One of the most common questions I am asked about my research is “Why did we switch from GNP to GDP? What difference did that switch make?” In this post, I sketch a quick and partial answer. First, I define GNP and GDP. Second, I note the historical emergence of GNP. Third, drawing on McNeely’s (1995) UN data, I show that the switch from GNP to GDP occurred in the international context in the late 1960s. Fourth, I report on the moment in which the United States switched (the 1990s) and the stated justifications for so doing. Fifth, I note that there has been a recent move to reconsider the choice of principal aggregate for analysis, as evidenced by the switch to GNI (Gross National Income) in the newest iteration of the UN Human Development Indicator (HDI). Sixth, I present my own analysis of Penn World Tables data to show the geographic distribution of divergent trajectories of GNP and GDP – that is, those countries in which the difference between the two aggregates varies most. Last, I reflect on the significance of the geographical distribution of GNP-GDP discrepancies and, in particular, how it showcases the limitations of relying on any single measure of national income, welfare, or “size of the economy.”

1. Definitions

GNP stands for Gross National Product. “Gross” refers to the absence of estimates of capital consumption. “Net National Product” is equivalent to GNP except for the subtraction of an estimate of the value of capital used up in production. Thus, GNP reports the total market value of goods and services produced in a given period of time. NNP reports the same total, less the “cost” of used up capital equipment. Note that neither currently accounts for the depletion of natural resources, so neither is truly a measure of sustainable production, but NNP is slightly closer.

“Product” signifies that GNP measures the value of goods produced. GNP is theoretically equivalent to GNI – Gross National Income – but the two are measured differently. GNI records income received (e.g. paychecks and dividends and retained earnings) while GNP records income produced (e.g. sales). The two estimates differ slightly, and modern convention notes this difference in the national accounts as a “statistical discrepancy.” Recently, some economists have argued that (at least in the US) income-based measures are more reliable (revisions of each number tend to move the product estimate closer to the income estimate), and thus that we should focus on GNI or GDI instead of GNP or GDP.[1]

“National” refers to the set of actors whose production GNP measures. To amend the above definition, GNP reports the total market value of goods and services produced in a give period of time by the nationals of a particular nation. GNP thus counts production by citizens (and, more importantly, companies) of a nation even if this production occurs in another country.

“Domestic” aggregates differ from “National” aggregates by measuring production that takes place in a given region rather than by a given set of individuals. In other words, the 2011 US GDP measures the total value of goods and services produced in the United States in the year 2011. The 2011 US GNP measures the total value of goods and services produced by Americans and American companies in the year 2011. The growth of multinational corporations presents challenges for interpreting GNP and GDP, one example of which will be discussed below.

2. The Emergence of GNP

As discussed elsewhere, national income statistics date back to the 17th century work of William Petty. In the first half of the 20th century, organizations and governments took over from lone scholars as the producers of national income statistics, and their production increased dramatically. In the 1920s-1940s, countries produced a heterogeneous mix of national income statistics, mostly focused on National Income (which is conceptually most similar to Net National Income and Net National Product, though the exact differences varied from country to country and year to year). During World War II, national income statistics took on new significance as tools for wartime planning. In the United States, national income statisticians successfully used their data and analysis in order to produce a more “feasible” mobilization plan, over the objections of much of the military establishment (Lacey 2011). In the midst of these wartime efforts, economist Milton Gilbert (chief of the National Income Division in the Department of Commerce) argued that economists and statisticians should shift their focus from national income to Gross National Product. Given the immediate needs of all-out wartime mobilization, the total value of outputs in a given unit of time was seen as the most relevant question, and tricky issues of capital consumption were seen as largely irrelevant (Gilbert 1942, Gilbert and Bangs 1942).[2]

During and immediately following the war, Gilbert transitioned the official national income calculations to a focus on Gross National Product. In 1947, the US implemented the first version of the National Income and Product Accounts (NIPA) that, in line with similar developments in England and Canada, used a double entry accounting system (with an “income” side and an “expenditure” side, which were made to equal using the statistical discrepancy). The primary aggregate reported was GNP.

On the international scene, 1947 saw the publication of an influential memo commissioned by the League of Nations and written by Richard Stone, a mentee of Keynes and one of the economists responsible for the official English accounts. This memo laid out international guidelines for the production of national income statistics. In 1953, Stone produced a revised version of this memo for the United Nations which was published as the first UN System of National Accounts. The UNSNA, like the English and American systems, emphasized GNP.

3. The International Switch from GNP to GDP

In 1956, the UN began sending questionnaires requesting national accounts data (McNeely 1995). These first questionnaires asked countries for the GNP (both real and nominal), as well as a few subsidiary data points (such as government receipts and expenditures). In 1956, 57 countries (of 77 surveyed) reported GNP data to the UN. The 1968 update of the System of National Accounts, in addition to other changes, shifted the emphasis of the accounts to GDP over GNP. By 1970, the UN questionnaires (presumably following the shift in the SNA) abandoned GNP in favor of GDP as the principle aggregate of interest. In 1975, 108 countries reported GDP data to the UN (of 113 surveyed).

These changes in reporting standards affected the use of GNP and GDP. The following charts show shifts in the relative frequency of GNP vs. GDP in Google NGram’s sample of 4% of all books published, and JStor’s corpus of indexed academic journal articles. The graphs tell a consistent story – GNP rose to prominent usage in the 1950s, GDP entered usage in the 1960s, and eclipsed GNP starting in about 1990.

Figure 1: Relative frequency of books containing GNP and GDP in Google NGram’s corpus, 1940-1995. Note, not all hits refer to the macroeconomic statistics. Source: Google NGrams.

Figure 2: Absolute counts of articles containing GNP and GDP in JStor’s corpus, 1940-1995. Note, not all hits refer to the macroeconomic aggregates, but the bulk do occur in economics journals. Source: JStor Data for Researchers.

4. The US Switch to GDP

Although most countries switch to reporting GDP in the 1960s-1980s, the United States was a late adopter. In 1991, just as mentions of GDP were overtaking GNP in the academic literature (see figure 2), the US finally changed its reporting practices to emphasize GDP. An article by the Bureau of Economic Analysis (BEA) in the Survey of Current Business explained the rationale behind the switch:

GDP refers to production taking place in the United States. It is, therefore, the appropriate measure for much of the short-term monitoring and analysis of the U.S. economy. GDP is consistent in coverage with indicators such as employment, productivity, industry output, and investment in equipment and structures. In addition, the use of GDP facilitates comparisons of economic activity in the United States with that in other countries. GDP is the primary measure of production in the System of National Accounts, the set of international guidelines for economic accounting that the U.S. economic accounts will be moving toward in the mid-1990’s, and virtually all other countries have already adopted GDP as their primary measure of production. (BEA 1991: 8)

Notably, the BEA justifies the switch explicitly in terms of compliance with international norms, in addition to practical concerns about the coherence of GDP with other measurements of the American economy that focus on production in America, rather than by Americans. The BEA goes on to note, however, that GDP and GNP are nearly identical in the US – differing usually by less than 2% – but that the equivalence does not hold in all countries. I will discuss this last point more in part 6, below.

5. Recent Moves Back to “National” Aggregates

After their proliferation across the world in the mid-20th century, national income statistics have come to be used in a wide range of government, business and academic settings. One of the most controversial uses has been treating GNP or GDP as proxies for overall wellbeing. In the 1990s, the United Nations pioneered the Human Development Index (HDI), an alternative measure of development that combined measures of health (life expectancy at birth), education, and economic standard of living. The first Human Development Report defined the HDI and its purpose specifically against the growing reliance on national income measures:

The idea that social arrangements must be judged by the extent to which they promote “human good” goes back at least to Aristotle.
…
But excessive preoccupation with GNP growth and national income accounts has obscured that powerful perspective, supplanting a focus on ends by an obsession with merely the means. (UN 1990: 9)

Although the report itself referred to GNP at various points, the initial HDI calculations used Real GDP per capita as its primary measure of economic standard of living. In 2010, inspired by new research by development economists, the UN moved away from GDP and instead adopted Gross National Income (GNI) per capita as its primary measure of economic wellbeing. The 2010 Human Development Report explained the switch:

To measure the standard of living, gross national income (GNI) per capita replaces gross domestic product (GDP) per capita. In a globalized world differences are often large between the income of a country’s residents and its domestic production. Some of the income residents earn is sent abroad, some residents receive international remittances and some countries receive sizeable aid flows. For example, because of large remittances from abroad, GNI in the Philippines greatly exceeds GDP, and because of international aid, Timor-Leste’s GNI is many times domestic output. (UN 2010: 15)

It is too soon to know how wide-reaching the consequences of this transition will be for the reliance on GDP as the main aggregate discussed in popular sources and used in most academic studies.

6. Geographic Variance in GNP/GDP

Why does it matter whether a given study, report, or index uses GDP or GNP? As noted by the BEA, for the United States, as for many richer, developed nations, GDP and GNP are nearly equal in all years. As I show, however, the same is not true for all nations. The map below uses data from the Penn World Tables to examine which countries have a large – and variable – discrepancy between GNP and GDP.

To start, we’ll focus on GNP/GDP * 100, or the percentage ratio of GNP to GDP in a given year. For example, in 2006, the US GNP was 102% of GDP – just a bit higher. In 2009, on the other hand, GNP was just a bit lower – 99% of GDP. These differences are small, but not insignificant given the intense political scrutiny of the US national accounts – growth rates of 2% vs. 3% may have radically different public interpretations.

For a more extreme example, consider Ireland. Favorable tax laws combined with creative accounting tricks allowed foreign-owned companies to report large profits in Ireland in the 1980s, using a maneuver known as “the Double Irish.” [3] Additionally, Ireland attracted more foreign investment in the 1990s (partially due again to its favorable tax laws). In 1980, Ireland’s GNP/GDP ratio was 100% – the two were very nearly equal. By 2009, the GNP/GDP ratio had fallen to 81% – GDP captured much of this foreign investment, while GNP did not, and thus the two grew increasingly separated.

To create a rough measure of the possible distortions caused by focusing on just GDP (or just GNP), I created a variable that measures the range of the GNP/GDP ratio over all years covered by the Penn World Tables. So, for Ireland, the maximum GNP/GDP ratio was 106% (in 1960), and the minimum was 81% (in 2009), and thus the range is 25%. For the US, the range is just 3.8%. In figure 3, I display these results on the world map, with larger circles corresponding to a larger range.

Figure 3: Range of GNP/GDP ratio. The smallest circles correspond to ranges of 0-10%, the largest correspond to differences over 50%. An interactive version of this graphic is available here, as is the associated data.

Unsurprisingly, countries in Latin America, Africa and the Middle East have a lot of variation in their GNP/GDP ratios. Perhaps more surprisingly, several countries in Europe do as well – Ireland’s 25%, but also Iceland’s 15%, Switzerland’s 16%. In general, large, rich countries seem to have quite small ranges. For example, the US, Canada and the UK, the three countries most associated with modern national income statistics, have a very tight range. Smaller, rich countries sometimes have larger ranges, as do many poorer countries, large and small. Notably, some countries in Africa have extreme values – Angola’s is 46%, Lesotho’s is 84% – which may reflect problems in the underlying national income data as much as they reflect real shifts in foreign investment (see Jerven 2009 on contemporary problems with African national income data).

What does it mean? For researchers, the conclusion is reasonably obvious: the choice of GNP vs. GDP may produce substantively different results. Discrepancies are especially likely when examining rich and poor countries over time in the same analysis, although divergence between GNP and GDP occurs even in relatively prosperous countries. For policymaking, the conclusions are a bit less obvious: without specific knowledge of the question at hand, it’s hard to say whether GNP or GDP is “better.” All we can say is that they are different, and that they may follow different trajectories across time even within a single country.

7. On the Impossibility of Perfect Measures

Why do we quest for one aggregate to measure them all, one aggregate to rank them? Why, confronted with two different measures of two slightly different things, do we have to fight to determine which is the best to emphasize, and then push for conformity? The difference between GNP and GDP is itself an important economic fact about various countries – especially given its variability both within and across countries. And yet, because we so tightly associate national income statistics with the “size of the economy,” it becomes a pressing matter to determine a “true size” or best aggregate.

Simon Kuznets, the economist most associated with the development of national income statistics in the United States, warned against fetishizing a single measure. Instead, analysts interested in different aspects of economic productivity or welfare must rely on different measures:

The valuable capacity of the human mind to simplify a complex situation in a compact characterization becomes dangerous when not controlled in terms of definitely stated criteria. With quantitative measurements especially, the definiteness of the result suggests, often misleadingly, a precision and simplicity in the outlines of the object measured. Measurements of national income are subject to this type of illusion and resulting abuse, especially since they deal with matters that are the center of conflict of opposing social groups where the effectiveness of an argument is often contingent upon oversimplification.
…
A student of social affairs who is interested in the total productivity of the nation, including those efforts which, like housewives’ services, do not appear on the market, can therefore use our measures only with some qualifications. … A student of social affairs who conceives of a nation’s end-product as undistorted by the existing distribution of income, would again have to qualify and change our estimates, possibly in a marked fashion.
…
The abuses of national income estimates arise largely from a failure to take into account the precise definition of income and the methods of its evaluation which the estimator assumes in arriving at his final figures. (Kuznets 1934: 5-7)

Perhaps we should abandon this quest for a single best aggregate, as Kuznets wanted so many years ago. Admittedly, something is lost: focusing our attention on one measure is a heuristic device that speeds up discourse and discussion. There are, indeed, payoffs to selective attention. But there are costs as well, and I believe it’s long past time to weigh these costs and benefits, and when that weighing occurs, it may well come out in favor of a larger plurality of measures, not conformity with a single standard. The differences between GNP and GDP are small compared to some of the changes proposed by environmental economists or feminist economists (e.g. the inclusion of environmental pollution as a cost to GDP or the inclusion of the value of unpaid housework), but they are still significant, and offer a simple case to show how we have historically approached conflict within national income statistics by searching for the truest, best aggregate rather than allowing the flourishing of multiples. The question is not – and cannot be – whether GNP or GDP is a better measure, but rather, whether GNP or GDP (or GNI or HDI or …) is a better measure for some particular purpose.

Notes

[1] See, for example, Nalewaik (2010).
[2] I am eliding here other differences between National Income as measured in the 1930s and Gilbert’s GNP, including “factor cost” vs. “market price” and the treatment of indirect business taxes. See Carson 1975: 169-171 for details.
[3] For a description, see “How Apple Sidesteps Billions in Taxes,” The New York Times, 4/28/2012, available here.

Bibliography
Bureau of Economic Analysis. 1991. “Gross Domestic Product as a Measure of U.S. Production.” Survey of Current Business. August: 8.
Carson, Carol. 1975. “The History of the United States National Income and Product Accounts: The Development of an Analytical Tool.” Review of Income & Wealth 21(2):153–81.
Gilbert, M. 1942. “War Expenditures and National Production.” Survey of Current Business 22(3):9–16.
Gilbert, M., and R. B. Bangs. 1942. “Preliminary Estimates of Gross National Product, 1929–41.” SURVEY OF CURRENT BUSINESS (May) 9–13.
Jerven, Morten. 2009. “The relativity of poverty and income: How reliable are African economic statistics?” African Affairs.
Kuznets, Simon. 1934. National Income, 1929-32. U.S. Congress.
Lacey, Jim. 2011. Keep from All Thoughtful Men: How U.S. Economists Won World War II. Naval Institute Press.
McNeely, Connie L. 1995. Constructing the Nation-State: International Organization and Prescriptive Action. Greenwood Press.
Nalewaik, Jeremy 2010. “The Income- and Expenditure-Side Estimates of U.S. Output Growth [with Comments and Discussion].” Brookings Papers on Economic Activity 71–127.
United Nations. 1990. Human Development Report 1990.
United Nations. 2010. Human Development Report 2010.