Jump to content
House Price Crash Forum
Sign in to follow this  
Extradry Martini

A Study Of House Price Mean Reversion

Recommended Posts

I have a chart here, which I would love to put in the body of this post, but I don’t know how, so I have attached it instead….

You all know the Nationwide graph showing house prices adjusted for inflation reverting to a slowly rising trend (from which point we are a very long way away). A few years back, I saw a chart produced by Dresdner Bank which showed UK house prices adjusted for take home pay since 1945*.

Using data from Nationwide itself, the ONS and a document called “British Labour Statistics, Historical Abstract, 1886-1968", I have been able to produce something similar in the attached graph. It is house prices expressed as a multiple of average earnings including bonuses since 1953. I believe there are 3 things to note here:

1. That house prices are ultimately a function of pay and mean revert to a multiple of pay (as we always suspected)

2. That were prices to mean revert to fair value from here (at roughly 4 times), they would fall by around 42% relative to earnings.

3. That after a period of higher-than-mean prices, prices fall to below the mean and the higher they rise above the mean, the lower they subsequently fall below it. So, were prices to fall to around 3 times average earnings, they would fall, relative to earnings, by about 57%.

OK, so how much are house prices going to fall in nominal terms – i.e. how much are earnings going to rise while the market falls? I think the starting point here is the MPC. We know that they target 2% inflation and that despite some recent wobbles (and gross incompetence in another area of the central bank), they have performed more or less ok in this respect, so we can assume inflation to be around 2%.

The next question is by how much will wage growth exceed inflation? The last data for this is 2%, which is roughly the middle of the recent (0.5%-3%) range. However, in an economic slowdown, it tends to slow to 1% over inflation (as id did in 1989-1994 and 2000-2002). So, we can expect average earnings to be about 3% over the period in which house prices fall.

But how long will that be? In the two previous peaks, the return to fair value took 2 years. However, they were both during times of very high nominal interest rates. Despite the (partially valid) argument that increased leverage means that interest rates are effectively high once more, my belief is that this will be a function of how quickly inflation falls: If inflation stays low then the MPC can be quite aggressive in cutting rates while the economy slows. This will have the effect of drawing out the bear market longer than previously. So let’s say it takes 50% longer to return to fair value – i.e. 3 years. Compounding the 3% annual earnings increase over 3 years, we get 9.27%, meaning that house prices fall 35% in nominal terms while they are falling 42% in relative terms.

How long the market then takes to fall to the low is anyone’s guess, and the total fall in nominal terms may not be much more than 35%, but remember that last time house prices “crashed”, they only fell by 15% in nominal terms.

(* It was interesting that in the Dresdner chart there was another peak at above 5 times earnings in the 1945-1950 period which then fell back to where the attached chart begins – if anyone can give me pre-1953 house price data I can incorporate it)

Lastly, I'd like to say thanks to Si et al from the old FT forum days - nice to hear you're all doing so well!

Share this post


Link to post
Share on other sites
I have a chart here, which I would love to put in the body of this post, but I don’t know how, so I have attached it instead….

You all know the Nationwide graph showing house prices adjusted for inflation reverting to a slowly rising trend (from which point we are a very long way away). A few years back, I saw a chart produced by Dresdner Bank which showed UK house prices adjusted for take home pay since 1945*.

Using data from Nationwide itself, the ONS and a document called “British Labour Statistics, Historical Abstract, 1886-1968", I have been able to produce something similar in the attached graph. It is house prices expressed as a multiple of average earnings including bonuses since 1953. I believe there are 3 things to note here:

1. That house prices are ultimately a function of pay and mean revert to a multiple of pay (as we always suspected)

2. That were prices to mean revert to fair value from here (at roughly 4 times), they would fall by around 42% relative to earnings.

3. That after a period of higher-than-mean prices, prices fall to below the mean and the higher they rise above the mean, the lower they subsequently fall below it. So, were prices to fall to around 3 times average earnings, they would fall, relative to earnings, by about 57%.

OK, so how much are house prices going to fall in nominal terms – i.e. how much are earnings going to rise while the market falls? I think the starting point here is the MPC. We know that they target 2% inflation and that despite some recent wobbles (and gross incompetence in another area of the central bank), they have performed more or less ok in this respect, so we can assume inflation to be around 2%.

The next question is by how much will wage growth exceed inflation? The last data for this is 2%, which is roughly the middle of the recent (0.5%-3%) range. However, in an economic slowdown, it tends to slow to 1% over inflation (as id did in 1989-1994 and 2000-2002). So, we can expect average earnings to be about 3% over the period in which house prices fall.

But how long will that be? In the two previous peaks, the return to fair value took 2 years. However, they were both during times of very high nominal interest rates. Despite the (partially valid) argument that increased leverage means that interest rates are effectively high once more, my belief is that this will be a function of how quickly inflation falls: If inflation stays low then the MPC can be quite aggressive in cutting rates while the economy slows. This will have the effect of drawing out the bear market longer than previously. So let’s say it takes 50% longer to return to fair value – i.e. 3 years. Compounding the 3% annual earnings increase over 3 years, we get 9.27%, meaning that house prices fall 35% in nominal terms while they are falling 42% in relative terms.

How long the market then takes to fall to the low is anyone’s guess, and the total fall in nominal terms may not be much more than 35%, but remember that last time house prices “crashed”, they only fell by 15% in nominal terms.

(* It was interesting that in the Dresdner chart there was another peak at above 5 times earnings in the 1945-1950 period which then fell back to where the attached chart begins – if anyone can give me pre-1953 house price data I can incorporate it)

Lastly, I'd like to say thanks to Si et al from the old FT forum days - nice to hear you're all doing so well!

My big fat bottom they only fell 15%

Share this post


Link to post
Share on other sites

Well, it just adds fuel to the theory that its lax lending that has led to this HPI- however- if lax lending is to be continued, dont expect a reversion to mean anytime soon

Share this post


Link to post
Share on other sites

This is comparing average earnings against house prices. Traditionally only 1 parent worked, however in a lot more households now, both parents are working, so the mean level has been pushed up. I don't know what the new mean level would be, but we're certainly along way above it.

Share this post


Link to post
Share on other sites

a better graph would be to compare average house hold take home to the average cost of servicing an average mortgage. Interest rates, Miras, duel incomes and houseprices all play a part when you want to display how much money (as a percentage of income) is left at the end of the month over the past couple of decades:

currently just over 30% of household average take home is needed to service an average mortgage. We really need interest rates to double to match previous levels of unaffordability.

2_25nw.jpg

post-552-1191505589_thumb.jpg

Edited by moosetea

Share this post


Link to post
Share on other sites
This is comparing average earnings against house prices. Traditionally only 1 parent worked, however in a lot more households now, both parents are working, so the mean level has been pushed up. I don't know what the new mean level would be, but we're certainly along way above it.

My mother worked when I was a child and so did the mothers of many of my friends, ok many of them were part time, and now mothers are more likely to work full time I agree (but by no means all of them).

BUT

Is the phenomenon of working mums the cause or effect of high house prices?

I don't think we can know for sure.

Share this post


Link to post
Share on other sites
a better graph would be to compare average house hold take home to the average cost of servicing an average mortgage. Interest rates, Miras, duel incomes and houseprices all play a part when you want to display how much money (as a percentage of income) is left at the end of the month over the past couple of decades:

currently just over 30% of household average take home is needed to service an average mortgage. We really need interest rates to double to match previous levels of unaffordability.

Interesting to observe how quickly this ratio shot through the roof during the last house price boom in the 80's, obviously because of interest rates rising to 15%. This was after a sustained period hovering at around 30%, which is where we are now....

I'm interested to know what effect on your trend line would small or medium interest rate rises have now?

Or are there any other variables that are likely to occur today that would also have a dramatic effect on this ratio?

Share this post


Link to post
Share on other sites
This is comparing average earnings against house prices. Traditionally only 1 parent worked, however in a lot more households now, both parents are working, so the mean level has been pushed up. I don't know what the new mean level would be, but we're certainly along way above it.

No, that argument is a myth. Were it to be true we would have seen a huge increase in the labour force, which we have not. The total number in work in the UK has increased by 19% since 1960. Even if there had been no immigration, and all that increase was taken up by people living in houses with another employed person, it would still get nowhere near the increase you describe. It’s one of those fallacies (much like the one about the UK tending to buy more houses than other Europeans) that market bulls trot out every so often…

Share this post


Link to post
Share on other sites
a better graph would be to compare average house hold take home to the average cost of servicing an average mortgage. Interest rates, Miras, duel incomes and houseprices all play a part when you want to display how much money (as a percentage of income) is left at the end of the month over the past couple of decades:

currently just over 30% of household average take home is needed to service an average mortgage. We really need interest rates to double to match previous levels of unaffordability.

The true affordability of a mortgage is versus real interest rates (or, more accurately, wage-adjusted interest rates). If you use nominal rates, all you are doing is comparing the first month of the mortgage, not the mortgage over its term. In the 1980’s everyone’s salary went up because of high inflation, making all successive mortgage payments a lot easier. Real interest rates are actually higher now than they were in 1989…

More broadly though – are you actually making the point that you think house prices will continue to rise?

Share this post


Link to post
Share on other sites
Interesting to observe how quickly this ratio shot through the roof during the last house price boom in the 80's, obviously because of interest rates rising to 15%. This was after a sustained period hovering at around 30%, which is where we are now....

remember that interest rates are not the only factor in play in 1989 - there was a recession which meant a lot of folk were out of work. Our 30% figure could shoot up again if such factors come into play even without interest rises.

Share this post


Link to post
Share on other sites
Actually it makes a lot of sense once you see it in real terms…. Anyway, the point is that the next one is going to be at least twice as bad.

You said prices fell 15% in *nominal* terms though, not real.

Prices in many parts of the UK fell by 30/40% in nominal prices.

Share this post


Link to post
Share on other sites
You said prices fell 15% in *nominal* terms though, not real.

Prices in many parts of the UK fell by 30/40% in nominal prices.

Yes - and I saw some fall by 70% - but I think two things were happening: Firstly, some areas barely moved (K&C in %age terms had barely risen so didn't fall that much), and secondly there was a temporal shift - prices were still rising in some of the provinces after they had begun to fall in London. Once London starting going up again, prices outside the capital were still falling. This meant that the averages show more shallow falls from peak to trough than the regional data would. Either way, when I said it made sense on a wage adjusted basis, I meant that the fall didn’t look too out of line with other periods, and that I trust the data (or at least as averages). It is possible of course that same thing happened in the previous periods, distorting the data there as well.

Share this post


Link to post
Share on other sites
a better graph would be to compare average house hold take home to the average cost of servicing an average mortgage. Interest rates, Miras, duel incomes and houseprices all play a part when you want to display how much money (as a percentage of income) is left at the end of the month over the past couple of decades:

currently just over 30% of household average take home is needed to service an average mortgage. We really need interest rates to double to match previous levels of unaffordability.

The graph shows an 'unaffordability' ratio much higher at the last crash, a statistic I find v. difficult to believe.

For an 'average' 150K mortgage it would suggest take-home average of 3K per month - I definitely disbelieve that. ;)

Average pay is a notoriously unreliable statistic as >75% of people earn below it (skewed by high earners). What I want to see is MEDIAN household take-home pay versus median mortgage cost for FTBs with 90% or higher LTV. As an affordability index, who cares about the average cost for those who bought 15 years ago, or OO on their 4th trade-up - that's relatively meaningless.

Share this post


Link to post
Share on other sites

The nominal price fall during the 1989-95 was 20.2% not 15%, from a nominal high of £62,782 in the third quarter of 1989, to a nominal low £50,128 in the first quarter of 1993.

There's four statistical problems with your chart. Firstly it's household income that counts, not personal income. The last twenty-five years have seen a significant increase in the number of working women, if that continues we could reasonably expect to see a permanent change in house prices relative to income. Furthermore women's pay is still behind men's pay in absolute terms, if this continues to correct then we could see household income increasing faster than average male pay.

The second issue is that incomes are only half the equation, interest rates and availability of mortgages are equally significant, after all that's what people are using their incomes for. So if a £200k mortgage is made cheaper through low interest rates then it can offset a higher house price.

The third issue is BTL, about a third of recent property sales have been to BTL landlords. What determines their willingness to buy? There's no definitive answer to this question but I guess that their income isn't a primary factor. And if they exit the market that would precipitate a crash no matter where house prices are in relation to incomes.

Finally to revert to the mean implies a fairly constant level of owner occupancy. This went from about 11% in 1900 to over 70% today, but who knows what it will be in the future? UK owner occupancy is already well above French or German levels, but behind Spanish or Irish levels, which of these is the benchmark norm that we'll evolve towards?

I agree that there will be some connection over the long term between income and house prices. But it's not as precise a linkage as you suggest, and there's still have some big unknowns. What will interest rates and mortgage availability be in the future, what will happen to rents, will the percentage of two income households continue to rise, and will women's pay continue to rise faster than mens?

Share this post


Link to post
Share on other sites
Yes - and I saw some fall by 70%

Slightly tangential, but nice to see someone else bringing up some of the more extreme drops from last time. My parents road had recorded sales for detached four-beds before, during and after of 58k, 175k, and 75k. I was starting to wonder whether we were living in freakland at the time, as the change in prices experienced in their area (a genuinely nice area) seemed so far away from the ickle humpette shown on the national average graph.

Share this post


Link to post
Share on other sites
The true affordability of a mortgage is versus real interest rates (or, more accurately, wage-adjusted interest rates). If you use nominal rates, all you are doing is comparing the first month of the mortgage, not the mortgage over its term. In the 1980’s everyone’s salary went up because of high inflation, making all successive mortgage payments a lot easier. Real interest rates are actually higher now than they were in 1989…

More broadly though – are you actually making the point that you think house prices will continue to rise?

in a nutshell yes its possible we could see more rises, if the government doesn't create or suffer a massive economic shock that doubles the cost of servicing a mortgage within a couple of months. If the cost of housing is managed and increased and credit isn't totally wiped up, i think the population can manage being boiled alive for a few more years before we see a Japanese style decline for 15 years.

The graph i posted is accurate although i appreciate everyone not trusting it, it was reproduced by a forum member from data many months ago but i cant find the thread. The cost of servicing a debt is down to lenders and the interest they charge, mortgage lending was traditionally higher than the base rate.

At the end of the 80s the final bit of the boom you see in the graph was caused by a policy change:

1. removal of double MIRAS meant you had to pay more tax after it was removed

2. there were no buyers after miras was removed as everyone had rushed to buy before the policy change

3. Interest rates had been risen to combat rising prices

4. Houseprices had risen rapidly because of people panic buying

the result was the last crash. IMHO the powers that be learnt that MIRAS was a very stupid thing to do and if it was repeated today we would get a uturn (housing in a SIPP was uturned for exactly that reason, the government through it would create a boom, which would in turn create a crash)...

Personally i think we need to see IRs at 10->12%, but the current credit crunch could do it IF and only IF most lending dries up and we see falling wages (average 20->40%)

Edited by moosetea

Share this post


Link to post
Share on other sites
The nominal price fall during the 1989-95 was 20.2% not 15%, from a nominal high of £62,782 in the third quarter of 1989, to a nominal low £50,128 in the first quarter of 1993.

This is the only part of your post which is correct - though I'm not sure what you are trying to prove. Anyway, apologies for getting it wrong earlier.

There's four statistical problems with your chart. Firstly it's household income that counts, not personal income. The last twenty-five years have seen a significant increase in the number of working women, if that continues we could reasonably expect to see a permanent change in house prices relative to income. Furthermore women's pay is still behind men's pay in absolute terms, if this continues to correct then we could see household income increasing faster than average male pay.

Not true - see my earlier post in this thread on this subject.

The second issue is that incomes are only half the equation, interest rates and availability of mortgages are equally significant, after all that's what people are using their incomes for. So if a £200k mortgage is made cheaper through low interest rates then it can offset a higher house price.

Temporarily true, but not over time - that's the whole reason why they mean revert (and why I started this thread).

The third issue is BTL, about a third of recent property sales have been to BTL landlords. What determines their willingness to buy? There's no definitive answer to this question but I guess that their income isn't a primary factor. And if they exit the market that would precipitate a crash no matter where house prices are in relation to incomes.

Finally to revert to the mean implies a fairly constant level of owner occupancy. This went from about 11% in 1900 to over 70% today, but who knows what it will be in the future? UK owner occupancy is already well above French or German levels, but behind Spanish or Irish levels, which of these is the benchmark norm that we'll evolve towards?

These two points contradict each other – think about it....

Share this post


Link to post
Share on other sites
Trying to re-attach the attachment now...

A trendline through these points would be useful... and would probably show "a rising mean", as it were, which brings reversion into question.

Share this post


Link to post
Share on other sites
There is no such thing as reversion to the mean. There is no 'elastic band' pulling prices or anything else back to the mean. This is basic maths. End of story.

Er…. ok, I’ll be gentle here: The reason why they mean revert is that they hold a constant relationship over the very long term with wealth. Wealth is what we use to buy things, so it ought not to be that surprising that the value of something like housing (which does not depend in any meaningful way on production costs) is governed by wealth.

Share this post


Link to post
Share on other sites
There is no such thing as reversion to the mean. There is no 'elastic band' pulling prices or anything else back to the mean. This is basic maths. End of story.

When a term exists that many people understand, then denying that existence is pointless.

Reversion to the mean isn't necessarily a mathematical construct, just an observed tendency in many walks of life.

Example: Take a 100 people with normal blood pressure mean 120/80. Add (inflate) it by giving some of them stressful situations. Mean rises. Take stress away, blood pressure reverts to mean.

Common concept, observed everywhere in all sorts of situations. It exists though you may disagree with it's application to HPI.

End of story. ;)

Share this post


Link to post
Share on other sites
When a term exists that many people understand, then denying that existence is pointless.

Reversion to the mean isn't necessarily a mathematical construct, just an observed tendency in many walks of life.

Example: Take a 100 people with normal blood pressure mean 120/80. Add (inflate) it by giving some of them stressful situations. Mean rises. Take stress away, blood pressure reverts to mean.

Common concept, observed everywhere in all sorts of situations. It exists though you may disagree with it's application to HPI.

End of story. ;)

Is that really what he meant - that nothing mean reverts ever? Amazing…. Even the people in the BTL forums don’t say things as crass as that....!

Share this post


Link to post
Share on other sites
When a term exists that many people understand, then denying that existence is pointless.

Reversion to the mean isn't necessarily a mathematical construct, just an observed tendency in many walks of life.

Example: Take a 100 people with normal blood pressure mean 120/80. Add (inflate) it by giving some of them stressful situations. Mean rises. Take stress away, blood pressure reverts to mean.

Common concept, observed everywhere in all sorts of situations. It exists though you may disagree with it's application to HPI.

End of story. ;)

The blood pressure example is not relevant but even so I'll try to answer. The 120/80 mean is made of 100 people who have their own individual resting blood pressures. Some will have higher and some will have lower readings than the mean. What about someone with a 100/60 blood pressure. Add a bit of stress, oh dear his reading goes up to 110/70. Take stress away and it falls back to 100/60. Doesn't look like he's reverting to the mean does it?

Edited by youthoftoday

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.

  • 354 The Prime Minister stated that there were three Brexit options available to the UK:

    1. 1. Which of the Prime Minister's options would you choose?


      • Leave with the negotiated deal
      • Remain
      • Leave with no deal



×
×
  • 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.