Gross domestic product (GDP) per capita is used by economists and politicians as the most significant indicator of a country’s average income and consequently as measure of well-being. Real GDP per capita is a measurement of the total economic output of a country divided by the number of people and adjusted for inflation. It is used by international organisations such as the World Bank to make international comparisons of the standard of living between countries. It is used within countries to measure progress in the income of its inhabitants over time and to estimate the extent of regional inequalities. Unfortunately, it is a seriously flawed statistic for a number of reasons, some well documented others not, which will be summarised below. Notwithstanding these conventional criticisms World Economics research shows that the data on which it is based is inaccurate and therefore unreliable for comparative purposes. In this paper we development an index to assess the reliability of published official GDP per capita data.
Since the first attempts by the United Nations to measure GDP in 1947, countries have pursued the policy of raising GDP per capita as a means of improving the quality of life. The theory behind the obsession with the ratio is that rising real incomes, a flow variable, allow consumers to achieve higher levels of satisfaction from being able to consume a greater volume of goods and services without reducing the stock of wealth. Rising satisfaction means that rising real incomes enhance economic welfare.
Unfortunately, real GDP per capita is a defective proxy measure for real income, per head and as such is only loosely correlated with economic welfare. Numerous studies have been conducted to attempt to more directly measure welfare and most have given results very different from the story provided by rising GDP per capita. The results of one study, for example, show that although developed countries have demonstrated high rates of growth in GDP per capita over the past 50 years, the reported happiness of citizens in developed countries has stagnated1. This has also led to a number of international attempts to replace GDP per capita with alternative measures of economic welfare.2
Furthermore, GDP per capita says nothing about the distribution of income. Given that an average measure of the resources generated from economic activities per person does not show how these resources are distributed, which is one of the three main economic questions, it would be better for economists interested in measuring income to put more focus on median household income rather GDP per capita. The growth or stagnation in median household income tells us more about how the typical person is doing better and through the public choice median voter theorem would be a batter predictor of policy trends. Unfortunately, medians are harder to calculate than per-person averages since median household income can only be calculated through in-depth household surveys that many countries do not conduct regularly. In the countries that do conduct surveys they show a different picture of income growth than GDP per capita data. In the United States, for example, although GDP rose significantly from 1999-2008, most individuals saw a decrease in income over the same period when adjusted for inflation.3
There are, however, much more fundamental flaws with the concept of real GDP per capita which affects its reliability even if it were a useful proxy for income and economic welfare. The economic indicator consists of three concepts which require measurement: GDP at current market prices, or the market value of goods and services produced in a given period such as a year; GDP in Constant prices, which is Current GDP adjusted for inflation; and the population of a country in the same period. Unfortunately, all are subject to measurement problems which can produce serious biases in published estimates.
The accuracy of estimates of a country’s GDP at current market prices differ enormously between countries and over time which makes many inter-country GDP comparisons of dubious validity. Comparisons between France and the USA are feasibly made since the statistical offices of both countries adhere to the best practices of national income accounting and allocate significant resources to the process, but comparisons between countries such as China and Chad have little validity.
There are sound reasons why the quality of GDP at current market price statistics differ between countries. First, despite that fact that the United Nations recommends that countries update their base years every five years in many cases the base year used to estimate the GDP are seriously out of date. The base year reflects the structure of the economy in a particular year from which GDP is estimated at current prices, but since the balance between sectors changes rapidly over time, particularly in emerging markets, a failure to update a base year can completely omit growing industries such as telecommunications and finance. Updating an out-of-date base year often leads to large upgrades in GDP estimates.
Second, GDP is based on recorded economic activities that are priced. In many economies, informal activities are not measured and even in the developed much of the output of digital industries such as Google or Facebook are free. Data from the IMF estimates that accounting for informal activities would lead to significant uplifts in GDP. In addition, many countries do not apply the same vintage of the United Nations’ recommended national income accounting which specifies what economic activities to include in GDP and how to measure them. However, underlying all of these measurement issues are the resources, training and manpower that national statistical offices use for national income accounting purposes. The resources available in many emerging markets are sparse compared to those applied to estimate GDP among the OECD economies.
The second concept involved in the calculation of GDP per capita is real GDP, which is GDP adjusted for inflation. This requires the calculation of a GDP deflator and the IMF’s regular country reports on country’s statistical output shows large discrepancies between countries in the quality of their price indexes. Finally, reported statistics are often manipulated by governments to flatter real GDP growth. Articles in World Economics on Argentina and China have demonstrated how real GDP can be overestimated by underestimating inflation.
In recognition of these data problems, World Economics has devised a Data Quality Rating (DQR) which assesses countries according to a series of weighted variables and ranks countries on an index from 0 to 100. For example, New Zealand is ranked with a DQR index value of 99.5 compared to 79.5 for China, 65.0 for India and 58.4 for the Cote d’Ivoire. Countries are then ranked from A to D based on the underlying reliability of their data for cross-country comparative purposes.
The third concept needed to calculate real GDP per capita is population, the denominator of the GDP per capita ratio and here again there are serious differences across the world in the accuracy of population estimates. The most reliable means of calculating population is by a regular census and in between census years population estimates are based on migration statistics, the number of reported deaths and live births. The accuracy of the results depends again on the statistical capacity of the country undertaking it and many countries do not undertake regular censuses. In order to assess the reliability of population estimates World Economics has produced a Population Quality Index whose value depends on the length of time between censuses. The index varies from 0 for the Lebanon which has not conducted a population census for decades to 100 for Azerbaijan whose last census was conducted in 2019.
GDP per capita data is, therefore, seriously flawed since it is subject to three sources of error: mistakes in GDP at current market price calculations, incorrect use of price indexes to estimate real GDP and errors in estimates of population size. Therefore, how much trust should be placed upon a comparison of GDP per capita data between countries. Research by World Economics has shown that data quality issues in general, apart from government manipulation, tend to bias official GDP data estimates downwards, the same direction as population estimates, but does this mean that since numerator and denominator are biased in the same directions, the ratios should minimize the bias. Unfortunately, not since data errors are multiplicative not additive so a better means of testing the reliability of a country’s GDP per capita data would be some weighted multiplicand of the two estimates of data quality.
Since there we have no reliable way of comparing the relative quality of the estimates of GDP with population, a GDP per capita index has been calculated for each country giving both the DQR and PQR equal weights. The GDP per capita index is shown in the chart below for a sample of countries, but the full dataset can be downloaded which shows that GDP per capita comparisons between Canada, Korea and Singapore are relatively reliable while comparisons between India, Nigeria and Chad are not.