I’m a full time stock market investor for good reason, but it has to be recognised that there are several major advantages to owning your own home and that fortuitous timing when buying your first one, when upsizing and when downsizing can make a significant difference to your wealth.
This article has been inspired by the long running website and forums at housepricecrash.co.uk, and the following chart:
Two possible takeaways from the chart are:
- House prices are not currently overvalued. Indeed, they appear to be undervalued to a similar extent they were in the mid 1990s trough that followed the late-80’s asset value boom.
- House prices have offered incredible returns over the last nearly 50 years, with average prices growing from £10,388 to £258,115. Although the small print shows this is merely equates to an RPI+2.4% growth rate, the availability of high LTV, uncallable, unmargined and very cheap debt makes it a unique investment proposition.
But, I see two major problems with the above graph.
Problem – Linear rather than Log Scale
The linear scale exaggerates the peaks and toughs in later years – was 2007 really as over valued as 1989? Is the apparently undervaluation now really as great as 1996?
The graph below uses the same data as the above, but with an logarithmic scale and also with a trend line calculated by Google Sheets superimposed:
2007 vs 1988
From the logarithmic scale it is clear the the 2007 overvaluation was actually worse than in 1988, but it is the subsequent crash than differs the most.
Many argue the 1989 overvaluation was driven by a race to qualify for double MIRAS, although the timing doesn’t quite seem to match, however the subsequent crash was very clearly driven by lower affordability and higher unemployment as the UK base rate doubled from 7.5% in May 1988 to 15% in October 1989. The 1992 sterling crisis (aka Black Wednesday) led to a further lurch downwards (highlighted on the chart) and a general loss of confidence in the property market let to prices flatlining in nominal terms (falls in real terms) for several years even as interest rates fell.
The 2007 overvaluation is generally blamed on excessively lax mortgage lending exemplified by the 125% mortgage. The 2007-8 financial crisis hit the housing market at first due to the loss of confidence, withdrawal of mortgage products and higher unemployment, but the interest rate background was very different, with the base rate falling from 5.75% in July 2017 to 0.5% in March 2009 and staying thereabouts until last year.
2022 vs 1988
From the above it should be clear that while confidence drives house prices in the short term, it is interest rates that drives them in the medium term, and that therefore the current situation is more akin to the aftermath of 1988 than of 2022. The first leg down (highlighted) appears to be a mixture a) the boom simply running out of steam akin to 1988, b) interest rate rises, again like 1988, c) loss of confidence most akin to 1992, and d) temporary withdrawal of mortgage products akin to 2007. The cost of living crisis was not seen in 1988 or 2007 but balancing this is full employment with pay rates of many unskilled and semi-skilled jobs broadly keeping up with inflation.
So far in this cycle the level of apparent undervaluation is less than that seen in 1996 and the lesson from history appears to be that, at best, prices will be flat in nominal terms for the next 5 years. As I have frequently said on Small Caps Live, my opinion is that the BoE have repeatedly and clearly signalled that the medium term inflation target is around 4.5-5.0% and so nominally flat implies significant real-terms falls.
Problem – Inflation rather than Average Earnings
The next problem with housepricecrash.com’s chart is the use of inflation. The BoE have told us we need to accept that we’re all poorer as a result of recent high inflation, so why should we expect house prices to keep pace with inflation? In a market characterised by long-term undersupply where the median behaviour is to buy the most expensive property you can afford, it seems to me more logical that long-term price trends would reflect long-term trends in affordability, and I argue the best measure of this is household disposable income.
It is important that the figures used are per household because that is what mortgage affordability is effectively based on and because the use of average earnings per person would introduce a significant long term distortion due to the strong and long-term trend for increases in income from a second household earner.
It is also important that disposable income is used because there has been a long term shift from direct to indirect taxation (e.g. income tax and National Insurance to VAT, IPT and also Council Tax) which has caused a sustained net increase disposable income available for housing
There are many other factors in affordability, including nominal interest rates (determining initial monthly payments), real interest rates (determining total interest cost) and the long term trend of a decreasing proportion of earnings spent on other essentials such as food and utilities (despite recent reversals). Counteracting these have been an above inflation increase in labour intensive expenditure such as childcare. The taxation shift and these other factors are probably behind the long term movement in lending multiples from the traditional 3.5x first salary + 1x second, to 4.5x joint income as standard, with up to 5.5x currently available.
Although there are a large number of factors involved, and my suspicion is that it understates somewhat the natural trend, I am confident that household disposable income is a significantly better trend indicator for house prices than inflation. This is especially when comparing valuations now with those 15 years ago given the historically weak average earnings growth during post-financial crisis austerity. As earnings have historically risen faster than RPI, using the former should require less of a “magic fiddle factor” than RPI when deriving the trend line.
Handily the ONS provides real-terms disposable income figures found here. As suspected, the effect of austerity is very visible in the graph of real earnings. More recent data is not yet available and so I will assume that real household disposable income fell 3% over the last year.
Income distribution is also a factor. One argument is that the income of those likely to be owner occupiers should be used, perhaps then excluding the bottom quartile and being closer to the mean than the median in this case. However house prices are also set by the rent than non homeowners can afford to pay residential landlords. Therefore I will use the median value.
Conclusion
The graph above shows an unexpectedly strong correlation between average household income and house prices over shorter periods with clear dips in the former corresponding to every crash within the time period and also various other minor features in the period from 2009 onwards. As ever, a simple causal relationship is not a credible hypothesis and it is likely that 1) house prices have a speculative element that causes them to move in an exaggerated manner ahead of expected changes in incomes, and 2) incomes are affected by a downturn in house moves and house building that drives significant parts of the economy, and also 3) a general economic downturn drives both falls in average incomes and house prices to some independent extent and with different degrees of lag.
The above graph also shows particularly clearly how, since the financial crisis, average earnings growth has fallen behind a simple fitting of a long term trend to house prices, suggesting that house prices may now be less affordable than the graphs looking at inflation + 2.4% suggest.
One further unavoidable flaw in the original graph is that the long term trend line for house prices is fitted to the prices starting and ending at an arbitrary time. Using today as the end date is particularly problematic and every quarter of weak growth going forward will pull the trend line downwards such that in five or ten years time its position as at 2023 Q1 could be very different. As an example see the difference between the red and violet lines in the left-hand graph below.
In these final two graphs I model house prices (and average earnings) falling in line with conditions in the early 1990s, assuming we are (A) at the point in the cycle of 1991 Q1 with another ERM-type shock to come, and (B) that the abortive Kwarteng budget was analogous to ejection from the ERM in September 1992.
For the very little it is worth, I think:
- House prices are currently at or just below the true trend line taking into account affordability and the limitations of the measure selected
- The future direction of house prices and earnings is likely to fall somewhere between projection A and projection B
- The best time to sell is now and the best time to buy is in the range Q4 2025 to Q1 2028
- The most important determinant of which projection is closer to the truth is whether a sustainable round of interest rate cuts commences in mid-late 2024.
Data and graphs are available here.
LeoInvestorUK's blog 