Eating the Seed Corn? Consumption in the American Economy Since 1929
Following up on some work I did a while back (Kuznets Revisited: Investment in the American Economy Since 1929), I got curious about what consumption has looked like in America over the last 80 years.
I’ll give you the results first, as a proportion of output, or GDP, followed by explanation and discussion. Click for larger.
Thinking About Consumption
This all requires some explanation. First off, when I use the word “real” herein, it doesn’t mean “inflation-adjusted.” It means real-world, nonfinancial. In national-account-speak, all consumption and investment spending is about purchases of real goods and services — things that are produced and consumed. Financial “goods” or “assets” — which aren’t/can’t be consumed by humans — aren’t even part of the accounting. (We’re “inside” the NIPAs.)
Next: when you hear economists talk about “consumption,” they’re almost always talking about something somewhat different: consumption spending. Definition: purchases, within a period, of goods and services that are consumed within that same period. (If the period you’re looking at is long enough, everything is consumption. If it’s short enough, everything’s investment — breakfast is an investment in the afternoon’s work. I’ll just talk about one-year periods here to keep things simple.)
But in any period, we’re also consuming stuff that was produced in the past, and so is not included in measures of consumption spending. We’re depleting inventories (these fluctuate up and down over the years, sort of a buffer stock), but more importantly we’re consuming real, long-term productive assets by using them, and through the inevitable decay of time (and obsolescence — a tricky technical accounting issue that I won’t explore here). When you run a drill press you’re using it up. You’re consuming it. It’s even true of living in your house. Absent maintenance and remodels, it eventually becomes a worthless heap of rotting lumber; you’ve consumed its value by living in it.
This is accounted for in the national accounts as Consumption of Fixed Capital (CFC).* (Fixed capital is broken down into equipment, software, and structures, and structures are further broken down into residential and non-residential. Consumer durables — cars and refrigerators — aren’t included in fixed capital because it would require national accountants to model households as production centers).
But even with that big C-word at the beginning of its name, economists rarely include CFC as part of “consumption.” In the national accounts you’ll find it accounted for as part of investment — Gross Investment minus CFC is Net Investment, or the net amount added to our national horde of fixed assets. (When you hear macroeconomists talk about investment, it’s almost always shorthand for “gross investment.”)
So: Consumption of Fixed Capital is not spending, but it is (actual, if estimated) consumption.
I was wondering how much of our real output we actually consume each year, so I added CFC to consumption spending to get what I’ll call Gross Consumption (with Net Investment being the remainder of national spending).**
Coming back to the graph, we see:
• A denominator-driven spike in these measures during the depression (GDP plummeted).
• A rapid decline in the pre-war and war years driven by 1. rising GDP and 2. a massive temporary increase in government consumption spending. (Factoid: In just seven years 1940-1947, government’s share of national consumption spending went from 13% to 39% and back to 16%.)
• A steady period of relatively low consumption from the late 40s until the mid 70s or early 80s (depending which measure you look at).
• Consuming an increasingly large portion of our national output since the late 70s/early 80s — with Gross Consumption just shy of 100% ’08-’10.
(The timing of the 70s/80s breakpoint is somewhat unclear because of the high and variable inflation of that period, which had a significant impact on estimates of capital consumption. Changed amortization and depreciation schedules, applied against the whole existing stock of fixed assets, results in significant moves.)
The depression and war years are pretty to easy to understand. What’s interesting here is the trend change since the 70s/80s. That change is not generally driven by more rapid capital consumption: that’s been pretty steady over the decades (though business capital consumption has accelerated as the capital proportions have shifted from longer-lived structures to more rapidly depreciating hardware and software):
You can see that in the first graph as well; the two lines have moved pretty much in synch.
So the long-term trend change for the nation as a whole is simply toward more consumption spending and less investment spending. That may seem obvious to some, but it’s nice to know it for sure.
It’s the same 70s/80s inflection point and secular shift that we see in so many other measures, especially various measures of economic growth (pointed out repeatedly by so many, notably on Angry Bear by Mike Kimel, moi, Jazzbumpa, and others, and serving as the impelling premise of Tyler Cowen’s The Great Stagnation).
I can think of a dozen different explanations, interpretations, and rhetorical riffs off of this three-decade trend. While most have a perjorative and pessimistic import, not all do. Perhaps we’ve gotten better at supply-chain management, so we require an ever-smaller buffer stock of real stuff relative to consumption and output. Or maybe we don’t need as much productive stuff — as it’s measured in dollar terms in the national accounts — to produce the same dollar-denominated quantities for consumption.
These strike me as pretty charitable and speculative explanations, though. The most obvious interpretation is suggested by the title of this post: we’re consuming more and more of our seed corn (and tractors), leaving less for future consumption and production (and growth). That may be true, or it may be simplistically reductive.
For the moment I’ll leave that determination to my gentle readers, and simply leave you with a long-term view of what seems to be an important and not-widely-discussed measure — Gross Consumption — describing how our economy has been operating over the last eighty years.
* I really wish they said “fixed assets” instead of “fixed capital,” because economists are so profoundly confused about what “capital” means. (Many seem to think that fixed “capital” and financial “capital” are equally parts of some synonymous, homogenous, or vaguely contiguous undifferentiated blob.) Many everyday folks are equally confused about “assets,” but most economists seem to understand at least dimly that real (or at least fixed) assets and financial assets are decidedly different things that can’t be thought about, analyzed, or modeled in the same ways. The key in discussion is to always specify what type of assets or capital you’re talking about, especially differentiating between financial and real (with fixed being a subset of real that happens to be fairly easily measurable by totting up money transactions).
** Here ignoring the trade balance for the moment; we purchase 3 or 4% more than we produce these days — part consumption spending, part investment spending — because in toto we import more than we export. Don’t be confused by the GDP = Production = Expenditures = Consumption Spending + Investment Spending = Income identity; that’s only true in a closed economy.
Cross-posted at Asymptosis.
Over time how has the percentage of investment in IT technologies changed? One thought is that with the decline in the cost of a given transaction, the reinvestment required for static levels of consumpution would decline over time.It may be just a coincidence but the change in slope is about the time the PC arrived.
Excellent post. I’m going to Kimel (verb: to ask for simple graphs which make Republicans suffer). I really would like to see a scatter of net investment vs the top marginal income tax rate, and vs the capital gains tax rate. Eyeballing the decade averages, I guess an astonishingly close to perfect positive correlation. This wouldn’t settle any interesting debate about the economy (simple regressions generally don’t) but it would show that supply siders are totally full of it.
I’m going to opt for the optimistic answer, for now, and guess that the chief factor is a growth in the productive efficiency of fixed assets, which includes, I suppose, a shift in the share of production devoted to goods production vs. service production.
At the imaginary endpoint where we produce vast numbers of consumables from low-cost Star Trek style matter hacking, the consumption share of output should be extremely high.
I’d say reduced investment in property plant and equipment timed relatively well with NAFTA, and the boomer housing peak. Basically you’re hitting that point where amortization and depreciation are outpacing cash spending.
It’s the CAPEX section from cash flow, and right now, it is a “source” of cash, instead of a “use” of cash.
Other possibility is that in this calculation, assets are shown with a depreciable life not equal to (less than) their useful life, or using some accelerated method. that similarly causes expense to be inequitably allocated (most of the tax depreciation methods, which are dictated by policy preferences rather than accounting principles).
Robert, what I’d really like to do, and have poked at a bit, but it’s a huge job:
Provide a presentation that looks like Andrew Gelman’s Red State/Blue State work:
http://andrewgelman.com/movabletype/mlm/secretweaponsuper.png
A display method he calls the “secret weapon” — “fit a statistical model repeatedly on several different datasets and then display all these estimates together”:
http://andrewgelman.com/2005/03/the_secret_weap/
Now imagine that each of those images consists of one of these, showing the results for many overlapping periods:
http://www.asymptosis.com/wp-content/uploads/2012/05/Screen-shot-2012-05-11-at-1.08.26-PM.png
Each of these tables would be comparing an independent variable (the hypothetical “cause”) and a dependent variable (a hypothetical effect).
Each cell shows a correlation coefficient between the two variables.
Here are some of the independent and dependent variables one might compare:
Taxes, Investment, and Growth in the American Economy
Independent variables
Top marginal tax rate
x-year change in top marginal tax rate (additive or percent?)
Capital gains rate
x- year change in capital gains rate
Effective tax rate for top earners
Composite of corp effective, cap gain, and dividend rate
Dependent variables (show multiple x-year lags)
Net investment
Gross investment
GDP/capita growth
Productivity
Also show net and gross investment as IVs, lagged productivity and GDP/capita growth as DVs
We’ve got the secret weapon (showing many periods) embedded inside the secret weapon (showing many variables and lags)!
This would result in a heck of a lot of tables. But I think it would be a valuable presentation –both for the big-picture scannable overview and the drill-downable details — for feeding our human judgments as to what causes what.
It would still face the eternal problem, of course: is it just displaying effects caused by some external force, i.e. leaving the gold standard in ’71? (Maybe Republicans underperform because they were in power more during a secular slow period that was not their fault.) It doesn’t solve that problem any more than multiple regression does. You can’t include all the possible causes/IVs.
But still. I think it would be an amazing contribution to the discussion and body of knowledge.
Who’s in?
Another confounding factor *may* be the effects on consumption/fixed asset ratios of increasing reuse and recycling of production inputs as we attempt to become more efficient in our use of materials and energy in production. What would a shift from our current reuse and recycling practices toward those long effective in Germany imply for our fixed capital requirements relative to production of goods for consumption?
The decision to ignore trade balance effects, and thus the share of consumption attributable to net imports, is should have no impact if the balance condition is constant over the period. But if it changes significantly, say through an increase in imports relative to total consumption, it would perhaps be reflected in a decrease in fixed asset requirements and in the propensity to invest in fixed assets. Whether that is independent of, or reflected in the disadvantaging of domestic production by the overvaluation of the dollar accounting for such an increase in imports is altogether beyond me.
Not sure why you want to make up new meanings for words in order to write about economics. Words like “real” and “consumption”. It’s uneconomical. Here’s the thing. Nobody in your intended audience has the vocabulary in their head that you want to use. Either they are naive about economic lingo, or they aren’t. If they’re naive, you end up confusing them if they ever read standard economic lingo. If they are not naive, you are requiring the effort on their part of holding a well-known meaning for words away from their thinking while remembering novel meanings.
I’m pretty sure every aggregate you want to consider has been considered before, and has a name in standard economics. By using some other lingo than the standard, you avoid the work of digging out the standard terms, but imposing work – or the risk of confusion – on readers. One person avoids effort, but lots of people have to do extra work as a result. Not economical.
(This, by the way, is the best counter-argument I know against the libertarian view that we all can learn what we need to know about cars, chemicals, medicine and the like, and so don’t need government to protect us from consumer products. That view requires all of us to learn the same thing that a handful of regulators could learn on our behalf. Bad use of human resources.)
To broaden the point about re-inventing stuff that already exists, isn’t there a vast body of work on just these issues over decades and decades? Isn’t this pretty much the insight of production functions based on ratios of labor and capital inputs? Solow? Soviet growth model? Where you going with this?
kharris:
I’m not redefining “consumption.” I’m using it in its real sense, and explaining how that sense relates to a the term as (generally and silently) used by economists: consumption spending. The definition I give of that term is stright out of the textbooks and national accounts.
Ditto net and gross investment — as defined in the NIPAs.
And “real” in the sense I use it here is far from uncommon in the economics literature. I just stated clearly the meaning I was intending to make sure everybody knew–especially those who might not know the “codes.”
Accounts can be looked at in many different ways. Capital consumption is displayed in the NIPA tables under investment, but that doesn’t mean that is must be or that it’s the only useful way to position or look at it. It is consumption, by any standard.
It’s crucial when you use terms to be very precise. Most economists know that “investment” almost always means “gross investment.” But wouldn’t it be better if they always said “gross investment” when that’s what they mean?
I think your view of economists’ uses of terms is excessively sanguine. My impression is that there’s a lot of sloppy speaking out there; I see economists talking past each other based on misunderstanding of terms all the time. Like watching I Love Lucy sometimes.
Not taking a lot of time here, I’ll just give you these as an example, from those as knows a whole lot of economics and economics history:
http://worthwhile.typepad.com/worthwhile_canadian_initi/2012/01/why-saving-should-be-banned.html
http://www.themoneyillusion.com/?p=12617#comment-126456
This was indeed an excellent post and I hope more experts will focus on this issue. The solution is difficult, and neither wing, left or right, is likely to endorse it. But we need to move back toward more of a hands-on manufacturing and agricultural society and away from not only entitlements and unproductive, wasteful and corrupt public investment but also from financial services, professional services, and the security state.
You don’t own the language. Your claim that your sense is “real” doesn’t address the points I made. Your claim is nothing more than “does so!”
I would agree there is lots of sloppy economic speech. You are participating in that sloppy speech. It is one of those odds tricks of human irrationality that you see your own faults in others.
Let’s take just one example, that you have kindly offered above:
“Accounts can be looked at in many different ways. Capital consumption is displayed in the NIPA tables under investment, but that doesn’t mean that is must be or that it’s the only useful way to position or look at it. It is consumption, by any standard.”
OK, the first sentence is true, but vacuous. We aren’t talking about “looking”, but rather about using words in their accepted meaning.
The, there’s the recognisiton that by one standard, capital consumption is “investment” followed directly by the claim that “by any standard” capital consumption is consumption. No. By NIPA standards, it is investment. You have decided that a label IS the meaning. Another term for capital consumption is “replacement investment”. If you had settled on that as the thing that is true “by any standard”, you’d be claiming that everybody who doesn’t realize that we are talking about investment is wrong.
As to the links, the first one is a claim that the way the profession does things is wrong. Telling me that a claim that some other term “savings” is not used correctly doesn’t make your claim that “consumption” isn’t used correctly true. You are making a similar claim about a different term. but there is another difference. The “Worthwhile” claim is extensively argued, with reference to something other than the linguistic preferences of the author. Your argument seems largely based on your view that between “capital consumption” and “replacement investment”, the former is somehow more right. Your view is not important in making the claim.
While I’m at it, I recommend to you (and anybody else who demands that specialists speak some other way than they do, to bring specialist speech in line with anybody else’s preferences) the introduction to Siri Hustvedt’s “Living, Thinking, Looking”. Hustvedt’s writing often involves making sense to non-specialist of specialist speech. Her view is:
“Specialized laguages make certain conversations possible because the speakers have refined their definitions and can then share and work with them. The problem is that the circle of speakers is closed unto itself, and the expertise of one field is not available to those in another…”
There is virtue in makings sense of specialized langugae for those not conversant in the language – a work of translation. There is virtue in refining the specialized language when it is found to mislead specialists. Demanding that specialized language comport with the common parlance or the ideosyncratic preferences of bloggers? There really isn’t much benefit to be had there.
@kharris: “you (and anybody else who demands that specialists speak some other way than they do”
I think you’re kind of off the rails here.
All I did was display the accounting in a different way (and yes: I coined “Gross Consumption” cause it seemed to easily encapsulate the concept):
(Consumption Spending + CFC) + Net Investment = Total Expenditures = GDP
vs.
Consumption Spending + (CFC + Net Investment) = Total Expenditures = GDP
Is one more “right” than the other? No. They’re just different ways of looking at the same thing.