Friday 2 May 2008

Halifax house price index down five percent from peak

Today, the Halifax published their price index for April. It showed that house prices are down over 5 percent from their peak of autumn last year.

I also compared the Halifax's seasonally adjusted data with the non adjusted counterpart. Since 2005, the seasonaly adjusted data has consistently understated the true level of price inflation. It missed the peak, but now also appears to be falling faster than non-seasonally adjusted prices.

Personally, I think seasonal adjustments are overused, particularly with data series like house prices. Recent data from the Halifax amply illustrate why we should be so skeptical of these kinds of adjustments.

5 comments:

Anonymous said...

I agree about seasonal adjustments... they worked well when the effects were - erm - seasonal. The credit crunch is not seasonal - so it is an inappropriate technique today.

What I found obscene about the Halifax data was that the year-on-year didn't compare monthly averages a year apart, but rather quarterly means of monthly averages. This means that the annual decline has been falsely spun at 0.9% year on year - when, in fact, it is 3.9% seasonally adjusted or 3.7% really.

The Halifax data will be amusing over the next few months... their deceptions in picking self-serving techniques will bite them back.

Anonymous said...

I agree about seasonal adjustments... they worked well when the effects were - erm - seasonal. The credit crunch is not seasonal - so it is an inappropriate technique today.

What I found obscene about the Halifax data was that the year-on-year didn't compare monthly averages a year apart, but rather quarterly means of monthly averages. This means that the annual decline has been falsely spun at 0.9% year on year - when, in fact, it is 3.9% seasonally adjusted or 3.7% really.

The Halifax data will be amusing over the next few months... their deceptions in picking self-serving techniques will bite them back.

Anonymous said...

Seasonal adjustments make sense for consumption data and wages, largely because of the christmas effect. Agricultural production also needs seasonal adjustments. With other data, it is better to look at the original series.

Alice

Anonymous said...

Fed and ECB agree new growing rolling 2 weekly liqued auctions for Emu

Anonymous said...

Debt isn't seasonal. ;)