Updated: March 30, 2022
America’s GDP could be 51% bigger than China’s, or alternatively China could be 24% larger than America, according to World Bank data.
Price data are crucially important. They are used to measure changes in the cost of living for government pensions and many other services; to assess real economic growth; to make inter-country comparisons of national income; to estimate real value- added by economic sector to estimate productivity, and for many other serious tasks.
Price data are central to many key functions of economic activity and public policy, but there are very good reasons for believing it’s a near impossible task to produce good price data.
This paper discusses seven major issues that undermine the reliability of official price indexes.
Is China the largest economy in the world or is the United States? According to the World Bank in 2019 GDP at current prices of the United States was US$21.433 trillion, 49% larger than China at US$14.280 trillion. However, when measured in Purchasing Power Parity (PPP) terms, (again in 2019 and according to the World Bank) the GDP in current prices of China was US$23.488 trillion, 9.8% larger than the United States at US$21.433 trillion.
The distortions in measured size and rank by applying different price indexes are not confined to the relative sizes of China and the United States as the Table below shows for the largest five economies.
Country GDP comparison require the use of price data. The simplest method is to use the prices of domestic currencies in terms of a standard currency – usually the US$ - based on average exchange rates over a period. But currencies appreciate and depreciate over time so international comparisons over time are more usefully made by removing such distortions and using detailed Purchasing Power Parity (PPP) price indexes which take account of differences in the relative domestic prices of a comparable basket of goods across countries.
The need for price indexes to adjust for these fluctuations in exchange rates is important, but the sheer scale of the changes that occur from the application of the price indexes produced by the United Nations and World Bank’s International Comparison Program (ICP) which calculates PPP price indexes, cast doubt on the reliability of the comparisons.
For example, comparing GDP in 2019 at current market prices with GDP measured via PPP rates, nearly doubles the size of Chinese GDP while inflating the Indian economy by nearly 4 times.
These large differences call into question the validity of all price indexes used to adjust economic data. In theory, Purchasing Power Parity prices should reflect real purchasing power and hence be closer to reality than US Dollar market price based data. In practice, it’s difficult to have much faith in either measurement. And hence the GDP data, based on these Price Indexes.
Calculating price indexes to adjust current to constant price economic activity in order to measure growth in consumers expenditure requires data on prices. However, a rapidly growing and important part of many developed economies consists of the services of internet based companies such as Google and Facebook ( now among the very biggest companies in the world in market capitalisation terms) which provide services at no charge to one set of consumers, subsidized by another set or by advertisers. This makes it impossible to define the value or the price of the product and consequently to measure the real output, economic growth and productivity relating to these services.
Furthermore, apps which are free to download, blogs, online new sources, open source content and volunteer digital services, such as contributing to Wikipedia or educational content on YouTube, are all outside the ‘production boundary’ that defines GDP from non-market human activity, but all these instances substitute from paid for services such as proprietary software, classified newspaper advertising ,publications or educational services. The growth in the digital economy, therefore, has an impact on value added economic activity not picked up by price surveys, biasing price indexes towards only calculating inflation in directly measurable output. This must underestimate the growth in real output, given the large size of this un-measurable sector of economic activity.
Technological progress and the introduction of new products reduce the reliability of price indexes. In order to use a price index to deflate current values into constant prices, price surveys require that the nature of the goods remains stable. Technical progress, new products and fashion changes means that price indexes based on this assumption become increasingly meaningless over time. In an often quoted example William Nordhaus showed how technical progress in one industry – the provision of lighting – have been so significant that traditional price indexes vastly overstate the increase in lighting prices over the last two centuries with consequent underestimates of the true rise in living standards. He concludes that “the growth in the frontier volume of lighting has been underestimated by a factor of between nine hundred and sixteen hundred since the beginning of the industrial age.”
The unreliability of price indexes used to deflate values in current market prices is not confined to such long run changes in the real output of a product. A computer laptop or a mobile phone bought currently may be similar in price to one bought two years ago but will much improved technical features in terms of speed, memory, screen resolution, camera etc. A smart phone of today , with the power to produce in seconds a vast array of music, human knowledge, detailed maps of almost any corner of the earth ,and a thousand other utilities, cannot be compared at all to a telephone of yesterday. Statisticians make adjustments to calculate quality-adjusted prices (or ‘hedonic prices’) for some items, but measured price indexes used by national statistics offices are not generally based on these adjustments. Errors in the calculation of price indexes used to deflate current GDP to constant GDP values can have serious distorting effects in sectors where technical change is rapid such as in telecoms. For example, between 2010 and 2015, for example, data usage in the UK expanded by around 900% but real measured output in the industry fell by 4%. According to a recent paper the problem lies in using an incorrect price index to adjust current GDP to constant GDP value added. The authors conclude that the “current deflator is upward biased and that telecommunications services prices could have fallen between 35% and 90% between 2010 and 2015.”
In summary, accurate and meaningful price indices are virtually impossible to calculate for many new services.
There are measurement problems with the price indexes used to measure inflation and changes in living standards of living. In particular, central banks employ measurements of a basket of consumer prices – Consumer Price Indexes (CPI) to target inflation for the purposes of monetary policy, but these indexes generally fail to adequately include measures of asset price inflation and notably, for the cost of housing. This produces not only inaccurate price indexes, but it has serious social and political consequences.
Over the last decade, since the financial crisis, measured inflation in terms of the costs of goods has been very low. However, adding in estimates for asset price inflation over this period for housing, to estimates of consumer price inflation produces a significantly higher level of real price inflation faced by an average consumer in countries like the UK and the USA where there are high levels of owner occupation. Daniel Alpert of Cornell University believes that over the last decade the United States’ Federal Reserve has “failed to appreciate the economic feedback loops that join past and present interest rate policy to housing inflation, and the fundamental differences between inflation in interest rate-dependent capital goods, on the one hand, and that of other consumer goods and services (and wage rates), on the other.”
Given that housing costs can amount to 50% or more of the after tax disposable income of many income groups, it is clear that price indexes which fail to take account of asset price inflation are unlikely to represent anything close to reality.
Modern marketing techniques undermine the meaning of using standard price indexes to measure inflation and to adjust current into constant values. The calculation of consumer price indexes rely on regular in depth price surveys across a region in a time period which take average prices as representative of the prices paid by shoppers. This is no longer a valid assumption. The growth of online shopping has created a so-called Amazon effect which is driving wedges between the actual prices paid by many consumers and the prices displayed in store. One United States study demonstrates that internet shopping with algorithmic pricing technologies and the transparency of the Internet “can also change the pricing behaviour of large retailers and affect aggregate inflation dynamics. “
A recent in-depth analysis of actual price data in the United Kingdom found large degrees of price dispersion within product categories casting serious doubt on the use of average prices derived from price surveys to construct price indexes. Richard Davies analysed a dataset of 32 million prices of a wide range of consumer goods including food and drink, homewares, furniture and appliances, motoring supplies and fuels as well as a range of services recorded between January 1996 and December 2018. These are the consumer products that are used in the calculation of the CPI, but Davies found a large dispersion of prices within items and by product. For example, taking vodka shots and pushchairs (strollers) as two examples. For vodka shots measured inflation at the 10th, 50th and 90th deciles of the price distribution was 114%, 138% and 159%. However for pushchairs Inflation at the 90th deciles was 138%, while at the median (over 20 years) it is just 20%, and at the 10th percentile prices fell. Price dispersion can seriously distort measurements of aggregate inflation, again casting doubt on the accuracy of price indexes.
The United Kingdom publishes two main price indexes – the Retail Price Index (RPI) and the Consumer Price Index (CPI). They are calculated using different methods and provide significantly different estimates of underlying inflation. This has led to ‘index-shopping’ by the government which uses the CPI in order to reduce payments on index linked items such as welfare payments and pensions, while using the RPI to appease powerful economic groups such as financial institutions when indexing government bonds. According to Sturgess: “The RPI is not only inaccurate as a measure of price inflation, but imposes costs on different segments of society.” The problem is even worse in the developing world where many countries in Asia and Africa employ out of date and poorly measured indices. India, for example, uses four price indexes to measure inflation.
Finally, in some countries price data has been deliberately manipulated by politicians by putting pressure on national statistics offices to publish inaccurate price indexes in order to flatter GDP growth in constant prices or to reduce measured inflation in the cost of living. In a paper for World Economics, Christopher Balding using published data from the Chinese National Bureau of Statistics found that Chinese inflation measurements were understated as a result of manipulation of the weighting of urban to rural private housing. Instead he estimated that between 2000 and 2011 the annual CPI in China should be adjusted upwards by approximately 1% reducing real Chinese GDP by 8–12% or by more than US$1 trillion in PPP terms.
The publication of false inflation figures to underplay falling living standards has been used in many countries, particularly in Argentina and Venezuela. In the latter country an expert on hyperinflation, Steve Hanke, found that under pressure from the government, the Central Bank of Venezuela (BCV) first stopped reporting inflation data, and then restarted publishing massively distorted price index data. In a paper published in World Economics he writes: “instead of reporting Venezuela’s real ‘open’ inflation rate, the BCV has attempted to measure supressed inflation.”
The measurement of price, and price change, is extremely difficult, even for well financed national statistics offices in developed countries. The inadequacy and inaccuracy of price data means that most officially published price Indexes need to be treated with the greatest caution.
In addition to the inherent difficulties involved, price indexes are sometimes deliberately manipulated by Governments either to flatter reported GDP growth in constant prices or to supress published inflation. Since international organisations such as the World Bank that publish international GDP data rely on official country statistics offices for their data, any intercountry GDP comparison or attempt at measuring GDP growth or absolute size is very likely to be inherently flawed.