Economists and statisticians have long understood that economic data is only of limited reliability. The average man on the street, on the other hand, almost always assumes the unemployment or economic growth figures provided by the government and the financial media are accurate and at least close to the gospel truth.
In a November 24th report from UBS Global Research, Paul Donovan and Sophie Constable suggest that the problem is really two-fold. First, society, the workforce and the workplace are evolving and old methods of economic data collection may no longer capture the same data they used to. Second, the average American (or European) is placing a growing amount of trust in the growing flow of economic data continually cranked out into the public domain by the financial media, and experts say this trust is clearly misplaced.
Donovan and Constable also identify four issues for investors regarding the unreliability of economic data:
“(1) Rapid structural change in the global economy today is poorly captured by economic data often designed over half a century ago
(2) Survey evidence is increasingly unreliable – which matters not just to issues like sentiment data, but also to data items like GDP that do not appear to be surveys (but which are).
(3) Economic statistics are reported as single data points, often on the spurious reasoning that this is what is demanded, which gives a false sense of confidence as to the precision of the data
(4) Big data may make the problem worse by giving the impression it is reporting macro-economic trends when in fact it is generally covering more micro (and biased) economic data.”
Structural changes in modern economies impact economic data collecting and reporting
One key point made in the UBS report is that structural changes in the functioning of modern economies means that traditional economic data may actually be significantly misreporting the underlying rates, trends, etc. being measured by the data.
Donovan and Constable highlight the recent “A Secret Capex Story” , which claims the flexibility of modern technology and the rise of self-employment meant that capital spending is likely occurring as consumer spending, making the current “traditional” capital spending data misleading.
The growing trend towards elf-employment is almost certainly impacting employment and income data. The UBS analysts point out if the proportion of self-employment in an economy is rising, then traditional employment statistics based on hiring by businesses may not accurately reflect today’s labor market. For example, an increase in self-employment is likely to change how many people are compensated for work (ie, proprietors’ own income instead of salary). Therefore, hours worked becomes much more difficult to accurately determine, and various measures of labor productivity also become less reliable.
Unfortunately, this issue is a virtual certainty to get worse in the future. With automation, robotics and the virtual economy growing apace today, more and more social dislocation is almost inevitable.
As Donovan and Constable note: “Changing society and changing working practices in society will make traditional data releases at best misleading and at worst obsolete – unless they adapt.”
Surveys are increasingly unreliable
The growing unreliability of consumer and corporate sentiment surveys is a trend that UBS has noted in the past. The problem is that the volatility of sentiment survey data has tended to rise relative to the volatility of underlying economic fundamentals.
The UBS team suggests this trend “reflects the role of a more sensationalist media in influencing sentiment readings, and the possibility that some survey respondents will “game” the system (knowing the economic impact of surveys, they give answers that they believe will secure a favorable policy response rather than which reflect their own economic circumstances).”
Surveys are used for much more than just measuring sentiment, and there has been a notable trend of surveys becoming less accurate over the last few years. In fact, the growing inaccuracy of political polls has become a cultural meme in the UK these days, what with the poor performance of opinion polls in the last several elections.
Keep in mind that surveys are often used to capture real economic activity data, that is, measuring income or employment status, for example.
Donovan and Constable argue there are three problems with survey based data today:
- People are less likely to bother to answer surveys at all
- If people do answer a survey, they are likely to miss questions in the survey
- When people do answer the questions, they are less likely to give an accurate answer (more people intentionally (to manipulate survey) or unintentionally giving a wrong answer).
Economists shifting to ranges
Economists and statisticians know there is a problem with economic data today, which is why they are increasingly shifting to reporting in ranges rather than as a single figure.
The Bank of England‘s charts from August of 2015 reported the UK GDP growth rate in 2012 averaged around 1%. That said, the Bank of England was only really officially prepared to commit to a rather broad range — ie, that growth was either above trend, or in a recession, or somewhere in between. Note that three years later, the Bank of England is still assuming there is a 3% to 4% range of possible growth numbers.