国民经济核算体系

Criticism of GDP

The most popular criticism of national accounts is made against the concept of gross domestic product (GDP).

In part, this criticism of GDP is misplaced, because the fault is not so much with the concept itself. It is useful to have a measure of a country's total net output, and its changes over time – that's better than having no measure at all.

The fault is with the actual use that is made of the concept by governments, intellectuals, and businessmen in public discourse. GDP is used for all kinds of comparisons, but some of those comparisons are conceptually not very appropriate.

GDP measures are frequently abused by writers who neither understand what they mean, how they were produced, nor what they can be validly used for.

Economists like Joseph Stiglitz argue that a measure of "well-being" is needed to balance a measure of output growth.[7]

Feminist criticism

SNA has been criticised as biased by feminist economists such as Marilyn Waring[8] and Maria Mies[9] because no imputation for the monetary value of unpaid housework, or for unpaid voluntary labor is made in the accounts, even though the accounts do include the "imputed rental value of owner-occupied dwellings" (the market-rents which owner-occupiers would receive if they rented out the housing they occupy). This obscures the reality that market production depends to a large extent on non-market labour being performed.

However, such criticism raises several questions for the statisticians who would have to produce the data:

• whether an international standard method of imputation for the value of such services is feasible, given e.g. that the conditions under which the market equivalents for unpaid household services are supplied vary a great deal internationally ;
• whether making the imputation would result in truly meaningful, internationally comparable measures;
• whether attaching a price to voluntary labor, done primarily by women, itself actually performs an emancipatory or morally propitious function or has a general useful purpose beyond academia.

The intention of those who would like to produce this kind of standard data might be perfectly honorable, but the production of the data has to be practically justifiable in terms of technical feasibility and utility. Attaching an imaginary price to housework might not be the best data to have about housework.

In most OECD countries, statisticians have in recent years estimated the value of housework using data from time use surveys. The valuation principle often applied is that of how much a service would cost, if it was purchased at market rates, instead of being voluntarily supplied. Sometimes an "opportunity cost" method is also used: in this case, statisticians estimate how much women could earn in a paid job if they were not doing unpaid housework. Typically, the results suggest that the value of unpaid housework is close to about half the value of GDP.

Christine Lagarde, the head of the International Monetary Fund, claimed at the IMF World Bank annual meetings in Tokyo in October 2012 that women could rescue Japan's stagnating economy, if more of them took paid jobs instead of doing unpaid care work. A 2010 Goldman Sachs report had calculated that Japan's GDP would rise by 15 percent, if the participation of Japanese women in the paid labour force was increased from 60 percent to 80 percent, matching that of men.[10] The difficulty with this kind of argument is, that domestic and care work would still need to be done by someone, meaning women and men would need to share household responsibilities more equally, or rely on public- or private-sector provided child and eldercare. According to the ILO, there are over 52 million domestic workers in the world, who mostly work for little pay and with little legal protection.[11] They are mainly servants of the wealthy and the middle class.

Marxist criticism

Marxian economists have criticized SNA concepts also from a different theoretical perspective on the new value added or value product.[12] On this view, the distinctions drawn in SNA to define income from production and property income are rather capricious or eclectic, obscuring thereby the different components and sources of realised surplus value; the categories are said to be based on an inconsistent view of newly created value, conserved value, and transferred value (see also double counting). The result is that the true profit volume is underestimated in the accounts – since true profit income is larger than operating surplus – and workers' earnings are overestimated since the account shows the total labour costs to the employer rather than the "factor income" which workers actually get. If one is interested in what incomes people actually get, how much they own, or how much they borrow, national accounts often do not provide the required information.

Additionally, it is argued by Marxists that the SNA aggregate "compensation of employees" does not distinguish adequately between pre-tax and post-tax wage income, the income of higher corporate officers, and deferred income (employee and employer contributions to social insurance schemes of various kinds). "Compensation of employees" may also include the value of stock options received as income by corporate officers. Thus, it is argued, the accounts have to be substantially re-aggregated, to obtain a true picture of income generated and distributed in the economy. The problem there is that the detailed information to do it is often not made available, or is available only at a prohibitive cost.

US government statisticians admit frankly that "Unfortunately, the finance sector is one of the more poorly measured sectors in national accounts".[13] The oddity of this is, that the finance sector nowadays dominates international transactions, and strongly influences the developmental path of the world economy. So, it is precisely the leading sector in the world economy for which systematic, comprehensive, and comparable data are not available.

Statisticians' criticisms

Statisticians have also criticized the validity of international statistical comparisons using national accounts data, on the ground that in the real world, the estimates are rarely compiled in a uniform way – despite appearances to the contrary.

For example, Jochen Hartwig provides evidence to show that "the divergence in growth rates [of real GDP] between the U.S. and the EU since 1997 can be explained almost entirely in terms of changes to deflation methods that have been introduced in the U.S. after 1997, but not – or only to a very limited extent – in Europe".[14]

The "magic" of national accounts is that they provide an instant source of detailed international comparisons, but, critics argue, on closer inspection, the numbers are not really so comparable as they are made out to be. The effect is that all sorts of easy comparisons are tossed around by policy scientists which, if the technical story behind the numbers was told, would never be attempted because the comparisons are scientifically untenable (or at the very least rather dubious).

Both the strength and the weaknesses of national accounts are that they are based on an enormous variety of data sources. The strength consists in the fact that a lot of cross-checking between data sources and data sets can occur, to assess the credibility of the estimates. The weakness is that the sheer number of inferences made from different data sets used increases the possibility of data errors, and makes it more difficult to assess error margins.

The data quality has also often been criticized on the ground that what pretends to be "data" in reality often consists only of estimates extrapolated from mathematical models, not direct observations. These models are designed to predict what particular data values ought to be, based on sample data for "indicative trends". One can, for example, observe that if variables X, Y, and Z go up, then variable P will go up as well, in a specific proportionality. In that case, one may not need to survey P or its components directly, it is sufficient to get trend data for X, Y, and Z and feed them into a mathematical model which then predicts what the values for P will be at each interval of time.

Because statistical surveys are very costly or may be difficult to organize, or because the data has to be produced rapidly to meet a deadline, statisticians often try to find cheaper, quicker, and more efficient methods to produce the data, by means of inferences from data that they already have, or from selected data which they can get more easily.

But the objection to this approach - although it can sometimes be proved to provide accurate data successfully - is that there is a loss in data accuracy and data quality.

• The extrapolated estimates may lack any solid empirical basis, and the tendency is for fluctuations in the magnitudes of variables to be "smoothed out" by the estimation or interpolation procedure.
• Any unexpectedly large fluctuation in a variable is difficult to predict by a mathematical model since ultimately the model's descriptions assume the future trend will conform to the law of averages and the patterns of the past.
• Without adequate, comprehensive observational data from direct surveys, many of the statistical inferences made are simply not truly verifiable. All one can then say about the estimates is, that they are "probably fairly accurate, given previous and other concurrent data."

A typical reply of statisticians to this kind of objection is that although it is preferable to have comprehensive survey data available as a basis for estimation, and although data errors and inaccuracies do occur, it is possible to find techniques that keep the margins of error within acceptable bounds.

See also