Tag Archives: Statistics

Lies, Damned Lies, and Unemployment Figures

Brian Gaynor in the NZ Herald wrote a good piece on Lies, Damned Lies, and Statistics. Share prices and bond markets are moved by variables including market indicators and statistics on interest rates, inflation and unemployment. The variables are important in other ways, too. Interest rates affect mortgages, inflation harms those on fixed incomes and rising unemployment breeds job insecurity. Yet do many of us really understand how the figures are calculated? Do we really know how to interpret them? If we don’t, should we strive to know more?

We certainly should. Many commentators make a good living out of describing and interpreting statistics. Often their comments are commonplace and uninteresting.The pundit industry would collapse if we all knew more about these numbers and could judge them for ourselves. Interpreting market indicators is a minefield. Many indicators are official statistics that relate to economic performance. Although official statistics are usually honestly compiled, their accuracy varies.

While you can generally trust data published by, for example, Statistics New Zealand, you need to be aware of the limitations of the numbers. Unemployment and the ways that it is calculated varies considerably. It is a very influential indicator – the US Federal Reserve and its monetary policy is impacted by the levels of unemployment. They have said that if unemployment hits the 6.5% threshold it may consider raising interest rates.

In New Zealand the official measurement is the “Household Labour Force Survey (HLFS)” which is conducted every 3 months, covers about 15,000 private households and about 30,000 individuals.

Another measure is the unemployment benefit which counts those looking for work or in training who are eligible for payment of the benefit.

A third measure,the registered jobseeker statistic, is an administrative statistic affected by seasonal fluctuations and by changed administrative practices. Historically, it counted those who registered with the Department of Labour to find work (from late 1998, Work and Income New Zealand). As well as those out of work and available for work it includes people who work up to 29 hours a week who seek to increase their hours of work and does not include specific job availability or search criteria.

The difference between these three measures can fluctuate dramatically. See the graph below from the Parliamentary Library.

NZ Unemployment 2013

Lies, Damned Lies, and “Argentinian Inflation” Statistics

Share prices and bond markets are moved by variables including market indicators and statistics on interest rates, inflation and unemployment. The variables are important in other ways, too. Interest rates affect mortgages, inflation harms those on fixed incomes and rising unemployment breeds job insecurity. Yet do many of us really understand how the figures are calculated? Do we really know how to interpret them? If we don’t, should we strive to know more?

We certainly should. Many commentators make a good living out of describing and interpreting statistics. Often their comments are commonplace and uninteresting.The pundit industry would collapse if we all knew more about these numbers and could judge them for ourselves. Interpreting market indicators is a minefield. Many indicators are official statistics that relate to economic performance. Although official statistics are usually honestly compiled, their accuracy varies.

While you can generally trust data published by, for example, Statistics New Zealand, you need to be aware of the limitations of the numbers. Take the consumer price index, a measure of “inflation”. Many commentators attach much greater significance to CPI figures than they deserve. The change in CPI from one month to the next may not be highly significant. It may be influenced by unusual factors that only applied in that month. Deducing a trend from one month’s figures can be dangerous. Statistics can also be adjusted and manipulated. Sometimes the adjustments are helpful. In the case of the CPI, for example, seasonal adjustments can be made. These reduce or eliminate the effect of items that vary sharply in price at different times of the year. They smooth out the figures, giving a better view of a trend. A moving average of several months’ figures may give a better view of an underlying trend. But neither of these is a panacea when it comes to interpreting the numbers. The less scrupulous might “annualise” one month’s or one quarter’s figures. But annualising can cause problems. If this month’s figures are sharply at variance with last month’s, then drawing conclusions from the annualised figure could be misleading.

Argentina’s Inflation Rate

For the last few years the Argentinian government has published inflation figures that a lot of people find hard to believe. Expansionary fiscal and monetary policy has caused the economy to grow too quickly which eventually led to higher prices. In order to conceal the higher inflation rates the government resorted to price controls and tampering with the official figures. Some employees of the statistics institute, INDEC, were told to omit decimal points, not round them. According to The Economist, although this doesn’t seem much, when you do the calculations you get the following:

1% monthly rise in the CPI = 12.7% annual rise
1.9% monthly rise in the CPI = 25.3% annual rise

Unions in Argentina use independent statisticians when negotiating pay increases. Surveys from a university show inflation expectations running at 25-30%. When you compare the official (Government) and the unofficial (PriceStats – private provider of inflation rates) you get the following (see also chart from The Economist):

Unofficial annual rate – 24.4% and cumulative inflation since the beginning of 2007 at 137%
Unofficial annual rate – 9.7% and cumulative inflation since the beginning of 2007 at 44%

Confidence in the present government’s economic policy has taken a hit and it will have to earn back the trust of not only its people but also the global community.