Counting on numbers

OECD Observer

Click to enlarge.

Who would question the value of numbers? They can add credibility to an argument, clinch a deal, or simply illuminate an issue. But they can also deceive, through misreading or even manipulation. Can we really rely on the statistics we read? In the build-up to a major international forum on “Statistics, Knowledge and Policy”, to be held 10-13 November in Palermo, Italy, we asked Enrico Giovannini, the OECD's chief statistician, for an “unofficial” view.

OECD Observer: The public and the media in Europe often complain about official data, for instance, that prices appear to be rising faster than government figures show. How would you explain this gap between public perceptions and official data?

Enrico Giovannini: This is an old story. Ten years ago, the Boskin Report showed that the consumer price index (CPI) in the US was overestimating inflation, and urged statisticians to improve their methods. Now, consumers living in euro-area countries believe that today’s inflation figures underestimate reality.

If you compare the “official” inflation with the so-called “perceived” inflation rate measured by opinion surveys, you will see two well-aligned curves until 2002, when the euro came in. But after that, the curve for perceived inflation rises more steeply than the official one. Now, is it possible that a system based on collecting hundreds of thousands of individual prices for each country and that is scrutinised by the IMF, Eurostat and the OECD could fail, while the experience of shoppers buying what is probably a limited range of goods and services, and probably bought in one city, provide a truer picture? My answer is no.

If we exclude the existence of an international organisation conspiracy, the only possible explanation for the difference regarding the euro is that consumers must confuse changes in prices between different goods with the average level of prices generally. In other words, just because a weekly shopping basket of food seems more costly does not mean there has been a rise in the overall price level, which may include furniture, computers and so on. With the introduction of the euro, some prices, especially of some frequently bought items, rose more sharply than others. This made it more difficult to assign correct values to the new notes and coins.

According to some news reports, households faced unusual liquidity constraints before the end of each month and naturally they blame inflation. But how many of them accept that they might simply have spent more than usual because of the switchover? It is not easy to understand prices in a new currency, and overspending is a risk.

If there was collusion, it was in the media, which delighted in exaggerating the incidence of “local” price jumps. In herd-like fashion, this added further price pressure, even among those producers who had avoided raising their prices in the first place.

But could the experts be getting it wrong? For instance, couldn't a price index place more weight on bread and banking charges, and less on the falling cost of technology?

EG: For a start, if anyone dislikes unreliable statistics, it’s the statisticians themselves. It’s in our nature to want to improve them. After the criticisms made by the Boskin Report regarding the US price index, all countries began to update the weight structure of their indexes. In 1998, when I was Director for Economic Statistics at the Italian Statistical Institute, we decided to update them yearly, for instance.

On the other hand, there are some differences in household spending, depending on income level, sociodemographic conditions, etc, but they are not large enough to explain wide differences in overall inflation rates. In some countries, consumer associations have estimated a “true” inflation at 3 or 4 times higher than the official rate. But there is no way of obtaining such results using reasonable weight structures. Also, what would be the aim of measuring separate inflation rates for 10 or 20 categories of households? Such an approach could be useful to analyse the impact of the euro on income distribution, but they cannot be used separately to disprove the overall inflation rate.

But the real clinching point is this: let’s imagine actual inflation had been three or four times higher than the officially published rate. In that case, households’ disposable income in real terms would have plummeted, hurting consumption, employment, and so on. Euro-area countries might have tipped into a serious recession! This did not happen. Overall, it seems that official figures are broadly correct.

You are hosting a major international forum on statistics, to be held in Palermo, Italy in November. What do you expect this conference to achieve?

EG: We see the forum as a unique opportunity to bring statistics to the centre of national policy debates, by adding an international perspective. We live in an information-rich world, and there is a real risk of confusion. Do we really know whether society as a whole is improving or declining? My hope is that by identifying some key indicators to assess country performances, taking into account economic, social and environmental aspects, we can find clear answers. And we can then use those indicators to build a true knowledge base for better action.

The selection of any credible “key indicators” must involve society broadly – academics, NGOs, business and the media, and so on. Statistical offices will gather the data and communicate them regularly. Technical and policy questions, best practices, useful initiatives: all these issues will feature at the Forum. We expect results. After all, more than 150 world-class speakers from 30 countries and 12 international organisations will be present. As George Washington said in 1790: “Knowledge is in every country the surest basis of public happiness”. Reliable statistics can help us build such knowledge. This is what the Forum is about.


Giovannini, Enrico (2003), "Damned lies and statistics: Helping numbers make sense", in OECD Observer No 237, May 2004. Available here

©OECD Observer, No 244 September 2004

Economic data

GDP growth: -1.8% Q1 2020/Q4 2019
Consumer price inflation: 0.9% Apr 2020 annual
Trade (G20): -4.3% exp, -3.9% imp, Q1 2020/Q4 2019
Unemployment: 8.4% Apr 2020
Last update: 9 July 2020

OECD Observer Newsletter

Stay up-to-date with the latest news from the OECD by signing up for our e-newsletter :

Twitter feed

Digital Editions

Don't miss

Most Popular Articles

NOTE: All signed articles in the OECD Observer express the opinions of the authors
and do not necessarily represent the official views of OECD member countries.

All rights reserved. OECD 2020