We publish below the second part of the article The trouble with GDP published by the magazine "The Economist", which discusses possible and eventual new ways of calculating GDP more in line with an economy that has moved far away from a scenario in where it was the manufacturing sector that drove the economy and consequently the growth and prosperity of a nation. The conclusion of the London weekly is rather disheartening and is entrusted to the cold and even cynical humor that characterizes this publication which is also one of the most important global think-tanks.
Attempts to update GDP
Despite what has been said, there are many things included in GDP that fall outside the market economy. Many public services are provided at no cost to the consumer and for decades the value attributed to these resources in GDP has been the price of the provision. It is only recently that statisticians have begun to directly measure some parts of public sector services, by counting, for example, the number of interventions performed by the health service or the number of students attending schools.
Some areas of the private sector are also measured indirectly. Real estate is one of them. This occurs regardless of whether the owners rent or memo the property in which they live. Leases measure the value transferred by tenants as well as the income of the landlords who make the properties available. But in cases where the owner himself lives in the property, a large part of the real estate value must be imputed.
Finance is another activity that is mostly measured obliquely (and poorly). Typically, financial services are not paid for directly by the consumer: banks obtain a large part of their revenues by charging more interest on loans than they pay on deposits. To capture the added value, statisticians use to enter a figure, the "spread", ie the gap between a risk-free rate and the effective rate of the loan: then they multiply this value by the number of loans granted. The problem with this measurement is that the loan "spread" measures the risk that the bank bears. For example, at the end of 2009, the financial sector in Great Britain was on the verge of collapse. But as fears of bank failures were sending spreads skyrocketing, GDP figures recorded a spike in value added by the sector to national GDP.
As statisticians try to capture ever new forms of economic production in their models, new activities are continually added to GDP. In 2013, a European-level agreement on the standardization of GDP included the value deriving from the sale of soft drugs and that produced by sex workers. In Britain the change added 0,7% to GDP. How much credibility one should give to these numbers is up for debate. Statisticians have to resort to pretty trivial and crude phenomena to figure out what's going on: it is assumed that the commercial sex market can expand in line with the size of the male population and the price of admission to dance clubs belly is taken as a parameter of the price of sexual performance.
Leaving aside the appropriateness of these approximations, Paul Samuelson might be prompted to ruminate on the GDP implications of marrying a woman to her gigolo. Robert Kennedy might have wondered whether a nation is really doing better when its drug and sex trade is as good as it is thriving.
The puzzle of adjusting the price correctly
A further complication is that, despite all the statisticians' recommendations not to take GDP as a measure of well-being, the two are intertwined in the most deceptive of ways, namely through inflation adjustment calculations. Inflation measures the amount of money you have to pay more than in the previous year to reach the same level of income. It's really hard to measure it as an output.
First of all, a change in the price of a product will affect how much consumers can buy. If the price of red apples goes up, people will buy more green apples: if the price of beef goes up, people will buy pork. There are ways to capture this kind of substitution when measuring prices. One is the geometric mean aggregation of price quotes. By adding the price of "n" goods and then taking the nth root of the product, we obtain the aggregation from which to derive the degree of proportional switching of the variation in relative prices. It sounds bizarre and it is: but if you do this correctly, you have the effect of lowering inflation by half a point or similar. More extensive changes in consumer preferences are detected by updating the weight of each category of goods in the general price index.
Then, there are the adjustments related to changes in product quality. The latest model of smartphone may cost more than the year before, but if so, it must be better. If statisticians focus only on nominal price changes, they may overestimate the inflation rate and miss improvements in performance. An advisory committee of prominent economists, set up by the US Senate in the mid-0,6s and chaired by Michael Boskin of Stanford University, estimated that failure to adjust the quality of new products meant that real inflation was overstated by at least XNUMX .XNUMX%.
This adjustment also requires greater use of "hedonic" estimation, a technique that captures the implied value of any particular product attribute by assessing how much each change in this attribute affects the price of the product: for example, how much a consumer pays more for a more efficient light bulb? Once the implied price of each attribute is established—computing speed or memory, say, a phone—the prices are adjusted accordingly.
Hedonic evaluation
Hedonic evaluation helps. But this is labor intensive and demanding because the implicit values must be updated often to achieve any accuracy; at the end of the day only a small fraction of the prices are adjusted in this way. Furthermore, many problems arise when the quantitative aspect expands to the point of becoming qualitative. A modern flat-screen television is an entirely different "beast" from the small pot-bellied CRT television set of the XNUMXs.
Such adjustments are even more difficult to make for services, which tend to be increasingly personalized while goods, for the most part, are still standardised. The value of a dinner, for example, depends on the cuisine and the ingredients, but also on the speed of the service, the noisiness of the dining room, the distance between the tables and so on. Each of these factors can change over time.
The true value of public sector services is even more difficult to calculate over time. The number of health interventions can be counted quarter by quarter. Their effect on the patient's health and longevity cannot be appreciated until years or decades later.
As the Boskin commission has shown, new products are truly a puzzle. In theory, their value for the consumer is given by the difference between the reservation price (that is, the price that consumers are ready to pay) and the actual price; this difference is known as consumer surplus. It happens that new products enter the consumer price index without such an adjustment.
Then there are the novelties to expand the range of new products. For example, the number of TV channels or over-the-counter pain relievers in America is enormous. In 1970, five each were considered. Although people complain that there are too many, this huge variety is a big plus. But it remains completely invisible to the measurement of GDP. For GDP, the output of one million shoes of one size and color is the same as that of one million shoes of different sizes and colors.
The benefit of so many new products is not simply collected by GDP. The initial costs of digital service platforms, such as Facebook and Twitter, are exorbitant. But the marginal cost is close to zero and the consumer price is usually non-existent. By global convention, zero-priced goods are excluded from the GDP count. As are all voluntary forms of production such as Wikipedia and open-source programs. Some of this free activity is included in the count; While there is no cost for a Google search, consumers pay a hidden price for providing information and attention, which advertisers buy. But the revenues from advertising fall far short of the benefits consumers get.
New estimation types: Usage time and Internet traffic
A survey conducted by Sir Charles Bean has outlined two possible approaches to evaluating digital services. One is to estimate the value of time spent on the Internet. The Bureau of Economic Analysis, the main American statistical institution, used the market wage level to estimate the value of household activities such as cooking, cleaning and ironing. Taking a similar approach Erik Brynjolfsson and Joo Hee Oh of MIT have estimated that the prosperity benefit of free Internet products reached 0,74% on an annual basis to US GDP between 2007 and 2011 (other studies have proposed a lower estimate, for example 0,3%).
The other approach uses Internet traffic growth as a benchmark. The Sir Charles Bean survey mentions research which found internet traffic in Western Europe to grow by 35% year on year from 2006 to 2014. If the output of the IT sector grew by the same proportion, the official GDP of the UK is expected to be 0,7% higher for each year over the period. However, it does not happen that all the services are provided free of charge; certainly are some that used to be paid for such as long distance and international phone calls. Some physical products have become digital services whose value is difficult to track. It has happened, for example, that more and more music is listened to, but the record industry's revenues have shrunk by a third from its peak in the pre-internet era. Consumers bought city maps, street maps and newspapers. They paid an agency to book their holidays. Now they are alone, an activity that does not pass into the GDP.
As commerce migrates online, less and less is spent in physical stores, which again translates into less GDP. Just as rebuilding after an earthquake (which pushes GDP higher) doesn't make people richer than before, opening fewer shops than before doesn't make people poorer.
These problems do not affect the use of GDP. But given the direction of technological change in an increasingly digital world, these problems are becoming increasingly serious and their solutions are becoming increasingly complicated and approximate. Measuring consumer surplus from new products or free products rests on bold assumptions; estimates vary widely from those that have been used previously. Being consistent over time requires the ability to measure consumer surplus of goods and services that are well defined in the consumer basket. The problem is that the consumer changes tastes and reference products more and more quickly.
Measuring a revolution
An understanding of the difficulty of the task can be had by looking at the estimates of the growth of the economy in another period of impetuous technological change - the industrial revolution.
GDP is primarily used to measure contemporary economies, but some economic historians have ventured to apply it to the past as well, concluding that there was a sudden take-off in economic growth after 1750; a landmark postwar study estimated that GDP per worker grew 1,4% annually in the first half of the XNUMXth century, an unprecedented rate.
In the 0,5s, research by Nicholas Crafts of the University of Warwick found that XNUMXth-century industry's surplus of transformative inventions was underestimated: the mad growth actually occurred in only a few sectors of the economy. Thus he lowers the value of productivity to a less than revolutionary XNUMX% per annum.
A later generation of Crafts colleagues, led by Steve Broadberry, published research that pushes the valuation even further down.
Even considering more recent times, it has been difficult to agree on GDP estimates in times of strong economic change. For example, the change in consumer surplus due, for example, to the development of the railways and related industry is not correctly considered.
One number for all purposes?
“It is a big mistake to think that one number serves all purposes,” says Sir Charles. The problem is that, that being the case, GDP risks serving all these purposes increasingly poorly. The Bank of England has become so cautious about GDP estimates that it publishes a range of numbers for both forecasts and time series. Its latest projection puts current GDP growth in Britain in the range of 0 to 4%. Such hyper-skepticism may seem a little silly. But isn't it more absurd to proclaim, with great assertiveness, that China's GDP fell from 6,8 to 6,7 percent in the first quarter of the year, when it's fairly certain that it didn't?
If the comparison of the GDP of one quarter with another is not advisable, that of 10 years with the previous 10 years is dangerous to say the least. The America's Census Bureau calculates that the median inflation-adjusted home yield in 2014 was just slightly higher than it was 25 years earlier. This means that the standard of living of a typical American has been stagnant for a quarter of a century. But for the average American citizen, has the cost of medical care really remained unchanged between 1989 (at 1989 prices) and today (at current prices), asks Ken Rogoff of Harvard University?
Whether GDP figures really measure what they are trying to measure is the question to ask and also the question to find a rational answer. The challenge, says Nordhaus in his paper on light, is to construct measurements that "account for vast changes in the quality and range of goods and services we consume." But that means finding ways to compare e-mail with the fax machine, the driverless car with the 1910 car, vinyl records with streaming services, and custom prosthetics with health insurance crutches. Maybe only Einstein could do it.
The likelihood is, though, that after taking a quick look, you'll immediately go back to a science as simple as physics.
