CSO Property Indices and Consumer Sentiment

Following the very useful suggestion from Terra Incognita to provide charts showing CSO property indices combined with consumer sentiment indices, the following charts show this information.

The ESRI maintains the consumer sentiment index -
esri.ie/irish_economy/consum … sentiment/

The latest data is available from: esri.ie/irish_economy/consum … June12.pdf

Their methodology is described in this document: esri.ie/docs/CSI_METHOD.PDF.

I have included the three ESRI indices here:

• Consumer Sentiment Index
• Index of Current Economic Conditions
• Index of Consumer Expectations

The ESRI data spans a longer interval than the CSO RPPI.

This chart shows the various CSO RPPI indices and the three ESRI consumer sentiment indices. The left scale applies to the CSO indices. The right scale applies to the ESRI indices.

This chart shows the various CSO RPPI indices and the three ESRI consumer sentiment indices three-month moving averages. The left scale applies to the CSO indices. The right scale applies to the ESRI indices.

This chart shows the various CSO RPPI indices and the three ESRI consumer sentiment indices back to Jan 1997. The left scale applies to the CSO indices. The right scale applies to the ESRI indices.

This chart shows the various CSO RPPI indices and the three ESRI consumer sentiment indices three-month moving averages.back to Jan 1997. The left scale applies to the CSO indices. The right scale applies to the ESRI indices.

Because of the nature of property purchases, the CSO indices reflect a past situation. The duration of the lag cannot be determined and also the lag will probably vary over time. Property purchases were probably completed more quickly during the bubble and now progress considerably more slowly.

The ESRI indices measure an immediate sentiment. The information is quite variable. The recent increase in sentiment is interesting and is not reflected in property prices, again at least because of the lag. Chart 2 would seem to indicate the lag is about five months.

I may try to analyse this information in more detail if I get time.

Of course, if the ESRI published a housing sentiment index, it would be jammed at 11… even now.

Interesting, though, that the Index of Expectations having bounced in 2009 has not resulted in a bounce in prices. I can think of two offhand possibilities:

  1. The stickyness of house prices means that they are only now getting to the point of meeting - when expectations collapse, sales volume collapses, but prices stay sticky. Eventually prices will hit an expectations level (which may have happened, but I don’t think so, expectations are fragile at the moment).
  2. There’s an expectations floor below which prices gain no upward traction - until expectations hit that floor, prices will continue to decline. It’s interesting to note that during 2002, house prices did decline before the credit bubble kicked into it’s final red giant expansion.

Whether these or something else, thank you for the time and effort you put into the charts jxbr.

Thanks jxbr there is not an immediately obvious prdictable link.

Why are the “mood” indexes so high and flat prior to 2000?

Obviously the dot com burst pulled the indexes down bottoming in July / October 2003, then indexes rose with the boom peaking with a dip between April 2005 / 6 and then fell.

Whats interesting is that the peak in the sentiment index peaked way before house prices and this was well in the throws of the Bertie governments and as far as I can remember this was well before any “bad” economic news.

The logical extension of that is that the index appears to have bottomed in October 08, I assume this was the Leemans / IMF bailout effect dragging it lower than it otherwise would. It rose after the bailout but fell presumably with the Cowan debacle and rose again with the FG election. It has been bouncing along the bottom for a while.

Presumably a sentiment index is not linier and is more likely to be sigmoidal ie very difficult to get it very low but also very difficult to get it very high but in-between it probably has some linier aspects.

It would be fair to say that the index has on average been rising sine the FG election in Feb 2011 and its just about the level of where it was in July October 2003 at the start of the last boom. We are clearly in a different place than we were in 2003.

All that said improvement in consumer sentiment is probably required prior to any resumption fo a normal housing market.

We are not there yet and I bet the sentiment index will be down again in before the year is out.

Here’s a chart that takes the year on year change in the CSO National Property Index and compares it to the ESRI Consumer Sentiment from ONE YEAR EARLIER.

Nice correlation, and I might post the chart that ‘predicts’ the year ahead if I get a chance tomorrow.

Or I might not.


Now that is an interesting graph.

Thinking logically about this: consumer sentiment reflects a willingness to spend money as an over all national measure. If the index is falling or rising, property prices should also fall and rise with a time lag. This lag is about 1 year according to the graph above.

It looks like the sentiment index is bouncing along the bottom now which could indicate that in 1 year the CSO Property Price stats will also bottom out.

For property prices to record a rise, sentiment would have to also be increasing for about a year.

Or maybe the causality runs the other way. The change in the value of your house compared to a year ago has an effect on your net wealth, ie, how much of your future income you will have to spare over paying down debt. This has an effect on how much you will spend today.

Very interesting stuff Jxbr
One thing that you might expect is consumer expectations to increase ahead of any significant price increases but the correlation may not be too strong.
Supply of properties and investment yields will also have an impact if prices stabilise this year
I would expect though that sentiment to turn strongly negative in the autumn if a EU recession deepens and the US weakens

That’s what’s sold by those saying that house prices should be supported. Personally, I think it is a pup. If that was the case, you wouldn’t expect to see a lag between sentiment and prices, the lag would be the other way round.

Good catch.

Based on taking an average of the high and low differences between the two, the inference is that property prices will be around 5% lower in 12 months. This is all based on the CSO national average.

So if we believe the chicken before the egg, that suggests a floor for sustained price rises of consumer sentiment around the 77 mark?

The closest correlation between the two is obtained at a lag of seven months between price change and consumer sentiment, a bit more than the 5 months I estimated at the start.

However, this assumes homoscedasticity (same variability/dispersion) along the time of the two series.

In the bubble, credit was readily available so constraints on availability of money to fuel property prices were not present. Now, there are considerable constraints on credit. Credit and property prices are also highly correlated:

This chart suggests that the National All Properties figure will be 10% lower by this time next year. I think it’s fair to assume that a large majority of that fall will be made up of decreases in price outside Dublin.

I’m quite excited by the Sentiment index correlation with the CSO house price index. There clearly is a correlation.

For someone deciding to buy a house there must be a point where they decide that they can afford it, get their ducks in a row and go out to buy it. The time between that decision and completion of a sale and then its inclusion in the CSO index must be of the order of 7-12 months. Ease of of credit may affect that but I dont think is as big as the ease of credit during the boom and now ie I think the biggest time lag is caused by the decision “yes I can afford a house” and the completion of a sale.

The lag may be different with ease or tightening of credit. Presumably shorter with easing of credit.

Rising sentiment has to translate in to sales and price rises

Falling sentiment has to translate into lack of sales and falling market.

Stable sentiment has to indicate a stable market.

Jxbr / Colles2 can you look at the bias of the sentiment index from the y/y % change in CSO index with an appropriate lag? this would allow you to calculate the 95% confidence intervals for your prediction for the next year. ie is it -5% or -10% whats the 95% confidence (2 Standard deviations from that mean)?

great work jxbr and coles2 - some fascinating graphs there.

The Credit graphs below would suggest to me a crossover prior a CSO bottom (however with only 1 cycle its hard to be definitive).

The other thing is that commentators are looking for changes in the direction of the consumer sentiment as indicators - I think more attention need to be paid to an absolute level - doesn’t the sentiment index already include directional information - I.e. anything below 100 means consumers believe the direction is negative - so if you are looking at changes in the direction of consumer sentiment you are actually looking at a second derivative. (I remain open to correction on this).

All the ususal caveats apply about such technical projections. This is based on the CSO National Index so it does not take into account local variations.

    Month        High    Estimate         Low
    Jul 12     -16.34%     -12.44%      -8.91%
    Aug 12     -16.94%     -11.57%      -6.60%
    Sep 12     -17.35%     -11.53%      -6.12%
    Oct 12     -17.55%     -11.84%      -6.58%
    Nov 12     -17.54%     -12.16%      -7.25%
    Dec 12     -17.33%     -12.25%      -7.65%
    Jan 13     -16.94%     -11.97%      -7.51%
    Feb 13     -16.39%     -11.31%      -6.77%
    Mar 13     -15.73%     -10.37%      -5.57%
    Apr 13     -15.02%      -9.36%      -4.29%
    May 13     -14.31%      -8.60%      -3.51%
    Jun 13     -13.67%      -8.53%      -4.03%

Any model about house prices has to take into account such factors:

Availability of credit
Availability of supply
Propensity to buy (sentiment)

The ESRI CSI is based on responses to the following questions:

Q.1. How do you think the economic situation will develop over the next 12 months?
(get better/stay the same/get worse)
Q.2. Do you think the number of people out of work in the country in the next 12 months will
(increase/remain the same/decrease)?
Q.3. How does the financial situation of your household compare now with what it was 12 months ago?
(got better/stayed the same/got worse)
Q.4. How do you think the financial position of your household will change over the next 12 months?
(get better/stay the same/get worse)
Q.5. In view of the general economic situation at the present time, what do you think about people buying
large items such as furniture, washing machines, TV sets etc. Do you think that for people in general the
present time is (good/neither good nor bad/bad)?

Consumer Sentiment Index = (Q1+Q2+Q3+Q4+Q5)/Base

Good job, jxbr.

Thanks - you’re not so bad, yourself.

Nice job both of you.

All good science makes testable predictions.

Even with the most generous prediction it still shows house prices nationally down next year by 4%

It will be interesting to see how accurate we are.


I suppose I should share this with you guys.

I’ve taken on board your observation that 1 year was not the ideal delay, so I’ve reduced it to 6 months and I’ve gathered up the data to look back over the last 15 years. What’s interesting is that the correlation is still quite good, but the periods of ‘irrational exuberance’ in the property market are clearly seen. I’ve named these 3 distinct periods as A, B, C. The largest of them, B, started in 2002 with the introduction of very enticing tax incentives for the property market. ‘C’ was the blow-off-top.

Any other observations would be welcome.

Folks, if you’re going to start making predictions based on statistical relationships you should first be happy with the quality of the data.

If you look back over the consumer sentiment index it does correlate loosely with consumer behaviour over the period. The monthly volatility, however, is something that is very hard to explain. What factor caused the lagged index to dip twenty points between January and February 2006 before making up nearly all the ground in the following months? I’ve never seen a convincing explanation (I have my own suspicions).

Given how much it bounces around on the month - would you think its predictive power is very strong?