The debate over the accuracy of Thomas Piketty’s book on wealth inequality now often cites Martin Feldstein’s Wall Street Journal piece – but commentary on Feldstein’s statements seems to me not to analyze what Feldstein wrote, to the point where I question whether anyone actually has read carefully what Feldstein said. I have – and, frankly, my jaw dropped at both the apparently poor quality of the analysis and the seemingly huge disconnect with what things are like in the real world.
So, since no one else per Google seems to be doing so, let me set forth Feldstein’s core argument and why it seems not only flawed but wrong-headed. Feldstein says that (a) wealth disparities in the US are much less than in Piketty’s analysis, and (b) there has been no increase in these disparities over the last 35 years. Let’s take each of Feldstein’s core supporting arguments in turn.
First, “Ever-increasing inequality –true if people lived forever; but they don’t. Cumulative effect diluted by estate taxes and sharing among children/grandchildren”. To a first approximation, 40-50% of the US has negative or zero net worth, 49.9% has positive net worth (wealth) primarily in housing, and 0.1% has wealth primarily in stocks and bonds – those being the wealthy. As the last few years have shown, the trend has been towards a slightly greater percentage of the US having zero or negative net worth over the last 35 years, while housing over that period (if we recognize that for most of the 49.9%, housing is also for shelter, and hence has major expenses) has little net return, and per Piketty the wealthy are seeing 7% return on investment – while if they had invested in Vanguard S & P 500 index funds during that period, they would have seen about 8.5% return on investment.
So let’s assume (reasonably), on average, that the megarich earn this return for 40 years before they die. At, let’s say, a 6% investment advantage over the rest of us, the megarich have 10.2 times the net worth ratio at the end as they have at the beginning. Now slice this by 35% under the old estate tax formulas, adjusted to reflect tax-avoidance strategies, or 25% under the new ones. You’re still up by 6.7 or 7.7 times after those forty years compared to the rest of us. Even with dilution by (on average) 2.2 beneficiaries, each beneficiary starts at 3 or 3.5 times where his or her predecessor did relative to the non-wealthy – the only thing that changes is a slow expansion of the “very rich” category, from, say, 0.09% over that 40 years to 0.11%.
Think about it: all that “dilution” simply means that every year, 1/40 of the total rich get a 70 % penalty if you worry about multiple beneficiaries, or 30% if not (which makes more sense), meaning that the rich get less than 25 % of their relative increase lopped off by estate-tax effects. In other words, dilution doesn’t get close to affecting the fact of ever-increasing inequality.
Second, “total wealth grows over time roughly in proportion to total income [3.2% per year].” Feldstein cites problematic evidence: Federal-Reserve-calculated flow-of-funds since 1960. He doesn’t cite any details, so I have to guess at the best scenario for his argument, but clearly using Federal Reserve data meant for an entirely different purpose and not vetted against Census wealth figures is really questionable.
More importantly, this says nothing about the relative share of that supposed increase going to the wealthy. Consider, for example, the median wage ($30,000) versus reported dividend income from $100 million in an index fund -- $2.4 million. We know that for the bottom 80%, over the last 35 years, real income has been flat or lower, and for the top 0.1%, it has increased more than 10 times. We also know that 90% of the increase in productivity in the last 35 years has effectively accrued to the wealthy. Either way, it says that even if the Fed is right about the rate of growth of wealth and income, 90-120% of that growth is going to the wealthy. Again, increasing inequality, and apparently ever-increasing inequality.
And note that since the figures are since 1960, 30% of Feldstein’s analysis takes place in a world where inequality is not increasing – stacking the deck, as it were. Finally, note that if the super-rich were investing in stocks, and therefore reported only dividends as income (until end of life, if not later), and dominated the wealth distribution, then you would see total income and wealth increase in tandem in the same way.
Third, Feldstein gets to the part where he is arguing that the appearance of increasing inequality is an artifact of the wealthy reporting more of their income – not wealth, income. He says, more or less, that with the passage of changes around 1980 plus tax reform in 1986 there was a strong incentive to shift investments from low-yielding tax-exempt investments to higher-yielding taxable investments. This is ridiculous: bonds were and are priced to yield the same net return on taxed and tax-exempt investments for the wealthy who buy them, and so we can be sure that something like ½ of the bond income was already reported – and, of course, this was a smaller proportion than stocks. In fact, for the wealthy, over time, it makes no sense except for diversification to go for (taxable or otherwise) bonds instead of stock index funds. And so, by any stretch of the imagination, the effect would apply to no more than 15% of the wealthy’s investments – a one-time effect over thirty-five years, for a net bump of 0.42%. And, of course, what about what’s happened since then?
Fourth, Feldstein cites the income tax rate being lowered from 50% to 28%: result, more taxable income reported (I guess, from the wealthy, although why this doesn’t apply in the same proportion to the rest of the top 10% I haven’t a clue). This greater taxable income, claims Feldstein, comes from more work incentive, more payment as salaries, and reducing deductions.
Oh, really? On what planet is this guy living? Over the last 30 years, we have seen more payment as stock options, continued heavy use of deductions, and (here we go again with the flawed “raise the income, they work less” income effect on labor) that lowering of tax rates had no effect on either the work incentives of the wealthy or, for that matter, those with subzero net wealth (40-50% of the whole). If anyone was affected, it was the rest of the top 10%.
Fifth (again with the taxable income), Feldstein sees a “shift from business income to subchapter S” – i.e., what was reported before as business profits is now being reported as income to the wealthy from new subchapter S corporations. Is Feldstein seriously pretending that most CEOs are the owners of their businesses, or that most wealth involves small corporations suitable for subchapter S? Even if he isn’t, he again conflates wannabe wealthy who are getting high income to get into the charmed circle with those who are already wealthy and are getting most of their income from stocks. And, of course, this is a one-shot effect, even if it existed.
Sixth, Feldstein gets back partially to wealth. “Calculations of income [and, I have to guess, wealth] exclude Social Security [actuarial value], health benefits, and food stamps”, says Feldstein, resulting in much more in the way of relative income gains (and, apparently, relative wealth gains) to the non-wealthy, and much less to the wealthy. By all means, Feldstein, undercut your previous argument about income rising at the same rate as wealth. But don’t count Social Security, health benefits, and food stamps as wealth. These mainly accrue to the bottom 50%, and have done so since the 1930s and 1960s, respectively. Negative wealth is not the same as positive wealth. With negative wealth, you spend all of the benefits you get when you get them, leaving you still in the general category of “negative to zero wealth”. So trying to create a giant sum and saying, well, by decreasing negative wealth, this increases the relative share of the wealth of the bottom 50%, is not even close to a good way of calculating net worth.
Let’s run this one into the ground. What happens if you take away SS, food stamps, and health benefits? A lot of those people die – as they did before 1933. In that case, they aren’t counted as part of total wealth and total income. A reasonable figure might be a 50% difference, when these benefits are used, i.e., primarily after retirement – remember, only 1 in 10 potential SS beneficiaries lived to retirement age in 1933. And yet, at that point, the negative-net-worth folks were still apparently around 40-50%. And so, the maximum 25% reduction in the debt of the negative-wealth non-wealthy is partially to wholly counterbalanced by a 10% increase in the retired negative-to-zero-net-worth percentage of the population. We may also note that the use of an overall actuarial value almost certainly ignores that the poor die much younger than the wealthy.
Then Feldstein makes the jaw-dropping undocumented assertion that this putting of SS and health benefits into the income column results in a much smaller rise in relative income of the wealthy. Let’s go back to our previous example: Adding on today’s median yearly (actuarial) Social Security (and probably health benefits), and factoring in the fact that this would be received by 1 in 2 of today’s median workers who survive to retirement age and in old age are not receiving other income, would result in income in today’s terms of a maximum of $60K instead of $30K. So now, we are talking about comparing $2.4 million in income from the $100-millionaire to $60K instead of $30K. That’s a smaller ratio, but certainly not a smaller rise. And it is by no means significant.
[extended note: health benefits includes benefits during work days (through much of this period, available only to ½ of Americans, and decreasing over time) and Medicare after retirement – which is not guaranteed to rise with inflation. Thus, translating this into money, especially without factoring in the 1/3 share of company health insurance that the large corporations are asking workers to pony up, means that health benefits run around $7,000 per year or less before retirement, and more later – but probably not the equivalent of $7,000 more per year. Then we have “pensions”, apparently including 401(k)s and IRAs. Obviously, company pensions these days are worth far less as an annuity than promised, and both IRAs and 401(k)s suffer from both eventual income-tax application for those who actually have significant amounts in them. Moreover, IRAs and 401(k)s have had 2 ½ % fees and poor diversification choices, leading to a 3.5-4 % investment penalty relative to a Vanguard fund, or perhaps a 2% return in real terms, if you’re lucky, over an average of about 20 years. In effect, this allows a withdrawal of maybe $10,000 a year over your 10 years of life after retirement – and that’s optimistic. Total: $14,000 plus $7,000 plus $9,000, or $30,000 on top of your median wage. Again, we are moving after-retirement income to working days, and comparing it to rich whose investment income operates at the same pace before and after retirement.]
Then Feldstein completes the argument by saying “Count SS, health benefits, and pensions in wealth figures”. I’m sorry, but this person is really clueless. I just disposed of that one.
But really, we could have skipped the entire analysis above and applied some simple real-world observations. This is one case where examples really are representative of the whole, because there are pretty few wealthy – I estimate 150,000 people (not households) with more than 8 million in net worth, and so looking at the top 10 or 100 billionaires gives a pretty good picture of that group. So we have, in the top few, the Waltons, Trump, and the Koch brothers, all of whom inherited plenty of wealth, and increased it. We also have the upper ranks dominated by CEOs of major corporations and hedge fund upper management – and it should be clear that both have seen compensation including stocks (a large proportion, for CEOs) grow by 10-30 times over the last 35 years – in contrast to almost all of the rest of the population. Or I can cite the experience of an acquaintance who is seeing far more income and a far faster increase in net worth from a multi-million-dollar inheritance converted to index funds than from previous periods of top-0.2% job income and little investment income.
The kindest thing I can say about Feldstein’s analysis is that it seems to take what he did as part of Reagan’s economic team, using unexamined assumptions and figures available at the time, and assume that these figures and assumptions can be extended beyond 1992. The nature of the corporation, the types of and returns from investment, and government taxation (or lack thereof) were changing as he served, and have changed dramatically since then, and yet there is no real trace of these changes in his argument. Instead, we have conservative shibboleths justified by 1980s statistical analysis that even then was not satisfactory.
So don’t, as Prof. deLong does, say that it’s a good analysis with a poor WSJ title. Let’s lay the blame where it really is: Martin Feldstein did it to himself.