Jump to content
House Price Crash Forum
Sign in to follow this  
beerhunter

House Price Graphs... A New Take On Them

Recommended Posts

I'm sure all of us have looked at the HPI graphs on HPC and other sites, and some of us have even plotted our own. But once you have looked at them and tried to predict what will happen in the next few years, what other use is there for them?

Another common grip is why are the values so different, there meant are measuring the same thing.

Well here’s a new take: rather than look at just one house price index over time, I’m going to use 2 of them (Nationwide and Halifax Quarterly figures) and look at the correlation.

Correlation is a statistical technique that can show whether and how strongly pairs of variables are related (see http://www.surveysystem.com/correlation.htm for a simple explanation).

In this case both Nationwide and Halifax HPI’s are measuring the same (theoretical) house price, so we would expect a strong correlation between them (i.e. the correlation coefficient would be very close to 1.0)

However I had a hunch that perhaps this is not always true, because sometimes one index goes up, whilst another goes down... i.e. sometimes there is a period of uncertainty in which way the market is going.

So can we use this uncertainty in the House Price Indexes to our advantage?

The attached picture at the bottom of this post shows a graph;

- The green line is the Nationwide Quarterly House Price Index (Right-hand Scale)

- The blue line is the Halifax Quarterly House Price Index (Right-hand Scale)

- The red line is the correlation between the Nationwide and Halifax House Price Indexes (over the previous 4 Quarters).

When the red line is;

- Positive: Nationwide and Halifax agree that prices are going up (or down)

- Negative: Nationwide and Halifax disagree on the direction of prices.

The closer the red line is to 1.0 (or –1.0) the stronger the correlation.

As expected the red line is around 0.8 to 1.0 most of the time, except for 3 notable spikes;

- Q4 1989 to Q4 1991 : The last crash

- Q4 1993 to Q1 1996 : The bottom of the last peak

- Q4 2000 to Q2 2001 : The affect of 9/11?

In the last few years, there has been overwhelming correlation between Nationwide and Halifax, until Q4 2004.

Is this the start? Maybe... it’s probably impossible to tell from this graph, since I used quarterly data to plot the graph. But compared to Q4 1989 there hasn’t been such a quick divergence of HPI’s... yet

correlation_between_nationwide_and_halifax_price_indexes.gif

post-388-1127341414_thumb.jpg

Share this post


Link to post
Share on other sites

beerhunter,

Inspired work. Halifax and Nationwide have their own in-built regional differences - bubbles propagate (ripple effect) so inflection points would be shown on the stats. I think that is what your correlation chart is showing.

Share this post


Link to post
Share on other sites

Very interesting. The drops may be points where the volumes are very low and therefore the statistical errors become large.

Share this post


Link to post
Share on other sites

Excellent thinking. From this we can at least calculate the error implicit in their lies, sorry I mean stats. I believe the error is at least 1% and the constant 0.2, -0.4, 0.6 is pure froth.

Heisenburg constant is kind of what I'm thinking about, but it's probably irrelevant.

Share this post


Link to post
Share on other sites

Does anyone know if either bank have a greater presence in the North or South? If they did, then that might explain any potential variations at the time crashes begin as the North always lags behind by a year or two.

Share this post


Link to post
Share on other sites
Does anyone know if either bank have a greater presence in the North or South? If they did, then that might explain any potential variations at the time crashes begin as the North always lags behind by a year or two.

That was my unserstanding from what I have read elsewhere - that there is a significant enough difference in their core lending profile to account for some of the variation in their figs. One note, this may have been more relevant in the past when lending has a very significant high street component - it may not be quite the same now as smaller lenders have been subsumed, greater volume from broker network and telesales/internet sales.

Share this post


Link to post
Share on other sites

Just an extra thought - It would be good if you could do a London graph. The reason I say this is that at present they are showing opposite movements so it might be a good example of what you think might happen on the uk one.

From Homepage

Halifax -2.5

Nationwide +3.5

Share this post


Link to post
Share on other sites

Beerhunter

Have you thought of extending the analysis to include sets of data that might be seperated in time, but recording similar events. I was thinking of leading and lagging indices. Does the Hometrack data corellate with the Halifax if the time is shifted. Would this be a pointer for where the lagging indicators are headed. Apart from downwards obviously!

Share this post


Link to post
Share on other sites

Thanks to everyone for the kind comments.

Sorry about the brief reply... I should be working :(

Halifax and Nationwide have their own in-built regional differences - bubbles propagate (ripple effect) so inflection points would be shown on the stats. I think that is what your correlation chart is showing.

I agree, and I hope that's what's being shown on the graph.

I was (possibly dumbing it down) by calling uncertainty... ie if both hpi's say prices are going up (or down), then there is a degree of certainty (assuming you believe their methods). But when one is going up, and the other down... this is what I termed uncertainty... which is caused by ripples (and probably other things too).

The drops may be points where the volumes are very low and therefore the statistical errors become large.

I'm not convinced of this... for example the attached graph uses the monthly figures (as opposed to the quarterly figures used before).

Note that currently, even though volumes are low, there is still a reasonable correlation between the 2 hpi's.

From this we can at least calculate the error implicit in their lies, sorry I mean stats

An interesting idea.. and for example I've always felt the ODPM figures seem to follow a random trend :)

How could this be measured / presented?

My initial though is to work out the correlation with each other index (eg if there are 3 HPI's called A, B and C work out correlation between each pair (so you get 3 coefficients) and use these as a basis for the "distance" between the points. Ie you would end up with each hpi plotted on a graph, such that closely correlated indexes would be close to each other, unrelated indexes would be further away.

Does anyone know if either bank have a greater presence in the North or South?

I have heard in the past (don't remember the source at the moment) that the Halifax as a majority of mortgages in the North.

I don't think its just about if / where they have a greater presence.. it could also relate to the relative number of types of mortgages (FTB / MEW / BTL) and of course the type of housing.

London graph
Can you test this with other sources of data?

I took the Nationwide and Halifax Quarterly figures because they covered back to 83ish... enough to see what happened last crash.

I'd certainly like to get a definitive graph (using the sources on the front page of HPC.. including London).. to test the theory out :D

Meantime I've used the monthly data (as opposed to the quarterly used in the first post) to generate a new graph which shows in more detail what's happened since 1991.

correlation_between_monthly_nationwide_and_halifax_price_indexes.gif

post-388-1127399557_thumb.jpg

Share this post


Link to post
Share on other sites

On looking at these graphs further it seems that they are probably picking up changes in momentum or direction of house prices.

This is probably because the indices use different lengths of time for their smoothing so one graph is more responsive and the other is smoother.

ie in 2 graphs a 2 month moving average would change direction faster than a 6 month moving average with a change in direction of the data.

Looking at the graphs though its hard to see any consistent patterns between the data so maybe I'm talking crap.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
Sign in to follow this  

  • Recently Browsing   0 members

    No registered users viewing this page.

  • 301 Brexit, House prices and Summer 2020

    1. 1. Including the effects Brexit, where do you think average UK house prices will be relative to now in June 2020?


      • down 5% +
      • down 2.5%
      • Even
      • up 2.5%
      • up 5%



×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.