Seasonally corrupted
Statistical quirks in figures for the US Bureau of Labor Statistics may have produced price signals and valuations that are far from reality.
breakingviews.com
Reported US consumer price inflation remained under control in March and April. But that's only because of huge and unprecedented downwards seasonal adjustments, of 0.6 per cent in March and 0.4 per cent in April, reducing reported inflation to a tame 0.3 per cent and 0.2 per cent, respectively. Whatever the theoretical justification, these adjustments raise questions about the credibility of the figures and risk creating a false market. Stocks jumped after the April report.
Seasonal adjustments in consumer price indices are intended to show price trends free of seasonal effects. They do not, however, represent changes in actual prices, so social security and Treasury bond indexation are done using unadjusted figures. Over the 10 years to 2007, the largest monthly seasonal adjustment was 0.3 per cent, the average was zero (as one would expect) and the standard deviation was 0.14 per cent.
Taking account of the normal adjustments in March and April over the previous decade, it's hard to see how these latest changes were derived by the same process as their predecessors. The April figures also have a problem in the energy area. According to the Bureau of Labor Statistics, petroleum-based energy costs declined by a seasonally-adjusted 0.7 per cent between December and April. That seems an extraordinary result, given the 20 per cent increase in both crude oil prices and US gasoline prices between those two months.
The seasonal adjustment process was changed in January 2008, and appears to have gone seriously adrift. If the past decade's average seasonal adjustments are used, consumer price inflation in March and April looks to have run at an annual 7.4 per cent rather than the 2.3 per cent for February-April claimed by the BLS.
That is a very serious difference. If inflation is really over 7 per cent then US interest rates are hugely negative in real terms and the stock market correspondingly overvalued. Statistical quirks in the BLS figures may have produced price signals and valuations that are far from reality. It is a great pity that these quirks should have occurred now, when inflation is the principal uncertainty in US markets.
For further commentary visit www.breakingviews.com
Reported US consumer price inflation remained under control in March and April. But that's only because of huge and unprecedented downwards seasonal adjustments, of 0.6 per cent in March and 0.4 per cent in April, reducing reported inflation to a tame 0.3 per cent and 0.2 per cent, respectively. Whatever the theoretical justification, these adjustments raise questions about the credibility of the figures and risk creating a false market. Stocks jumped after the April report.
Seasonal adjustments in consumer price indices are intended to show price trends free of seasonal effects. They do not, however, represent changes in actual prices, so social security and Treasury bond indexation are done using unadjusted figures. Over the 10 years to 2007, the largest monthly seasonal adjustment was 0.3 per cent, the average was zero (as one would expect) and the standard deviation was 0.14 per cent.
Taking account of the normal adjustments in March and April over the previous decade, it's hard to see how these latest changes were derived by the same process as their predecessors. The April figures also have a problem in the energy area. According to the Bureau of Labor Statistics, petroleum-based energy costs declined by a seasonally-adjusted 0.7 per cent between December and April. That seems an extraordinary result, given the 20 per cent increase in both crude oil prices and US gasoline prices between those two months.
The seasonal adjustment process was changed in January 2008, and appears to have gone seriously adrift. If the past decade's average seasonal adjustments are used, consumer price inflation in March and April looks to have run at an annual 7.4 per cent rather than the 2.3 per cent for February-April claimed by the BLS.
That is a very serious difference. If inflation is really over 7 per cent then US interest rates are hugely negative in real terms and the stock market correspondingly overvalued. Statistical quirks in the BLS figures may have produced price signals and valuations that are far from reality. It is a great pity that these quirks should have occurred now, when inflation is the principal uncertainty in US markets.
For further commentary visit www.breakingviews.com
Share this article and show your support
Call 1300 880 160
Start a chat
Schedule a call
