Spline's Graphs And Analysis Becoming Mainstream?

[Guest wrongmove]

UK mortgage approvals weaken - but outlook for UK housing market is still rosy

"Last week it was announced that the number of new UK mortgage approvals fell to 113k, resulting in much media speculation that the UK housing market is in danger of collapsing. I strongly disagree with this view and expect the UK housing market to remain buoyant over at least the next half year.

The image here (click to enlarge) shows the close correlation between mortgage approvals and UK house price growth. There is a lag of seven months between mortgage approvals and the rate of house price changes, so I've moved the mortgage approvals data forward by seven months so you can predict where house prices will be in seven months time. While mortgage approvals have fallen slightly, a level of 113k has historically corresponded to annual house price growth of over 15% seven months later...."

I notice another close correlation. Here is the graph from the article

And here is the graph from Spline's excellent website

Coincidence?

Edited May 11, 2007 by wrongmove

[bazzzzzzz]

Coincidence?

Or is Richard Woolnough, Spline?

[Guest wrongmove]

Or is Richard Woolnough, Spline?

Maybe! The article reads as if Spline of KingOfNowhere has written it. So far, the analysis has proved correct. I turned from bear to neither in late 2005 because I was convinced by these trends. Correlations this strong are rare in nature. So far, HPI predictions based on BoE mortgage approvals have been proven correct.

Mortgage approvals are still high (113k), but at least they are dropping. But while they remain above 90-100k, I can't see price drops within 6 months.

[DabHand]

Seems to me that since 04 it requires a proportionally higher number of approvals to sustain HPI. Looks like a divergent trend as it runs out of steam.

[drrayjo]

I wouldn't really call this mainstream but, whether or not you do, I have to admit it's quite convincing reasoning. Assuming accuracy of the data, it does look like there may be more of 'The Madness' left to go.

[Guest wrongmove]

Seems to me that since 04 it requires a proportionally higher number of approvals to sustain HPI. Looks like a divergent trend as it runs out of steam.

Indeed - if this was not the case, I would be a bull.

[Guest wrongmove]

I wouldn't really call this mainstream but, whether or not you do, I have to admit it's quite convincing reasoning. Assuming accuracy of the data, it does look like there may be more of 'The Madness' left to go.

I linked to this report from an article in a more mainstream source - I can't find the original article now.

[Guest wrongmove]

I linked to this report from an article in a more mainstream source - I can't find the original article now.

Here is the original article - it's our old friends, Sh1ttyWire!

http://www.citywire.co.uk/Blogs/Property/E...VersionID=91932

[drrayjo]

Here is the original article - it's our old friends, Sh1ttyWire!

http://www.citywire.co.uk/Blogs/Property/E...VersionID=91932

Fair enough. Plenty of folk will be being advised that all is well in 'house-wealth growth' land then. But the divergent trend that Dabhand highlighted should be cause for at least a little caution from even the bulliest of bulls.

To me this says that we're well into the current (supply/demand fundamentals and +ve sentiment) vs (affordability, -ve sentiment and credit availability) battle for house price influence.

So far it's been easy campaign success after success for HPI in the price war but is the fighting on 3 fronts starting to tell?

I think so.

[JustYield]

I notice another close correlation. Here is the graph from the article

And here is the graph from Spline's excellent website

Coincidence?

Spline's graph you show here doesn't overlay the lagged HPI (although I know he has shown that).

When do we cross the line, yoy<0? Certainly not this year by the looks of things. Sometime in 2008 I reckon.

[thedebtisreal]

Spline's graph you show here doesn't overlay the lagged HPI (although I know he has shown that).

When do we cross the line, yoy<0? Certainly not this year by the looks of things. Sometime in 2008 I reckon.

I thought £113k approvals meant 4-5% HPI later on.

Interestingly look at the mortgage approval figures, they are slowly markedly. Back of the fag packet calculations I think we will be <90k approvals by around September time, making MOM falls likely in q1 2008. YOY not until later in 2008.

I think I'm in agreement with d23 on this one. Does that make me a neither or him a bear?

[Guest wrongmove]

Spline's graph you show here doesn't overlay the lagged HPI (although I know he has shown that).

When do we cross the line, yoy<0? Certainly not this year by the looks of things. Sometime in 2008 I reckon.

I think you can lag the data on Spline's website - it was the analysis, and most of all, the same choice of line colours that struck me.

[ae589]

Shame we can't see approvals by region - since we already have neg MOM in some regions, we could see (er... guess) each regions position in its own cycle.

[Landagan]

Also interesting to see how mortgage approvals dropped like a stone in mid 2005, and later that year things almost turned YOY negative.

There's the same feeling in the air, so could another sharp drop be possible?

And then will they rescue it as before with a non-logical interest rate reduction?

I remain bearish, but not as bearish as I was in 2005. Once bitten ..

[Guest wrongmove]

Also interesting to see how mortgage approvals dropped like a stone in mid 2005, and later that year things almost turned YOY negative.

That was when I chnaged from bear to neither. everything was going to plan up till then - approvals dropping below 90k, and sure enough, HPI dropped to zero.

Then it all went pear shaped

But approvals continued to be an accurate guide to future HPI. And they are the lowest now for nearly a year. May this trend continue......

[Landagan]

That was when I chnaged from bear to neither. everything was going to plan up till then - approvals dropping below 90k, and sure enough, HPI dropped to zero.

Then it all went pear shaped

But approvals continued to be an accurate guide to future HPI. And they are the lowest now for nearly a year. May this trend continue......

Can't help feel they may try the same trick again. Certainly tests the bear in you.

The quicker that moment of decision arrives the better. For I am certain it will arrive.

Good graphs, hard to argue the facts there. But a sudden steep decline is possible, and perhaps will be steeper than the last with all the added circumstances.

10-12 Months till YOY neg. I'd bet my house on it.

[spline]

Hi wrongmove – I’m amazed that the secret inner working of house price mechanics are sometime able to attract a bit of press coverage, although I have to add that I’m not Richard Woolnough. But it’s good that the argument is presented along with the graphs in a way that’s essentially the same (and as we've been through on threads here) – so it stands a good chance of being be right.

JustYield’s point - correlation and the lagged HPI graph:

This is the HPI graph from 1990-2007 using the Haliwide prices, lagged in this case by 6 months and with HPI implied by the correlation added in yellow. The divergence is relatively recent with the approvals slightly overestimating the HPI – possible reasons are the change in the ratio of cash/mortgage sales (fewer cash) hence using approvals as a proxie for transactions would need a slight tweak, or possibly that the total approval numbers since 2003 now contain a significant added contribution from “specialist” lenders. This post-2003 tweak is included in the model curve that overlays the “House Price” graph but not in the HPI graph.

From: http://www.houseprices.uk.net/graphs/

[select HPI annual %, click on 'prediction: BoE …', 'Shift YoY by 6 months', and push the start data back to 1990]

Edited May 11, 2007 by spline

[Guest wrongmove]

Hi wrongmove – I’m amazed that the secret inner working of house price mechanics are sometime able to attract a bit of press coverage, although I have to add that I’m not Richard Woolnough.

Hi Spline. i'm glad you could confirm that! I loved the way he even appears to have half-inched your colour scheme (as well as your argument).

[spline]

Yeah, he could at least have listed his references. I guess I’ll have to build a proper predict-o-meter now.

[needle]

Maybe! The article reads as if Spline of KingOfNowhere has written it. So far, the analysis has proved correct. I turned from bear to neither in late 2005 because I was convinced by these trends. Correlations this strong are rare in nature. So far, HPI predictions based on BoE mortgage approvals have been proven correct.

Mortgage approvals are still high (113k), but at least they are dropping. But while they remain above 90-100k, I can't see price drops within 6 months.

All of this graphology is exactly that - phrenology for the mathematically inclined.

I think spline would acknowledge that these graphs show nothing of the quality of credit extended or the amount loaned or the structure of the finance.

I also find it curious that a 1/2 point drop in rates seemed to cause such a bounce. Surely, if we were working on pure logic, the recent rises will have the opposite effect?

Again these graphs are interesting and instructive but they are only part of the HPI/HPC puzzle and, in my view, should hold no more weight than the figures showing record indebtedness.

[Guest wrongmove]

All of this graphology is exactly that - phrenology for the mathematically inclined.

I think spline would acknowledge that these graphs show nothing of the quality of credit extended or the amount loaned or the structure of the finance.

I also find it curious that a 1/2 point drop in rates seemed to cause such a bounce. Surely, if we were working on pure logic, the recent rises will have the opposite effect?

Again these graphs are interesting and instructive but they are only part of the HPI/HPC puzzle and, in my view, should hold no more weight than the figures showing record indebtedness.

I think there is more to it than that. The graphs are certainly a simplification, but we need simplifiaction in order to try to get a handle on things. IMHO, this is a useful simplification, and it has shown good predictive qualities over a long period of time.

Take your example of record indebtedness. What predictive power does that have? Every day that goes by sets a new record of indebtedness, and this has been the case for decades now. I don't see what this proves in itself, or predicts about the future of house prices. 2000 was a record year for indebtedness, so was 2001, 2002 etc.

What mortgage approval figures do is simply give us a sneak preview of the levels of demand and activity in the housing market, and the levels of demand and activity tend to set the price. Nothing more, but nothing less.

ps: it was only a 0.25% drop btw

[spline]

All of this graphology is exactly that - phrenology for the mathematically inclined.

I would say that these graphs are specifically about house price dynamics and what might be used to predict house prices into the nears future – and it appears that HPI correlates very strongly with both the BoE approvals data and the RICS sales-to-stock ratio.

You can interpret this as evidence of an underlying relation between HPI and some measure of transaction volume, or at least that a large part of the behaviour is captured by the correlation – either that buyer volumes reflect anticipated HPI gains/falls, or the reverse, that prices rise a certain percent for each house sold, e.g. EAs mark everything up a fixed amount on the last sale. There is, of course, the usual health warning – correlations do not always imply a causal relation – but in this case the argument seems to be a good one and it has held (apart from the minor post-2003 drift) over all the different parts of the last cycle.

The BoE numbers can be treated simply as a timely proxie for transactions, and the RICS ratio adjusts this a bit to take into account the supply side and does a slightly better predictive job post 2003.

Edited May 11, 2007 by spline

[JustYield]

JustYield’s point - correlation and the lagged HPI graph:

This is the HPI graph from 1990-2007 using the Haliwide prices, lagged in this case by 6 months and with HPI implied by the correlation added in yellow. The divergence is relatively recent with the approvals slightly overestimating the HPI – possible reasons are the change in the ratio of cash/mortgage sales (fewer cash) hence using approvals as a proxie for transactions would need a slight tweak, or possibly that the total approval numbers since 2003 now contain a significant added contribution from “specialist” lenders. This post-2003 tweak is included in the model curve that overlays the “House Price” graph but not in the HPI graph.

Another obvious reason for the divergence is that the market is peaking and turning - a different type of transaction volume, if you know what I mean. In other words you might need to reverse the sign of the correlation if the market is falling.

[spline]

The interesting thing is that the correlation held throughout the last boom, peak, and fall (at least on the annual approval data, the monthly BoE series goes back only to around 1993) and the recent divergence appears to be something that’s “different this time”, for example, the rise of the specialist lenders (these add to the numbers after about 2003) or the recent drift/drop in the proportion of cash purchases.

Ideally, you would expect neutral HPI (or the HPI = earnings growth point) to correspond roughly to the long-term average rate at which houses are transacted, although possibly normalised by some measure of demographics. The numbers from the last cycle suggest that neutral (nominal) HPI appears to correspond to a monthly approvals rate of around 80-90k

Edited May 11, 2007 by spline

[JustYield]

The interesting thing is that the correlation held throughout the last boom, peak, and fall (at least on the annual approval data, the monthly BoE series goes back only to around 1993) and the recent divergence appears to be something that’s “different this time”, for example, the rise of the specialist lenders (these add to the numbers after about 2003) or the recent drift/drop in the proportion of cash purchases.

Ideally, you would expect neutral HPI (or the HPI = earnings growth point) to correspond roughly to the long-term average rate at which houses are transacted, although possibly normalised by some measure of demographics. The numbers from the last cycle suggest that neutral (nominal) HPI appears to correspond to a monthly approvals rate of around 80-90k

The question I'm asking is what do we expect, in broad terms, to happen to transaction volumes as y.o.y first slows, then falls, then goes negative?

At the peak, I'd expect to see a flurry of activity, then a pause where the mainstream outlook becomes predominantly bearish, then a dry up of volume to a trickle until sellers get with the programme and volume returns at lower prices. At the peak you would expect the lagged approvals vs. HPI to diverge, IMO, which is exactly what your graph shows.