income tax

Tax tail wags investment dog

Capital gains and ETFs

Tax considerations drive the rapid growth in ETF assets according to a recent paper by three business professors, Rabih Moussawi, Ke Shen, Raisa Velthuis, “The Role of Taxes in the Flow Migration from Active Mutual Funds to ETFs,” (December 2020). The paper is interesting to a tax geek like me. It is worth reading, even though it is pretty long (over 70 pages with tables, graphs, notes etc.). Disclaimer: the paper is posted on SSRN, but has not been peer reviewed or published by a journal. However, the research methods look solid to me, although I’m unfamiliar with the data they use and am not an econometrician.

ETFs are more tax efficient than standard mutual funds because their structure and operating practices allow deferring capital gain taxes (or entirely avoiding them for those affluent enough to use the Buy/Borrow/Die avoidance plan, as described by Professor Edward McCaffrey). Their results were surprising to me and are concerning on a couple fronts. If accurate, they augur further erosion of the income tax base and will reinforce the trend toward more inequality by further undercutting the taxation of high-net-worth individuals.

ETFs came on the scene in the early 1990s and were initially a niche product, an open-end mutual fund that trades on the stock market. In the old days, only closed-end funds traded on stock exchanges; open-end funds were purchased with their prices set once per day. ETFs have grown dramatically in popularity over the last decade and US ETFS now have over $5.5 trillion in assets (traditional funds have about $18 trillion). Some observers think ETFs will eclipse and may supplant traditional open-end funds. See John Rekenthaler, “Farewell Mutual Funds” (Morningstar 1/21/21) for one account (1/25/21 follow up column responding to critics).

I am modestly familiar with ETFs and have personally invested in them for over a decade. I had always thought their tax advantage was a minor attraction because ETFs initially consisted mainly of passive, index products and traditional index mutual funds were already tax efficient. The main attractions of ETFs, in my mind, were their lower expense ratios (why they attracted me) and the ability to trade them in real time, not just once per day as with traditional open-end funds. The article says otherwise and seems convincing (at least to me).

ETF’s tax advantage

ETFs’ tax advantage stems from their ability to avoid making capital gain distributions when the ETF trades the securities it holds. Traditional mutual funds pass most of their net realized gains through to their shareholders, even though the shareholder has not sold any of her mutual shares. Both traditional mutual funds and ETFs are taxed under the same federal tax rules (i.e., as regulated investment companies) on a pass-through basis like partnerships and S corporations. So, when capital gains are realized, they are deemed to be distributed pro rata to their shareholders. In theory both ETFs and traditional mutual funds can avoid distributing capital gains by making “in-kind” distribution of the securities they wish to trade to their shareholders (rather than giving them cash for redemptions) or authorized participants (for ETFs). But for structural reasons only ETFs routinely do that, thereby avoiding distributing most of their realized gains to their shareholders. Why that is so requires a little explanation.

Two typical events trigger mutual funds to realize capital gains. If a fund’s shareholders sell or redeem their funds (i.e., withdraw their money), the fund may need to sell some of its holdings to raise cash to pay them. That typically occurs only if a lot of shareholders “head to the exits” selling their shares when a fund underperforms or otherwise falls out of favor and the redemptions exceed their usual cash holdings. Obviously, the typical shareholder wants money, not to be paid off “in-kind” (i.e., to be tendered shares of the securities the fund holds). If the shares that are sold have appreciated over the fund’s tax basis in them, capital gains are triggered. Second, repositioning the fund’s portfolio may trigger gains. For a managed fund, this occurs when the investment managers no longer wish to hold the securities. For an index fund, it happens because the index changes. For example, Tesla gets added to the S&P 500. That is a more common case with smaller or niche indexes, such as a small cap, smart beta, or sector funds, since the composition of narrower indexes is more fluid than the S&P 500 or a total stock market index fund.

By contrast, an ETF shareholder does not get money directly from the fund, but instead sells her shares on the stock exchange. To maintain liquidity in the ETF shares and to deal with circumstances when investment in the ETF increases or decreases, the ETF contracts with an authorized participant or AP (or market maker) to supply and take the securities the ETF invests in and to make a market in the ETF stock – such a mechanism is necessary to allow an open-end fund to trade in real time. Thus, when an ETF’s investors sell shares on net, it will tender some of the portfolio’s underlying securities to the AP. Low basis securities are tendered (consistent with the portfolio construction obviously) to bleed away the capital gains. This structure and practice mean when a lot of an ETF’s shareholders head to the exits, capital gain distributions will typically not be triggered for the remaining shareholders, as is the case with traditional mutual funds. Somewhat counterintuitively, a lot of shareholders selling creates an opportunity for the ETF to tender its low basis securities and to reduce the potential for future capital gains. (That connection is not something that had occurred to me but is obvious on a moment’s reflection.)

What I did not realize and what the article uses to test tax sensitivity is a second technique, “heartbeat trades.” If capital gains would be triggered by realigning the portfolio (e.g., a managed portfolio or because of changes in the index) or other events, the ETF can engage in a second mechanism whereby the AP buys ETF shares, pumping money into the ETF that is used to buy more securities for its underlying portfolio. Shortly thereafter (but at least two days later to satisfy the tax law) the AP sells ETF shares. That outflow is satisfied by the ETF tendering low-basis securities (i.e., not the ones just purchased) to the AP, draining potential capital gains away from the ETF. These are referred to as “heartbeat trades” because when you graph them, they show up as spikes similar to graphing an EKG heart monitor’s output.

The research and findings

ETF assets have grown dramatically. The paper reports US equity ETFs had assets under management of about $2.4 trillion in December 2019. A year later that had grown to $3.2 trillion. The authors’ hypothesis is that ETFs’ tax advantage (the ability to defer capital gain taxes and to entirely avoid them for those passing on ETF shares at death) is a principal driver of the growth. To test the hypothesis, they analyze a sample of mutual fund data assembled from the SEC, Center for Research in Security Prices, Lipper, and other sources, including data on heartbeat trades and holdings by advisors to high-net-worth investors. The strategies use sophisticated statistical analysis to determine (1) the general relationship of fund performance, expense ratios, and taxes on investment flows (i.e., to which fund types is money flowing), (2) differences between high- and more average net-worth investors, and (3) the impact of the capital gains tax rate increases in 2012-13 as a natural experiment. All the results point to taxes as the big factor in driving the flow of money into ETFs, particularly out of traditional active or managed mutual funds (where investment managers pick stocks to “beat the market”).

The first step obviously is to calculate/estimate the relative tax advantage of ETFs versus managed mutual funds and index funds. To do this they determine the differences in capital gain distributions (broken down by short and long term, since the former are subject to higher taxes) relative to the funds’ net realized gains, adjusted for investment styles. They then convert the differences into estimates of tax for top-bracket taxpayers and express it as a percentage of the funds’ assets. That makes the burden like an annual property tax and conveniently makes it equivalent to a fund’s expense ratio. The descriptive statistics showing the ETF advantage are quite striking. I expected that for managed funds, where higher portfolio turnover typically generates more capital gains, but it is also true for passive or index funds. Over the period they analyze, the tax advantage compared to actively managed funds was 86 basis points or bps (0.86% of assets) and for the last five years (2013 – 2017) was 112 bps. (pp. 22 – 24) Obviously, the difference for index funds was smaller. They make the point that the advantage roughly is of the same size as the expense ratio advantage that index funds and ETFs have compared with actively managed funds. (As an aside, reputable investment and personal finance advisors have been harping on the importance of seeking out low-expense funds as the one sure way to increase returns – finding high-return investments, by contrast is a fraught exercise. Finding a fund or advisor who is consistently above average is next to impossible over the long run, unless they office in Lake Woebegone. By contrast, expense ratios are transparent and easy to compare.) As they observe, “For all investment styles, ETFs distribute less than 0.14% in capital gains, a trivial amount compared to [traditional] mutual funds.” (p. 24). Heartbeat trades are a big contributor to this, particularly in reducing short term gains, which otherwise would generate the highest tax burdens.

Their graph of the annual amounts between 1999 and 2017 shows the stark advantage and how ETFs (solid blue line) systematically keep capital gain distributions low, even though they have realized net gains that are about the same as comparable traditional funds according to the authors’ data.

Effects on flows by fund type. The authors use data on distributed and realized gains by funds and information on fund holdings to test whether performance, fees or taxes are the biggest determinant of which type of fund attracted money using regression analysis. They found that taxes were the strongest factor driving the outflow from managed funds and into ETFs. Managed funds with larger distributions of capital gains were more likely to have outflows. Tax sensitive investors headed for the exits after funds announced their capital gain distributions but before the actual distribution dates.  One of the anomalies of mutual fund mechanics is that traditional funds announce capital gain distribution amounts well before distributing them. As a result, investors with a high basis in the fund shares can exit before the distribution date and avoid tax on the distributions (if their bases in fund shares are well below the market this strategy doesn’t work obviously). The authors used this data to analyze outflows.

Sorting active traditional funds by those with higher capital gain distributions (i.e., those triggering more tax liability for their shareholders) revealed that these funds tended to have larger outflows, while ETFs of the same style/investment type experienced large inflows. By contrast active funds with the lowest distributions paired with flows into traditional passive, index funds. As the authors observe:

This tax-related flow migration is not an active to passive phenomenon, but strictly a flow migration into ETFs.

Rabih Moussawi, Ke Shen, Raisa Velthuis, “The Role of Taxes in the Flow Migration from Active Mutual Funds to ETFs,” (December 2020), p. 34

ETFs and high-net-worth investors. To further test the hypothesis, the authors analyzed data on the use of ETFs by advisors to high-net-worth individual investors who they (and I) presume are tax sensitive. They found a strong correlation:

[A]llocations to ETFs by investment advisors of high-net-worth clients are nearly four times more than investment advisors with low or no high-net-worth clients and have reached 32.4% of the overall 13F [investment manager] assets managed by these advisors in 2017, compared to less than 9% for other investment advisors, respectively.

Rabih Moussawi, Ke Shen, Raisa Velthuis, “The Role of Taxes in the Flow Migration from Active Mutual Funds to ETFs,” (December 2020), pp. 35 – 36.

Their graph, perhaps, shows it best.

2012-13 capital gain tax increases. Finally, they used the two capital gain tax increases Congress enacted in the Affordable Care Act and the American Taxpayer Relief Act of 2012 as a natural experiment to test the effect on flows into ETFs (difference-in-difference regression analysis with ETF allocations of advisors of high-net-worth individuals as the dependent variable). They found a big effect:

[W]e document an overwhelming increase in allocations and flows into ETFs [by advisors of high-net-worth individuals] relative to advisors with lower fractions of high-net-worth clients especially after the increase in capital gains tax rates after 2012. * * * Overall, our evidence points to the dominant role of ETF tax efficiencies behind the massive outflows from active mutual funds and the dramatic surge of flows into ETFs in recent years.

Rabih Moussawi, Ke Shen, Raisa Velthuis, “The Role of Taxes in the Flow Migration from Active Mutual Funds to ETFs,” (December 2020), p. 39.

They finally conclude that:

Without a doubt, the tax efficiency of ETFs is likely to continue exacerbating the active-to-ETF flow migration and inevitably lead to more mutual fund conversions for tax purposes.

Rabih Moussawi, Ke Shen, Raisa Velthuis, “The Role of Taxes in the Flow Migration from Active Mutual Funds to ETFs,” (December 2020), p. 41.

Policy implications

The authors note the resulting unevenness of the taxation of investors in different types of fund structures. A main concern is the resulting inefficiency. Of course, the fund industry (Investment Company Institute position) has for years advocated to exempt reinvested capital gain distributions from capital taxes until the fund shares are sold. That would resolve the inefficiency and inequity by blowing an even bigger hole in the income taxation of capital income. So far Congress has not appeared to be interested; at least, the provision did not make the cut for any of the various versions of the TCJA Congress seriously considered.

They estimate that the current level of domestic equity ETFs will result in reduced federal tax collections (by deferring capital gains) of $400 billion to $679 billion over the next decade. (pp. 40 – 41) Those calculations are based on ETF AUM of $2 trillion and the authors say they ignored future flows into ETFs. (p. 41) Since that equity AUM has already increased to $3 trillion as of December 2020, I assume that their numbers should be increased by 50% or so to adjust for just that growth. If they are correct about the migration to ETFs continuing, the federal revenue impact will grow. The amount could easily exceed the amount of the estimated tax cut under TCJA, if their numbers are right (I don’t have a high degree of confidence in that) and flows into ETFs continue, substantially replacing traditional funds in most taxable accounts. There are limits to that migration, of course – primarily low-basis investments in traditional funds whose redemptions would trigger big capital gain liabilities.

My observations

  • The wisdom of using ETFs, rather than traditional funds, for one’s taxable investments seems obvious. I plan to take that to heart as I convert RMDs to taxable investments.
  • Their results provide yet one more illustration of how difficult it is to tax capital and particularly capital gains.
  • Using an ETF structure will allow high-net-worth individuals a workaround for one of the biggest challenges of holding low basis securities until death (the lock-in effect and resulting lack of portfolio diversity).
  • It is one more piece of evidence supporting the need to eliminate the step-up in basis on death, which creates a big hole in capital taxation.  Ideally, Congress would tax gain on death as both Obama and Biden had/have proposed, but I don’t expect it to happen any time soon if ever. A more modest fix would be to make heartbeat trades more difficult to do (expensive because APs will bear more market risk and charge more for their services) by increasing the minimum 2-day holding period to something much longer (why not apply 30 days, which is used for wash sales, e.g.?).
  • The relatively low effective tax rates on capital income are, of course, an obvious contributor to the growing inequality. The ETF effect will only make matters worse. However, my instinct is that society would be better off following the model used elsewhere in the developed world – i.e., a healthy consumption tax (VAT obviously) that funds a more generous social safety net – and reduce the focus on trying to make the income tax ever more progressive as the primary way to address income inequality. It simply doesn’t work for reasons of practical political economy. I think Ed Kleinbard was generally right that focusing on the overall fiscal system, rather than the progressivity of the tax system, is preferable We Are Better than This (chapter 12)
  • Minnesota relies more heavily on revenues from taxing capital gains than most states (nothing like California, of course). It relies heavily on the income tax and taxes long term capital gains at the same rate as ordinary income. An obvious and often discussed drawback of that is revenue volatility, since revenues hang on the ups and downs of the stock market. But the paper illustrates another drawback – the potential for long-term erosion. Fortunately, capital gain distributions from regulated investment companies are not a huge part of the overall tax base, but they are important and every bit matters.
estate tax income tax

January miscellaneous stuff

This post is a cats-and-dogs collection of stuff that I read last month and wanted to archive and highlight. A couple relate to one another, the rest are just random pieces that interested me:

  • Proposal to use the SALT deduction tax expenditure for a State Macroeconomic Insurance Fund
  • Taxpayer Advocate’s 2020 annual report
  • Taxing corporate stock buybacks
  • Taxing capital gains on death
  • Due process limits on state estate taxation of QTIPs
  • PPP loan forgiveness taxation (yet again)

Redeploying SALT deduction tax expenditure

I previously blogged about the cap on the SALT deduction. I don’t like either the current or the earlier, uncapped version of the deduction. Neither passes a rudimentary test of good policy. Restoration of the uncapped SALT deduction has history, inertia, and congressional Democrats behind it (restoration was in the HEROES Act but is not in the Biden virus relief plan).

Len Burman, Tracy Gordon, and Nikhita Airi, three TPC staffers, have a proposal out to instead use those dollars (the tax expenditure cost of restoring the uncapped deduction) to fund a countercyclical state aid fund, which they call the State Macroeconomic Insurance Fund (SMIF). You can read the outline in this Tax Vox blog post or listen to Gordon present it in this Brookings webinar (the rest of the presentations are also worth a listen). Their proposal is well-thought out and makes sense as a way to address the policy problems created by the impact of recessions on state and local governments, given their balanced budget requirements, the practical and political difficulties of maintaining adequate reserves, the feedback effect of laying off government employees deepening recessions, and the negative macro-economic effects of state tax increases in a recession. (Federal conformity, as I have pointed out, can also create problems – especially for states with rolling conformity but even for those who conform on a static basis.)

Their proposal puts the amount of revenues that otherwise would be lost from restoration of the uncapped SALT deduction in a federal fund that automatically pays federal aid to states in recessions, scaled to how hard the recession affects the state under neutral measures such as increasing unemployment. The aid could be used to leverage or encourage the maintenance state reserves by requiring states to fund their reserves to qualify for more federal aid. Their blog post is short for anyone interested in more detail.

Like many good tax policy ideas this one has no political viability (the webinar discusses that with a wishy-washy Pollyanna response):

  • It will be a no-go for Republicans; they will likely view it as a blue state bailout (if their response to the proposals to pay coronavirus aid to replace lost states revenues is any guide) even though much of the aid would go to red states they represent (same would be true of coronavirus aid). Moreover, after 2025 (when the cap on the SALT deduction expires), it would violate Grover Norquist’s tax pledge (i.e., they would view it as a tax increase because the cap would stay in place and the resulting revenues used to fund the account would be a tax increase for government spending) and be verboten on that basis.
  • Democrats will oppose it because most of the revenue will come from blue state taxpayers who are the predominant beneficiaries of an uncapped SALT deduction and the benefits of the spending will be spread over all states, red and blue (a good thing, of course, but not what Chuck and Nancy would want). Democrats will also not like it because it strips away the buffering effect that an uncapped SALT deduction has on high state and local taxes for their affluent constituents (donors). That is likely a prime reason that Republicans capped the deduction in the first place.

But it is still worth reading and dreaming about what could be possible in a parallel universe – e.g., if Congress focused on good policy, rather than what is politically appealing and what each party’s political priors and base tell it to do. The Minnesota legislature did that in the late 1960s and early 1970s when it came together on a bipartisan basis and enacted fiscal disparities, the Metropolitan Council, the Pollution Control Agency, the Minnesota Miracle, and so on. It is nice to dream but then you wake up and move on.

Taxpayer Advocate’s annual report

The Taxpayer Advocate’s 2020 report is available on the IRS website; it is great reading for tax nerds like me. I always learn something, often a lot, by skimming through parts of it. For example, it costs the IRS $4.78 to process a paper return and $0.18 for an electronic return. The IRS annually spends $37 million handling, shipping, storing, and retrieving data from paper returns. More than one-half of paper returns are prepared using tax software. (I assume most of them are filed as paper returns because they did not meet the IRS efiling tests or the taxpayers were unwilling to pay efiling fees to software firms.) IRS employees manually enter data from returns, such as the 1040, leading to transcription errors. (All the preceding is from this document.) The challenges the IRS faces in this environment are daunting to put it mildly. Woe be it to you, if you are owed a refund and your return was flagged for some (potentially innocent) reason. It may be a while before you get your refund.

The Purple Book is chocked full of good suggestions for improving the IRS’s performance and the tax system. The recommendations to Congress typically go unheeded, such as to appropriate adequate funds for IRS’s operations and to modernize its antiquated IT systems. At a minimum the IRS needs a system that scans returns so employees no longer need to manually type in names and numbers off of 1040s and its myriad schedules (at least for those prepared by software).

Interesting but sobering reading – core administrative functions for our tax system, essential to the operation of government, are teetering on the brink while Congress fiddles. Maybe with the change in administrations and in control of the Senate, things will improve. I’m not holding my breath. In any case, it is going to take a sustained, multi-year effort to rebuild the agency. Politics militate against any kind of sustained effort with a low-visibility payoff.

Time to tax stock buybacks?

Several of the Democrats running for president (but not Biden) proposed taxing corporate buybacks. In the last few years I was working I occasionally got questions about the possibility/advisability of doing that (a few times from legislators). I always pooh-poohed it and counseled against it – if excessive buybacks were a problem (unclear to me), my perception was that it was best not to address through the tax system and certainly not at the state level. Once capital gains and dividend taxation were equalized (in 2003 at the federal level and in 1987 in Minnesota but in different ways), I didn’t think there was any tax inequity or problem involved.

This article by two tax professors, Daniel Hemel and Greg Polsky, Taxing Buybacks, Yale Journal on Regulation, vol 38 (2021), has caused me to rethink that and conclude that there is good case to be made for changing the rules on how buybacks are taxed. It’s obviously not going to happen with this Congress or probably ever, but the article is worth reading if only to understand the nuances of the tax rules and the economics of buybacks. They resurrect and tweak a proposal by a legendary tax professor, Marvin Chirelstein, from a 1969 Yale Law Journal article – back then the differential treatment of dividends (ordinary income) and long-term capital gains (partial exclusion) provided justification for changing the rules. Since that is no longer the case, they need to and do repurpose his proposal.

Hemel and Polsky point out that adapting Chirelstein’s proposal (essentially taxing buybacks as deemed dividends paid to all shareholders pro rata; they add a tweak of requiring cash payment of a dividend equal to the tax on the deemed dividend to address the “phantom income” resulting from a deemed dividend) would help address two problems:

  • The “Zuckerberg Problem” (a term coined by Ed Kleinbard) – i.e., that large amounts of labor income embedded in founders’ stock (hence, the term from the founder of Facebook; you could use Elon Musk, Bill Gates, or a host of others with large or small fortunes) goes untaxed because of the step-up in basis on death or on charitable contribution of the stock to tax exempt foundations. Those folks largely escape income taxation on vast fortunes by following what Ed McCaffery calls the Buy/Borrow/Die tax avoidance plan. I was aware of this problem but never thought of taxing buybacks to get at it. See the next topic below for a more direct and complete way to address that, though. Hemel and Polsky point out that the two approaches are complementary to one another. Carried interest presents a similar problem that has attracted more political attention and Congress appears unwilling to act – just to provide a dose of political reality as an aside.
  • The “Panama Papers Problem” – that is, that tax haven investors hold large amounts of US publicly traded stock (about 9% by some estimates). These holders are often tax cheats as the Panama Papers have revealed. When dividends are paid, they don’t escape taxation because dividends paid to foreign investors are subject to mandatory 30% withholding. Capital gains, by contrast, escape taxation. Converting buybacks to dividends (per Cherelstein’s proposal) for tax purposes would end that and generate a fair amount of revenue. The authors estimate $27 billion/year (probably high because behavioral responses would cut into that as they migrate to other avoidance mechanisms). A surfeit of revenue would also be generated by tax on other foreign investors in OCED countries.

Taxing unrealized gains at death or on gift

In their article, Hemel and Polsby observe that “The most straightforward way to address ‘Buy/Borrow/Die’ is to repeal section 1014, the code provision that allows for stepped-up basis at death. We agree with that prescription, and Chirelstein did too.” (p. 300, footnotes omitted). Harry L Gutman, Taxing Gains at Death (Tax Notes Federal, Jan. 8, 2021), describes how to do that in way that may be more politically palatable than his experience with the 1970s enactment of carryover basis. Unlike most Tax Analyst content, the Gutman article is ungated.

Carryover basis was enacted in the 1976 Tax Reform Act but was repealed before it became effective. (Taxing gains at death and adjusting the estate tax is preferable to carryover basis in Gutman’s view. I agree.) Gutman was at Treasury during the Carter administration and directly involved in the attempt to implement and unsuccessfully defend that regime. As a result, he brings a host of knowledge of what is practical and political possible. However, he recognizes the political challenges, characterizing his effort as a decision “to mount Rocinante and tilt at this particular windmill.” Gutman has been a private tax lawyer at a DC law firm since leaving Treasury (I believe without checking that).

This – like taxing corporate buybacks – was a proposal of various Democratic presidential candidates, including Biden. Gutman points out that it has been proposed by both Democratic and Republican administrations (most recently by Obama). Carryover basis was signed into law by President Ford. But Ford would likely be unwelcome in today’s Republican Party.

None of the campaigns provided any details on how they would have actually done it, of course. Gutman does that and his experience defending carryover basis makes him especially competent to focus on the possible and practical (second best solutions), rather than the typical tax professor type who focuses on the theoretical best solution. None of this will happen, of course, with an equally divided Senate and the thinnest of majorities for the Democrats in the House. In my opinion, it would be one of the most desirable reforms Congress could enact and would allow further deemphasizing or even repealing the estate tax. If enacted, it would enable states like Minnesota to conform and, then, repeal or dramatically reduce their estate taxes; stepped-up basis is the strongest argument for maintaining an estate tax. It’s why the double taxation argument typically raised against estate taxation is flawed – unrealized appreciation on which income tax was never paid is the largest component of taxable estates, especially the really big ones. A prime function of the estate tax is as a backup to the income tax; a function it doesn’t perform well with the current gargantuan exemption ($11.7 million). Because Gutman’s version would apply (as would any sensible alternative proposal) to inter vivos gifts, it would solve the problem that almost all state estate taxes have – no complementary gift tax (Connecticut is the exception). The lack of a gift tax allows the very richest to dramatically minimize state estate taxation by transferring much of their estates via gift.  My 2019 post, Tale of Two Billionaires, provides an example of this (i.e., Carl Pohlad v. James Binger).

Gutman focuses a lot of his attention on transition rules and hard to value assets, such as closely held businesses, other than marketable securities. On the latter, he would defer tax until the property is sold or transferred. His experience (consistent with my legislative experience) suggests that being flexible on transition rules is often a key to a politically viable proposal. Tellingly to me, he observes (in the last footnote to the article): “It is my understanding that had Treasury agreed to apply carryover basis only to assets acquired after the effective date, the provision would not have been repealed. And today, 40 years after its repeal, it would be virtually universally applicable.” If valid, that is a testament to short-sightedness of its advocates in the 1970s, that is, of the best being the enemy of the good. When Canada implemented its system, it gave up taxing appreciation that had occurred before enactment (Gutman does not mention that; his proposal is even more generous in grandfathering assets, not just appreciation).

A couple of Gutman’s quotes are worth repeating:

The Joint Committee on Taxation lists more than 230 income tax provisions as tax expenditures. An economic or social policy objective can be cited for virtually all of them. However, try as one might, no one can create a plausible tax, social, or economic policy justification for tax-free step-up.

That might be a bit of an overstatement (there are some other tax expenditures with little or no justification for them, in my opinion) but not much.

In addition, to addressing the “Zuckerberg problem” of billionaire’s whose labor income embedded in founders’ stock goes untaxed, Gutman provides an example that captures the essential inequity (horizontal flavor) of stepping up basis on death:

A and B are siblings. Each bought Stock X for $100,000. It is now worth $3 million, and each has decided to sell. A meets B in the street outside their broker’s office just after A has executed her trade and before B is going to do the same thing. A car hits them and both die. Assuming a 20 percent income tax rate and no estate tax, B’s heirs receive $2.42 million. A’s heirs get the stock with a new basis of $3 million and can sell it the next day and pocket the entire $3 million. That’s indefensible.

If Congress ever gets serious about dealing with this problem, Gutman’s article is a good roadmap describing how to do it and is an easy, interesting reading for anyone interested in the issue.

Due process restrictions on estate taxation

Since the Supreme Court decision in Kaestner Trust case, serious legal questions lurk about states’ ability to impose income taxes on trusts’ retained income if the trust is not clearly domiciled or managed in the states, even if the settlors or beneficiaries are or were residents. The Court’s opinion mainly reaffirmed, rather than clarifying, the murky status of prior law (a status I mistakenly had thought obsolete because of due process decisions made in other contexts). Thus, the constitutional limits imposed by due process remain unclear, at best. Minnesota is in the middle of this, having lost a somewhat comparable case to Kaestner in the Minnesota Supreme Court (Fielding). Governor Walz’s budget proposes to address that (pp. 20-21) in some way that is not yet clear to me. I’ll wait to see a bill draft.

The challenges posed by Kaestner, however, are not limited to income taxation. Implications for estate and gift taxation also lurk. A November State Tax Notes article by a giant in state taxation, Walter Hellerstein who is a coauthor of the standard treatise, addresses this question. Walter Hellerstein and Andrew Appleby, “State Estate Taxes and the Due Process Clause” Tax Notes State, vol. 98, pp. 771-77 (November 23, 2020). I had put off reading it and was disappointed when I finally got around to doing so. I had hoped Hellerstein and his coauthor would analyze and apply Kaestner but the article largely consists of their summarizing (with some minor commentary) four state tax decisions on states estate taxes and QTIP trusts. Still useful, but less than I had hoped. Oh well. Only one of the opinions explicitly addresses Kaestner’s implications.

The fact patterns of the cases are similar, and all the courts reached the result that the states can tax the QTIPs’ intangible property. QTIP trusts are an estate planning device that operate as follows.  When the first spouse dies a limited interest trust is left to the surviving spouse. Because it is a partial interest, the trust property would not normally qualify for the marital deduction. But because it is “qualified” (QTIP stands for qualified terminal interest property) it does. So, the QTIP is not taxed when the first spouse dies and the survivor gets an income interest. In all the cases, the surviving spouse moved to a different state in which she (all the survivors were women) died. At that point, the QTIP property passed to the ultimate heir(s).

The issue is whether the surviving spouse’s state can tax the QTIP’s intangible property under its estate tax (any physical property would be taxed in the state in which it is located). The survivor’s state’s only connection with the trust was the residency of the income beneficiary under the trust, that is the surviving spouse who died. That is roughly the same pattern as in Kaestner – a beneficiary who was entitled (ultimately) to undistributed income was in the state but the trust retained the income. The state courts all concluded that there was a second transfer when the survivor died and that connection (residence of the second spouse) satisfied due process.  It is not clear to me that that is consistent with Kaestner or how income tax, due process principles map onto estate taxation. Would the first spouse’s state also have authority to tax it upon the second death (to my knowledge no state attempts to do something like that)? (It clearly could on the first death if it had so chosen.) That seems unlikely based on Fielding and probably Kaestner for which the passage of time seems to negate the ability to tax. The time period in Fielding was exceedingly short. The connections of the state of the second spouse seems tenuous too – she/he did not own the property or “make” the transfer.

The bottom line is that this is a murky area and fraught with the potential for litigation with unclear results – particularly in Minnesota with a court that appears to be very protective for these due process principles in dealing with abstract entities like trusts that can be located/administered virtually anywhere – totally separated from where the real life people who established/funded them or who will benefit from them are. The reasoning of Kaestner is an unsatisfying stew of minimum contacts, fair play, formalism, metaphysics, and similar in unclear amounts or weights.

One could easily take the Court’s summary of its holding and conclude that its rationale may apply if the taxing state’s only contact is the residence of the surviving spouse who will get nothing more from the QTIP and who did not her or himself make the transfer being taxed (unless one considers dying to be the same as making a transfer; the first spouse in executing the QTIP in a will or trust document likely made the transfer or the trustee, neither of whom are in the state imposing the tax):

We hold that the presence of in-state beneficiaries alone does not empower a State to tax trust income that has not been distributed to the beneficiaries where the beneficiaries have no right to demand that income and are uncertain ever to receive it.

Slip opinion, p. 7

My general observation is that it would be more straightforward to impose tax (called it an accession, inheritance, or income tax) on the beneficiaries who receive the property, rather than on the estate/trust on the property or transfer. Taxing transfers made at indeterminate times and/or locations is fraught with the potential for endless disputes. Taxing the receipt of property under the income tax (over some basic exemption amount) would eliminate any due process concerns, as well as being simpler and clearer, while avoiding the epithet of being a death tax, perhaps.

PPP loan forgiveness

I have blogged about this issue previously (ad nauseum) as a federal tax issue, making clear my views on its merits (I do not favor exempting the income and allowing the expenses to reduce other income). Since the feds enacted it, it is now up to states to decide whether to conform. It appears likely that the Minnesota Senate is headed down the path of conforming. S.F. No. 268, authored by Senator Bakk, would conform. Since the bill is coauthored by Senators Nelson and Rest, respectively the chair and ranking DFL minority member of the Senate Taxes Committee, it appears that tax leadership of all three caucuses are onboard with conformity, although it is often dangerous to read too much into decisions to coauthor bills. S.F. No. 268 also proposes to allow pass-through entities to elect to file as C corporations to make their state taxes deductible in computing federal income tax. So, that might be a reason Nelson or Rest co-authored the bill, I guess.

In any case, I hope they think more carefully about that or that the House decides to not conform and prevails in conference. To beat a dead horse, the following are some of the reasons why I would opt to not conform:

  • Forgiveness of PPP loans is income just like any other type of income that presumptively should be taxable on horizontal equity grounds. Deviation from that bedrock tax policy principle requires some compelling policy justification. I cannot think of anything compelling, although I recognize the political appeal of giving a tax benefit to businesses suffering from the shutdown and recession. (As an aside, there is plenty of evidence that PPP loans and I presume forgiveness went to businesses who do not appear to be the most deserving candidates for pandemic relief. Some of them have returned the money.)
  • Conforming to federal provisions with weak policy justifications (like exempting PPP loan forgiveness) can be justified on the basis that conformity promotes ease of compliance and administration. That is particularly true for complex, multiyear provisions (like depreciation or retirement plan provisions) that impose ongoing headaches and costs on both taxpayers and the state. Failing to conform on PPP loan forgiveness poses few of those problems – it will require a one-time add-back to AGI or FTI and add modest complexity, as conformity items go. But see caveat below regarding the effects on NOLs.
  • The state budget is likely to be very tight. The current forecast shows a gap of $1.6 billion. Senate leadership has made it clear they oppose tax increases to close the gap, making identifying the highest priorities for either spending or tax cuts more important. The revenue loss from conformity is very large (more than $400 million).  Surely, there are better uses for that money, whether spending for Democrats or better targeted tax cuts for Republicans. In my view, helping individuals or businesses hard hit by the recession and the public health measures but who have not been lucky enough to score a forgivable PPP loan is a higher priority or just maintaining existing government services for that matter (the Senate GOP plans to cut them).
  • It would treat employees and businesses asymmetrically. PPP loans were primarily intended to help employees (“Paycheck Protection” is right in the name) – a good portion of the loans must be used for payroll. Note that the benefiting employees must pay tax on the resulting wages they get from the PPP loans. Why should we exempt the employers/business when the loan’s forgiveness generates profits (net income) from paying tax on that income? That is exactly what both exempting the loan forgiveness and allowing the expenses paid with it to be deducted does. I would think that Republicans – who are regularly subject to the old trope that they favor businesses over employees – would be wary of championing a provision that does exactly that. This really is a just a way of restating the horizontal equity point. So, rather than continue to beat the deceased nag, I’ll stop.

To be somewhat even-handed (my long history as a legislative staffer is hard to shake), some arguments support conformity (aside from its raw political appeal):

  • It is conformity, after all. So, it will help keep the tax simpler and easier to comply with and administer – even if there are better ways to do that. The long-term effect on NOLs is concerning (but not $400 to $500 million worth).
  • It is a one-time provision, so it will not permanently reduce the tax base and impair the state’s ongoing ability to provide services.

DOR has estimated the cost of conformity at $438 million in reduced revenue over the biennium. I do not have much feel for how accurate that estimate is but a couple of points are worth noting about it. DOR assumes that only 32% of businesses receiving PPP loan forgiveness. The rest of them are assumed to not have enough taxable income to immediately benefit. As an aside, it is not clear to me how they reached that conclusion, but it is not out of line with what I would have expected. Two points should be noted about that reality:

  1. Although the revenue loss is one-time, it will be spread over many years. Businesses without sufficient taxable income to use the deduction for PPP paid expenses will have NOL carryovers. So if the current cost is (as DOR estimates) $411 million in Fiscal Year 2022, the 68% remainder could reduce future year taxes by as much as $870 million through NOL deductions. Of course, it won’t be that high because many of the businesses will be unable to use the NOLs because they go out of business or just never generate enough other income use their NOLs. DOR estimates an annual ongoing cost of about $20 million. This ongoing effect tempers my argument that nonconforming is not that complex; many businesses will be untangling their NOLs for years. The future cost will go up if Minnesota conforms to the CARES Act NOL provisions that temporarily repeal TCJA’s NOL haircut.
  2. The fact that about one-third of recipient businesses are estimated to benefit illustrates how poorly targeted this benefit it. These businesses are still profitable, notwithstanding the pandemic’s effects.  Because forgiven PPP loans were used to pay deductible expenses, exempting the loan forgiveness and allowing the deduction reduces the tax on other income. The implication to the contrary (i.e., that businesses that lost money even with the PPP loans would be hit with a tax obligation) at the Senate hearing by testifying business owners is incorrect. The tax is an income, not a gross receipts, tax. Getting a government grant to pay deductible expenses is tax neutral. Exempting the loan forgiveness (grant) and allowing deduction of expenses paid with the loan/grant reduces the tax on other income. To benefit from both the exemption and the deduction, you need income over and above the grant/forgiveness. If your business is otherwise losing money, there will be no tax. It’s as simple as that. I hope the senators understand; I’m not sure they do.

Polarization: I hope it’s not this bad

As I noted in my book report on Ezra Klein’s book, Why We’re Polarized, I have been watching for years with trepidation the growing political polarization. My former vantage point as a legislative staffer provided a view of how the trend was manifesting itself in the behavior of legislators: more strident views, less working across the aisle even on issues that traditionally were not partisan, less willingness to compromise, fewer centrist members, and so forth.

To try and better understand this phenomenon I have been engaged in a reading project on what’s going on the Republican Party because my perception is that the right (Newt Gingrich, Pat Buchanan, and Trump as prime examples) has been driving the trend. Of course, action sparks reaction, so it’s a bipartisan phenomenon. The right was just the initial and bigger mover. I’ve been in engaged in the reading project for months (January 6th obviously heightens the relevance) and may do a multi-book report (the count is at 5: Max Boot, The Corrosion of Conservatism; John Fea, Believe Me The Evangelicals Road to Donald Trump; David French, Divided We Fall; Robert F. Saldin and Steven Teles, Never Trump; Stuart Stevens, It Was All a Lie; with a couple more on my list to go). Or more likely not, since contemplating going over my notes and collecting my thoughts is too depressing and so far I do not have any special insights. My reading has mainly reinforced my prior perceptions and observations. As an aside, I found the Saldin and Teles book the most interesting with more information new to me. Stevens’ is the most brutally honest about the GOP dynamics by a high level insider that I have read.

I have tended to view the problems of polarization as mainly a matter of politics and effective governance. That is sobering enough since the country increasingly appears ungovernable, given a constitutional structure based heavily on checks and balances with multiple minority vetoes and how closely divided the country is. See the collections of academics’ opinions in this piece by Thomas Edsall in the NYT on ungovernability.

What is really scary to me is this piece at 538, Maggie Koerth and Amelia Thomson-DeVeaux, Our Radicalized Republic, suggesting the polarization goes deeper than what I think of as politics. Rather than being about policy or controlling government, it’s a matter of tribalism, identity, and stuff that is harder to “fix” or change. I really hope that thesis is wrong (probably not) or at least overly alarmists about how bad things are (maybe). The thought that we may be in self-reinforcing feedback loops is particularly frightening. As sociologists have observed when two groups have multiple factors that distinguish them from one another (race, language, religion, cultural traditions, and so forth) the potential for those factors reinforcing each other and leading to conflict, potentially violent, escalates. That may be where we’re going with partisan identity playing a key role. Gives a whole new meaning to “identity politics” – something Klein points out but in a somewhat more benign context and way.

Some telling quotes (to me anyway) from their article:

In this study and others, Mason [Lilliana Mason, a University of Maryland political scientist whose work the article leads with and heavily relies on] found that the increasingly neat alignment between our party loyalties and other parts of our identity — race, religion, education — has made politics an integral part of the way we perceive our own moral character and that of others. [Me: in other words, the “other guys” are immoral!]

Despite that, the way we all think about public disagreement has shifted, said Jennifer McCoy, a professor of political science at Georgia State University. There’s a difference between “I don’t like your ideas” and “I don’t like you.” There’s also a difference between “I don’t like you” and “You have no legitimate claim to political power and don’t deserve it.” Eventually, you get to a place where fewer and fewer people believe in government by and for all the people.

Of course, this deeply personal form of polarization has developed alongside other divisive trends we talked about earlier, like deepening social segregation and isolation, rising income inequality and eroding trust in institutions. Americans’ political identities were being fed by — and, in a sense, absorbing — those changes.

A nation divided into Hatfields and McCoys largely by geography (e.g., urban v. rural) and segregated social groups engaged in political and social war over often irrational and irreconcilable disputes and potentially willing to fight about it (even physically as suggested by January 6th) is not a recipe for a healthy society. Matters of race, of course, play a central role as the article makes clear. Frightening and depressing thoughts.

The worst part is the article’s last section which they title “No Way Out” and which makes clear the difficulty of righting the ship. As far as I’m aware, there is no example of a large multi-racial, multi-cultural democratic society elsewhere in the world to serve as a template.

My initial and now abandoned reaction was primarily to blame Republican elites (particularly the 2016 presidential candidates) who failed to adequately respond to Trump’s candidacy in a way that put leadership and moral values above their own interests. (I recognize and understand that they legitimately underestimated his threat, because I did too. But I would have thought they would know the dynamics of the Republican base better than I or than they obviously did. Politicians, especially at the level occupied by presidential candidates and their consultants, must be more out-touch than I judged – spending too much time with donors rather than their voters. Some of the problem is wishful thinking, obviously. Stuart Stevens’ book, It Was All a Lie is instructive on this dimension.) This is deeper than Trump; he was merely a symptom or the latest manifestation. Yes, he was a major accelerant who put us in a much deeper hole than we were in, but we were in a hole before he ramped up the birther nonsense and ultimately took over the GOP (temporarily, I hope). More fundamentally, it is unclear if actions by a goodly number of elite Republicans would matter – the reaction to Liz Cheney (or Mitt Romney etc.) is telling (re: Cheney see this Politico story the implication of which I take to be that elites are pretty powerless and will end up like Jeff Flake and Bob Corker). There’s a strong element of antiestablishment (w/o regard to party) and even more tellingly “Mason’s research found that people who saw the opposing party as evil were three times as likely to wish death on opponents within their own party.” Yikes. The tendency of most elected congressional Republicans to just keep their heads down and hope this passes (someone else or events fix the problem) is understandable but deeply regrettable and pathetic.

On the bright side, things were much worse before the Civil War, because slavery created a massive policy and economic division and the country was divided neatly by geography (the Mason-Dixon line). Now, the policy differences are relatively minor (in fact, one can make the case that Trump succeeded, in part, because his campaign abandoned Republican orthodoxy on trade and immigration – unfortunately, in my view) and geographical divisions are more like a marble cake across the entire country with sizable minorities of the the other party’s partisans in all states, no matter how red or blue they are. I don’t think we’re headed to another civil war or a breakup as some think, but that’s little comfort.

538 also has a podcast episode (“Partisans Don’t Just Disagree, They Hate One Another”), which includes Nate Silver and Perry Bacon as well as the two authors of the article. It provides some additional color and detail. The discussion of polarization starts at about the 18 minute mark.

income tax

Paradox – stimulative tax increase?

I recently stumbled on this article, Laura E. Jackson, Christopher Otrok, and Michael T. Owyang, “Tax Progressivity, Economic Booms, and Trickle-Up Economics,” (Nov. 18, 2020, St. Louis Federal Reserve Bank Working Paper), with results so unexpected (to me, anyway) but somewhat logical when I thought about them, that I had to note it.

Conventionally we think of tax increases are contractionary. If you raise taxes, it discourages work and investment lowering total economic output or activity. (Obviously, what the government does with the money also matters. But most analyses ignore that side of the equation, essentially assuming the resource are dumped on a bonfire or similar. Okay, all you need to do is assume the government spending is less productive than what the private market would have done, a given for most economists.) Makes logical sense, following basic economic incentives. You tax something you get less of it. Similarly, if you shift who pays taxes by enacting revenue neutral changes (increasing Peter’s taxes to reduce Paul’s, so to speak), one would expect the result to be neutral or mildly contractionary.

Classic “trickle down” proposals assume that if you cut taxes on “job creators” (aka, the rich), that will increase economic output and economic activity benefiting a larger group (those down the economic scale). That premise has been the GOP’s heart and soul for decades and is hotly debated with mixed evidence supporting it. Where the money comes from is obviously an issue. Most federal tax cuts are largely deficit financed (e.g., the Bush tax cuts and the TCJA), so did the fed accommodate it and how did that affect interest rates, etc. becomes an issue and so on.

The Jackson et al paper analyzes a different, more complex or nuanced scenario by decomposing tax cuts and increases into two components – their level (total tax amount) and their progressivity (who pays by income strata). That allows analyzing the expansionary/contractionary effects of changes in the progressivity of the changes. They did this by breaking down tax rate changes into level and distribution (progressivity/regressivity) components (they’re analyzing federal income taxes) and, then, analyzing their macro-economic effects. Disclosure: their methodology is complex economically and mathematically. Evaluating it is way beyond my pedestrian economic expertise. The math is at a much higher level than the college math classes I took. With that caveat, here is what they found:

  • The conventional wisdom is correct. Increasing the level of taxes is contractionary. No surprise. Their other two findings are what took me by surprise, but have some logical and intuitive sense in thinking about the economics.
  • Increasing progressivity is expansionary. That is, raising taxes on the high income folks and using the revenue to reduce taxes on lower income earners increases total economic output.
  • Increasing tax progressivity increases income inequality. So, even though the government is taxing high-income Peter and using the revenues to cut low-income Paul’s taxes, Peter ultimately comes out ahead in the income/wealth race on net!

The authors characterize their finding of expansionary effects of increasing progressivity as a “striking result.” (p. 13) That’s putting it mildly. As a result, they go through a number of exercises to test the validity, which I’m not qualified to evaluate. But here is why their finding makes some logical sense, based on economic principles. Low-income and high-income individuals respond differently to increases in income and the return on work effort. As the authors put it, their responses are heterogenous. Low-income individuals have a higher propensity to consume. Increasing their incomes may well increase the overall economy more than deploying the same dollars to high-income individuals. The former is more effective in driving up consumption that leads to more production, etc. (Me: increasing high income individuals’ income and net worth could generate investment increases but more commonly they bids up prices of existing assets, stock buybacks and so on, which likely have a much lower or negligible impacts on output. Buying existing assets is not investment that generates output expansion; that requires building new durable assets.) Moreover, most evidence suggests top earners do not reduce work effort (labor) much, if at all, in response to higher tax rates. So, there is little downside there.

The second of their surprising findings, that the increased tax progressivity also increases inequality, follows from the first finding. The economic expansion (caused by the increased tax progressivity upping consumption) drives up asset values, the capital owned by high-income earners. Because there is a multiplier effect (e.g., think price earnings ratios) and the benefits mainly flow to the top who own the vast majority of the capital, this effect swamps the tax redistribution. Those at the lower end are not worse off; it’s just that the top is even better off. (The effect on consumption equality is more ambiguous, though, the authors note.)

This is why the authors describe it as “trickle up” economics:

Trickle-down economics suggests that lowering tax rates on those with high incomes spurs an expansion. To the contrary, our empirical results suggest that the opposite is true: Lowering the tax burden on lower incomes sets off an expansion that also raises the incomes of those at the high end of the income distribution. This is consistent with trickle-up not trickle-down economics. This result can be understood by considering the fact that the change in progressivity is on wage income while income inequality is measured with both wage and capital income.

Laura E. Jackson, Christopher Otrok, and Michael T. Owyang, “Tax Progressivity, Economic Booms, and Trickle-Up Economics,” p. 19.

Unexpected and interesting – I cannot vouch for the reliability or robustness of their econometrics, but the piece is thought provoking and worth waiting to see the reactions of those more competent to evaluate their methods and results.

SALT angle. Obviously, the validity of their results likely do not apply at the subnational level, where migration to more tax friendly states and a variety of other tax minimization efforts are more readily available to those at the top.


Virus bowl: Gophers v Badgers

After some initial posts pointing out the (then) lack of population adjustments in most media presentations of state COVID data and on Minnesota’s poor long term care facility record, I have refrained from writing about COVID out of respect for my lack of expertise. But COVID keeps dominating the news and much of my attention. So, I couldn’t refrain from doing one last post.


Because this post is ridiculously long and I cannot imagine anyone will read it (certainly not all of it), I will start with a bullet point summary of its highlights:

  • Minnesota and Wisconsin are similar states that have adopted different policies for addressing COVID-19. Minnesota has modest public health restrictions; Wisconsin has very few after its Supreme Court invalidated the governor’s executive order in May. That provides an opportunity to assess the effects of their respective actions, a “natural experiment.” This post presents some raw data comparing the two states’ experiences. A full evaluation must await more complete data and sophisticated statistical analysis by experts who know what they are doing (not me). Preliminary raw data present, at best, an impressionist painting of the situation.
  • For all of 2020, Minnesota has had many fewer cases (about 100,000 less) of COVID-19 but more deaths than Wisconsin after adjusting for population differences. Wisconsin has had more deaths than Minnesota following the invalidation of its public health restrictions but many fewer relative to its case rate than Minnesota.
  • Minnesota’s unexpectedly higher death rate is not explained by the age of its population, which is modestly younger than Wisconsin’s. The lethality of COVID-19 increases with age, particularly for the elderly, so that would suggest Wisconsin should have a higher fatality rate. It does not.
  • Minnesota’s higher minority population, groups who statistically are more susceptible to contracting and dying from COVID-19, also does not appear to explain its higher death rate.
  • My best guess as to the culprit is that Wisconsin’s long term care industry practices and regulatory policies are besting Minnesota’s, based on sketchy data.
  • On balance, Wisconsin’s looser health public health restrictions have resulted in much more sickness and modestly more deaths than in Minnesota.
  • But they also have led to more economic activity than in Minnesota – smaller drops in employment, consumer sales, and small business revenues. The overall differences are small with much bigger differences showing up in certain sectors (e.g., leisure and hospitality).
  • The imponderable is whether trading off more sickness and death (albeit mainly among the very old) for small increases in economic activity is a good choice. Much subjectivity (e.g., in assigning dollar values to pain, sickness, and death) is involved and my instinct is that where one comes down devolves to their philosophical priors and/or identification with a partisan tribe. Available data does not justify the vociferous self-assurance of many of the commentators and elected officials and should inspire more modesty, compromise, and cooperation.

COVID-19 data for the two states

This post compares data on Minnesota’s and Wisconsin’s experience in dealing with COVID-19. I have not seen these comparisons presented elsewhere but may simply have missed it. The two states have taken different policy paths to address the pandemic, mainly because of a Wisconsin Supreme Court decision (text of opinions) that nixed its governor’s statewide public health mandates. Since that decision Wisconsin has largely been “open for business” (starting May 14th) other than a patchwork of local restrictions and a statewide mask mandate adopted by the governor to which legal challenges in process but have not yet invalidated. By contrast, Minnesota has taken a somewhat more activist approach, but well short of what states in the northeast and on the west coast have done.

In dealing with SALT issues during my career, Minnesota and Wisconsin occasionally presented opportunities for “natural experiments” in social science research speak. The two bordering states are similar in size, demographics, and other factors with some modest differences in their business profiles, rural/urban breakdown (Wisconsin is slightly more rural), minority populations (Minnesota’s is a couple percentage points higher), and similar. Overall, they are similar. Social science research typically cannot run controlled experiments, since there is no opportunity to give placebos to a control group and see how they differ from the treatment group. Thus, when the two states’ public policies diverge (e.g., Wisconsin long has had a capital gains exclusion while Minnesota has not; Wisconsin has only minor homeowner tax incentives while Minnesota’s are generous), it presents an opportunity to study what effects those differences have. The COVID-19 policies present a similar opportunity. Background differences between the two states that affect public health outcomes may be much greater than is the case with SALT policies – here again my ignorance counsels caution in reading much of anything into this exercise.

The following are some readily available data without analysis or conclusions, just my commentary and speculation. I have never studied epidemiology at even the most fundamental level. (As an aside, I have noticed a fair number others who are similarly unqualified but appear smugly confident in reaching conclusions. More troubling, some mainstream media seem happy to lend credibility to their fairly wild assertions – looking at you Strib editorial page for publishing, e.g., Lennes, Tice, Kersten – all of them critical of the state government based mainly on their political priors and not credible data and analysis. Qualified experts with similar views do exist, but the Strib found it easier go with unqualified locals, I guess.) In any case, the basic data are interesting and suggestive – even if potentially misleading to the uninitiated like me. I assume after the dust settles (2023?), competent people will do careful analyses that control for relevant factors, are peer-reviewed, and will be informative – even if each pandemic is sui generis. Cross-country comparisons (using Sweden, which has consciously taken an approach even looser than Wisconsin’s accidental policy, and the other Scandinavian countries, e.g.) will certainly be done.  See this FT story on the Sweden policy, which suggests its sponsors may be losing their nerve. (Quote: “Sweden has reported more than 2,000 Covid-19 deaths in a month and 535 in the past eight days alone. This compares with 465 for the pandemic as a whole in neighbouring Norway, which has half the population. As Sweden’s King Carl XVI Gustaf said just before Christmas: “We have failed.””)

The data are from the Minnesota Department of Health, the Wisconsin Department of Health Services (some of which are usefully aggregated by the Journal Sentinel), and the COVID Tracking Project.

Basic data on case and death rates show Wisconsin has more cases but fewer deaths, adjusted for population. The table shows population, testing, case, and fatality data (as of December 31, 2020) for the two states.

Population (000)5,611,1795,822,434
Persons tested2,972,8042,822,063
Total tests5,574,962NA
% of population tested53.0%48.5%
Positivity rate14.0%18.4%
Number of cases415,302520,483
% of population7.4%8.9%
Total number hospitalized21,86421,207
Total in ICU4,6202,034
Number of deaths5,3235,242
% of population0.09%0.09%
Case fatality rate or CFR1.3%1.0%
Minnesota and Wisconsin COVID-19 cases and deaths

The two states have similar populations, so population adjustments only make modest differences. Minnesota has proportionately fewer cases (about 17% less on a population adjusted basis), but more deaths (about 5% more on a population adjusted basis). Cases include those confirmed by both PCR and antigen tests. The number of cases is sensitive to the level of testing and how the testing is done (i.e., who is being tested). Since Minnesota is testing at a higher rate than Wisconsin, testing differences are unlikely to explain Wisconsin’s higher case rate. With higher testing rates, one would expect Minnesota to have higher case rates all else equal. Of course, Wisconsin’s testing policy might be directed at individuals more likely to be infected; one would expect fewer tests to be better targeted. So, on the surface it appears that Wisconsin has a higher incidence of infections than Minnesota and more community spread. That would be consistent with Wisconsin’s looser public health restrictions.

But cases in Minnesota are more likely to result in death, as reflected in its higher number of ICU admissions, deaths, and higher CFR (the percentage of positive tests that end in death). Death statistics are less subject to testing levels than case levels are, even though lower testing levels may cause some COVID deaths to be misattributed to other causes. Given that, I would trust the numbers more after the CDC statisticians have reviewed and adjusted death certificate data. But even preliminary death data are more reliable and obviously more consequential than case rates. So, that looks superficially like a modest advantage for the Badgers.

Interestingly, Wisconsin despite its higher case rate has lower hospitalization rates and much lower ICU rates. Those differences could be attributable to medical care practices or simply to the fact that Wisconsin has fewer cases with severe symptoms because more younger people are infected. The lower ICU rates are consistent with a lower death rate, but the difference in ICU rates is much larger than in the death rates. So, something else must be going on.

The age distributions of the state populations do not explain the differences. The lethality of the virus is strongly correlated with age; the older you are the more likely contracting the virus is to be fatal.  The power of this age effect is shown by the two tables below showing case and death rates by age group for Minnesota and Wisconsin. The death rate consistently rises with each successively older age group – by a lot for those over 70 (close to 10 percentage points per decade). The age distribution of Minnesota cases and deaths, as of December 31st (note that the case total is lower than the state total in the table above, because MDH did not yet have age date for a few cases on December 31st when I grabbed this data):

Age GroupCases% of totalDeaths% of totalCFR
0 – 19 years67,45016.2%10.0%0.0%
20 – 29 years79,53419.2%90.2%0.0%
30 – 39 years68,15016.4%300.6%0.0%
40 – 49 years59,89614.4%711.3%0.1%
50 – 59 years59,54714.3%2224.2%0.4%
60 – 69 years41,32210.0%57710.8%1.4%
70 – 79 years21,3185.1%1,14521.5%5.4%
80 – 89 years12,4793.0%1,88035.3%15.1%
90 – 99 years5,2021.3%1,30924.6%25.2%
100+ years2920.1%791.5%27.1%
Minnesota 2020 COVID-19 cases and deaths by age group

The age distribution of Wisconsin’s cases and deaths, as of December 31st (note: the Wisconsin death data is limited to confirmed deaths, which is why the total number of deaths is lower than in state total table which shows both types – I was lazy and used the Journal Sentinel table, rather than trying to construct my own from Wisconsin Department of Human Services API data and J-S reports on cases and deaths from PCR tests only for some unknown reason):

Age GroupCases% of totalDeaths% of totalCFR
0 – 19 years72,36615.2%20.0%0.0%
20 – 29 years91,53219.2%160.3%0.0%
30 – 39 years74,50315.6%350.7%0.0%
40 – 49 years68,01214.3%821.7%0.1%
50 – 59 years73,06015.3%2665.5%0.4%
60 – 69 years51,99610.9%63513.2%1.2%
70 – 79 years27,2515.7%1,20825.1%4.4%
80 – 89 years13,3872.8%1,52131.6%11.4%
90 + years5,1851.1%1,05321.9%20.3%
Wisconsin 2020 COVID-19 cases and deaths; source: Journal Sentinel

Thus, the age distributions of the two states’ populations could be a factor. A state with an older population, all else equal, is likely to have a higher death rate for the same infection rate. The table below shows the relative age distributions of the two populations. As can be seen, they do not differ much. Minnesota’s population distribution skews slightly younger (higher percentages in the under 20 group and lower in the 60 and over groups), so it moves in the opposite direction that one would expect if the age distribution explains the death rate difference. With proportionately more of its population in the younger groups, one logically would expect Minnesota’s fatality rate to be lower; it is higher. Blind alley.

Age groupMN populationMN %WI populationWI %
0 to 19   1,444,18625.7% 1,420,57424.4%
20 to 29      736,59913.1% 762,03613.1%
30 to 39      768,08113.7% 739,24512.7%
40 to 49663,49711.8% 682,90811.7%
50 to 59749,49113.4% 788,81813.5%
60 to 69      654,00511.7% 746,40212.8%
70 to 79      368,8526.6% 435,8007.5%
80+      226,4684.0% 246,6514.2%
Total5,611,179100.0% 5,822,434100.0%
Distribution of Minnesota and Wisconsin by age group; source: US Census Bureau

Age distribution of cases and deaths: Minnesota’s higher death rate is concentrated in the oldest age brackets. Of course, the issue is not simply the age distribution of the population, but the age of individuals who are infected with the virus and who ultimately die. Here, we are stuck with the vagaries of testing data because that is the only way we know whether someone is infected or not. Of course, the real infection rate is some unknown multiple of the case rate (i.e., the number of positives/population), because many infected individuals are not tested. This multiple could be 5 to 10 times the case rate and is sensitive to the level of testing and the protocols used to select whom to test. The graph shows the comparable Minnesota (blue bars) and Wisconsin (red bars) case rates by age group as a percentage of each state’s respective populations. Since Wisconsin does not report probable deaths by age group and Minnesota does, I distributed its probable cases and deaths to age groups in proportion to the confirmed cases and deaths to be consistent with the Minnesota data.

Minnesota and Wisconsin COVID-19 case rates as a % of population by age group

Wisconsin’s higher cases are more concentrated in the lower age groups than Minnesota’s. Except for the oldest age group, the red bars are consistently longer than the blue bars. The percentage of the population that tested positive in each of the age groups below 60 are about 2 percentage points higher for Wisconsin than Minnesota. For age groups 60 and older, the effect starts to reverse. For 60- and 70-year old’s, Wisconsin’s case rate is about one percentage point higher. For those above 80, Wisconsin’s case rate is less than a half percentage point higher than Minnesota’s. This concentration of more Minnesota cases in those 80 and older group almost certainly explains why it has more deaths, despite its lower case rate. As shown in the tables above, death rates are much higher in the older age groups, especially those 80 and older.

The graph below shows the two states’ COVID death rates by age group (again, as a percent of the population of the age group). Aside from dramatically showing the higher death rates for older age groups, the graph shows that Minnesota’s death rate is higher than Wisconsin’s primarily in the oldest age group (80+). Its rates are still higher for those between 60 and 79 but reverse with slightly lower rates than Wisconsin for those below 60. This likely reflects Wisconsin’s higher case rates in those age groups. Since Minnesota is testing at higher rates than Wisconsin, its infection rates may be even higher for those younger age groups. (Note that is pure speculation, since the states likely have different testing protocols that could be a factor in the relationship between and distribution of case rates or positives relative to actual infection rates.) In any case, despite its higher testing rates, Minnesota’s CFRs are higher for the oldest age groups. That may suggest that more of Minnesota’s most vulnerable elderly are contracting COVID-19 than in Wisconsin.

Minnesota and Wisconsin COVID-19 death rates as a % of population by age group

On the surface, this does not look good for Minnesota’s more restrictive public health policy, as compared with the laisse faire Wisconsin Supreme Court’s approach. Minnesota’s restrictions appear to be better at controlling community spread of the virus but are not in preventing deaths among its elderly as effectively as Wisconsin’s. The latter seems more important and is what conservative critics have been harping on, albeit largely based on uninformed speculation. Minnesota’s success in minimizing community spread has not carried through to preventing its more vulnerable elderly from becoming infected and dying. If that is so, why is an important public policy question for legislators and executive branch public health officials. The next section explores the most obvious candidate, long term care facilities.

Long term care facilities (LTCF) may explain the two states’ differences. Why does Minnesota do a better job of preventing general community spread than Wisconsin, while many more of its most vulnerable population – those 80 and older – contract the virus? One possible answer lies in regulatory and business practices in the two states’ LTCFs (i.e., nursing homes, assisted living and memory care facilities) or in differences in the demographics and health status of the populations of those facilities. I have blogged about Minnesota’s abysmal LTCF COVID-19 record and the media has covered it extensively in many stories, including multiple stories in the Strib.

Unfortunately, as far as my unexpert eyes can tell, comparable state-by-state data on COVID-19 infections and deaths in LTCFs are not readily available. Data are available from several sources, but they are not comparable because of differences in reporting, state LTCF regulations and reimbursement practices that causes institutions and their resident populations to vary from state to state, and so forth. The CDC requires (as of May) reporting by skilled nursing facilities. But even those facilities likely vary considerably in their practices and populations from state to state. Moreover, reporting for other facilities, such as assisted living and memory care, is totally inconsistent. Some states report this data (e.g., Minnesota), while others do not (e.g., Wisconsin). Moreover, these facilities because they are more lightly regulated, probably vary even more than skilled nursing facilities, making comparisons of available data more problematic.

Despite all those caveats, available state-by-state data show that Minnesota and Wisconsin have such wide differences that a good part of the story of Minnesota’s higher COVID-19 death rate among the elderly must lie in LTCF policies, practices, and regulations. At least, that seems to be a reasonable conclusion. Data from the COVID Tracking project, for example, report that Minnesota has had 15,320 cases in LTCFs and 3,220 deaths; Wisconsin, by contrast, has had 5,976 cases and 1,109 deaths. The differences in both cases and deaths are staggering; Minnesota’s cases and deaths are more 2.5X higher. Some of difference is explained by reporting differences – i.e., because Wisconsin does not include assisted living facilities in its reporting and Minnesota does. But it seems very improbable that that accounts for the full difference.

I could not find an ongoing data source that breaks out Minnesota’s cases and deaths between nursing homes and assisted living and other care facilities. This weekly CDC MMWR (Nov. 20) reports that Minnesota had 1,744 COVID-19 cases in assisted living facilities as of October 15th (Table 1). Minnesota deaths are not reported by the MMWR and it includes neither cases nor deaths for Wisconsin. CDC says it gathered this data from state websites. I have been unable to find on the MDH website a breakdown of cases between skilled nursing homes and assisted living facilities. So, I am unsure where CDC got its Minnesota data. In early June, MDH released data by type of facility under threat of a legislative subpoena. It showed that about 68% of the then LCTF COVID-19 deaths were in skilled nursing homes. I have been unable to find more recent breakdowns, now that LTCF COVID-19 deaths in Minnesota are more than triple the then June number of 896. That suggests most (maybe two-thirds) of Minnesota’s LTC COVID-19 cases and deaths are in nursing homes. If that is an accurate inference, Minnesota has significantly more cases in nursing homes than Wisconsin, despite Wisconsin’s higher population.

In any case, Minnesota is among the states with the highest proportions of its COVID-19 deaths attributable to residents and staff of LTCFs (64%) based on COVID Tracking Project data. Only four states had higher percentages. It seems safe to conclude that some set of differences attributable to LTCFs are a major explanation for Minnesota’s higher death rate among the elderly than Wisconsin’s. And that Wisconsin LTCF operators and regulators are doing a better job than their Minnesota counterparts.

A principal premise of reducing community spread is that doing so is essential to keeping the virus out of LTCFs. Otherwise, LTCF workers or visitors will bring the virus into LTCFs. That may be so, but if it is, Minnesota’s better job of reducing community spread appears to be, then, thwarted by some other factor or factors.

As an aside, see this WaPo story (Will England, For the first time, the U.S. will reward nursing homes for controlling the spread of infectious disease) on HHS incentive payments to LTCFs that have done a good job of controlling the virus in their facilities. The measure HHS uses is based on the differential between community spread and the level of the virus in LTCFs. Thus, CMS appears to have accepted the premise that controlling community spread helps LTCFs control infection rates. But as the article notes, this is controversial. It creates the opportunity for LTCFs in states with rampant community spread to get incentive payments by keeping their incidence low. Conversely, LTCFs in states that have done a good job of controlling community spread – e.g., Vermont and Maine – will rarely qualify. That is not obviously wrong to me, unless the measure rewards absolute differences. The article does not say. In any case, Wisconsin is getting a disproportionate amount of the payments – twice the rate its population would suggest it should get. This provides indirect support for the narrative that LCTFs policies and practices are an explanation for Wisconsin’s lower elderly death rates.

Minnesota’s higher minority population does not appear to be a factor. Minnesota has higher proportions of its populations and higher absolute numbers of minorities than Wisconsin. National data show that minorities suffer more severe COVID-19 cases, including deaths. For example, African-Americans experience death rates, when compared with whites, as if they were a decade older (Brookings Institution). Thus, Minnesota’s higher minority population (about 2 percentage points higher than Wisconsin’s) suggests that it should have a slightly higher death rate, all else equal. Both states publish case and death data by race (many states do not). The data reveal that despite its lower minority population, Wisconsin’s has essentially the same number of COVID-19 deaths of minorities as Minnesota and, of course, higher proportions relative to its total population. Minorities comprise about 13 percent of COVID-19 deaths in Wisconsin and 11.5 percent in Minnesota. Thus, this moves in opposite direction expected, suggesting the differences in the relative sizes of their minority populations do help explain the differences in death rates. Another dead end. This naturally points back to LTCFs as the likely culprit for Minnesota’s higher death rate.

What has happened since the two states’ policies diverged?

All the preceding data is for the entire period of the pandemic (for 2020 to be more accurate). But the two states policies began to diverge only after the Wisconsin Supreme Court invalidated the Governor Evers’ executive order in mid-May 2020. The natural experiment only really began in late May or early June. To control for this effect, the table and graphs below show the differences in cases and deaths from June to December. Because of the lag between exposure and when tests can detect an infection, June 1st seems like a reasonable starting point. December seemed like a reasonable cutoff since vaccination availability, distribution, and administration policies may begin to affect matters starting sometime in early January. The Table shows COVID-19 cases and deaths for June – December 2020.

per 100k of pop6,9328,585
per 100k of pop7680
COVID-19 cases and deaths, June through December 2020

Following the change in policy required by the Wisconsin Supreme Court, Wisconsin’s population-adjusted case rate is approximately 20% higher than Minnesota’s and its death rate is 5% higher. During March through May, Minnesota had higher case and much higher death rates. Post-May data continue to show Minnesota with a higher CFR than Wisconsin. Although Minnesota’s population-adjusted death rate is now lower than Wisconsin’s, it still seems too high given its lower case rate. Testing differences seems= unlikely to explain such a large gap. Again, my suspicions lie with LTCF differences between the two states.

The graphs show the weekly case and death rates for the two states for June through December. To normalize for population differences, I adjusted the Minnesota numbers upward so they would be proportional to Wisconsin’s higher population (about 4% higher).

Source: COVID-19 Tracking Project

The two states’ pattern of outbreaks have followed one another. Minnesota’s lower case counts (positive tests) are obvious (the large area between the two lines, of which the lower orange line represent Minnesota cases).  Minnesota’s better performance in deaths similarly show up in the next graph, albeit in more modest fashion, which follow the same pattern as cases with the characteristic lag (deaths typically occur two or more weeks after infection).

Source: COVID-19 Tracking Project

For the pre-June period, Minnesota’s COVID death rate was 84% higher than Wisconsin’s. Following the Wisconsin Supreme Court decision, it was 5% lower. One could hypothesize that the court’s nullification of Governor Evers’ executive order explains the difference. I would not leap to that conclusion since other factors may be at play. In particular, when and how extensively the virus appeared in the two states may be a factor. States subject to early and virulent outbreaks of COVID-19 (e.g., New York, New Jersey, and Louisiana) suffered much higher death rates because medical practitioners were still learning the best therapeutic techniques and death rates were generally higher. The Twin Cities, as a large corporate headquarters location and travel hub, likely suffered much higher early exposure to the virus than Wisconsin. If Minnesota was more heavily exposed in February through March, it naturally would have experienced higher death rates that regressed to the mean. If so, the big divergence in the two states’ experiences during March through May period may be partially attributable to that and the reversal in the differences thereafter less to Wisconsin’s change in policy. But Wisconsin’s wide-open approach undoubtedly also contributed to the acceleration of cases and deaths in the post-May period.

On balance, it is difficult to not infer that Minnesota’s modestly more robust public health mandates have reduced infections, as well deaths to a lesser extent. Factors other than the public health measures also affect the infections and deaths and may differ between the two states. Thus, the magnitude of the effect requires analysis by someone who understands what control variables will better identify the effects of the public health measures in employing statistical techniques like regression analysis. But it seems safe to say the Wisconsin Supreme Court decision resulted in increased sickness and death. I assumed that they (the Republican justices who struck down the governor’s order and the Republican legislators who brought the suit) knew that would occur but concluded it was justified, which brings us to the next issue.

Effect on Economic Activity

More thoughtful critics of Minnesota’s public health measures generally recognize that more sickness and death will result from the looser policies that they advocate. Their point is that the resulting expanded economic activity provides greater benefits than the costs of sickness and death. (As an aside, much of the public commentary I have read is clear about that; what the authors have left unsaid is how they value the “cost” of more infections and deaths, relative to some measure of the “benefit” of more economic activity or what they think the relevant magnitudes of each are. So, their assertions are highly general and ultimately unsatisfying, bordering on the tautological or meaningless. I say that because reading their stuff is often maddening for me – self-righteous and condescending toward the folks making life-and-death public health decisions – given the lack of real analysis or stated factual bases for their criticisms. That may sound harsh, but their commentaries are pretty harsh, in my opinion.) Thus, one needs to look at the other side of the cost-benefit equation: to what extent has Wisconsin’s policy resulted in higher levels of economic activity that justify the more adverse public health outcomes (deaths, short or long debilitating illness, crowding out others from access to health care, etc.).

As an aside, I would point out that putting a dollar value on human lives and sickness in such cost-benefit analyzes is controversial and can be highly charged. Most of the COVID-19 fatalities are old folks (really old, 80 or older). How does that factor into the value of loss of their lives? By implication, these hard-nosed conservatives implicitly would discount it, I assume, as Texas Lt. Governor Patrick colorfully asserted. How do you put values on sicknesses that do not result in death? To an extent, economic output may implicitly take that into account (i.e., people not working, higher health care expenditures, etc.) but that doesn’t come close to capturing the real “value.” An abstract way to do so would be to sum how much infected individuals would be willing to pay – after the fact – to avoid becoming sick. That is an unknowable number, of course. All of this just underlines the difficulty of the calculus that the critics are asserting are being miscalculated.

Economic benefits could appear as either a matter of level (total economic activity under some measure) or distribution (which businesses or individuals realize net benefits or losses). Restrictions affect businesses and individuals differentially as they cause some buyers to substitute other goods and services. Individuals who cannot go to health clubs may buy home exercise equipment. Savings from the inability to travel or to go to restaurants may cause more home remodeling or construction. If we can’t go to restaurants, maybe we should update our kitchen? Home construction and remodeling in the Twin Cities have had a remarkably good year. See Jim Buchta, “For Twin Cities builders, 2020 was year of the single-family home,” Strib, 1/5/21: “Single-family homebuilders in the Twin Cities had one of their best years since 2005.” Big box retailers, like Target, have also done well. Most of the rhetoric focuses on the level, but some on distribution as well. Republican opponents of the restrictions typically walk restaurant and health club owners up to the microphones at press conferences, so their (Republican) concerns likely have an element of distributional concerns.

There are multiple challenge in assessing the economic benefits. Some big factors are:

  • Data lags. Unlike reporting of COVID test results, death, hospital admissions and so forth, reporting of economic data, particularly the best measures (such as gross state product, income measures, etc.) lag considerably. As a result, it will take a while to get the data necessary for econometricians to analyze the effects.
  • Inherent complexity make assessing cause and effect difficult. Many background and other factors affect economic decisions, sorting that out and isolating the effects of varying public health restrictions will be a challenge. This is, in addition, to adding controls to make the natural experiment a better measure of the effects attributable to policy, rather than background factors that differ between the states.
  • Distributional effects resist evaluation. These effects are real and hurt or help individuals as an unintended side effect of the restrictions. But there is really no principled way, for example, to value the fact that the reduction of one business’s sales (e.g., a restaurant) has help another (e.g., a home builder). Peter’s cost may be Paul’s benefit.

Given that and the fact that my goal in doing this is simply to provide an impression or first look, I simply assembled some comparative data on basic measures – jobs, sales, etc. from available sources. The easiest way to do this was to use the data from the website, which assembles viturally real time data. Both MCFE and I (Webinar worth watching) have described this data. Since I’m doing this to get an impression, I took the easy route. The charts below are from that website.

The first shows the differences in Minnesota’s and Wisconsin’s employment. It shows Wisconsin with an initial smaller drop in total employment than Minnesota. The difference predates invalidation of Ever’s executive order (i.e., it starts showing up in April when the order was invalidated in May), so there must be a little more involved than the health policy differences. Minnesota’s heavier earlier exposure to the virus is a plausible explanation. Interestingly, the latest data show Minnesota has closed the difference and is doing slightly better than the Badgers in the last months of available data.


If one focuses on unemployment of low-income wage earners, the effect is quite different. Wisconsin is doing much better (8 percentage points) than Minnesota as seen in the graph below.  Many of the workers in the high touch industries (restaurants, bars, personal services, etc.) are low-wage workers and the effects of Wisconsin’s lack of restrictions on those businesses is obvious. Thus, if minimizing distributional changes or protecting low-wage workers is important, the Wisconsin policy appears preferable.


The next chart compares total consumer spending in the two states. It shows that Minnesota’s spending dropped considerably more than Wisconsin’s (through December 6th) – by almost four percentage points. The immediately following graph shows the drop in spending by consumers in low-income zip codes. Reversing the pattern shown in employment, spending in these Minnesota areas dropped less than in Wisconsin (-0.1 versus -2%). A paradox – probably because more of the spending is by higher income consumers? The final graph shows the drop in restaurant spending, which shows the dramatic difference in the two states’ consumer spending on that sector as one would expect (Wisconsin’s spending dropped by 14 fewer percentage points). It is worth noting, however, that spending in Wisconsin is still down a whooping 38 percent, more than double its advantage over Minnesota. So, the lack of a public health restrictions are not a panacea for those business – two-thirds of their problem is people simply choosing to avoid the high risk activity of dining out, even if they are open.


A final set of three charts shows the differences in small business revenues, which highlight the differences in the structures of the two states’ economies. Factors that a careful econometric analysis, when more complete data is available, would need to take into account. The first chart shows the change in revenues for all small businesses. The differences are small and follow similar patterns; revenues of small businesses in Minnesota dropped by 0.4-percentage points more than in Wisconsin. A very small difference for a 30-percentage drop. But the second and third charts show breathtaking differences by sector. The professional and business sector in Minnesota declined by under 6%, while that Wisconsin sector dropped by 26%. Minnesota’s better performance likely reflects the benefits of its headquarters economy with multinational firms (in finance, health care, food, consumer products and similar) that have not been as hard hit by the pandemic. That pattern reverses for the leisure and hospitality small businesses, which saw a 13-percentage point larger drop in Minnesota than Wisconsin. This, of course, reflects the effect of Wisconsin’s comparative lack of public health restrictions on those businesses. But those business are sucking wind with revenue declines south of 60% in both states. So, most of the cause is the pandemic and the consumer response it to, not the public health orders. It appears Minnesota’s public health orders increases its leisure and hospitality small business’s revenue loss by, perhaps, 20 percent.


When I was working, my point of reference was to consider the potential effect on state tax revenues. That point of view provides one additional data point confirming that Wisconsin’s looser public health restrictions caused a smaller decline economic activity than in Minnesota. The Urban Institute reports on state revenues and shows that for April through September Wisconsin’s revenues dropped by 2.6 percentage points less than Minnesota’s. (Getting full access to the Urban data is more expensive than I’m willing to pay.) That is an extraordinarily crude measure economic activity because Urban is simply reporting year-over-year net revenues, unadjusted for tax changes, and the two states’ tax structures differ somewhat. (This NY Times Upshot blog post graphs additional Urban data for another month, which shows a similar pattern although Minnesota appears to be catching up a bit.) But it is something, especially since my instinct and available data suggest that the pandemic has affected Minnesota’s underlying economy less adversely than Wisconsin’s, implying the effect on Minnesota’s revenues should be smaller. The phasing-in of the revenue reductions from Minnesota 2019 tax cut could be a small factor in the April to September revenues.

Overall, it appears clear that Wisconsin’s looser public health restrictions have resulted in somewhat smaller reductions in levels of economic activity (employment, consumer sales, and small business revenue), when compared to Minnesota’s more widely applicable and slightly more robust restrictions. While the overall effects appear small, the distributional effects (particularly on the leisure and hospitality sector) are more dramatic. That combination may suggest that the structure of Minnesota’s economy, principally its sector mix, has insulated it more from the pandemic’s effects and/or that public health restrictions are inducing more substitution effects. More data and statistical analysis are obviously required to reach meaningful conclusions about the economic effects.


On balance, I am comfortable concluding that both (1) Minnesota’s modestly more robust public health restrictions saved lives and reduced the incidence of sickness compared to Wisconsin’s and (2) Wisconsin’s approach reduced the adverse economic effects of the pandemic a bit. The distributional effects of Wisconsin’s policy are probably bigger than the overall level of economic activity, though. But the imponderable factor is how to value the saved lives and reduced pain and sickness in the cost-benefit equation. There is a large subjective element inherent in that calculus – how should the age of fatalities factor in, how much one does one value avoiding the pain and suffering of a bout of COVID, aside from the pure economic costs of health care expenditures and lost output, etc.? So, the apparent results (tentative as they are) do not tell us a lot about whether the tradeoff makes policy sense.

As a healthy 70-something, I am happy that I live in Minnesota rather than Wisconsin. By contrast, if I lived in an LTCF or had a loved one in an LTCF, all else equal, I rather be in Wisconsin (to paraphrase W.C. Fields). Similarly, if I owned a restaurant I would likely prefer to be on the other side of the river.

When I first conceived of doing this post, I imagined awarding a “trophy” to the winning state (like Paul Bunyan’s Ax for the football game). But I couldn’t decide on an appropriate one (Nurse Ratched’s (“The Big Nurse”) syringe filled with a million doses of vaccine?) and, in any case, it would need to stay in the trophy case in the middle of the Mississippi or St. Croix River.

Of course, as pointed out at the outset, this at best a mere impression with greater clarity awaiting more data and sophisticated statistical analysis by expert economists and epidemiologists. At best it is like a Monet painting, while later high-quality analysis by experts might be closer to a low-resolution black and white photograph. Both of which are far from the goal of a high-resolution color photograph. In my mind, all of this points out the nonsense of the vociferous and self-assured nature of the political debate that is going on over these public health restrictions. These are difficult decisions for which there are no clear or certain answers.

I guess that pushes people back to the self-assurance of their priors. If you are a Republican who is skeptical of any government interventions in the “market” or of limits on private behavior, you revert to that mode and are convinced the restrictions are a poor choice. As an aside, I would tend to think “conservatives” (in the Burkean sense of conserving or preserving what is good in the status quo) would be more conflicted and might lean to being careful about preserving health and life. Their lack of conflict probably says something about the flavor of conservatism that now dominates the Republican Party (if it actually is conservatism). Democrats, by contrast, instinctively favor communal efforts and have higher levels of trust in government and, not coincidentally, control the state executive branch. As a result, they default to favoring Governor Walz’s more restrictive approach. So, much of this pitched fight is likely little more than the usual partisan philosophical fight carried out on a new battlefield.