Author Archive

Buybacks Are Bad. But not for the Reasons You Think.

April 21st, 2020 1 comment

The megabillion-dollar corporate bailouts raining down in the coronavirus response are giving new urgency to voices on the left excoriating corporate stock buybacks. How can we pour money into these firms, even as they pump gushers of money out the other end that could be spent on hiring and investment?

Much of the pushback against that view is unabashed hippie-punching, claiming that lefties don’t understand basic finance and business. “Dividend payments drain money just like buybacks do. Why aren’t you complaining about those? And, the money distributed by either method gets reinvested in other companies. Are you saying we should eradicate shareholder corporations? Sheesh.”

Dividends do “drain” funds from firms — that’s their very purpose — so it seems like a telling question. But the finance guys claiming that buybacks are benign or even salutary (really, they’re almost always guys) merit a serious dose of punchback themselves. Because they’re foolishly (or intentionally) blind to the three big ways that buybacks are bad compared to dividend payments.

The first two are about tax avoidance. Buybacks let the top 10% of households (which own 88% of equity shares) extract cash from the firms they own, and pay vastly less in taxes than they would with dividend distributions.

When a firm distributes dividends, the whole disbursement is taxable for households. But with buybacks, households only pay taxes on their “profits,” or capital gains — the cash received for their shares, minus the shares’ original purchase price, or “cost basis.” If the firm buys shares for $25 and the selling shareholders’ average basis is $20, only $5 is taxed, versus $25 for a dividend distribution. (Dividends and capital gains are currently taxed at the same rates, much lower than taxes on earned income from working.)

But that’s just the tip of the iceberg. Even economists and tax experts, even on the left, seem unaware that most capital gains are never, ever reported as “income.” There are myriad ways that households effectively hide capital gains (mostly, legally) and protect them from taxation — too many and too complex to detail here, but the big-picture result is eye-popping:

Combine the “basis” deduction with all those shelter methods, and those buyback disbursements are barely taxed at all — again, compared to dividends, which all count as taxable income for households.

If you think progressive taxes are beneficial, that they’re necessary for widespread prosperity and well-being, society-wide economic security, and (pas possible) even greater economic growth, this is reason enough to think that buybacks are bad. But there’s another big reason that even economic-efficientists have gotta love.

Corporate insiders know about impending buyback programs before they’re publicly announced. So they know not to sell their shares. Call it insider not-trading, something that it’s essentially impossible for regulators to regulate. Then the announcement drives up share prices, and they sell. Insiders make a nice extra buck on the deal at the expense those who sold before the announcement, and of slower or uninformed shareholders — notably buy-and-hold retirement investors and pension funds.

Here’s the the smoking gun, courtesy of Robert J. Jackson, Jr., a commissioner of the Securities and Exchange Commission.

There’s a fivefold increase in insider selling (average) from the day before the announcement, to the day of.

If the finance guys don’t know the basic economic concept of “information asymmetry,” so well explained a half-century ago by George Akerlof in his seminal “The Market for Lemons,” their fingers-twirling-in-cheeks triumphalism might be the thing that merits punching. If they do know it (uh…they do), even more so.

Remember the LIBOR scandal? Traders manipulated global markets on a massive scale with far less information advantage than this.

You’d think that this would be obvious to the finance guys — I mean, it’s what they’d do given the opportunity, right? And in fact it used to be obvious to everyone. Buybacks were illegal until 1982, treated as a violation of anti-fraud provisions of the Securities Exchange Act of 1934 — because it was assumed they’d be used for market manipulation.

But in 1982 the finance guys convinced their (ideologically?) captured regulators to gut that prohibition, with the enactment of Rule 10b-18, providing legal “safe harbor” for buybacks as long as certain conditions are met. As it turns out, the SEC doesn’t even collect the necessary data to enforce those conditions. But even if they did, the rule doesn’t touch on the crux issue: insiders knowing when to sell — and especially, exclusively, when not to sell.

Whether it’s about rules or enforcement or both, Robert Jackson’s (and others’) research makes clear that the current system is completely inadequate to prevent buybacks being used for insider trading, market manipulation, front-running, skimming, and — let’s just call it what it is — institutionalized fraud.

So sure: dividends are just as bad as buybacks when it comes to “draining” cash from firms, money that in theory could be used for hiring and investment. (Though: maybe those disbursements will be invested in other firms, which will hire and invest?) Economic progressives should stop trying to grind that “drainage” ax, first because it’s so hippie-punchable. But more so, because it misses the two giant things that are so pernicious about buybacks: they’re a decades-long, many-trillion-dollar tax dodge, and they’re a vehicle for corporate insiders to enrich themselves while fleecing everyday equity holders.

Both of those mechanisms overwhelmingly benefit the ten-percenters, one-percenters, and those far beyond. And they leave ordinary working people who are trying to build a nest egg and “safe harbor” of their own in this predatory, precarious world, as usual, with the scraps.

Is U.S. Productivity Actually Skyrocketing?

March 11th, 2020 Comments off


Our standard measure of production, GDP, doesn’t even come close to explaining the accumulated wealth of nations.


How productive are we? How much stuff do we produce for every hour we work? It’s one of the central questions of economics, and per many economists, productivity growth is the ultimate determinant of our world’s centuries-long increase in material well-being. If we increase that ratio we can work less, or have more stuff, or some of each. Here’s what that measure looks like post-war.

By this measure (here in 2012 inflation-adjusted dollars), today we produce $73 worth of stuff for every hour we work, compared to $21 in 1948 — a 3.5x improvement.

Economists have been concerned of late because productivity growth has been moribund in the past decade or so (and sluggish since the 70s compared to previous decades). Here’s annual percent growth in productivity.

But the measure of production here poses a conundrum: production minus consumption — our “saving” — doesn’t explain all the increase in our collective wealth. It’s off by about 30 trillion dollars. See below.

So suppose we instead assume that increasing wealth is itself a (superior?) measure of how much stuff we’ve produced and not consumed. We can easily create a different measure of production: stuff we’ve produced and not consumed, plus stuff we’ve produced and consumed. Equals, stuff we’ve produced. (Not coincidentally, this is also the definition of Haig-Simons income — ∆NW + consumption.*)

Net worth in this figure is annual change. Here’s the cumulative sum of those series.

The difference may not look like much (it’s dominated by production of consumption goods, which have been…consumed), but it represents $30 trillion in additional accumulated wealth, net worth. For reference, households’ total net worth end of Q3 2019 was $114 trillion. (No: adjusting these two series for inflation-adjusted dollars barely changes this comparison — just the numbers and scale on the left axis.)

What we have here are two very different measures of production and accumulation, both based on market prices/purchases — one on prices paid for new goods and services over the years, the other on existing-asset markets’ price-estimates of what all our “saved” stuff is worth. 

Here’s what “real,” inflation-adjusted productivity growth looks like using those two different measures of production. (The consumption + ∆NW measure is subject to big swings and volatility based on wealthholders’ “animal spirits,” so it’s smoothed here to give the long view, with a ten-year rolling average.) Source data here.

Which of these market-price-based measures of production is “correct”?

If you think production minus consumption should equal “saving,” which should equal change in wealth/assets/net worth, you have to reject the GDP-based measure, based on purchase-prices in the markets for new goods. It doesn’t explain wealth changes. But that suggests markets have been underpricing new long-lived goods for years — at least according to later years’ asset markets.

But the net-worth based measure is pretty eye-popping and hard to swallow based on previous understandings. (Perhaps: it could be revealing the unpriced value of free internet services, this article for instance?) It’s tempting to reject it.

To be clear: doing so is to say that the new-goods markets/prices were right about what the purchased stuff was worth (and ditto, national accountants’ tallies of all those market purchases, tallies which include large and necessary but sometimes convoluted estimations like imputed rent and profits for owner-occupied housing). And, it’s saying that the asset markets are wrong, have been since the 90s. It’s saying we’re in the mother of all multi-decadal asset bubbles.

The net-worth approach to measuring production is attractive because production minus consumption equals change in wealth — an intuition that fails with the standard measure. And that net worth approach suggests that current productivity growth is very healthy indeed.

I’ll leave it to my gentle readers to consider the implications.

* For those who are fans of Godley and Lavoie’s work, check out their discussions of Haig-Simons income vis-a-vis net worth in the index for Monetary Economics, notably page 140, and importantly the top of page 490, in their appendix on sectoral balances (though they don’t call them by that name).

The Eighth Way to Think Like a 21st-Century Economist

February 22nd, 2019 Comments off

The teams at Rethinking Economics and Doughnut Economics have launched a contest for entries, asking “What’s the 8th Way to Think Like a 21st Century Economist?” It builds on Kate Raworth’s seven ways, here.

Here’s my entry:

8. Widespread prosperity both causes and is greater prosperity: From false tradeoffs to collective well-being.

“Okun’s Tradeoff” — the idea that inequality is necessary for economic prosperity and growth — is baked into 20th-century economic thinking. It probably carries some significant truth in a generally egalitarian economy. But in the 21st century, with wealth and income concentrations beyond even what we saw in the 1920s, with that era’s disastrous denouement, it just doesn’t hold water.

Today’s extreme concentrations cause us all, collectively — especially our children — to have less. (Excepting those few who are lucky enough to extract, hoard, and benefit from multigenerational dynastic wealth along the way.)

Broadly dispersed wealth and income offer up opportunity, prosperity, economic security, well-being, and a springboard for success to hundreds of millions, billions of people and families. And it uses less of our earth’s limited resources in distorted production markets delivering low-value, absurdly priced luxury goods and services demanded by those with astronomical wealth and income. With the same amount of wealth, broadly dispersed — and the increased spending that broader prosperity delivers (spending on higher-value goods) — we can enjoy a vastly better life for ourselves. And we can deliver likewise to those who come after us.

At least today, the equality-vs-growth tradeoff is wrong by 180 degrees. The choice is not a difficult one. In fact it’s not even a choice we have to make. Widespread prosperity both causes and is greater prosperity.

Safe Assets, Collateral, and Portfolio Preferences

January 29th, 2019 1 comment

Matthew Klein and Mayank Seksaria had an interesting Twitter conversation yesterday in response to a Stephanie Kelton tweet. Read it here.

Here’s my understanding of the financial mechanisms they’re talking about.

Government deficit spending deposits fixed-price securities (“money,” checking and money-market holdings) ab nihilo onto private sector balance sheets. These are perfectly “safe assets” in the sense that you always know what they’re worth relative to the unit of account (The Dollar). A $1 checking-account balance is always worth one dollar — if the holding account contains less than the FDIC-insured maximum. But for big finance players with big cash balances, they’re not as safe as…

Treasury bills/bonds. And since Treasury is required to “sop up,” re-absorb, burn those newly-created cash balances by swapping them for bonds (“borrowing”): deficit spending + bond issuance, consolidated, effectively deposits new Treasuries, ab nihilo, onto private-sector balance sheets.

Now to the portfolio effects, which are driven by the market’s portfolio preferences (a broader and IMO more aptly descriptive term than “liquidity preferences”).

If deficit spending delivers cash onto private-sector balance sheets, the market is overweight cash. (Assuming portfolio preferences are unchanged.) It can’t get rid of cash because cash is (by its very definition) fixed-price. There’s a fixed stock, unaffected by capital gains and losses. (Yes, net bank lending changes this stock, but very slowly.)

So to adjust their portfolios, market players bid up variable-priced assets: mainly bonds, equities, and titles to real estate. Voila, cap gains: there are more total assets, and portfolio preferences are achieved.

But wait: deficit spending + bond issuance, consolidated, doesn’t make the market overweight cash. It’s overweight bonds. Portfolio balancing is more complicated here, because bonds have variable prices (though they’re less variable than equities).

The market could just sell bonds, driving down their prices and reducing total assets, to achieve its portfolio preference — less bonds, same amount of cash and equities. Or it could sell less bonds but also bid up equities to hit its portfolio prefs; the first reduces total assets, while the second increases that measure. (As they say, further research is needed.)

But none of this, in my opinion, has a whole lot to do with the value of “safe assets” as “collateral” (except when asset prices are diving and all correlations go to one). That seems peripheral and secondary to me, a hamster-wheel of financial shenanigans, sound and fury signifying…

Another takeaway from this: Government deficit spending & bond issuance delivers new assets (Treasuries) onto private-sector balance sheets. But no new liabilities. So it creates more net worth.

But the portfolio balancing that ensues generally also drives up equity (and real-estate) prices, yielding a deficit-spending multiplier effect on wealth by driving cap gains that wouldn’t happen otherwise. One dollar of deficit spending/bond issuance results in more than one dollar in new private-sector wealth, assets, net worth.

That’s how I see it…

Why the “Money Supply” Is Conceptually Incoherent

December 28th, 2018 Comments off

Economists’/monetarists’ use of the term Money “Supply” reveals multiple levels of deep confusion.

1. Supply implies a flow. But they’re clearly referring to a “stock” of money: what’s tallied in monetary aggregates.

2. Even if you’re think of a stock of money: Supply is not a quantity, an amount, a numeric measure. It’s a psychological/behavioral concept — willingness to produce and sell — commonly depicted in a curve representing that willingness at different price points. (All economics is behavioral economics.)

But “supply” is necessary to validate the incoherent ideas of the “price of” and “demand for” money — a set of financial instruments like checking deposits whose price never changes (relative to the Unit of Account). That price can’t change — by definition, by construction, and by institutional fiat.

Likewise, the aggregate stock or so-called “supply” of fixed-price instruments, money, changes only very slowly via bank net new lending. (That change in lending is determined by myriad economic behaviors and effects.)

If so-called demand for money can’t change the (P)rice of money (it can’t), or the collective (Q)uantity of money outstanding (it can but not much and very slowly), what exactly are we talking about here in our imagined supply-and-demand diagram toy thought-experiment?

Actually, Only Banks Print Money

December 12th, 2018 2 comments

I’m thinking this headline will raise some eyebrows in the MMT community. But it’s not really so radical. It’s just using the word money very carefully, as defined here.

Starting with the big picture: 

You can compare the magnitude of these asset-creation mechanisms here. (Hint: cap gains rule.)

The key concept: “money” here just means a particular type of financial instrument, balance-sheet asset: one whose price is institutionally pegged to the unit of account (The Dollar, eg). The price of a dollar bill or a checking/money-market one-dollar balance is always…one dollar. This class of instruments is what’s tallied up in monetary aggregates.

A key tenet of MMT, loosely stated, is that government deficit spending creates money. And that’s true; it delivers assets ab nihilo onto private-sector balance sheets, and those new assets are checking deposits — “money” as defined here.

But. Government, the US Treasury, is constrained by an archaic rule: it has to “borrow” to cover any spending deficits. So Treasury issues bonds and swaps them for that newly-created checking-account money, reabsorbing and disappearing that money from private sector balance sheets.

If you consolidate Treasury’s deficit spending and bond issuance into one accounting event, Treasury is issuing new bonds onto private-sector balance sheets. It’s not printing “money,” not increasing the aggregate “money stock” of fixed-price instruments.

This was something of an Aha for me: If you look at the three mechanisms of asset-creation in the table above, only one increases the monetary aggregates that include demand deposits (M1, M2, M3, and MZM): bank (net new) lending.

Arguably there might be one more row added to the bottom of this table: so-called “money printing” by the Fed. But as with Treasury bond issuance, that doesn’t actually create new assets. The Fed just issues new “reserves” — bank money that banks exchange among themselves — and swaps them for bonds, just changing TheBanks’ portfolio mix. That leaves private-sector assets and net worth unchanged, and only increases one monetary aggregate measure: the “monetary base” (MB). 

I’ll leave it to my gentle readers to consider what economic effects that reserves-for-bonds swap might have. 

Fake News from the CBO? Some Very Dicey Numbers in the New Income Inequality Report

November 21st, 2018 Comments off

It didn’t take long to realize that something was very wrong.

The Congressional Budget Office just released its new report on The Distribution of Household Income, updated to cover 1979–2015. One thing in particular looked very dicey right off (source xlsx):

Household Capital Gains (per household, average)
2007: $8,800
2008: $4,400
2009: $2,200

Wait a minute. Households didn’t incur capital losses in any of those years? Like…trillions of dollars in losses, as real-estate and equity prices dove for the zero lower bound? Red flag, something’s wrong here. (And yes: the CBO does include cap gains in this household “income” measure. Below.)

The report gives zero explanation anywhere I can find of how cap gains are measured/estimated/calculated. But for certain, the CBO’s measure is wildly lower, and wildly less volatile, than other (well-documented) measures:

These other two measures move much more closely together. The CBO measure is the huge outlier. And besides being unexplained, it’s just obviously wrong on its face. It’s missing 60–75% of recent decades’ household capital gains.

Since the top 20% of households own 85% of U.S. wealth, cap gains go overwhelmingly to them. So this cap-gains under-estimate makes invisible a huge part of their income (increases) over decades — whether you’re talking before taxes and transfers, or after. #GotInequality?

State and Local Taxes

Next up: smaller but still pretty huge: when calculating its “after-tax/after-transfer” household income numbers, the CBO ignores state and local taxes $1.8T last year. If that measure incorporated those taxes, income would be about 15% lower over recent decades.

State and local taxes are regressive: lower-income households pay a higher tax rate (in some states, wildly higher). So with a complete measure you’d see lower after-tax incomes especially among those with…lower incomes.

The CBO measure does of course include federal taxes (which are progressive, especially in lower tiers), and it does include transfers from the states. It seems very odd to exclude taxes paid to the states.



The Real (Real) Wealth Effect: Do Wealth Changes Change Spending and Cause Recessions?

November 16th, 2018 Comments off

My gentle readers who have followed me over time will have seen this graph and statement far too many times by now:

Since 1970 in the U.S., (almost) every time you saw a year-over-year decline in real household assets or net worth, you were either just into or about to be into a recession.* It’s seven for seven. Though to be fair: there have been two recent false positives, following the 2000 and 2008 recessions. So nine for seven. But interestingly, David Andolfatto and Neil Irwin have pointed to 2012/13 and 2015/16 as “mini” or “invisible” recessions. Ditto the weak GDP growth in 2002?

In any case, it’s a pretty impressive track record of predicting recessions. The apparent takeaway: when people (suddenly) have less wealth, they spend less. Seems plausible.**

But I’ve long meant to look at this relationship more systematically: what’s the correlation between changes in real household wealth, and spending in the economy? In particular: is the correlation concurrent, or is there a lag? Do you see greater correlations with spending changes one, two, or four quarters after a wealth change? Or: earlier? Do spending changes precede wealth changes? Post hoc ergo propter hoc?

Or: which came first? The chicken or the egg? Looking at wealth and spending changes (for example), which one looks like the dependent variable, which the independent?

I had no real idea what results I’d see. This was pure curiosity.

I’ll start with the results. But one key explanation first: the “spending” measure in these graphs seeks to capture spending that households and firms have the discretion to change over short periods; it excludes housing and health spending — a great deal of which is spending by other parties, imputed to households in the national accounts (see Cynamon and Fazzari, below). It also excludes spending on new structures by firms, which involves long-term decision-making.

For the year-over-year series, bigger bars on the right. Wealth changes correlate (much more) with spending changes in later quarters. You can also read this in reverse: Changes in spending don’t correlate as much with ensuing wealth changes.

It seems especially notable: Changes in spending actually show a negative correlation with changes in wealth a year later. The naive takeaway — higher spending “causes” less wealth increase and lower spending causes more — should probably be eschewed. But…what’s with that?

The quarter-on-quarter changes show much less of a pattern than year-on-year, in both this graph and all the ensuing — perhaps just due to more random variation in the quarterly series. I’ll just show year-on-year in the following graphs, to remove clutter. (This means there’s overlap; in the +1 and -1 lags, for instance, the YOY change-in-wealth measure has a three-quarter overlap with the change-in-spending measure. +4 and -4 have no overlap.)

It would be great to see this graph elaborated, somehow depicting the correlations for wealth/asset-market runups versus downturns. I would expect higher correlations for the downturns, since they generally happen so much faster and so would have fast impact on spending.

The spending measure here includes both consumption and investment spending. Which type of spending seems to respond more to wealth changes?

The overall pattern is the same. But if wealth changes do affect spending, the effect seems to be greater on investment spending. (Though investment spending is of course only about 20% of total spending — 15–20% in this discretionary spending measure.)

Next, just for reference: by this method, consumption and investment spending seem to move together, concurrently. (Note the Y axis change below; perhaps not surprisingly, these are far bigger correlations.)

More curiosity: do we see the same apparent wealth effect on GDP that we do on spending?

Again the pattern’s similar. We even see the same negative correlation between GDP changes and wealth changes a year later. But the correlations are much lower. Not surprising: much of the spending that comprises GDP is not amenable to short-term discretionary changes by households, firms, and government.

For comparison: what kind of correlations do we see between changes in personal income and spending?

The correlations are high (note the Y axis) and strongly positive throughout. But the pattern is opposite to the apparent real wealth effect we saw above. Income changes have a weak(er) correlation with ensuing changes in spending. Put another way, spending changes tend to precede income changes, more than the reverse.

Finally, closing the loop, wealth changes vs income changes. This is a surprising result.

Note that personal income does not include asset-price-driven capital gains and losses, which are the prime movers in short-term wealth changes. So we’re comparing very different measures here.

The correlations are smaller than we’ve seen, but the left-to-right gradient shows big differences. Wealth changes correlate with ensuing income changes, but income changes have a negative correlation with ensuing wealth changes. Do with that what you will.

Knowing you’ll want to collect the whole set, here are all of these graphs combined into one.

The spreadsheet’s here. It’s pretty easy to add your own data series to compare other measures. It would also be great to see longer periods; the almost sixty-year data set might suffice for five-year lags?

Wealth, Consumption, and Spending

I can’t resist adding an overall comment on mainstream economics and economic modeling, which I think is pertinent to all this: it seems crazy to me that Keynes’ consumption function:

1. Isn’t a spending function. The proportion of spending that goes to consumption vs investment is arguably an important thing to look at, but it’s secondary and peripheral to the larger question of aggregate demand, or more aptly, aggregate expenditure. Or just…”spending.” So in addition to the never-ending befuddled saving-investment confusion that Keynes has delivered unto us (oh: “desired” saving and investment), the whole Keynesian “investment-led recovery” construct promulgates and participates in reifying the pervasive and pernicious mythos of noble, job-creating “investors.”

2. There’s no wealth term, or function, in the consumption (spending) function — something that’s SOP in advanced Godley/Lavoie-style stock-flow consistent (SFC) models such as this great one from Michalis Nikiforos, Genarro Zezza, and Marshall Steinbaum. There’s only an income term. Rather, econs bolt the rather gimcracky contraption of “budget constraints” onto the back end of their models. KISS.

Credit Where Due

This correlations approach comparing positive and negative lags is inspired by Arindrajit Dube’s, in his magisterial takedown of Reinhart and Rogoff’s sophomoric “government debt causes slow growth” claptrap. (Though his statistical sophistication vastly surpasses the freshmanic effort you see here.)

Further inspiration came from Roger Farmer’s article, “The Stock Market Crash Really Did Cause the Great Recession.” The work here perhaps generalizes the asset/wealth effect implicitly bruited in that title, and helps demonstrate it over a long period and multiple recessions.

The “discretionary” spending measure used here is inspired by the work of Barry Cynamon and Steven Fazzari (viz), deconstructing personal consumption expenditure measures to exclude imputed spending and etc. Hat tip to J. W. Mason for pointing me their way. I actually have an older spending/consumption series of theirs to hand, but only annual. A more recent and quarterly version could quite easily replace the consumption-spending series here.

The temerity to write this post — including its implicit assertion that in reality the monetary/financial economy (here: asset-price-driven wealth changes) is what drives the real economy (recessions) — owes much to a great tweet by Sri Thiruvadanthai:

Contra neoclassical econ money/finance are not epiphenomena but they are the real deal and the real economy is the epiphenomenon!

Also thanks to Jason Smith for his comments on Twitter. Thread.

To all my other interlocutors: many thanks.

* Interestingly, adding liabilities to assets, to derive net worth, adds no predictive value. This is perhaps not surprising; household liabilities are only about 15% of household assets, and they change slowly or in other words not much, compared to changes from asset-price runups and especially drawdowns.

** It’s also very much in keeping with Kahneman and Tversky’s Prospect Theory: people are especially sensitive and responsive to losses. So it seems plausible that this is a real economic effect driven by real human behavioral reactions. (Microfoundations!)

What Causes Recessions? A Physicists’ Complex Systems Model

June 4th, 2018 4 comments

I received some very interesting comments from Yaneer Bar-Yam to my recent Evonomics post — “Capital’s Share of Income is Far Higher than You Think.” He pointed me to his very interesting paper, “Preliminary steps toward a universal economic dynamics for monetary and fiscal policy.”

I’m using this space to reply with with some stuff that can’t display in that comments space.

I haven’t gotten to the full-boat, multipart reply that I have floating in my head, but wanted to get back on two items for the nonce, a question plus a response on recession prediction:

1. What is the function in this model that “causes” capital gains? This always strikes me as the core problem in a complete SFC model where flows (including holding gain “flows”) balance to and fully explain (change in) net worth: if you can write a reaction function that predicts asset-price changes, you’re a very rich person… 😉

2. The recession-prediction based on investment/consumption ratio misses a bunch of recessions (false negatives). Contrasted here with a personal favorite: every recession since 1970 has been preceded by a year-over-year decline in real household total assets/net worth. (Including liabilities to arrive at net worth instead of just using assets adds no predictive value). Click for FRED.

This predictor is seven for seven. Though: there are two recent false positives — shortly following the 2001 and 2008 recessions.

The investment:consumption ratio bruited as a predictor/cause in the paper is four for seven, and even there: the first year of decline in this ratio seems late in each case (as opposed to the measure’s peak) to suggest it as a cause:

Investment:Consumption ratio peak/first year of decline Year-over-year declines in real household net worth (CPI-adjusted; base year 82–84) Quarters of real YOY net worth decline – YOY % decline in first declining quarter – NW decline peak-to-trough % Beginning of NBER-dated recession
Q4 1969 – Q4 1970 4 – 3.9 – 5.6 Q1 1970
Q4 1973 – Q1 1975 6 – 3.8 – 9.9 Q1 1974
Q1 1980 – Q2 1980 2 – 1.5 – 1.4 Q1 1980
1981/1982 Q3 1981 – Q2 1982 4 – 1.8 – 1.0 Q3 1981
1989/1990 Q3 1990 – Q2 1991 4 – 3.2 – 1.9 Q3 1990
2000/2001 Q4 2000 – Q4 2001 5 – 2.0 – 8.1 Q1 2001
Q2 2002 – Q1 2003 4 – 1.8 – 5.2
2007/2008 Q4 2007 – Q3 2009 8 – 3.8 – 19.3 Q1 2008
Q3 2011 – Q4 2011 2 – 0.5 – 2.6

I have various ideas and explanations for all this, but apologies, haven’t found time to write them all up.

Are Holding Gains “Pseudo” Income? A Response to Martin Sandbu

April 12th, 2018 2 comments

I just noticed with pleasure that Martin Sandbu, whose work I much admire, has posted a response to a thread of posts between me and Matthew Klein. Here in chronological order:

Me: Why Economists Don’t Know How to Think about Wealth (or Profits)

Matthew: The virtues and pitfalls of putting capital gains into the national accounts

Me: Wealth and the National Accounts: Response to Matthew Klein


Martin: You’re not as rich as you think
Beware of treating pseudo-wealth as the real thing

I’ll start with Martin’s conclusion (emphasis mine), and reply to some of his particular statements below.

Similarly, we should talk of pseudo-saving and pseudo-income when talking about valuation changes in asset (and liability) values. “Pseudo” does not mean it does not have real effects. It is precisely because stock measures of wealth are perceptions that they have unpredictable effects on real economic activity — and that these effects can be bigger the more unwarranted the perceptions are. But it is still real economic activity — as captured by conventional national income flow measures — that we should ultimately care about.

This is basically a statement about variability. Holding gains/losses are extremely variable. And yes, that variability — at least over the short to medium term — seems to be heavily driven by perceptions, optimism, confidence …  “animal spirits.” So you, we, can’t really “(ac)count on” those holding gains being “real.” They might vanish this year or next as perceptions change.

But variability is a function of time: how much does a measure vary over X period of time, Y period, etc. Holding gains are quite variable across our arbitrary one-year accounting periods. But over decades or a lifetime — or a dynasty’s lifetime, or the lifetime of a social, economic class — they’re very reliable indeed. Over any period greater than five or ten years, at least in the U.S. since 1960, they are consistently and reliably the overwhelmingly dominant method of wealth accumulation.

J.W. Mason makes that point very well in this post commenting on Piketty, from a couple of years ago, recently and appropriately re-upped by Cameron Murray on Twitter. Holding gains are the primary way that people (and we, collectively) get “rich” in balance-sheet terms.

If valuation changes, holding gains, are pseudo income and pseudo saving, then most of our monetary wealth, our balance-sheet assets and net worth — which has accumulated overwhelmingly through holding gains — is also pseudo. Or at least you have to ask: when do those balance-sheet changes, and the accumulated monetary wealth from those changes, become “real”? At what point do you decide that perceptions have become reality?

Amazon is the poster-child example for this. Despite showing essentially zero accounting profits over a quarter of a century, it has delivered half a trillion dollars onto shareholders’ balance sheets via holding gains — notably including Jeff Bezos’ balance sheet. Was that “real” income and saving? Is it now? Is the accumulated wealth “real”? Jeff Bezos owns The Washington Post. He’s throwing rockets into space. That seems pretty darned real.

Is Jeff Bezos “not as rich as he thinks”? Are ETF-fund investors who focus on total returns just foolish mugs?

Another way to illustrate this is to consider the free (advertising-supported) online services that people enjoy. (Recently discussed in a great Twitter thread with Sri Thiruvadanthai and Brad Setser.) How do we account for those? How do they enter into GDP? (The domain of “real” income, as tallied in the NIPAs and the FFAs, with no consideration of holding gains.) Google and Facebook sure seem to be creating and delivering “value” of some kind with those services…

The short answer is, they don’t get counted. Advertising spending isn’t part of GDP; it’s counted as an intermediate input to production, so it gets “backed out” of the GDP measure. This seems like a problem; there’s surely value, consumer surplus, being produced and delivered to the household sector; shouldn’t that show up in GDP? But no accountant is going to feel comfortable posting the imputed dollar value of free cute-kitten surfing as household-sector monetary “income.”

What actually happens: The profits from those advertising revenues are posted to firms’ balance sheets, increasing their book value. (Note that Facebook, Amazon, and Google, like other techs, don’t distribute their profit as dividends; they keep it on their books.) The markets see that increased book value, and bid up the companies’ stock prices. Voila, holding gains: every holder of those companies’ equities has more assets/money.

The household sector, ultimately, owns all the equity in the firms sector, at zero or more removes. The firms sector is a wholly-owned subsidiary of the household sector. (Because households don’t issue equity; firms can’t own households — at least not yet. It’s an asymmetrical ownership relationship. The ownership-accounting buck stops at the household sector.)

So the consumer surplus from “free,” advertising-supported online services is delivered onto household balance sheets (equity-owning households, at least) — via holding gains. The surplus is hidden in those gains. But that very real surplus is invisible in GDP. Should we call those holding gains, derived from real production surplus, “real” income?

My answer: New claims from holding gains, posted to balance sheets to the tune of trillions of dollars a year, variable as they are, are real claims. They can be (are) employed to buy stuff — notably including other people’s labor. (Yes: the asset markets must be liquid, there must be enough people swapping assets for this to work in practice.)

I address this from another, wonky angle — book-value versus mark-to-market, market-cap accounting, here. (Includes empirical data!)

Replying to some particular points in Martin’s post. He characterizes my thinking as follows:

economists miss much of what goes on in the economy by focusing largely on flows of income, spending and saving rather than the stocks of wealth, assets and liabilities.

I think this misses the key question: what do we mean by a “flow”?

There are three proximate financial mechanisms that create new ab novo private-sector balance-sheet assets — monetary wealth: 1. government deficit spending, 2. bank lending, and 3. holding gains. (Plus rest of world.)

Holding gains are special, completely unlike the other two. Because while holding gains is a flow measure (measured over a period), there is no actual flow. The new assets don’t come from anywhere, from any other sector. When there’s a market runup, everybody just marks their balance-sheet assets up to market. Nobody posts any new liabilities that you could identify as a “source” or flow for those increases. This is why holding gains are (must be) invisible in the balance-to-zero circular flow of the NIPAs and the FFA matrix.

And as detailed above, that non-flow “flow” of holding gains can derive pretty explicitly from real production and surplus.

As an aside, personal saving — spending less than your income — is another of these non-flow flows. It’s a residual flow measure of two actual flows in a period — income minus expenditures. (Household expenditures are all or mostly counted as consumption expenditures; it varies across different national account tallies.) It’s a measure of what’s not spent — income that’s not transferred to others’ balance sheets, accounts. It’s “not-spending.”

A focus on “net worth” and capital gains and losses draws our attention to assets — but liabilities, and the composition of each, matter hugely as well.

I addressed this in my reply to Matthew. Short form: This is like saying that a focus on revenues (assume we’ve been ignoring, failing to measure or analyze them) draws our attention away from expenses (which we’ve been tallying and analyzing very thoroughly). No: actually paying attention to balance-sheet assets, and where they come from, doesn’t “distract us” from liabilities (which are tallied well in the FFAs, and deeply analyzed by econs).

Paying attention to assets and their accumulation, monetary wealth, just increases what we’re paying attention to.

And: “net worth,” obviously, doesn’t ignore our well-accounted-for liabilities. They’re what net worth is “net” of. But you can’t get to net worth without a tally of total assets — or change in net worth without a tally of holding gains.

“Saving” in the sense of valuation increases does not correspond to anything on the ground, as it were.

I disagree. Over the long term at least (assuming “animal spirits” ebb and flow), valuation increases are the existing-asset markets saying “Wow, it looks like the markets for newly-produced goods and services got it wrong when they priced these goods. They’re actually worth more than we thought they were. They’ll deliver more value (via consumption or as inputs and services to production of goods) than we thought they would.” That’s them looking at all the “stuff on the ground” and giving their estimate of what it’s worth. (See the accompanying post on these two accounting/estimation methods.)

an economy as a whole cannot spend out of its financial wealth without devoting more of its actual current production to consumption

I think this is a widespread error of economic thinking. “Spend out of” is the problem; it’s an error of composition. When you “spend out of your wealth” — transfer assets from your balance sheet to someone else’s — nothing “comes out of” collective wealth. The assets still exist; they’re just in different accounts, on different balance sheets.

This is another instance of the “real stuff” vs. money confusion, here confusing consumption with consumption spending. When you eat more corn — literally consume — we have less corn. If you spend more to buy corn, we have the same amount of money.

So an economy can quite easily “spend [more] out of its financial wealth,” turn that stock over more rapidly, at higher velocity, with that extra spending going to either consumption spending or investment spending. Whatever.

The consumption spending doesn’t reduce our stock of goods/stuff, because the spending doesn’t happen if equal production doesn’t happen. Produced/sold goods minus consumed/purchased goods = zero (with some inventory/buffer-stock fluctuation period to period). This especially in a 70% service economy, where most goods are produced and consumed simultaneously; in a service business, inventories don’t exist. There’s no stored “stock” of labor hours.

(Net) Investment spending (“capital formation” in the IMAs) does increase our stock of stuff, which is then collectively monetized/assetized (fitfully) via the three financial mechanisms listed above. So okay: a larger proportion of consumption vs. investment pending does forego some wealth creation via capital formation-and-monetization. But it doesn’t destroy or diminish wealth as implied in the statement here.

Quite the contrary: Faster turnover of wealth, higher velocity, causes more production, investment, and consumption (assuming price inflation is in check). There’s no “spend out of” involved.

people chose to save more in the only way they collectively can: by spending less

I think Martin means “in the only way they individually can.” Individual saving has no accounting effect on the collective stock of assets. It only affects which accounts/balance sheet hold those assets. When you don’t-spend out of income, it just means you’re holding the money/assets in your account instead of transferring them to another account (by spending). Full stop. (Household debt repayment does reduce the household sector’s, and the financial sector’s, stocks of assets, shrinking balance sheets on both sides. Liabilities also decline on both sides, though, netting to zero, so it doesn’t change private-sector net worth.)

My main point is, again, political. Income measures that don’t include holding gains, saving measures that don’t sum to changes in assets and net worth, make invisible the primary method whereby owners get rich, stay rich, and get richer (without having to work). Until recently, even the total wealth measures were unavailable or squirreled away in separate tables that are themselves reliant on yet more obscure (“Reconciliation”) tables.

Economists are deeply implicated in that politically pernicious depiction of economic reality — mostly unconsciously. That’s forgivable,  perhaps, because economists receive no formal training in accounting theory or practice. (Is that forgivable?) But the result: even a remarkable student of wealth like Thomas Piketty is unable to perceive that his own second law is accounting-incoherent. It presumes that wealth increases all come from “saving.” Which isn’t even close to true. (Again, see J.W. Mason’s great piece.)

Thanks as always to my gentle readers…