Things that grow

National accounts are a means for connecting public policy to empirical information. No other set of statistical data is referred to more often in political rhetoric. The most ubiquitous indicator in the national accounts, the Gross Domestic Product, usually holds center stage. The word growth has become practically synonymous with a positive change in GDP. A recent book published by the OECD states that

“The most important use made of national accounts is to forecast the following year in order to provide the macroeconomic framework for the government budget. The prime aim is to evaluate the volume growth in GDP for the following year (…).” (OECD p.308)


“All the main public finance indicators are expressed as percentages of GDP (…).” (OECD p.248)

Clearly then, the political importance of the GDP aggregate can hardly be exaggerated. However, I will argue in this essay that national accounting is a problematic enterprise. It is by no means clear what features of the economy the aggregates like GDP truly reflect.

GDP can be examined through definition on the one hand and through usage on the other. Surely, what we expect from an essential constituent of both the government budget and the public finance indicators is a clear and unambiguous definition. The statistical quantity in question should also be calculated and used in accordance with that definition. Unfortunately, GDP fulfills none of these conditions.

It is illuminating to begin with a comparison between the principles of national accounting and those of corporate accounting. In the OECD book the matter is stated succinctly:

“Despite their name, the national accounts bear only a partial similarity to the accounts of a company. The general frameworks are similar but the data sources are entirely different. The company accountant has at his disposal a ledger showing to the last cent all the transactions carried out by the firm during the period. The national accountant obviously has nothing similar for all agents, especially for households. [National accounting requires] acceptance of the notions of approximation, estimation and revision, things in which the national accountants excel but which are anathema for company accountants” (OECD p.34)

In corporate accounting the definitions are clear: company accountants keep track of economic transactions by adding up inflows and outflows. It’s also easy to see how the accounts are to be used: they provide the precise economic bottom line for the decisions the company makes. These decisions do not themselves dictate the account definitions.

But national accounting is not about keeping track of inflows and outflows at all. Instead it seeks to estimate a much more nebulous quantity, “production”. As we shall see below, this slippery concept quickly falls into a pit of vague definitions from which there is no escape. National accounting becomes a game of postulation, estimation and guesswork which indeed is anathema for people who think that statistics should reflect reality.

Why then cannot national accounts be run like company accounts, as summaries of import and export transactions? This is a question of usage. First of all, the state relies on taxation, and taxes are by definition a domestic affair. Fiscal policy in turn is dictated both by domestic and international events. So in order to make important economic decisions, policymakers need general information on the domestic economy. That’s what national accounts are for.

But it is worthwhile to reflect a little on how complex that information can be. For example, public funds are gained from taxing a variety of goods and services, income, capital, property, inheritance and so on, in different ways. Private consumption, on the other hand, consists of billions of transactions carried out by individuals and companies across all sectors of the economy. And the state also provides a whole spectrum of services for the benefit of the economy: banking, the legal system and infrastructure, to mention just a few. So clearly the economy is an exceedingly complex and heterogeneous system.

In describing a heterogeneous system, one can resort to two different methods. The first one is to construct an overarching theoretical edifice for describing the whole system on an abstract level. The second is to subdivide the system into subsystems and sub-subsystems before a theoretical description is attempted at a lower level of abstraction.

For the benefit of political decision-making, national accountants have gone the former way, the way of overarching theory. In this case usage has clearly dictated definition, not the other way. That’s why the accounts define “production” as an umbrella concept which supposedly applies to the whole economy. What we have here is a clear case of deliberate simplification intended for providing expedient information. Validity is a secondary concern.

Production – private and public

Let us then look more closely at production and GDP to assess their validity. The OECD report states that there are three different approaches to GDP: the output approach (GDP1 = the sum of gross values added), the expenditure approach (GDP2 = consumption + investment + net exports) and the income approach (GDP3 = compensation of employees + gross operating surplus + gross mixed income) (OECD p.29).

National accountants postulate that GDP1=GDP2=GDP3. It is motivated by the following idea:

“Any macroeconomist carries in his head a simplified model of the economy in which everything made by someone else is used by someone else, anything exported by someone is imported by someone else, anything saved by someone is invested by someone else, and so on.” (OECD p.266)

So when national accountants calculate GDP, the idea is first of all that the value created in a given product or service equals its selling price less the cost of production (GDP1). The sum of gross values added thereby becomes the total value produced in the domestic economy.

Secondly, in an economic transaction a product or service becomes the object of consumption (when it is used and discarded), investment (when it is partly or wholly saved for future use) or export – there is no fourth alternative (GDP2). And thirdly, since one man’s expenditure is another man’s income, the profits of economic transactions end up either in the pockets of employees or with the unit itself (GDP3).

In all of these approaches to GDP, production is measured as monetary value. If somebody manufactures an entirely useless product which cannot be sold to anybody, the work invested in it has produced negative value; it has consumed resources in vain. On the other hand, positive production produces value as profit.

The key assumption is that the products or services have been priced in a competitive market. This is the necessary condition for measuring value. When a company sells its production on the market, its profits provide a measure of the value the company has produced. Production in the private sector (excluding households) is therefore simple to calculate in the GDP1 approach. It’s just a matter of summing gross added values from the tax declarations which companies report to the state.

All products and services are comparable under the assumption of a free market, and so:

“Once products are expressed in monetary units, it becomes legitimate to add them together.” (OECD p.47)

I think this logic is valid as far as it goes, that is, as long as the assumption of a competitive market holds true. The aggregate value from the private sector is accurate because the corporate accounts are accurate measures of gross added value.

But when we include public sector “production” in GDP1, we are in trouble. We must express it in monetary units before it can be added to private sector production. But a significant number of products and services are of course produced in the public sector without any form of market. How could the value produced in the form of roads, healthcare, defense or education be estimated? The obvious answer is that it cannot be – comparing company profits to the profits of healthcare, for example, is pointless.

Nevertheless that’s essentially what national accountants are forced to do, fixated as they are on the idea of a national “product”. The method for including public sector production is deceptively simple:

“Whether individual or collective, as there are practically no sales, non-market output at current prices is conventionally measured as equal to the sum of its production costs (…).” (OECD p.105)

But this seems like a blatant fallacy. The mere cost of doing something certainly says nothing about the value of doing it. Good roads and bad roads can both be built either cheaply or expensively. The same holds for everything else produced in the public sector outside of the market. By what logic can it be legitimate to add the production cost of one product to the price of another? By none. I do not see how it can be theoretically justified. So what economic phenomenon does the aggregate production, GDP1, then represent after the private and public sectors have been lumped together in this manner? We’re beginning to see why that question is difficult to answer.

The logic of creative accounting

The fallacious logic outlined above can of course not be blamed on the accountants. It’s not their fault that they are forced to aggregate the prices of apples with the production costs of oranges. The problem is unsolvable because the theoretical model of “production” which they are forced to apply in their calculations is bad.

Consider the definitions of GDP1, GDP2 and GDP3 again. This is where we see the model at work. In many ways it resembles the idealizing postulates of pure economic theory on rational self-interest and so forth. First of all we have the producers of good things and the value they created by in their work (GDP1). These products are consumed by someone (GDP2) and the producers receive income (GDP3) in return.

Just like in pure economic theory, it can reasonably be argued that the model reflects some aspect of reality, but not all aspects. Unlike economic theoreticians, however, national accountants have to face the real world and are forced to creatively manipulate their data to arrive at the numbers which the model requires.

The methodological problems can perhaps be made clearer if we look a bit closer at GDP2 and GDP3. In principle, if “production” were really a scientific concept, it should be possible to calculate GDP1, GDP2 and GDP3 independently to verify the identity GDP1=GDP2=GDP3. But in practice this is far from being the case. Consumption in households, i.e. private persons, forms the majority of GDP2. But this quantity can of course not be monitored since private persons cannot be forced to keep detailed records of all their economic transactions. GDP2 is therefore “calculated indirectly by using statistics from other sources” (OECD p.35).

So what exactly constitutes an “indirect calculation”? To put it simply: the same data which yields GDP1, i.e. company tax declarations plus government records (p.34), is manipulated in alternative ways to create an estimate of household consumption. Similarly, GDP3 is not calculated from independent data sources either, but from essentially the same set of records.

But even though they have their roots in the same data sources, the calculated values for GDP1, GDP2 and GDP3 are never precisely the same.  This is hardly surprising, since the raw data is complex and each approach to GDP involves its own approximations and estimates. National accountants say that there is a “statistical discrepancy” between the three values. That’s why

“(…) those compiling [national accounts] are not necessarily anxious to have statistics on every single item in the supply-and-use tables.” (OECD p.267)

Why are they not anxious for a complete data set? Say for instance that we have GDP1= X1+X2+X3, GDP2=X4+X5+Y1 and GDP3=X6+X7+Y2 where the X’s are known and the Y’s unknown. Now instead of having to admit a statistical discrepancy, the statistician can just postulate GDP1=GDP2=GDP3, assign the corresponding values to Y1 and Y2 and be done with it.

It seems pertinent to add another illuminating quote in this context:

“[National accounts] are the result of combining a complex mix of data from many sources, many of which require adjustment to put them into a national accounts database and which are further adjusted to improve coherence, often using non-scientific methods.” (OECD p.34)

I’m guessing that the method of “let’s just assume that these values are equal, eliminate the unknowns and call it a day” belongs in the non-scientific bag of tricks.

But again, my intention is not to disparage the people who compile national accounts. If they begin with a bad model, they’re not going to arrive at an accurate account no matter how brilliant statisticians they are. And the model at hand is this: people produce value, consume value, and are paid accordingly. This is a model of commercial transactions.

This model just does not fit non-market output at all. Its “production” cannot reasonably be valued in monetary units (GDP1) and the postulated equality between expenditure (GDP2) and income (GDP3) just doesn’t make sense. Consumption and expenditure are not good words for describing how non-market output, say minted coins, waste management, banking, police services or international diplomacy, is used.

The essential problem is that the value of each public good or service has to be evaluated on its own terms by the people and institutions that make use of it. Even if such an evaluation, perhaps by surveying, would end up producing a number which expresses that value, it would still not be legitimate to compare the values of different services.

The simple fact is that such values are not in any way commensurable. They can perhaps be weighed against each other in political debate, but there is obviously never going to be an objective resolution to such debates. That’s why forcing non-market value into monetary units seems like a completely meaningless exercise.

Concluding thoughts

The question that now suggests itself is this: granted that the so-called production of the public sector cannot be compared with production in the private sector, and that the latter but not the former can be described by valid statistics, why not treat them separately by designating the private sector to be the target of precise accounting and evaluating government spending by some other method?

I think there are at least two pragmatic reasons why such an approach would be unsatisfactory. First of all, as a recent IMF Working Paper puts it:

“(…) most spending functions performed by government are also performed by the private sector. For example, the provision of services in health, education, social protection, and environmental protection (…) is shared between the public and private sectors. The extent to which the private or public sector supplies services in these areas varies considerably across countries (…)” (IMF p.4).

Secondly, some publicly owned corporations produce output which is sold at market price. And from another perspective:

“The borderline between “nonmarket” and “market” is far from distinct in many autonomous government agencies, which are authorized to own assets, incur liabilities, or engage in transactions in their own right, e.g., public hospitals, universities, research institutes.” (IMF p.4).

To all this I might also add that government is intimately involved with the private sector through monetary corporations such as the central bank, through direct contracting of private companies for public works and through the provision of subsidies for selected areas of private business.

So the state is intertwined with the private sector in a fundamental way, and the line between public and private is somewhat arbitrary and fluid. Changes in government policies shift that line and there is of course a constant exchange of value across that same line. That’s why it’s easy to understand why a unification of private and public under a single system of national accounting is such an an alluring idea.

But that doesn’t reduce the fundamental problem that I’ve discussed in this essay. Regardless of where private “production” shades into public, it is still illogical to think that the value of non-market goods and services can be added to monetary values from market transactions. The peculiar nature of the market is that it yields a number which expresses human preferences, however limited that number may be as a measure of human well-being. But non-market value is not in any way amenable to a similar mathematical treatment. So accountants have to resort to the strangest methods of data manipulation and freewheeling estimation to make up for the lack of theoretical justification.

Some might defend the GDP aggregate on the grounds that it is the best measure available and that its changes at least tell us something about changes in the general state of the economy. In a sense this means that GDP is a measure of something even though no-one can be quite clear about what that thing is – calling it “production” does not resolve the ambiguity.

The assumption behind this kind of thinking is that accounting practices remain constant. And indeed, a concern for constancy is repeatedly reiterated in the OECD book. If the yearly data is treated in the same manner every time, variations in GDP will be “real”. The corollary of this thought is that

“Changes [in GDP] are better known than absolute levels [and] international comparisons based on the levels (of GDP or other variables) are to be treated more cautiously than comparisons between variations in national aggregates.” (OECD p.316)

But frankly, this line of thought is bizarre. If a quantity isn’t really a measure of anything real, then changes in that quantity are not going to be any more real. Just because a meaningless value is calculated in the same way each time, that doesn’t mean that it will magically gain meaning as it changes.

That’s why I think we should be concerned about the GDP rhetoric we constantly hear from decision makers and reporters who are all too eager to throw caution to the wind when dealing with statistics. They don’t care how a value was calculated or what it really measures. In fact the esoteric origin of GDP probably just increases its rhetorical value because its meaning can be interpreted at will without having anyone interrupt and ask: what are you talking about?

In a world where information must be simplified to the point of absurdity before it can become an instrument of social policy, GDP is the icon of disguised ignorance. It is also a symptom of larger problems in representative democracy: that voters are expected to have an informed opinion on exceedingly complex matters and that representatives have to make complicated decisions in an overwhelmingly insecure environment which they struggle to comprehend. This state of affairs is not conducive for critical appraisals of political language.

So we are stuck with GDP for the foreseeable future and we will continue to see it paraded back and forth across the political stage as a vital and objective statistic, the fountainhead of a democratic and prosperous society. But as I’ve shown in this essay, the harsh reality is that it’s hard to find traces of valid reasoning behind this simplified statistical aggregate. So as long as they keep on emphasizing GDP and its growth above all else, I will assume that they don’t know what they’re doing.


Lequiller, Francois and Blades, Derek, 2006: Understanding National Accounts, OECD.

Lienert, Ian, 2009: Where does the public sector end and the private sector begin?, IMF Working Paper 09/122.