House Price Crash 5

RHemmings

Alan_M wrote:

Becasue they are as valid as a stand alone set of statistics, that's the very point I'm making, statistics don't have to be based on even reasonable compilations, they can be compiled from anything you want, they are still valid as statistics. Utterly meaningless but valid none the less.

Historically government average house price statistics for the past two decades have not included repossesed properties taken to auction, or properties where the negative equity reached such a level that the keys were handed back and the debt wriiten off, as in neither of these cases there was actually a house sale in the traditional sense.

Therefore if you look back at the factual government statistics , there was in fact no crash in the early 90's it must have been a figment of our imagination.

Now obviously we know different, and knowing that doesn't make the statistics any less valid, even if they are utterly useless in the form they are produced.

I disagree. For any statistic to be valid it has to be a "reasonably" unbiased sample from the population that the statistic is intended to describe. I can't see how you say "valid" about a statistic from an unrepresentative sample. Can you please say what your definition of "valid" is, and what you mean by "stand alone statistic"?

I also can't see how a statistic can be "utterly meaningless" but also "valid". Can you please explain further?

RHemmings

Hereward wrote:

Whilst it is true property prices will eventually stop rising, they do not necessarily have to crash, they just could have a period of zero prices rises. In addition, I have yet to see any evidence that property prices are forming a bubble: just because people here believe that they are doesn’t mean that a bubble is/has formed.

In theory, yes. But has this ever happened? Over a long period of time? In any market?

What would you consider evidence that a bubble has formed? I would say prices increasing over and above general inflation with it being clear that speculative investment, and an expectation that prices will continue rising justifying much of this investment, is the prototypical "evidence of a bubble".

RHemmings

sm9ai wrote:

They could do, but that would require severe wage inflation.

Which would be a real house price crash. I.e. nominal prices remain static while wages inflate.

tomstickland

AlanM is dead right about statistics. On their own they can't lie - they just tell you what they tell you. The way to manipulate them is to just decide measure things that suit your argument.

So if there was an argument about prison overcrowding then a government minister could take their pick of
-we've built 1 prison in 10 years
-prison overcrowding has increased by 1.5%
-we've provided 1000 more spaces this year,
All 3 might be true, but obviously the 3rd option gives the best impression, and if someone else chose to use option 2 then it might appear that the statistics were useless because they seem to "prove" opposing viewpoints.

Hereward

RHemmings wrote:

I disagree. For any statistic to be valid it has to be a "reasonably" unbiased sample from the population that the statistic is intended to describe. I can't see how you say "valid" about a statistic from an unrepresentative sample. Can you please say what your definition of "valid" is, and what you mean by "stand alone statistic"?

I also can't see how a statistic can be "utterly meaningless" but also "valid". Can you please explain further?

A quick survey of ten people about if it is better to rent or buy could throw up some interesting, and perfectly valid, statistics. For example, if all ten people owned hoses you are more likely to discover than buying is better than renting; conversely, if all ten were renters then you might discover that renting is better than buying. Both of these statistics would be equally valid, but they may not be a true representation of the population (hence being utterly meaningless) as a whole: this is where statistical analysis comes in. Statistical analysis would try to demonstrate how reflective of the whole population your sample is.

Hereward

RHemmings wrote:

In theory, yes. But has this ever happened? Over a long period of time? In any market?

What would you consider evidence that a bubble has formed? I would say prices increasing over and above general inflation with it being clear that speculative investment, and an expectation that prices will continue rising justifying much of this investment, is the prototypical "evidence of a bubble".

Food is a good market to demonstrate large prices rises (during WW2) followed buy a long period of small, to no existent, price rises.

If you consider a bubble to form if prices rise faster than general inflation, then there must be bubbles forming in many markets, as general inflation is a measure of the average change in prices.

RHemmings

tomstickland wrote:

AlanM is dead right about statistics. On their own they can't lie - they just tell you what they tell you. The way to manipulate them is to just decide measure things that suit your argument.

I strongly disagree. It's when statistics are presented without context (e.g. describing the method of sampling), "on their own", that they lie most effectively.

Consider the well known book "How to lie with statistics".

http://www.amazon.co.uk/How-Lie-Statistics-Penguin-Business/dp/0140136290

This page isn't so good, but mentions a few ways that statistics can be used to mislead.

http://faculty.washington.edu/chudler/stat3.html

tomstickland

It depends on what you thought I was implying about the lying. I think I was actually agreeing with you more than you realise. You can't change the mean of set of data, but you can decide to be very careful about what you measure and how you do it in the first place.

RHemmings

Hereward wrote:

Food is a good market to demonstrate large prices rises (during WW2) followed buy a long period of small, to no existent, price rises.

If you consider a bubble to form if prices rise faster than general inflation, then there must be bubbles forming in many markets, as general inflation is a measure of the average change in prices.

Were people investing in food? As I said in my post, the speculative component and in particular price rises creating an expectation of further price rises is a necessary component of a bubble. I would expect that the price of food rising during WWII and remaining static would be due to the laws of supply and demand.

The same point answers (I believe) your second point about me claiming that bubbles are forming in many markets. But I'll go further and say that for a bubble to form, prices must have risen considerably more than would be expected considering fundamentals.

RHemmings

tomstickland wrote:

It depends on what you thought I was implying about the lying. I think I was actually agreeing with you more than you realise. You can't change the mean of set of data, but you can decide to be very careful about what you measure and how you do it in the first place.

I see. But "on their own" is perhaps a bit confusing, given that what you need to understand and evaluate statistics properly is additional information and context. For example, the official USA statistics on house prices exclude properties with a mortgage of more than US$417,000. I'll leave it as an exercise for the reader to work out what that can do to the measurement of average house prices when prices rise and fall in areas with significant numbers of houses costing more than $417,000.