, yielding a 13.125 percent rate.) Actual foreign taxes paid, moreover, yielded foreign tax credits, and 80 percent of their value could be applied against GILTI. The system was far from perfect, but it was designed to tax CFCs’ income to the extent that it was “undertaxed” abroad, potentially (but not always) indicative of US tax avoidance rather than genuine economic activity in other countries.

Unfortunately, when states incorporated GILTI after the enactment of the TCJA, parts of this system immediately fell apart. The federal provisions were not adopted with states in mind, and states’ corporate apportionment rules weren’t designed to handle foreign income.

The foreign taxes generating the federal credits were paid by the foreign corporations that US-based multinationals owned or in which they had a substantial investment stake. As a pure matter of accounting, if the US shareholding entity is to be treated as having paid those taxes itself, an equivalent share must also be included in the company’s GILTI income, to avoid a double benefit. The 80 percent of credited taxes are first included in the US company’s income for GILTI purposes under the § 78 “gross-up,” and then the credit is applied. At the federal level, that all worked. But states rarely allow foreign tax credits, while they do conform to the gross-up, so their GILTI bases included 80 percent of the value of taxes paid by CFCs abroad, without the tax credits that gross-up was intended to facilitate. Rather than reducing tax liability based on foreign taxes having already been paid on the income, states’ GILTI regimes tax this income more because of the foreign taxes paid on it. The foreign taxes paid by the CFCs of corporations doing business in a state are not, of course, even remotely US corporate profits, which is what state corporate income taxes are intended to tax.

Converting to NCTI

For states, converting to NCTI makes the problem worse. To begin with, it eliminates the QBAI exclusion, bringing all the income of CFCs into the GILTI/NCTI base rather than just the “supernormal” returns. At the federal level, this base expansion is substantially offset through other changes, but for states, some of them turn into tax multipliers instead. Additionally, under NCTI, the § 250 deduction (which had been at 50 percent but was scheduled to fall to 37.5 percent in 2026), is made permanent at 40 percent, which has the effect of increasing states’ effective rates on this broader base.

But NCTI changes far more than this. Previously, under the GILTI regime, many expenses by US-based multinationals were sourced to their CFCs to the extent that they were deemed to benefit them. Since these business expenses would have ordinarily been deductions from the US company’s taxable income, these expense allocation rules (1) increased US tax liability for US-based multinationals, since they were denied deductions for some of their business expenses; but, at the same time, (2) provided deductions for the CFCs, reducing their taxable income potentially subject to GILTI.

On net, businesses would have preferred to have the deduction for their US-based corporations, as the ordinary rate is higher than the GILTI rate and because foreign taxes paid on ordinary business activity of CFCs abroad often yield credits in excess of GILTI tax liability, leaving some credits unused. (GILTI is, after all, a minimum tax. Foreign tax liability can often exceed its minimum.) Under NCTI, changes in expense allocation rules mean that more of these deductions are taken by the US parent corporation. Consequently, there are fewer deductions for their CFCs, yielding a larger NCTI base than the old GILTI base. That’s a welcome shift for many corporations with significant tax liability in other countries, because they get the benefit of the US deductions and can use more of their foreign tax credits against the new NCTI base. Simultaneously, the new law reduces the limitation on foreign tax credits (called the “FTC haircut”), raising the inclusion amount from 80 to 90 percent with a commensurate increase in the § 78 gross-up.

It’s not hard to imagine where this goes wrong at the state level. The NCTI base expands yet again, and the greater allowance for foreign tax credits, rather than offsetting liability, is picked up as additional income to be taxed. All four major changes—scrapping the QBAI exclusion, adjusting the § 250 deduction, trimming the FTC haircut with a commensurate increase in the § 78 gross-up, and revising expense allocation rules—make state taxation of NCTI more aggressive than state taxation of GILTI, whereas these changes yield a net tax cut at the federal level.

Obligatory Marketing Note

My new consultancy provides tax policy research, writing, and other services. If you are in the market for tax policy research or know someone who is, please let me know.

A Request: Please Share This Substack

If you find this Substack valuable, please do me a favor and share it with colleagues and others who may be interested, either by forwarding this to them or using the “share” button below. And if you haven’t yet subscribed (it’s free), please consider doing so!

Share The Salt Road - Jared Walczak