Is GDP Wildly Underestimating GDP?

January 23rd, 2015 31 comments

The markets have been showing a rather particular schizophrenia over the last dozen or so years — but not, perhaps, the one you may be thinking of. This schizo-disconnect is between the goods markets and the asset markets, and their valuations of U.S. production.

In short, the existing-asset markets think we’re producing and saving far more than we see being sold and accumulated in the newly-produced-goods markets. Take a look:

Screen shot 2015-01-23 at 6.17.38 AM

(See here for some ways to think about these measures. The spreadsheet cumulating saving is here. You can find all the data series on Fred here.)

A huge gap has emerged between what we’ve saved and what we’re worth.

Household Net Worth is the asset markets’ best estimate of what all our privately-held real assets are worth. It’s our best or perhaps only proxy for that value. (Household net worth includes all firms’ net worth, since households are firms’ ultimate shareholders. Firms, by contrast, don’t own households. Yet.) This is not just about assets like drill presses and buildings, but also skills, techniques, knowledge, organizational systems, etc. — all the tangible and intangible stuff that allows us to produce more stuff in the future. Household Net Worth at least provides us with an index of the change in that total value, as estimated by the asset markets.

As we increase our stock of real assets (“save,” by producing more than we consume), household net worth (wealth, or claims on those real assets) increases. The valuation jumps up and down as asset markets re-evaluate what all those real assets are worth — how much output and income they’ll produce in the future — but the two measures generally (should) move together.

Except: Since about ’98, and especially since ’02, that hasn’t been true. And no: zooming in on earlier periods doesn’t reveal the kind of anomaly we’ve seen since 2002.

There are two oddities here:

First, the flattening of cumulative savings: this measure was increasing exponentially for decades. Then it slowed significantly starting in the late 90s, and has gone flat to negative since The Great Whatever.

Second, the continued exponential growth of household net worth, and the resulting divergence of the two measures.

But bottom line: Net Worth and the cumulative stock of savings used to move pretty much together. They don’t anymore. What in the heck is going on?

There are three possibilities:

The asset markets are wrong. They’re wildly overestimating the value of our existing stock of real assets, and the output/income they’ll deliver in the future. See: “Irrational exuberance.”

The goods markets are wrong. The market for newly-produced goods and services is setting the prices for newly produced goods below the production’s actual value.

GDP is wrong. We’re producing something that’s not being measured by the BEA methods (tallying up what people spend on produced goods). There’s production the GDP methods can’t see in sales, so it doesn’t show up in saving (production minus consumption). But the asset markets can see it (or…sense it), and they deliver it to households in later periods, through the mechanism of market asset revaluation/capital gains.

Techno-optimists will like this last one. You’ve heard it before: The BEA has no sales-based method for estimating the produced value of free digital goods like Wikipedia, or the utility people derive from using them. They’re not purchased, so the BEA can’t “see” them. They could look at ad dollars spent on Facebook as a proxy for the value of browsing Facebook, but…that’s a pretty shaky estimation method, especially when many of those ad dollars would have been spent anyway, in other media. GDP simply doesn’t, can’t, measure that value, because nobody purchases it.

The timing sure supports this invisible-digital-goods story. The divergence takes off four to eight years after the release of the first mainstream web browser, and the global mainstreaming of the internet in general.

But it’s worth pausing before swallowing that explanation wholesale. You have to ask, for instance:

How does the internet/digital-goods story explain the flatlining of cumulative savings? Shouldn’t that continue to rise, though perhaps not as fast as net worth? Has the internet killed off sales (and accumulation) of traditionally measurable, purchased, goods to the extraordinary extent we see over the last dozen-plus years?

Are the asset markets seeing something else that GDP can’t see? Improved supply-chain management? More-efficient corporate extraction of profits from other other (less-developed?) countries? More-effective suppression of low-end wages? The rising costs of education and health care? (Which the BEA counts as consumption, extracted from saving, even though they’re arguably investment at least in part; they produce very real though intangible and difficult-to-measure long-term value/assets.) Or — here’s a flier — does it have something to do with the Commodities Futures Modernization Act and other financial “liberalizations” passed in the waning days of the Clinton administration? Something else entirely? In particular: would any of these explain the striking trend change in the cumulative savings measure?

Whatever the causes, the divergence of these two measures suggests a rather profound and singular economic shift of late — a shift that is not being widely discussed, even amidst the recent spate of commentary on Piketty’s Capital. (Piketty, by the way, defines wealth and capital synonymously — though his usages are not always consistent.) Prominent exceptions include the economists Joseph Stiglitz and Branko Milanovic, who are actively interrogating the troublesome theoretical intersection of wealth and real capital. The recent divergence of these two national accounting measures suggests that they’re tilling fertile ground for our understanding of how monetary economies work, and how we measure those workings.

Note: Technically one might add (negative) government net worth to the household measure to arrive at national net worth. But: 1. government net-worth estimates are inevitably dicey to meaningless. Government assets (and services) aren’t generally sold in the marketplace, so we have no observable sales information to base our estimates on. Liabilities are also very tricky: estimates vary massively based on your chosen time horizon and (necessarily) arbitrarily chosen discount and economic-growth rates. And 2. It barely changes the picture drawn above. Feel free to add government to the spreadsheet if you want; you’ll find estimates of net worth for the federal, and state/local, government sectors here. Net worth is — as it should be — the bottom line for each sector.

Cross-posted at Angry Bear.

Why You Probably Don’t Understand the National Accounts. In Pictures.

December 24th, 2014 5 comments

If anyone means to deliberate successfully about anything, there is one thing he must do at the outset: he must know what it is he is deliberating about. Otherwise he is bound to go utterly astray. Now, most people fail to realize that they don’t know what this or that really is. Consequently when they start discussing something, they dispense with any agreed definition, assuming that they know the thing. Then later on they naturally find, to their cost, that they agree neither with each other, or with themselves. That being so, you and I would do well to avoid what we charge against other people.

—Socrates, in Plato’s Phaedrus

A recent post of mine, How Do Households Build Wealth?, got a fair amount of attention (even a radio interview) because its takeaway graph seemed to surprise people (as it did me, when I put it together). Here it is again, presented more sensibly as a bar rather than an area chart. Click for larger.

Screen shot 2014-12-18 at 6.12.44 AM

Note: the revaluations shown here are not “realized” capital gains (which really only matter for tax purposes). They’re changes in asset values. If your portfolio’s value goes up by $20,000 this year, that bumps your net worth by $20,000 even if you don’t sell any assets. Ditto your house, but without the second-by-second reporting of prices.

What surprised people: capital gains and losses completely overwhelm Saving (net or gross) when it comes to changes in Net Worth. Here’s an even starker depiction, showing both magnitude and volatility:

Screen shot 2014-12-19 at 8.49.06 AM

I think these graphs are surprising to people because they don’t understand what Saving means in the national accounts, or how it relates to Net Worth. Thinking of households, for instance, many might be surprised to find that:

Starting Net Worth + Saving ≠ Ending Net Worth

Household Saving ≠ Change in Household Net Worth

As you can see in the charts above, it’s not even close. Why? In the national accounts, capital gains/revaluations aren’t counted as part of “Income.” It’s no wonder the Saving/Net Worth equality doesn’t balance; a whole bunch of household “Income” is missing from “Saving.” (Should capital gains be counted as Income and included in Saving? More on that below.)

The somewhat particular, stylized measure labeled “Income” in the national accounts serves an important purpose, but it also makes the word’s meaning opaque and confusing to many. That’s even more true of Income’s residual, “Saving.” The estimable Nick Rowe went so far as to title a post “Why ‘saving’ should be abolished.” Sez Nick: “it’s the most confusing concept in macroeconomics.”

But it doesn’t have to be confusing. The Integrated Macroeconomic Accounts of the United States (IMAs) — based on the modern international System of National Accounts (SNA) — calls out the particularity of “Income” by explicitly labeling the measure “Primary Income” (or more precisely, Balance of Primary Incomes). Drawing on that, the following chart may help render the relationships more transparent, adding a few named measures as aids to understanding.

Household Income
Comprehensive Income (gross contributions to net worth, before netting out expenditures) Non-Property Income (compensation for labor) Other Labor Income Social benefits and other transfers received (including employers’ social contributions)
Primary Income Primary Labor Income: Wages and salaries
Comprehensive Property Income (compensation for ownership) Primary Property Income: Dividends, interest, proprietors’ income, rental income, and operating surplus
Other Property Income Market asset revaluation (capital gains)
Other changes in asset volume

Using these terms, here’s how the national accounts tally up Income and Saving:

Primary Labor Income
+ Primary Property Income
– Interest Paid
= Primary Income

+ Benefits & Transfers
– Taxes
= Disposable Income

– Consumption Expenditures
= Saving (net or gross of capital consumption/depreciation, your choice)

We could call this final measure Primary Saving, because it’s a construct built from Primary Income. I discuss its meaning below.

This all balances, of course, but it’s confusing at least in part because it combines sources and uses of funds into Income, rather than keeping them distinct so Sources = Income. (See Interest Paid and Taxes — both “uses.”) If you’re not holding this whole construct clearly in your head when discussing Income and Saving — and likewise your fellow discussants — you (and they) are probably not adhering to Socrates’ dictum.

A different approach may help. Below is a breakout of Comprehensive Income sources for U.S. households over the last half century, smoothed with rolling ten-year averages to hide confusing noise and make long-term trends apparent. This presentation alters one named measure in the IMAs: Primary Income is displayed before deducting Interest Paid; that’s tallied with Uses, below. Interest Paid comprises 3–7% of Primary Income, 2–6% of Comprehensive Income. Note that everything balances. The spreadsheet’s here. See the balance checks in rows 86–93.

First, sources as nominal dollar values:

Screen shot 2014-12-19 at 9.48.51 AM

And here displayed as shares of the total:

Screen shot 2014-12-19 at 9.54.23 AM

Next, uses. Note that the total matches the total of Sources, with change in Net Worth as the balancing item. (Net worth is the starting and ending balancing item in the IMA/SNA presentation, which starts and ends with the balance sheet, Assets minus Liabilities.)

Screen shot 2014-12-20 at 10.15.39 AM

And here as shares of the total:

Screen shot 2014-12-20 at 10.16.06 AM

We could call changes in Net Worth “Comprehensive Saving,” since it’s a residual of Comprehensive Income. I find it to be a more straightforward and easier-to-understand residual than Primary Saving, partly because uses — here taxes and interest paid — are kept separate from the sources. But mainly because:

Comprehensive Income – Expenditures = Comprehensive Saving = Change in Net Worth

A necessary aside at this point:

I’m concentrating on the household sector here because household-sector net worth represents all private-sector net worth. The net worth of all firms is included in household net worth, because households ultimately own all those firms as shareholders or proprietors (or owners of those quasi-“firms,” households that own their homes).

Firms own (pieces of) firms which own (pieces of) firms (America’s founders would aghast at that, by the way), but ultimately households own it all.

You can think of the firm sector here as one big subsidiary of the household sector, with the accounting of its operations and net worth consolidated and telescoped into the accounting statements of their “parent company,” households. Details are obscured in the process, of course. But we’re looking at the big and arguably most important picture here.

By contrast, firms don’t own households. Households don’t issue equity shares — stock — and nobody, including firms, owns shares of, has ownership equity claims against, households. (Run don’t walk to read the Kollin brothers’ The Unincorporated Man.) This asymmetry means you can telescope firms’ value into households’, but you can’t do the reverse. The buck stops at households.

So household net worth comprises all of national net worth except for the net worth of government (plus/minus foreign net net ownership of U.S. assets/liabilities — equalling 1.5–5.5% of U.S. households’ net worth).

A further aside: in this telescoped, consolidated view, almost all household debt is owed to…households: the owners of the lending firms’ assets and liabilities. So this discussion ignores the important issues of lending, borrowing, and debt — important because different households are in different positions. Some (many) households are net debtors in this telescoped view, while others (few) are net creditors because they own the lending firms.

Returning to the key question I asked above: should asset revaluation, capital gains, be counted as income? It depends on our purposes, what we’re trying to display and understand. But the following simplified, stylized scenario may be useful in considering the question:

You buy a company that has a “book,” accounting value of $1 million. You buy it for $1 million. It produces stuff for ten years, makes $100,00 a year in profits, and retains them all. It doesn’t distribute any as dividends. A the end of ten years its book value is $2 million. You sell it for $2 million.

Capital gains: $1 million, with no Primary Income over those ten years. But: all of those gains resulted directly from ongoing profits from production. All of the “revaluation” of the firm as an asset was a result of the company’s increased book value due to retained earnings. The company’s value, and your net worth, increased by $100,000 a year, even if your household didn’t “realize” that by swapping the company for cash. Your annual $100K net-worth increases look decidedly like income from profits (quacks like a duck…), even if those  profits happened to be sequestered off on the books of your wholly-owned subsidiary.

There’s a pretty good case for calling that $1 million in cap gains “income.”

Now of course, not all revaluation/cap gains come from retained earnings and increases in book value. Asset values often go up just because the market has decided that assets are worth more than it thought they were — that they’ll produce more in the future (output and income) than was previously expected.

Here’s a useful way to think about it:

Primary Saving represents the net new stuff we’ve created in the period (the stuff we can count and measure, at least). Produced stuff minus consumed stuff. Net Saving = Net Investment — increases in the stock of fixed assets (and inventories) due to investment spending. We’ve saved in the sense of increasing our stock of stuff.

Comprehensive Saving (change in Net Worth) represents the net new wealth we’ve created. Sources minus uses, income minus (total) expenditures. New wealth ≠ new stuff, because market revaluation of existing stuff (not created within the period) causes wealth/net worth to increase and decrease even as those existing real-world assets remain unchanged. (See: We Have No Idea What Our Capital Is Worth.)

In the big, smoothed, long-term aggregate picture shown here, the market is consistently underestimating the value of current production — the net new created stuff. People are paying less for it than it turns out (in later periods) to be worth. Maybe the market’s undervaluing new human capital, or organizational capital, or future productivity increases, or…who knows. Whatever the case, the market catches on eventually, and (in volatile fits, starts, jumps, and reversals) revalues those previously created assets in later periods. That revaluation of previously produced assets adds to household Net Worth. So at any moment, household Net Worth (“savings,” or wealth — claims on future output) represents the market’s best guess at the value of existing assets and their future output.

When existing-assets markets go up, they quite literally create new wealth (we could call it “money”) out of thin air — new claims on those existing assets — to represent their best guess at the value of those real assets and the output they will produce.

So yes: capital gains are quite reasonably seen as income from production — production in previous periods. Even the parts that aren’t just book value increases due to retained earnings represent the value of production that we didn’t realize we’d produced. The market, in its infinite wisdom, realizes the error and in later periods, delivers that previously-unrecognized production value to households as…income, delivered via the mechanism of capital gains.

Going back to the chart showing income sources as shares of Comprehensive Income: we can see it as depicting the mechanisms whereby we (market and government) deliver and divvy up household income resulting from production. It shows the methods of slicing the production/income pie, and how the methods and slices have changed over time.

The first thing to notice: capital gains (and other volume changes) don’t dominate as they did in the “contributions to net worth” graph at the top of this post. That’s because that graph is somewhat unfair. It posts all household expenditures against Primary/Disposable Income (yielding Net Saving), and posts no expenditures against capital gains. It’s no wonder capital gains appear to dominate; they get a free ride in that picture — going straight to net worth, while Primary/Disposable Income are first drained through expenditures to yield Net Saving.

The Sources graph, by contrast, ignores expenditures; it just shows the Household sector’s sources of funds. Not uses.

This income/sources-only depiction still shows capital gains and other asset-volume changes comprising a significant (and volatile) 15–25% of comprehensive household income. And it displays some familiar features over time:

Compensation for ownership has increased as a share of comprehensive income, from 37% in the 60s to 45% in the decade ending 05/06. (Now recovered to 39% after the 08/09 decline, as ownership gains from recent years overwhelm the 08/09 losses.)

The wages and salaries share has declined from 51% of comprehensive income in the 60s to 40% today. (It’s currently on a four-year downtrend following the brief jump from the 05/06 low of 37% to 44% during the financial crisis.) Benefits and transfers have buoyed total labor compensation over the decades, but not enough to counteract the overall increase in ownership compensation.

That’s all useful and interesting, if not terribly surprising. We see the same results in many other measures. But this accounting construction lets us look at the economy in some other ways that my gentle readers may find valuable. For instance, changes in Net Worth as a percentage of Comprehensive Income:

Screen shot 2014-12-20 at 10.36.34 AM

Our income keeps increasing (yes, also when adjusted for inflation but of course more slowly). But post-Reagan, a smaller and smaller percentage of it has gone to wealth-building — from 30% in 1980 down to the low twenties in recent decades (and briefly, the low teens). Piketty watchers: what does this mean in the context of his thinking on capital and wealth?

Putting all the uses together in one place (rather than including some of them in the sources) also gives us a clearer picture of expenditures. We can remove the Net-Worth-changes balancing item, for instance, just looking at household expenditures, and see household consumption expenditures as a percent of total household expenditures:

Screen shot 2014-12-20 at 10.56.52 AM

This measure’s forty-year decline might give pause to economists (notably and vexingly, often liberal economists like Paul Krugman and Jared Bernstein) who pooh-pooh “underconsumption” explanations for The Great Secular Stagnation. (Though the runup over the last 15 years must also be explained.) Item: increased shares for interest payments and social contributions/current transfers were the two big contributors to this measure’s 8% decline (from 81% down to 73%), with each of those shares increasing by roughly 4%.

There are many other interesting depictions, which I’ll save for later posts. Before ending, though, I’d like to share a mental model that may provide a useful way of understanding all this.

Imagine that humans consume something we’ll call money. It’s the only thing they consume. And imagine that our whole economy, all our production processes, just produce money for humans to consume.

Next, imagine that we all produce for a year without doing anything else — no payments of wages, interest, or dividends, no stocks traded, no exchanges of any kind. We just put in 1. labor, 2. real capital services, and 3. natural resources, and money comes out.

We end up with a big reservoir of produced, consumable money at the end of the year.

The next day, the reservoir administrator (let’s call him Walras) transfers that reservoir of money to households. Wages and compensation are paid, dividends are distributed, interest is paid (to households; they’re not spending yet), and people furiously trade stocks, bonds, and houses.

At the end of the day the reservoir is empty; all that produced, consumable money has been divvied into the hands of households. (At this point, household net worth has skyrocketed!)

The next day, households spend. They buy food and other consumables for the year, pay interest to their creditors and taxes to the government, and give some money to the government to hold for them for the future. Whatever’s left over stays into their personal reservoirs of money, increasing their net worth compared to the previous year’s balance.

This stylized picture obviously misrepresents the economy in various ways, through omission at least. But I find it useful, along with the sources-and-uses bookkeeping depicted above, as one way to understand how markets and government divvy up our production pie over the years and decades. I hope others find it useful as well.

Merry Christmas!

Cross-posted at Angry Bear.

How Much Was Your Ballot Worth in 2014?

November 8th, 2014 1 comment

Amateur Socialist at Angry Bear asked me about how much was being spent per vote in 2014, and did the due diligence of finding me a spreadsheet showing how many ballots were cast per state. Ask and ye shall receive.

Based on that data, here’s a rough-and-ready calc of how much was spent on each ballot. Have your way with it…

dollars per ballot

Cross-posted at Angry Bear.

More Bad News for Dems: Total Total Total 2014 Spending Favored Them (Slightly)

November 7th, 2014 Comments off

If you’re like me, you’re often frustrated trying to find total (like, total) campaign spending by Democrats vs. Republicans. Outfits like the Sunlight Foundation do yeoman’s duty tallying spending, but you tend to get articles like this that (for fairly good reasons) don’t give you totals, rather breaking it down into campaign/party-committee spending vs SuperPAcs vs 501-whatever “social welfare organizations.”

What’s the bottom line? (Caveats follow.)

Screen shot 2014-11-07 at 6.46.58 AM

The Dems show a slight advantage (in the Senate and overall), but not much beyond the margins of estimation. Given the difficulties of estimation, the two parties spent about the same amounts this cycle on national elections.

To emphasize: this is an estimate. Three are undoubtedly some errors in the spreadsheet, both mine and others’. (Sunlight had Senate candidates Tim Scott of South Carolina and Bill Cassidy of Louisiana, for instance, tagged as a House candidates.) But fixing those errors would likely have little impact on the big picture:

Spending was roughly equal, probably a slight advantage for Dems.

The trickiest part of this estimate, based on Sunlight’s candidate spreadsheet, was allocating attack-ad spending by independent groups. If outside groups opposing Cory Gardner spent $30 million  in Colorado (tallied in the spreadsheet on Gardner’s line), I posted that as $30 million “spent” by/for his opponent, Mark Udall. And vice versa. My spreadsheet’s here.

Following are the Senate race-by-race spending totals. You may spot what look like anomalies, and you may be right. I obviously haven’t vetted Sunlight’s data. But if these numbers are close to correct, Democrats can’t claim a money avalanche by Republicans as a reason for the 2014 election results.

AK BEGICH, MARK 30,633,986
SULLIVAN, DAN 24,929,865
AR COTTON, THOMAS 33,199,597
PRYOR, MARK L 26,944,292
CO GARDNER, CORY 40,680,025
UDALL, MARK E 51,528,326
WADE, KEVIN L 123,614
PERDUE, DAVID 26,113,966
SCHATZ, BRIAN 5,688,359
IA BRALEY, BRUCE L 36,987,144
ERNST, JONI K 38,678,344
RISCH, JAMES E 977,987
IL DURBIN, DICK J 8,001,304
ROBERTS, PAT 15,937,833
LA LANDRIEU, MARY L 14,742,847
CASSIDY, BILL 7,506,478
MA HERR, BRIAN 857,332
MARKEY, EDWARD J 16,618,341
COLLINS, SUSAN M 4,864,766
MI LAND, TERRI LYNN 19,349,759
PETERS, GARY 28,049,683
MN FRANKEN, AL 20,172,311
COCHRAN, THAD 8,953,107
DAINES, STEVEN 6,313,452
NC HAGAN, KAY R 62,882,952
TILLIS, THOM R 41,752,166
NE DOMINA, DAVID A 1,143,959
NH BROWN, SCOTT 20,055,693
SHAHEEN, JEANNE 26,139,531
NJ BELL, JEFFREY 1,145,250
BOOKER, CORY A 16,980,057
NM UDALL, TOM 5,497,983
WEH, ALLEN 2,735,520
OK INHOFE, JAMES M 3,232,035
WEHBY, MONICA 5,378,129
RI REED, JACK F 2,454,090
HUTTO, BRAD 350,093
BALL, GORDON 1,180,680
TX ALAMEEL, DAVID M 10,217,029
CORNYN, JOHN 11,521,565
WY ENZI, MICHAEL B 2,486,637

None of this even glances, of course, at spending on state-level races. Given the condition of our campaign/electoral system and the amount of work it took to assemble these simple numbers, I tend to wonder whether that information will ever be known.

Cross-posted at Angry Bear.

How Do Households Build Wealth? Probably Not the Way You Think. Three Graphs

October 28th, 2014 12 comments

Work hard. Save your money. Spend less than you earn. That’s how you become wealthy, right?

That’s not totally wrong, but if you think that’s the whole story — or even a large part of the story — you may be surprised by this graph:

Screen shot 2014-10-28 at 8.45.12 AM

(Note: these are not realized capital gains, which really only matter for tax purposes. If the value of your stock portfolio or house goes up for twenty or thirty years, you’ve made cap gains even if you haven’t “realized” them by selling.)

Household “saving” — households spending less than they “earn” — contributes a remarkably small amount to increasing household net worth. And that contribution has shrunk a lot since the 90s.

Screen shot 2014-10-28 at 9.00.09 AM

The accounting explanation is simple: “Income” doesn’t include capital gains; it comprises all household income except capital gains. So capital gains are also absent from “Saving” — Income minus (Consumption) Expenditures. (This is why HouseholdSavings1 + HouseholdSaving ≠ HouseholdSavings2 — not even vaguely close.)

The capital gains mechanism appears to dominate the ultimate, net delivery of rewards to household economic actors. Earning more and spending less is weak beer by comparison.

What does this say about our understandings of how the economy works? Does economists’ fixation with “saving” provide a useful picture of macro flows in the economy? Since asset ownership is hugely concentrated among the wealthy (even real estate), can we think about the economy’s workings at all without looking at distribution? Does this dominant mechanism allocate resources “efficiently,” or deliver the kind of incentives that make us all better off? And etc.

There’s much more I’d like to say about this reality, but I’ll just provide one more graph for the time being and let my gentle readers ponder the bare facts.

Screen shot 2014-10-28 at 9.10.30 AM

The spreadsheet’s here. Have your way with it. (It’s kind of messy; drop a line with questions).

Cross-posted at Angry Bear.

Bill Gates Agrees with Me on Piketty

October 14th, 2014 3 comments

He really likes the book, but expresses frustration that Piketty (emphasis mine):

doesn’t adequately differentiate among different kinds of capital

Imagine three types of wealthy people. One guy is putting his capital into building his business. Then there’s a woman who’s giving most of her wealth to charity. A third person is mostly consuming, spending a lot of money on things like a yacht and plane. While it’s true that the wealth of all three people is contributing to inequality, I would argue that the first two are delivering more value to society than the third. I wish Piketty had made this distinction.

This is not exactly my point (here), but still it cuts right to the crux. There’s a conceptual flaw at the book’s core that makes it hard or impossible to think coherently about the levels, trends, and mechanisms that Piketty portrays:

Piketty defines capital as synonymous with wealth.

Wealth consists of all the tradable claims on real capital (specific ownership claims, generalized claims like dollar bills, and everything in between). The market constantly reprices those claims, resulting in a constantly-adjusted best-guess market estimate of what the underlying assets are worth — ultimately expressed as household net worth (with all firm net worth imputed to household shareholders).

The market reprices the claims, based on its revaluation of what the underlying assets are worth. If it thinks the assets are worth more (will produce more in the future), it bids up the prices on the claims.

Important: that stock of real assets is not just the “fixed capital” tallied (because it can be measured) in the national accounts; that’s actually a small part. Knowledge, skills, and abilities (think: education, training, health), business/organizational systems (this is huge), and similar unmeasurables constitute the bulk of real capital — the stuff that allows us to produce in the future. Most of that stock is not specifically claimed, but it is that whole body of real capital that the market it trying to value properly via pricing of claims — basically, holding up its collective thumb and squinting.

Piketty should have called it Wealth in the 21st Century. That’s what he’s really talking about, because we really have no idea (beyond the market’s best guess expressed in household net worth) what our real capital is worth.

“Financial capital” is an oxymoron.

Cross-posted at Angry Bear.

Are Poor People Consuming More than They Used To? Six Graphs

October 12th, 2014 5 comments

“Poor people today have air conditioners and smart phones!”

You hear that a lot. “You should be looking at poor people’s consumption, not their income. By that measure, they’re doing great.”

The basic point is very true. If poor people today have more and better stuff, can buy more and better stuff each year, maybe we should stop worrying about all those other measures that show stagnation or decline.

Does this measure tell a different story? Curious cat that I am, I decided to go see. I had no idea what I’d find.

The first thing I found: this simple data series isn’t available out there, at least where I could find it. Notably, the people who claim it’s so revealing don’t seem to have assembled it.

Consistent, high-quality expenditure data is available going back to 1984, from the BLS Consumer Expenditure Survey (CES). (Pre–2011 here. 2012 and after here. See “Quintiles of income before taxes.”) But you have to open a table for each year and pull out these numbers, which I did. The spreadsheet’s here. It’s simple, clear, and easy to work with, so please have your way with it. (Note: CES measures “consumer units” — “households” precisely defined for the purpose of measuring consumption. All the CES terms are defined here.)

The household spending measure is in nominal dollars. I converted the values to 2013 “real” dollars using the BEA’s deflator for Personal Consumption Expenditures. Here are the results (mean values; medians would be somewhat lower).

Screen shot 2014-10-11 at 5.11.24 PM

By this measure poor people’s consumption is up 5% since 1984 — not exactly the rocket-ship prosperity growth for the poor that consumptionistas proclaim.

But truth be told, this isn’t a very good measure as it stands — because households have gotten smaller. For the bottom 20% that looks like this:

Screen shot 2014-10-11 at 10.08.20 AM

Declining household size means:

1. Per-person spending would trend up (if income per household is the same, with less people per household).

2. But: you have to adjust because living with more people is cheaper per person because of shared rent, utilities, etc. If you don’t, it looks like the average person in a four-person household consumes vastly less than a person in a one- or two-person household — which clearly isn’t correct.

I corrected for that with the household-size adjustment method used by the Pew Research Center. This standard method is somewhat synthetic (more on that below), but it also clearly yields a more useful and accurately representative measure of poor people’s consumption. Here’s what those results look like:

Screen shot 2014-10-11 at 5.10.15 PM

By this measure, things have gotten more better: poor people’s consumption is up 14% since 1984, compared to 5% using the other measure. But really, that’s still nothing to crow about; real GDP per capita grew 63% in that period — four times as fast.

Screen shot 2014-10-12 at 7.43.01 AM

More comparison: the real price of a share in the S&P 500 has increased 370% over that period. That’s not counting dividends.

To get an apples-to-apples comparison, here’s a look at annualized growth rates for various periods:

Screen shot 2014-10-11 at 5.29.11 PM

Poor people have gained a little bit of ground in absolute terms. But they’ve lagged way behind the rest of the country. Growth in every period except the 1990s Clinton heyday was moribund or negative — notably ’84–90, the last twelve years, and (especially) the last four years of economic “recovery.”

The consumptionistas are absolutely right: this is a really good measure.

And it tells exactly the same story as the other measures.

No child left behind?

– – – – – – – – – – – –

Before I leave you, some proleptic responses to the predictable objections:

But, the Brookings study! Many point to this study (PDF) by the (liberal!) Brookings Institution. The study devises and tracks a measure they call “consumption poverty.” Here’s the money graph:

By this measure, very few people these days are living in “consumption poverty.”

Here’s the thing: this study uses the same data set you saw above. But you’re looking at it through a very synthetic lens: a somewhat-arbitrary “poverty threshold.” What percent of people are below that threshold? Which means you gotta ask: Is that a relative or absolute poverty threshold? How does it change year to year? How’s it calculated? Etc.

To say it another way: This measure is some calculation steps removed from — it’s a second or third or fourth derivative of — the data as measured by the BLS. I’m not saying it’s a bad or un-useful measure. I haven’t gone into the weeds with it. And Brookings tells you exactly how the threshold is calculated. But you have to understand the lens’s multiple assumptions and hold them in your head while you’re peering through the lens.

The graphs above, by contrast, are much closer to “the facts on the ground.” Assuming you’re interested in those.

Worth noting: the people who might prefer the story told by the Brookings poverty-threshold measure are the very same people who are forever complaining about measures that use arbitrary (and relative) poverty thresholds. Just saying.

The inflation adjustment misrepresents poor people’s reality. The “real” consumption spending in the graphs is also filtered through a lens: the BEA’s price index for personal consumption expenditures. What if that index is wrong? It is based on “hedonic” estimates, after all: what’s the value (as opposed to price) of today’s laptop compared to 1990’s? Cars are far more reliable than they used to be; knowing that your car will start every time you turn the key has real value. Air conditioning is more valuable than box fans. And think of all the free digital goods that poor people have access to now — from Google to online banking to… Those have no “pricing,” so they’re undercounted or uncounted in this measure.

This is basically saying, “you should create your own, different Consumer Price Index.” It’s exactly the same argument as the ShadowStats craziness, but in reverse: there’s not more inflation than is shown in the CPI, there’s less — more deflation. There’s a “true” index that’s misrepresented by the rather remarkable and diligent efforts of BEA statisticians.

Scott Sumner is rather the poster-child for this position. He says exactly this:

we should ignore all the official data, and use our eyes.  Travel around the country.  Go into poor people’s houses.  … I think I do have a rough sense of the different sorts of consumption bundles purchased by different classes of people.

You should construct your own CPI index by holding up your thumb and squinting, eyeballing poor people’s consumption bundles. Because…the official CPI is not saying what Scott Sumner would like it to say.

Scott’s been going on about his superior CPI estimates for years. Karl Smith probably gave the best response, a year back:

basically anyone with MS Excel and a rudimentary knowledge of the subject matter in question can create a workable index…. a task that brilliant people have devoted their life to

The thing is, Sumner doesn’t even use a spreadsheet. He does it in his head. (Now that’s brilliant.) We should clearly do likewise, or just adopt his index — if we knew what it was.

Finally, note: the comparisons above — to real GDP and S&P growth — use the BEA’s GDP-deflator and CPI indexes, which are only slightly different from the PCE index used here for consumption. Almost-identical apples to almost-identical apples. Feel free to mess with that in the spreadsheet if you’re so inclined. It won’t get you much of anywhere.

The household-size adjustment is invalid. This is another lens interceding between you and the measured data, on top of the inflation adjustment. No doubt about it. But as with inflation adjustment, you can’t get around it. You can’t ignore shrinking household size any more than you can ignore today’s less-valuable dollars. And you can’t just divide household income by people per household, or people in four-person households look like they have vastly lower consumption than people in one-person households. That just isn’t reality.

One part you might reasonably question: The Pew size-adjustment methodology uses a chosen variable that can be from 0 to 1; they choose 0.5 based on some decent research over the years. I tried values between 0.1 and 0.9. Lower numbers lower and flatten the red line, and show a consistent upward trend from ’84 to ’01 (flat thereafter). But the basic story is unchanged.

– – – – – – – – – – – –

Look: no method is going to give you the perfect gauge of human well-being — the”obvious,” magic-bullet measure that conservatives seem to forever be after in their eternal Search For The Simple. This consumption measure is no exception. But advocates of this measurement approach are absolutely right: it’s much simpler and easier to measure than most other measures, and it’s a very good measure of how poor people are doing.

Bottom line: Poor Americans’ consumption grew, very slowly, over the past three decades. Meanwhile the rest of the country grew four times faster, and a typical stock-market investment grew at least 27 times faster.

Do with that what you will. Adjust your priors as appropriate, if that’s something you do.

Then take a look at some closely related measures that might tell other important parts of the big story:

How much of this picture is about young versus old? The population is aging; do we need to change the story this measure tells based on changing demographics?

Did poor people take on more debt to achieve that higher consumption? How has that affected their lifetime well-being?

 How much of that consumption growth resulted from government spending, and how much from the market doing its job of allocating resources efficiently? (i.e., to those who will value them highly?)

Are poor people working more hours to get that consumption increase? Do they have more or less leisure and family time?

How economically secure are people? What is the typical person’s chance of falling into this bottom-20% consumption group? How has that changed?

I’ll try to address some of those questions in future posts.

Cross-posted at Angry Bear.

Amazon and Hachette: What’s Really at Issue?

October 5th, 2014 Comments off

I just sent the following to David Streitfeld, the main reporter covering this dispute for the New York Times. If any of my gentle readers has an answer, I’d love to hear it:

Dear Mr. Streitfeld:

What I have never found in any coverage (perhaps I’ve missed it?):

What is the precise dispute between Amazon and Hachette?

What does Amazon want from Hachette that Hachette is unwilling to concede to? (And/or vice versa.)

“Control over pricing” is vague, doesn’t answer that.

Prima facie, it seems that Hachette could charge whatever wholesale price it wants for books, and Amazon could charge whatever retail price it wants.

But that’s apparently not the case?

Absent a fundamental understanding of the specific matters at issue, I find it very hard to think coherently about the subject or draw any conclusions.

I have a long and diverse history in book publishing, by the way, from publisher to editor to author, and everything in between. I’d really like to figure out what I think about this.

Thanks for listening,


Is this Amazon saying, “Sure, charge whatever you want. But we don’t make money selling your books at our prices, so we’re going to downgrade their visibility in various ways to more prominently display books we do make money on.”

To which I’d respond, “Then raise the prices on Hachette books. If Hachette sells less books and sacrifices market share, why does Amazon care?”

Cross-posted at Angry Bear.

Explaining “The most important chart about the American economy you’ll see this year”

October 4th, 2014 1 comment

See update at bottom.

Pavlina Tcherneva’s chart has been getting a lot of play out there:

Vox/Matthew Yglesias labeled it “The most important chart about the American economy you’ll see this year.”

Scott Winship at Fortune came back at it on methodological grounds, with the headline “No, the Rich Are Not Taking All of the Economic Pie (In 8 Charts).” He ends up with what he calls the “money chart,” supporting his headline:

Yglesias responded, and Winship responded to him.

The basic contention at this point: who is actually “explaining” the situation?

Do Scott’s corrections explain the situation better? Do they paint a 1. more accurate, and/or 2. more complete picture? By those two standards, is he more honestly and fully informative? Let’s run through his changes.

1. The household method as opposed to tax-unit method is at least a useful additional measure, and is arguably more accurate and explanatory. It’s also less complete because the data only goes back to ’79. But he completes it with Tcherneva’s (by this standard less-accurate) earlier data. It’s a useful addition to understanding, but with little change in the story it tells.

2. The full-business-cycle approach (as opposed to just showing expansions) is also arguably more accurate hence informative. But it (necessarily?) ignores the post-2007 period because the current business cycle isn’t over yet. By Tcherneva’s measure this period is far and away the most egregious demonstration of the inequality trend we’re examining. Scott could have included recent years, with visual and verbal caveats explaining that the cycle is not complete, so the measure that period is not directly comparable. It has less explanatory power, but that doesn’t mean it has none. Omitting the very period that by Tcherneva’s measure are the “money proof” of the trend (and so omitting explanations of that period) arguably explains less about that trend.

3. Looking at the non-elderly population is arguably more accurate and is at least quite informative. It paints roughly the same picture, though less extreme.

4. Post-tax-and-transfer measures are arguably more accurate and informative, but again with the completeness problem. And he should explain that the pre-’79 picture would look quite different if it displayed post-T&T data; the bottom 90% would have been getting more of the pie, which would make the inequality trend look more pronounced than it does in his graph.

I do wish he’d shown a graph as he suggested, including health/medical benefits in T&T (assigning a value other than $0), despite the methodological difficulties he points to. I have no idea how or how much this would change the picture.

5. Omitting capital gains (because they’re hard to measure) for the final “money chart” — suggesting that it’s the most accurate, complete, informative, and definitive chart — is a massive hit to accuracy and explanation. Cap gains are the very vehicle, the primary means, by which the increasing inequality we’re trying to understand is realized. “The data might not be accurate” begs the question: Is excluding that data more accurate? To use Scott’s own words, “assigning a value of $0 is surely not right.”

So some of Scott’s corrections help to usefully and informatively explain the situation better, or at least more. But to summarize the changes that are less informative or downright misinformative:

• The blatant inaccuracy of ignoring cap gains in #5. It completely misrepresents the situation.

• The omission of recent years in #2 — the very years where the trend is arguably most apparent and egregious. Hiding the elephant under the rug?

• The blithely dismissive headline of Scott’s first post.

With these combined, I hope Scott can understand why many see his post as an effort to pooh-pooh and obfuscate the whole subject — the very antithesis of “explaining.”

Part of that hand-waving, obfuscation, and general chaff-dispersal is the proleptic but of course you’re right” rhetorical ploy, right up front in Scott’s second paragraph:

Let’s stipulate that income inequality is at staggering levels in the U.S., and that income concentration at the top has probably risen (probably)

One really must ask: if income inequality is at “staggering” levels, how did it get there…if it hasn’t risen?

How do you square that staggering stipulated reality with Scott’s headline assertion: that “the Rich Are Not Taking All of the Economic Pie.”

I can’t see how to draw any other conclusion from this direct self-contradiction: he’s talking out of both sides of his mouth. I’ll leave it to my gentle readers to decide why.

Takeaway: obfuscation is the opposite of explanation.

Update: Scott has taken issue with my only-barely-implicit imputation of his motives. He’s right on that. I both regret that and apologize for it. I still think the import of his post (especially the “money chart” and title) — that inequality’s not that bad and not that important — contradicts his “staggering” stipulation, and is rhetorically pernicious. But that’s not the same as bad faith. I withdraw and apologize for any suggestion of the latter.

Cross-posted at Angry Bear.

Again: Saving Does Not Increase Savings

October 1st, 2014 40 comments

I’m reprising a previous (and longer) post here in hopefully simplified and clarified form, for a discussion I’m in the midst of.

“Saving” and “Savings” seem like simple concepts, but they’re not. They have many different meanings, and writers’ different usages and definitions (often implicit or even unconscious) make coherent understanding and discussion impossible — even, often, in writings by those who have otherwise clear understandings of the workings of financial systems.

I’m going to talk about a particular meaning of Saving here: Personal Saving (by households) as defined in the NIPAs. Quite simply, it’s household income minus spending on newly produced goods and services. (It doesn’t include so-called spending to buy already-existing assets like deeds, stocks, bonds, or collectibles like art.) It’s a very different measure from household-sector Gross or Net Saving, which I won’t describe here.

Now think this through with me:

Your employer has $100K in its bank account. You have zero.

Your employer transfers $100K from its bank account to yours to pay you for work. You’ve saved (in the Personal Saving sense).

But is there more Savings in the banks? Obviously not.

Now you buy $100K in goods from your employer, transferring the money from your account to its. You’ve dissaved (spent).

Is there more or less Savings in the banks? Obviously not.

Now say instead that, being frugal, you only transfer $75K to your employer for goods. You’ve “saved” $25K. That’s “Personal Saving” in the NIPAs.

Is there more or less Savings in the banks compared to the first scenario? Obviously not.

When households save money, that (non-)act doesn’t add to the stock of monetary savings (the mythical stock of “loanable funds”).

Thinking about the accounting entries may help explain this. “Saving” is a flow, as opposed to a stock. Every accounting measure must be one or the other. (A flow is measured over a period; a stock is measured at a particular moment.) Saving is an accounting “flow” in that sense, but it doesn’t represent an actual transfer of funds from one account to another. It’s an accounting residual of two sets of actual transfers: income minus expenditures. You could say it’s nothing more than an artificial accounting construct — though a useful one for thinking about balance sheets and flow-of-funds and income statements.

The very essence of Personal Saving, its sine qua non, is that it’s not a transfer of funds between accounts. It’s leaving your money sitting where it is, instead of spending it by transferring it to others. It’s not-spending. (Your transfers between your checking and brokerage account, or your portfolio rebalancing, notwithstanding.)

Since Personal Saving doesn’t transfer anything anywhere, it can’t increase the stock variable, Savings.

What does cause Savings to increase? Spending.

Cross-posted at Angry Bear.