Mods, Fix This Nonsense For Once And For All Please

[happy_renting]

Yet another graph showing a nonsense '2.9%' trend by selectively extrapolating data from a date arbitrarily before the end-point, and then ignoring the data after it.

Disgraceful that the HPC website promotes this overstated 'trend', on it's own front page, too.

From the graph (as far as I can read it),

Price = £80k Approx in 1975

Price = £162k Approx in 2013.

Rise of £82k (102.5%) in 28 years

= rise of 2.55% pa over the period shown (1975-2013). AND FALLING, if one selects only the most recent data for a trend.

A calculation of a trend over a given period MUST take into account all the data for that period.

A trend over a period must, by definition, match the start and end points of the data.

If the data is weighted, the weighting should be weighted to give greater importance to the most recent data.

Also, as new data arrived, the trend line should be adjusted accordingly to admit that there is new data. Instead HPC shows a continuing 'trend' in denial of the fall over the past few years. Trends are not set in stone. If the data changes, the trend should be reassessed accordingly.

Unless the red line is supposed to show something else? Or did the crash of the past few years not register as the slightest blip on the trend? It's HPI nonsense.

[The Knimbies who say No]

Yet another graph showing a nonsense '2.9%' trend by selectively extrapolating data from a date arbitrarily before the end-point, and then ignoring the data after it.

Disgraceful that the HPC website promotes this overstated 'trend', on it's own front page, too.

From the graph (as far as I can read it),

Price = £80k Approx in 1975

Price = £162k Approx in 2013.

Rise of £82k (102.5%) in 28 years

= rise of 2.55% pa over the period shown (1975-2013). AND FALLING, if one selects only the most recent data for a trend.

A calculation of a trend over a given period MUST take into account all the data for that period.

A trend over a period must, by definition, match the start and end points of the data.

If the data is weighted, the weighting should be weighted to give greater importance to the most recent data.

Also, as new data arrived, the trend line should be adjusted accordingly to admit that there is new data. Instead HPC shows a continuing 'trend' in denial of the fall over the past few years. Trends are not set in stone. If the data changes, the trend should be reassessed accordingly.

Unless the red line is supposed to show something else? Or did the crash of the past few years not register as the slightest blip on the trend? It's HPI nonsense.

It's nothing more than a reproduction of the graph in Nationwide's release. It's fair enough to do an accurate reproduction of it IMV, we all know it's a load of cobblers as far as a fundamental description of pricing goes. If anything it'll be interesting to see at what point Nationwide decide to remove it themselves.

I say keep the unadulterated version.

[Bruce Banner]

It's nothing more than a reproduction of the graph in Nationwide's release. It's fair enough to do an accurate reproduction of it IMV, we all know it's a load of cobblers as far as a fundamental description of pricing goes. If anything it'll be interesting to see at what point Nationwide decide to remove it themselves.

I say keep the unadulterated version.

Yes, one thing is certain, that's the best the Nationwide can make it look.

[zugzwang]

They must be anticipating the next peak in 2025, 3x higher than 2008.

[@contradevian]

A Fab Fibonacci retracement would be better or even some simple moving averages.

http://www.housepricecrash.co.uk/forum/index.php?showtopic=145405

[snowflux]

I've often wondered about the rationale behind fitting an exponential curve to real house price data. I can't see any obvious justification for doing so. As far as I can see, there is no particular reason to expect house prices to follow such a curve.

[happy_renting]

It's nothing more than a reproduction of the graph in Nationwide's release. It's fair enough to do an accurate reproduction of it IMV, we all know it's a load of cobblers as far as a fundamental description of pricing goes. If anything it'll be interesting to see at what point Nationwide decide to remove it themselves.

I say keep the unadulterated version.

I see that Nationwide is the source. is HPC able t produce their own? I would anticipate some copyright problems, but then, there could be copyright issues copying the NW graph unaltered as it is anyway.

More practically, HPC could add a caveat, and regularly give the real HPI figure derived from the NW data underneath the graph.

i would find a house price graph together with mean/median/average incomes or RPI or CPI superimposed more informative.

(As if one can really believe RPI and CPI figures either )

I am a NW member... I should be raising this complaint with them,I guess. If I do, i'll keep people posted.

[enrieb]

I'd like to see the volume of sales added to the graph.

[R K]

I've often wondered about the rationale behind fitting an exponential curve to real house price data. I can't see any obvious justification for doing so. As far as I can see, there is no particular reason to expect house prices to follow such a curve.

It's a straightline not a curve.

It appears to be a curve because of the linear rather than log scale (imo)

[“Nasty Piece of work”]

And I wonder why Joe Public thinks property only goes up, and the great unwashed thinks bricks and mortar are safe and The Mail is a paper.

[Wurzel Of Highbridge]

I was going to produce a graph of total value of sales on a monthly basis. I think that this would be a good snapshot of the market, it will fluctuate seasonally but I feel that overall it is a better market indicator.

It is 'just' a mater of getting the land registry sold data and adding up all of the sale prices on a monthly basis. That way you can easily see if it was just 10 sales in London that moved the average sale price up.

In effect you would be looking at total market volume.

[interestrateripoff]

The exponential function.

[porca misèria]

I've often wondered about the rationale behind fitting an exponential curve to real house price data. I can't see any obvious justification for doing so. As far as I can see, there is no particular reason to expect house prices to follow such a curve.

You can try and model them based on all kinds of real-life factors to get a "rational HPI" curve, deviations from which are then irrational. But that begs lots of questions of how important different factors are:

• Interest rates: long-term gradual downward trend.

• Wealth effect: rapidly falling proportion of income required for non-housing living costs.

• Financial incentives, particularly tax changes in 1989 to crash the then-bubble and in 1997 to bring in a new flood of money.

• Tax-and-benefits system, though it's not clear to me how changes over time have worked.

• Changes in rental market, and its rise as a viable alternative to buying, and most recently people actively choosing to rent rather than buy FTB flats.

• Last but not least, substantial real improvements in houses that add to their real value.

[The Knimbies who say No]

It's a straightline not a curve.

It appears to be a curve because of the linear rather than log scale (imo)

?

[TheCountOfNowhere]

sounds like someone doesn't want to live in the real world.

[snowflux]

You can try and model them based on all kinds of real-life factors to get a "rational HPI" curve, deviations from which are then irrational. But that begs lots of questions of how important different factors are:

• Interest rates: long-term gradual downward trend.

• Wealth effect: rapidly falling proportion of income required for non-housing living costs.

• Financial incentives, particularly tax changes in 1989 to crash the then-bubble and in 1997 to bring in a new flood of money.

• Tax-and-benefits system, though it's not clear to me how changes over time have worked.

• Changes in rental market, and its rise as a viable alternative to buying, and most recently people actively choosing to rent rather than buy FTB flats.

• Last but not least, substantial real improvements in houses that add to their real value.

True. I suppose an exponential curve is going to be the best fit if you're restricting yourself to two degrees of freedom though. As Nouveau Leech says, such quantities as GDP and wages have followed similar curves. However, I can't see that it has much predictive value, given that we appear to have hit the bumpers of growth in 2007. I certainly doubt very much that houses are currently under-priced, as the graph would appear to imply!

[The Moderators]

Yet another graph showing a nonsense '2.9%' trend by selectively extrapolating data from a date arbitrarily before the end-point, and then ignoring the data after it.

Disgraceful that the HPC website promotes this overstated 'trend', on it's own front page, too.

From the graph (as far as I can read it),

Price = £80k Approx in 1975

Price = £162k Approx in 2013.

Rise of £82k (102.5%) in 28 years

= rise of 2.55% pa over the period shown (1975-2013). AND FALLING, if one selects only the most recent data for a trend.

A calculation of a trend over a given period MUST take into account all the data for that period.

A trend over a period must, by definition, match the start and end points of the data.

If the data is weighted, the weighting should be weighted to give greater importance to the most recent data.

Also, as new data arrived, the trend line should be adjusted accordingly to admit that there is new data. Instead HPC shows a continuing 'trend' in denial of the fall over the past few years. Trends are not set in stone. If the data changes, the trend should be reassessed accordingly.

Unless the red line is supposed to show something else? Or did the crash of the past few years not register as the slightest blip on the trend? It's HPI nonsense.

Thanks for your comments however the moderators have no control over the main part of the HPC website - this is handled by Fubra support. I have written to Fubra support on many occasions to see if we could get a better graph but none were available that we could use.

If you know of a better graph that is copyright free then please email Fubra support with the details and I am sure they will give you an answer.

[Tired of Waiting]

Yet another graph showing a nonsense '2.9%' trend by selectively extrapolating data from a date arbitrarily before the end-point, and then ignoring the data after it.

Disgraceful that the HPC website promotes this overstated 'trend', on it's own front page, too.

(...)

+ 1

Soon someone will write an article on these lines:

It's time to buy!

Even the website housepricecrash.co.uk admits, UK house prices are £30k below historical trend.

Average prices are below £160k, whilst trend indicates that fair prices now should be above £190k, as shown by a chart on its front page (... article continues)

[Tired of Waiting]

Thanks for your comments however the moderators have no control over the main part of the HPC website - this is handled by Fubra support. I have written to Fubra support on many occasions to see if we could get a better graph but none were available that we could use.

If you know of a better graph that is copyright free then please email Fubra support with the details and I am sure they will give you an answer.

Thanks for the reply.

Can't they just chart real house prices, without a trend line?

[vin rouge]

If one observes the increase in duration and amplitude of the 3 previous troughs one could extrapolate that the dip we are currenly in will eventually bottom out in 2026 with an average price of £125,000 and the next peak will occur around 2043 with a price of about £700,000

And whilst the price at present may possibly be below the extrapolated long term average a buyer would be much better off waiting a few years until nearer the bottom of the curve. Factoring in rent paid against the risk and benefit of waiting for the true bottom I would take a stab at 2020 as being a good year to buy.

[opt_out]

It's a straightline not a curve.

It appears to be a curve because of the linear rather than log scale (imo)

LOL. No.

[opt_out]

Yet another graph showing a nonsense '2.9%' trend by selectively extrapolating data from a date arbitrarily before the end-point, and then ignoring the data after it.

Disgraceful that the HPC website promotes this overstated 'trend', on it's own front page, too.

From the graph (as far as I can read it),

Price = £80k Approx in 1975

Price = £162k Approx in 2013.

Rise of £82k (102.5%) in 28 years

= rise of 2.55% pa over the period shown (1975-2013).

I'm sure we've been alll over this before.

But fitting a curve is valid.

Your calc is considerably less valid as you have picked what looks like a peak at the start, and some way off peak for the end. Peak to peak, or trough to trough would be reasonable choices.

example:

Lets say 1976 is 75k, and is at a similar point in the up/down cycle to 2013. Now Your calc gives 2.9%

Edited July 10, 2013 by opt_out

[opt_out]

The biggest problem with that graph is the truncated £ scale, that's what exagerates the trend.