Bryan Caplan  

Real Real GDP

Why Derivatives?... Access to Academic Research Pa...
Gross Domestic Product is staunchly atheoretical.  If someone spends money on X, X is GDP - even if "someone" is Congress, and X="a bridge to nowhere." 

There are exceptions; most notably, the stats supposedly exclude "intermediate goods" to avoid double counting.  I say "supposedly" because the list of "intermediate goods" is so inconsistent.  Insofar as police protection and the military protect firms from harm, aren't the police and military intermediate goods?  But despite these tensions, a big part of the philosophy of GDP is to eschew philosophical arguments about what's "really productive."

On reflection, though, the standard approach is anything but agnostic.  Official stats tacitly make an extreme assumption: waste does not exist.  Astrology counts, even though astrologers can't predict the future.  Every penny of health care counts - regardless of its efficacy.  The whole defense budget counts - even if it's provoking war rather than deterring it.  Indeed, if two countries' militaries mutually annihilate, both countries count the cost as a benefit.

Question: Suppose you had the chance to redefine the GDP formula - to create the real measure of real GDP.  How precisely would you change the formula - and why?  Your answer should contain the equation: real real GDP = something.

Extra credit: Create and share a graph that shows how your real real GDP measure compares to official real GDP over the last fifty years.

Comments and Sharing

COMMENTS (32 to date)
Nickolaus writes:

Could it be based off of savings, rather than the more Keynesian spending model?

Like, RealRealGDP = Amount Earned - Amount Spent

Brian writes:

Foundational axioms of economics include:

-Preferences are ordinal, not cardinal.
-Utilities are subjective and individual; they cannot be compared or aggregated between people.

So the enterprise of computing a personal measure of happiness as a linear quantity is impossible and hopelessly confused. The combination across a society of such a measure is doubly impossible.

The thing GDP purports to be to the public, a measure of nationwide production, is not a valid concept in any way. Sector economists add up what we call GDP as a political measure. It is made up of an aggregation of the opinions of government economists about what constitutes consumption and what constitutes value.

Now an aggregation of expert opinions can be valuable but reflects personal biases and cultural biases of the profession as much as it reflets reality. In the case of GDP it reflects the bias of politicians lobbying and spreading favors to get the numbers to reflect their preferred outcomes also.

The biases are expressed in the detailed segregation of consumption and intermediate goods; in the attribution of value to purchases, inventories, forward contracts, credit terms, long term subscriptions, insurance, theft and shrinkage, and risks of default; in the treatment of government spending and criminal activity; in accounting for research, experimentation, and investment; in the treatment of monopolies and non-market prices; in the pricing of exports and imports and internal corporate accounting transfers of governments and corporations; and many other fuzzy and flexible matters of opinion that cannot be removed from the supposedly well-defined measure.

But it is wrong to call those things biases because there is no "true GDP" to bias. There is no underlying reality of value to measure because interpersonal value comparisons have no meaning and personal value is strictly ordinal. The measurement of GDP is nothing at all beyond an aggregation of personal feelings of experts so even in the ideal case there is nothing there at all beyond biases.

The use of GDP also depends on price level comparisons and since there is no such thing as a single price level in an economy with more than one good, the idea of "real GDP" is also nothing but personal feelings. We construct many different CPI or PPP or GDP deflator price indicies according to an aggregation of feelings and opinions from experts so as to make them somewhat useful but that doesn't make them real, true, or objective. It just makes them somewhat consistent.

So there is no possible correct answer to the challenge. "Real GDP," "real real GDP," "true GDP," and the like are inherently self-contradictory concepts. You can construct an alternative measure but it cannot be more valid or less valid than the existing measures because standards of validity for a contradiction are impossible.

It's best to embrace the understanding that economic aggregates are just aggregates of personal feelings of economists and accept that that alone can still give them some value.

Djinn writes: do (__person)
__person.net_worth_at( -

John David Galt writes:

Real Real GDP = GDP excluding all government spending.

This represents the goods and services that each recipient actually wants enough to spend his/her/its own money on, voluntarily.

Real Real GDP would presumably be close to (say 80-90% of) GDP during peacetime before about 1880. After that, the percentage would steadily decline. I'd guess it to be approaching 25% now.

Bill Woolsey writes:

GDP doesn't measure welfare. It measures output.

To think about defense, think about opportunity cost. What is the value of the consumer goods that were sacrificed to produce the national defense.

As for intermediate goods, they are counted. All the wheat is counted, until it is used up in producing flour.

It is a mistake to frame the procedure as looking at all the goods, and then throwing out the ones that are deemed intermediate.

All goods are counted--unless and until they are being counted twice. You can't count the wheat in the bread when it is wheat and flour and bread.

Fralupo writes:
On reflection, though, the standard approach is anything but agnostic. Official stats tacitly make an extreme assumption: waste does not exist. Astrology counts, even though astrologers can't predict the future. Every penny of health care counts - regardless of its efficacy. The whole defense budget counts - even if it's provoking war rather than deterring it. Indeed, if two countries' militaries mutually annihilate, both countries count the cost as a benefit.

I'm struggling to interpret this in a way that doesn't say that you know better the value of goods traded in the market than the market participants themselves. If people want to buy $X worth of astrology services, who are you to say its worth less than $X?

Ghengis Khak writes:

RRGDP = Average Standard of Living * Population Count.

Seriously, these other measures are garbage. How can you take seriously some measure that includes EPA spending as part of some sort of national well-being metric but excludes the benefit that people get from things like the internet?

Kit writes:

I'll play the Sir John James Cowperthwaite card: just don't collect the bloody statistics in the first place. Economists and politicians will just have to find something else to fixate on.

Lori writes:

Before real real GDP can be calculated, real real transparency must be implemented. Every journal entry of every business concern must be entered into the public record.

Julie Novak writes:

I'd endorse John David Galt's suggestion.

Indeed, the question as to whether or not to include governmental activities in a quantitative measure of national income/output was a highly controversial topic in the early days of the modern national income accounting literature.

An excellent summation of, and contribution to, that debate, and which remains compelling to this day, was that provided by Forte and Buchanan in The Journal of Political Economy (1961). These debates seem to have been largely forgotten now, but ought to be revived if one wishes to ensure that economic indicators don't mislead.

One should also be mindful of the role that a growing post-war acceptance of hydraulic Keynesianism played in the fashioning of national income accounts as we know them today, with the inclusion of government activities (at cost).

Conversely, modern national accounts tend to give weight to spurious Keynesian arguments. This has enabled, for example, the Australian government to claim that its $A80 billion fiscal stimulus "saved" the economy, pointing to GDP growth during 2008-09, even though the majority of this spending (e.g., on roof insulation subsidies and overpriced school halls) surely detracted from the capacity of the economy to generate value.

Richard Allan writes:

Real Real GDP = Aggregate Social Utility. I'm afraid I have to disagree with the previous commenters and espouse the unfashionable view that utility is essentially cardinal. I'll agree that it cannot be measured, however. The closest you can get is to offer people citizenship in another country in exchange for time spent in some kind of pain machine to assess by how much they would prefer it.

The reason that intermediate goods are excluded from GDP is because they don't contribute to aggregate social utility. If we could get bread without going through the flour stage then we would, so money spent on flour is a pure social cost. "But what if we could sate hunger without going through the bread phase?" Don't ask.

Greg writes:

GDP just shows the output of a country, not whether that output is useful in any way!!!

GDP was originally used to calculate how much output the Nazis were producing and whether our bombing campaign was having an effect.

it is not that GDP is a bad measure it is just that you totally FAIL to understand what it measures and what use it has!!!

Shayne Cook writes:

How about ...

Returns to Capital
Returns to Labor
(change in Public Debt + Private Debt)

Dave writes:

C + I + (G/5)

Mrs. Davis writes:

[Comment removed for supplying false email address. Email the to request restoring this comment. A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

John Hall writes:

I don't see a need to re-define it. Real GDP does not equal living standards or value or welfare. It equals what's spent on goods over a certain period. It need not make any value judgement on why or what that money is getting spent on.

Tim Worstall writes:

Going off slightly at a tangent.

I don't want real real GDP to exclude government. I very much want it to include government in fact.

But I want it to include it in the correct manner.

Private sector GDP is valued at market prices. ie, at the value that consumers place upon it.

Government is not so valued as a GDP component. Government is valued at what is spent on it. NOT, I repeat not, at the utility derived from that spending, not at the value that consumers place upon the goods and services that government delivers.

And what I'd like to see is government spending being valued the same way that private provision is. At the value consumers place upon it, not the value that producers do.

Of course, this is nakedly idological: it would enable us to see where government adds value (I would certainly argue that a criminal justice system is worth more than is spent on it currently) and where it subtracts it (politician's salaries for example).

Dave Tufte writes:

I'd like to see two systems of accounts — one tracking flow variables as GDP now does and one tracking stock variables like national wealth. Essentially an income statement and balance sheet for the country.

A lot of the nonsense we have about government policies and quasi-governmental public goods provision would be seen a lot more clearly if we went in this direction.

For example, borrowing to fund social programs would increase income but not the asset side of the balance sheet ledger. Diverting some spending to build a bridge to nowhere wouldn't change our income statement much, but would make national "owner's equity" drop. Building a bridge to somewhere would make that equity go up.

OneEyedMan writes:

I'd like to see a core NGDP analog not to replace NGDP but to look at along side.

I'd exclude transactions where end use is not at market prices. Notably, G, Housing Services, changes in inventories (at least those produced and not purchased), in-house R&D, and in kind transfers from government. I want to see the core not because those things have no value, but because they are hard to value.

Becky Hargrove writes:

I am okay with GDP for what it does: tracks monetary flows throughout the economy. The problem is not with GDP but with the necessity of money to define optimal productivity in human capital utilization to a single standard for global competition. Increasingly, any deviation from the standard means further exclusion from the system, which sooner or later breaks the system unless an alternative means of measurement is used simultaneously.

Monetary systems can be stabilized by measuring the human labor capacity outside monetary production requirements (except for the human labor utilized within money). By long term knowledge based matching capacity on an infinite scale (person to person) additional wealth can be created that has the capacity to go far beyond what GDP creates, in that monetary GDP is limited by the necessity of adhering to a single optimal standard.

Silas Barta writes:
GDP just shows the output of a country, not whether that output is useful in any way!!!

Useless output isn't.

I want to second the votes for excluding government expenditures. Even if you aren't ideologically anti-government, you should recognize that the purpose of government expenditures is to augment the productivity of the rest of the economy, and it is the easiest to directly manipulate (and thus least informative about real conditions).

Also, I would want a RRGDP measure to gauge off-market (including household) production, so that you don't see a spurious "gain in productivity" when someone goes to buying something rather than making/doing it on their own, or at least don't count the full cost as part of the productivity gain.

Though I admit I don't know a straightforward way to do this.

P.S.H. writes:

Rather than exclude government spending outright, I think I would prefer to report GDP figures in the form of GDP = [GDP - G] + [G]. Outside of a few limited areas, I can't think of a sensible way to determine the "waste factor" for G, so it seems better to isolate it.

P.S.H. writes:

Sorry, I botched the wording on that last comment. Make that:

Rather than exclude government spending outright, I think I would prefer to report GDP figures in the form of GDP = [GDP - G] + [G]. Much of government spending is wasteful, but much of it is not. Outside of a few limited areas, I can't think of a sensible way to determine the "waste factor" for G, so it seems better to just isolate the G portion.

aaron writes:

RRGDP=GDP-increase in government debt

joeftansey writes:

RRGDP = sum(Cost Good Last Year/Cost Same Good This Year) + sum(Cost New Goods for this year), where the sums are over all final consumption products.

It doesn't matter that this is narrow and might ignore some consumption goods that are also intermediate goods. It is explicitly *qualitative*, as GDP only can be.

I won't draw a graph, but RRGDP is significantly higher than RGDP.

Steven H. Noble writes:

How about
National wages + National dividends received
That is payments received by residents of a nation not related to delayed consumption. This is not quite right because some dividends will be from corporate profits that came from interest payments.

I couldn't get wolfram alpha to come up with US dividend payments (but those numbers must be out there). However, this is probably pretty close to the relationship

ikitov writes:

There are several ways to introduce a quantitative variable, including real GDP. All of them have to match some basic requirements.

1. Compatibility through time. The current version of GDP has a break in 1978 when the CPI and GDP deflator started to deviate. In that sense one must not use the time series published by the BEA as it is -

2. To be useful, a definition of real GDP should not be "true". When the same portion of the "true" real GDP is measured one can use it for quantitative analysis. In linear equations, it's just the matter of coefficient. For example, the accuracy of Okun's law does not depend on the portion of the measured real GDP in the true real GDP -

3. A sound definiton of real GDP has to be delineated together with other macrovariables. The above example of Okun's law demonstrates that one measured macrovariable (unemployment)is correlated at the level of Rsq=0.95 with another one (real GDP per capita) in developed countries. Therefore, one can consider their definitions as consistent.

It would be counterproductive to introduce any measured variable by voting for its components. One should develop a sound procedure to test any definition before introduce it.

Liberal Roman writes:

These discussions are stupid. I agree with the couple of commenters who chimed in and said the obvious: RGDP is just a measure of output. Plain and simple.

The composition of the output and whether or not we should be using our resources (land, labor & capitol) to produce said output is a totally different debate.

However, even by having this argument, we are falling into the Keynesian trap.

Maurizio writes:

how about:

Any voluntary transaction should add to the GDP.

Any involuntary transaction (tax) should SUBTRACT from the GDP.

And any rule which prevents a free exchange should subtract from the GDP.

ChrisA writes:

RRGDP = Annual Consumption by end consumer less annual increase in debt of end consumers, both depreciated by inflation (hedonically adjusted).

This is the only real measure of how an economy is really doing - is it really increasing the living standards over time? We need debt in there to deal with the concern that consumption over short periods can be driven by increased debt. I am not entirely sure about debt, since some of that is offset by capital elsewhere, but in the long run that would show up in consumption data. I think we can ignore corporate and government debt, they are not owed by any individual person, and if the results are an increase in taxation, that shows up in the consumption data as well.

John Hawkins writes:

I think the entire concept of GDP should be eschewed and replaced simply with multiple measures. I would replace it with the multiple measures, all of which measure higher and more volatile numbers

From smallest and least volatile to largest and most volatile:

1. Consumption (Stable)
2. National Income (semi-stable)
3. GDP (nix it, for the most part. The following two are fully capable of capturing economic health for the most part without all the distortions that governments easily make for GDP)
4. Gross Domestic Expenditures (as conceived of by Mark Skousen, essentially a measure of the area of the Hayekian Triangle) (This falls hard in a recession, a la Hayek's Prices & Production)
5. Net Asset Turnover (the trading of a good that is not changed, like a stock price or real estate, essentially what Georgians call "land" plus a few other things, measured by the PQ of all exchanges) (This falls hardest in depressions, a la Fisherian Debt Deflation)

Measuring these measures not only by themselves is enlightening, but also in proportion to themselves. Measure Gross Domestic Expenditures ÷ Consumption, we would see the number rise over time, indicating more and more "roundabout" methods of production.

If Net Asset Turnover ÷ Gross Domestic Expenditures has started growing parabolically, we may start being concerned that our economy has become overly financialized (more involved with secondary markets and asset transfers than with value-adding projects. This may be because market signaling and price making becomes more and more important as society evolves or it may be unhealthy, I am not sure, but my bias is towards thinking it's unhealthy) and look at policies that may be promoting it, such as interest rates that are too low.

Other proportions would be enlightening for other reasons

T M Colon writes:

I am not an economist or anything like. I don't know exactly how GDP is now calculated. Based on what I think I know, I offer the following.

I suggest these four changes would make GDP more accurate:

1. Resale of used goods should subtracted.
2. Government should be treated as equivalent to a corporation.
3. Taxes should be considered equivalent to consumption.
4. Transfer payments should be subtracted.

The new GDP formula:

Consumption - Resale + Taxes - Transfers + Investment + Exports - Imports = GDP

My reasoning.


What is essentially different economically about government compared to a business corporation? Consider the following description:

An organization buying capital equipment and paying employees to provide products and services which the recipients pay for with money.

Does that describe a corporation or a government? Actually, it describes both.

A private school buys capital equipment and pays employees to provide a school and education to customers paid for with tuition. A government school buys capital equipment and pays employees to provide a school and education to citizens paid for with taxes.

As they function the same economically, there need be no separate category for government. Governments can be included and treated in the calculation as another corporation exchanging goods and services for money. What counts as investment for business counts as investment for government, and so on.

As government spending equates to business spending, taxes paying for government goods and services are equivalent to purchases buying business goods and services. The economic effects are the same. Taxes should be considered equivalent to consumption for GDP.

A private school collects tuition in exchange for an education. A government school collects taxes in exchange for an education.

However, governments give away a lot of money for nothing in return, businesses not so much. If the government sends someone a welfare check, the government gets nothing in return, nothing was produced. This spending is a transfer of money from one entity to another which does not equate to production.

Funding bailouts, welfare, social security, unemployment benefits, and the like are transfers of money from taxpayer to government to recipient which don't produce anything. The final recipient of the transfer might add to GDP by spending the money, but this is already added to GDP under consumption or investment. To also add it as government spending is double-counting.


If you buy a house built 100 years ago how does that boost current GDP when the production happened 100 years ago? How many times can you claim the same house was produced?

Imagine an island where ten families each bought a new house 40 years ago. This contributed ten houses worth to GDP. These houses are all now fully paid for. Then one year all ten families buy the house next door. Everyone moves one house over. Ten houses were produced 40 years ago. No houses were produced from the resales. How does that add to GDP?

The people moved from house to house, there was no other work done. No houses where built, no labor or materials were used, there was no increase in housing. Goods changed hands but there was no production from the resale. The same applies for a used car, used furniture, etc.

A used house can be in better or worse condition than when built. In which case you're not selling the exact same house. Still, the improvements to the house were already added to GDP under consumption when the improvements were done. You can't add them again when the improved house is resold.

As GDP is supposed to count production, resale of used goods doesn't add to GDP. A particular good can only be produced once, it is not produced again when resold.

Anyway, that's what I think. Thanks for the venue to let me propose it.

Comments for this entry have been closed
Return to top