How “macro” should national income accounting be?
St. Louis University – Madrid Campus
It is trivial to say that any cardinal number can be added to another to answer the question of “how many” or “how much” of something there is. The trick lies in teasing some relevance out of the resultant sum. The relevant tradeoff in constructing any aggregate is between the advantage of reduced complexity of a unique figure summarizing data and the disadvantage of the information lost from any individual figure within the set.
This realization was what Friedrich Hayek had in mind as he reviewed John Maynard Keynes´s 1930 Treatise on Money. Disappointed with the turn that he saw economics taking, Hayek lamented that “Mr. Keynes’ aggregates conceal the most fundamental mechanisms of change” (1931: 277).
There is an enduring appeal to macroeconomic theorizing. The urgency of the problem of 2.8 million unemployed Americans (5.8 percent of the workforce) seems to be a more important problem than, say, your neighbor losing his job. Likewise, the fact that Canadian per capita income was $52,000 in 2013 seems a more useful statistic than the salary of, e.g., Mike, from Canmore. Aggregates are appealing. They´re also dangerous when misapplied.
Although the misapplication of macro-aggregates is well-known, there is a deep-rooted commitment amongst the economics profession to accepting them at face value, cognizant of their limitations. In this vein, I´ve (2013) written about how the velocity of money commonly discussed is one such misapplication. The benefit of the equation of exchange is the simple relation between money and expenditures. What the common reckoning of money’s velocity loses sight of is that not all expenditures are paid for with money – purchases on current account and initial funding provided by fractional-reserve bank created money are two such examples. It’s not incorrect to say that velocity is the ratio between nominal expenditure (PY) and some monetary aggregate (M), but what this velocity means (if anything) is the key to understanding its implications.
The rate of unemployment provides a similar example. It may be that 5.8 percent of Americans who want a job don´t have one, but for any one individual there are only two relevant unemployment rates. You either have a job or don´t creating own-unemployment rates of either 0 or 100 percent. Instead of discussing an unemployment “rate”, perhaps we should be discussing the unfortunate phenomenon in terms of the rate of job finding, as I´ve suggested in Howden (2014). This, too, is an aggregate, though one that hits a little closer to home for any one unemployed individual.
It´s not that we need to rid ourselves of the baggage of macro-statistics completely. Rather, we need to 1) critically assess whether they are appropriate, if so, 2) for what role, and finally 3) whether they are consistent as a tool we can use to tell the story we want.
Consider the national income accounts. That gross domestic product is analogous with the aggregate income of an economy is something every first year economics student learns. GDP certainly appears to be important, even to those who think money doesn´t matter. The author of the textbook I use in my Principles class, Greg Mankiw, reconciles this paradox by noting that
[a] a large GDP does in fact help us to lead a good life. GDP does not measure the health of our children, but nations with larger GDP can afford better health care for their children. GDP does not measure the quality of their education, but nations with larger GDP can afford better educational systems. GDP does not measure the beauty of our poetry, but nations with larger GDP can afford to teach more of their citizens to read and to enjoy poetry… In short, GDP does not directly measure those things that make life worthwhile, but it does measure our ability to obtain the inputs into a worthwhile life. (Mankiw and Taylor 2010: 485)
Incomes themselves and aggregate income in particular, are useful to the extent that they allow one to keep “score” of their monetary contribution to society. More to the point, they let you see how this contribution varies over time. Quite literally, they tell you when you are faced with a change of fortune.
Surely any objection to aggregates cannot be solely on the basis of losing sight of the trees for the forest. Businessmen score their success or failure over the previous year by aggregating sales and costs to determine net income. Even vulgar unit-based expressions, such as sales per customer, are useful. This usefulness is not in spite of, but because of the fact that no one customer will be the average. In a former life I found myself as a clerk in a clothing store. Total sales and profit concerned me little, but I could depend on the average customer to spend $x and earn me $y in commissions. I motivated myself accordingly.
Aggregates are much more useful than the seeming insignificance of my own pay. During the socialist calculation debate, Mises (1985: 71-72) made clear that it was “[o]nly because the prices of all goods and services in the market can be expressed in terms of money [that] it is possible for them, in spite of their heterogeneity, to enter into a calculation involving homogeneous units of measurement.” And, as Joe Salerno reminds us along a similar vein, it is only through the use of homogeneous units, themselves created and made possible through monetary aggregation, that economic calculation can be successful (2010: 469). Some form of aggregation is the building block of the rational allocations provided for on the market.
Austrian economists are reluctant to use aggregate income measures, such as GDP, not because they are necessarily incorrectly constructed, but because they are not useful for the task at hand. The scope of transactions taking place at any one time is much broader than those focused only on the payment of “final goods and services”, as GDP focuses on. While Hayek’s critique of the aggregate might be the most famous, Skousen (1990; 2010) has provided the most useable alternative (one which many Austrian economists strongly support) in his gross domestic expenditure (GDE) figures. (In fact, through his efforts the U.S. Bureau of Economic Analysis now publishes quarterly data on the figure dating back to 2005Q1 under the guise of “gross output”.) Skousen’s GDE includes all monetary transactions, and generally shows that intermediary goods (excluded in standard GDP calculations) represent about double the value one obtains by focusing exclusively on final transactions. (In 2013, American GDE totaled $29.7 tr. versus $16.8 tr. for GDP.)
GDE might be a step in the right direction, but it is unclear what direction that is. David Colander (2014) views it as merely another statistic to confuse students, though that is no reason to exclude it as a tool of useful analysis. (Many things confuse students, as well as economists.)
Two criticisms of GDE spring to mind, both with Austrian-type origins. In the early days of the development of an aggregate income statistic there was vibrant debate as to whether government services should be included. On the one hand, it is questionable how much value the government portion of expenditures actually represents. After all, “the government´s tax revenue and deficit revenue are both burdens imposed on production, and the nature of this burden should be recognized. Since government activities are more likely to be depredations up, rather than contributions to, production, it is more accurate to make the opposite assumption: namely, that government contributes nothing to the national product and its activities sap the national product into unproductive services” (Rothbard 1962: 1293). Such a criticism may be valid in some instances, but just because an expenditure is inefficient or at the expense of other productive enterprise cannot be a reason to exclude it. (Otherwise we could count an innumerable amount of private expenditures that would satisfy both criteria.) In a similar vein, one could exclude government services since they are not transacted at market prices whether because they are paid through taxes, or because they are sold through monopolized services (Coyle 2014: 29-30). On the other hand, government services usually represent intermediary goods that would otherwise be excluded. In any case, Austrian economists should be wary of rejecting one measure because it includes government services (GDP) while embracing another with a similar inclusion (GDE).
From a technical point of view, the number of monetary transactions in an economy (the monetary sum of which GDE measures) is itself a product of the structure of businesses. Larger businesses, conglomerations, and the result of mergers and acquisitions all bring an increased emphasis on intra-firm transactions at the expense of those taking place between firms. Hayek alluded to this in his discussion of the difficulties of varying the money supply to meet the needs of the economy, because it is not entirely clear what such needs would be. At one extreme, one large firm (or a completely socialized economy) will, per definition, have no need for any extra-firm transactions (or money to conduct such transactions in). For the same amount of output, one firm per production process would lead to the maximum amount of extra-firm transactions. Business size, and the total number of businesses, matter for the construction of GDE.
In a sense, income matters in a way that the sum of expenditures does not. I am not sure whether any individual considers the total amount of purchases or sales he makes over the course of a year, though he certainly will often consider the amount of income he earns.
What needs to be done is a reconstruction of aggregate income statistics in a more meaningful way. I will suggest three such figures.
The first disaggregates income between that which is privately procured and that which is through government activities. Rothbard’s (1963: 296) use of “gross private product” (GPP) is useful. Effectively GPP subtracts government expenditures from the calculation of GDP. Normalizing GPP by allocating it among the private workers who create it gives an idea of what is the average income created by a privately employed worker. (Alternatively, since the national income accounts are determined by expenditure classes this figure will amount to the average cost of each private worker.)
Just as Rothbard’s GPP/private worker is helpful in teasing out the rise and fall of the private sector’s fortunes, taking the total of government expenditures and normalizing it for the number of public employees (G/public worker) gives one a reckoning of how well the government’s fortunes are faring. Again, since the national income accounts are able to treat income and expenditures more or less as interchangeable variables, this figure will tell us how expensive each public worker is, or how lucrative it is to be employed as one.
Finally, it is instructive to compare these two figures to the average income earned per worker, both public and private. In distinction to the more common GDP/capita figure, this statistic uses only those employed as the denominator (instead of the whole population). The result is a figure that shows the average income generating contribution per worker, or the average expense incurred by the employment of each employee. Figure 1 gives a snapshot of these figures since 1960.
Figure 1: Aggregate income figures for the United States, 1960-2014
Source: Federal Reserve Bank of St. Louis
I will conclude with a brief interpretation of these aggregate income figures. Since 1960 the average income generated by a private worker has grown at an annual rate of 4.57%. This is slightly lower than the average per total workers (public and private) at 4.77% and far lower than the average for public employees at 5.20%. In some sense the growth in private income has lagged that of the greater average, and has performed far worse than its public counterpart. Not only that, but the volatility in the growth of private income has also been far more pronounced than in the public sector. Recessions are far more damaging to private workers. Between 2008Q3 and 2009Q2 the average private worker saw his income reduced by more than 5%. (The average public worker, in contrast, saw his income contribution increase by nearly 9% over the same period.)
There is one alternative way to look at the above figures. Instead of treating them as the income generated by each worker, one can think of them as the expense created by each worker. Under this interpretation one can point to one of two facts. The first is that the average public employee is nearly three times as expensive as the average private worker. This way of interpreting the data is not so telling, as it could be that the government specializes in services that are inherently more expensive to produce (e.g., national defense, law and justice enforcement, healthcare, etc.). Alternatively, and of more relevance, one could point to the fact that the average growth rate in expenditures per public employee has risen much more rapidly than that of private workers. More pointedly, this divergence has grown especially wide since the year 2000.
Colander, David. 2014. Gross Output: A Revolutionary Way to Confuse Students about Measuring the Economy. Eastern Economic Journal 40: 451-55.
Coyle, Diane. 2014. GDP: A Brief but Affectionate History. Princeton and Oxford: Princeton University Press.
Hayek, Friedrich A. 1931. Reflections on the Pure Theory of Money of Mr. J. M. Keynes. Economica 33: 270-95.
Howden, David. 2014. Measuring Unemployment, Take Two. Mises Canada daily article, June 21st. [Available] http://mises.ca/posts/articles/measuring-unemployment-take-two/
Howden, David. 2013. The Quantity Theory of Money. Journal of Prices & Markets 1(1): 17-30.
Mankiw, N. Gregory, and Mark Taylor. 2010. Economics, 2nd edition. Andover, UK: South-Western.
Mises, Ludwig von. 1985. Liberalism in the Classical Tradition, 3rd ed. (trans.) Ralph Raico. Irving-on-Hudson, NY: Foundation for Economic Education.
Rothbard, Murray N.  2009. Man, Economy and State with Power and Market. Auburn, AL: Ludwig von Mises Institute.
Rothbard, Murray N. 1963. America´s Great Depression. Kansas City: Sheed and Ward.
Salerno, Joseph T. 2010. Money, Sound and Unsound. Auburn, AL: Ludwig von Mises Institute.
Skousen, Mark. 1990. The Structure of Production. New York: New York University Pres.
Skousen, Mark. 2010. Gross Domestic Expenditures (GDE): the Need for a New National Aggregate Statistic. (Economics Working Paper no. 113). Centre for Comparative Economics, SSEES, UCL: London, UK.
 Excluding marginally attached or over-qualified workers. Even the binary own-unemployment rates of 0 and 1 are suspect if considered relative to whether one wants a job or not. Somehow the own-unemployment rate of 1 in the case where I want a job cannot be much different than my similar un-ownership rate of a Ferrari of 1, though I doubt many economists place much importance on the second figure.
 The classical statistician in me would relish a poor sales week for the implications it had for the next week´s commissions.
 One could imagine what the Austrian response would be to a surging GDE figure attributed to a surge in government expenditures in the face of a drop in private transactions.
 In all three figures cited, GPP/private worker, G/public worker, and GDP/worker it is not strictly accurate to equate the resultant figure with the average cost or income generated by the relevant worker. There are many sources of income (e.g., interest, dividends, etc.) not related to wages, and likewise many expenses (e.g., capital costs, interest payments, etc.) not related to wages. In all cases the figure given should not be taken as synonymous with what the average worker incurs, but rather what the relevant total cost /income associated with that worker is.