As early as 2009, commentators have been asking whether Hong Kong property is a bubble. Since then, the debate has continued as, apart from a brief pause in the second half of 2011, prices have continued to climb inexorably upwards; despite efforts by the Hong Kong government to cool the market. Prices have more than doubled from their recent low point at the end of 2008. In fact, for most people, whether or not the property market is a bubble is not in doubt. The question now being asked is when the bubble will burst.
In a series of posts, I plan to take a detailed look at the Hong Kong property market, to try to assess whether it is overvalued and, if so, by roughly how much. I will primarily focus on the residential property market. Specifically, this is relevant to my investment Henderson Land (12:HK) and potentially other future investments. In this first post, I will look at the background and, in particular, how prices compare to historical trends. In subsequent posts, I will look in more detail at the reasons for the recent run-up in prices and current valuations.
Price trends and asset bubbles
Just looking at Chart 1, it is easy to conclude that Hong Kong property must be a bubble, however you choose to define it. Prices recovered quickly after the financial crisis and have more than doubled since then. Looking a bit further back, prices have increased by around 300% since their cyclical low point in the middle of 2003, during the SARS outbreak. Longer term, the market has been a real rollercoaster ride: property prices increased strongly from the mid-1980s up to the previous peak in 1997, before falling almost continuously for a six-year period following the Asian financial crisis, losing two-thirds of their value. After 1997, Hong Kong suffered five years of deflation, when consumer prices fell by around 16% in total.
One thing is clear, however: if Hong Kong property is a bubble, it is not just confined to the residential market. Prices of both office and retail property have risen even faster than residential property (see Chart 2), increasing by more than 500% over the last decade, and the rental yield in all three markets is now below 3%. This makes it hard to argue that the recent run-up in prices is solely due to buyers’ irrationality, given that investors in commercial property would perhaps be expected to take a more value-orientated approach.
There is no agreed definition of what constitutes an asset bubble. Some commentators focus on the increase in price relative to fundamental value, or on speculative behaviour. Others claim that bubbles are impossible to identify until after they have burst. Some economists even deny that they occur at all, as the very idea contravenes the notion of efficient markets. From a practical perspective, a useful starting point is the widely-quoted definition propounded by Jeremy Grantham, chief investment strategist at asset manager GMO, that a bubble represents an increase of more than two standard deviations above an asset’s long-term trend level. Looking at it another way, if price movements were completely random, such an event should only occur roughly once every 44 years.
The above charts, however, only show nominal prices and do not compare prices to a trendline. Furthermore, the fact that price rises are exponential can be misleading: it inevitably results in a sharply upward-sloping line, even when the growth rate is constant. To overcome this, Chart 3 below shows the same data as Chart 1 but using a log scale. I have also added a trendline, which represents annual growth of around 7% per year. On this basis, property prices are only around 19% above their long-term trend, compared with 146% at their peak in 1997 and 43% below trend in 2003.
Superficially, at least, the trendline might seem like a reasonably good fit. However, it is difficult to argue that property prices actually follow such a constant growth trendline. All this shows is that, ex post, since 1979 (the earliest date for which the index is available), residential property prices in Hong Kong have grown at around 7% per year.
The limitations of a fixed growth trendline can be seen more clearly in Chart 4, which shows the index rebased to the trendline. There is no logical reason why property prices should grow at a constant rate over several decades, given changes in demographic and economic factors, such as population growth, inflation, interest rates and economic growth. Before the 1997 Asian financial crisis, economic growth and inflation in Hong Kong were much higher than afterwards; using a constant growth trendline therefore exaggerates the degree of overvaluation in 1997.
What might represent a better trendline? Potential candidates would be inflation, income growth and growth in the economy. Chart 5 shows the residential price index rebased relative to consumer prices, nominal GDP and nominal GDP per capita. Whilst each of these provides a better fit than using a fixed growth rate trendline, GDP per capita provides the best fit over the entire period, although, given the decline in population growth in recent years, there is not a big difference with nominal GDP more recently. Based on this, residential property is currently around 60% above its trend level, almost identical to prior peaks. Interestingly, this figure is not particularly sensitive to the starting date chosen, or whether recent data is excluded. As with Chart 4 above, the market’s cyclicality means that prices tend to overshoot, cutting through the trendline and spending almost no time “on trend”.
Ominously, on both previous occasions when prices have been this far above trend before, they have fallen rapidly soon afterwards, relative to their trend, overshooting on the downside. Nevertheless, any conclusion that prices will inevitably fall in the near future should be treated caution. Although there is more than 30 years of price data, given the long length of each cycle, there are not enough previous examples to draw any firm conclusions.
There are a number of reasons why residential property prices might in the long-term grow in line with nominal GDP or GDP per capita. Nominal GDP, which is obviously a measure of economic growth, may be a logical benchmark, particularly where land availability is tightly constrained. GDP per capita is generally a good proxy for income; perhaps surprisingly, it actually provides a better fit for property prices than household income, albeit over a shorter time period, from the early 1990s. Furthermore, unlike a fixed growth rate, they are not just arbitrary ex post measures of historic growth.
Additional support can also be found from the UK housing market, where, similar to Hong Kong, land supply is also restricted. UK house price data is available over a longer period (since 1953 for the Nationwide House Price Index). Chart 6 shows the UK Nationwide House Price index since the mid 1950s rebased to nominal GDP and GDP per capita. Even with the more than 50 years of data available in the UK, any pretence that a suitable long-term trendline can be determined with any degree of precision is illusory. For the UK, nominal GDP actually provides a slightly better fit. UK house prices are currently somewhat less above trend using nominal GDP than GDP per capita: by only 8%, compared with 21%, mainly due to the acceleration in UK population growth since 2000.
The annualised standard deviation of quarterly price changes in Hong Kong, relative to a GDP per capita trendline, is around 11%. Based on this figure, property prices are currently more than five standard deviations above their long-term trend. However, quarterly price changes show strong serial correlation of around 0.6: i.e. house prices trend, so that an increase in one quarter tends to be followed by a further increase in the next. This is an inevitable consequence of such a slow-moving, illiquid market, but it does mean that in reality the market is more volatile than short-term prices movements might suggest. The calculated annual standard deviation is around 14%. Annual prices changes are also serially correlated, although the correlation is weaker at around 0.3. The cyclicality in Hong Kong property prices is a result of this trending.
Another consequence is the distribution of annual property prices changes, which are shown Chart 7 for the period since 1980. This might sound like a truism but property prices in Hong Kong tend to be either going up or going down. In contrast, the distribution of price changes over shorter periods more closely resembles the standard, normal distribution.
Divergence from rents
The second argument for Hong Kong property being a bubble is that, in recent years, prices have increased much faster than rents. In general, the overall costs of renting a property should not be substantially different from the costs of ownership. Taking into account the required investment return for an owner, there should be an equilibrium position, where neither alternative is obviously more attractive than the other. After all, a property investor receives income in the form of rent. Ultimately, property prices should be a function of rental income, although higher prices can lead to an increase in rents. Nonetheless, over the long-term, it seems reasonable to expect rents and prices to move broadly in line with each other.
From Chart 8, it can be seen that residential property prices and rents have generally moved together, although, not surprisingly, prices have been significantly more volatile. However, whilst short-term changes in prices and rents have been strongly correlated (the correlation in annual changes was 0.80), over the period as a whole, rents have grown at a much slower rate than prices. Chart 9 below shows prices rebased to rents.
Chart 9 shows property prices rebased to rents. This is in effect a measure of property valuation; and is equivalent to the price-earnings multiple for stocks (although obviously it is not scaled as an actual multiple). Over the period as a whole, there has been a clear uptrend and property prices relative to rents have more than doubled during this time. There was a big step change in 1991, when prices increased rapidly. The relationship between prices and rents was then relatively stable between 1992 and 2008. However, since 2009, prices have again increased much more quickly than rents, which were slower to recover after the financial crisis. In the absence of any stable level to which prices have historically tended to mean revert, it is quite difficult to suggest how current prices compare relative to rents. What can be said with confidence is that, relative to rents, prices are at their highest level since at least the early 1980s; and 25% above the previous peak in 1997. Prices are also around 60% above the average level during the relatively stable period between 1992 and 2008, although whether this can be regarded as a long-term trend level is debatable. This is not too far off the figure of 69% that The Economist estimated Hong Kong prices were overvalued in its most recent survey of global housing (12 January 2013), based on long-run average ratios between prices and rents.
The inverse of prices relative to rents is the rental yield. Chart 10 shows the rental yield on a typical small flat (Class B, which have a total floor area of 40-70m2). Not surprisingly, it is the mirror image of Chart 9 and shows a long-term decline in rental yields to below 3% currently. The obvious explanation for the fall in rental yields is the decline in interest rates during the period (I intend to look at the impact of interest rates in more detail in future posts). The rental yield has averaged just over 6% for the period as a whole. During the relatively stable period between 1992 and 2008, the yield varied within a range of 3.6% to 5.4%, averaging 4.6%.
Whilst rents have not increased as fast as prices over recent decades, it is apparent from Chart 11 that they have matched inflation quite closely during this time. This might, of course, just be a coincidence, although some link between the two might be expected, given that rental costs make up a substantial part (around 30%) of the consumer price index. Chart 11 also shows that since the late 1990s, rents have increased in line with GDP per capita too. Nonetheless, this still does not explain why, if there had been a step up in property valuations during this period as a result of lower interest rates, rents did not increase in line with GDP per capita during the earlier part of the period and prices increase at an even faster rate.
Affordability: prices and incomes
Finally, in addition to the magnitude of recent price increases and the divergence between prices and rents, the third major argument for a bubble is the decline in affordability. In the latest annual Demographia International Housing Affordability Report, Hong Kong was the least affordable of all the markets covered: more expensive than Vancouver, Sydney, London and San Francisco. The median price for Hong Kong property was 13.5x the median gross household income (where a multiple of more than 5x is classed as “severely unaffordable”). Vancouver, the next most expensive, had a multiple of 9.5x. (As an aside, it may also be just a coincidence that the most expensive cities are also amongst the most popular overseas markets for mainland Chinese buyers.)
Such aggregate ratios can, however, be misleading. The structure of the property market in Hong Kong is very different from other countries. In Hong Kong, only around 36% of households live in private owner-occupied housing, very low by international standards. This low level of private home ownership, and the fact that close to half of all households live in subsidised public housing, clearly reflects the high property prices. However, the level of ownership has not changed significantly in the last few years and has been stable at around current levels for 20 years or so. A number of other factors also need to be taken into account. Firstly, income tax rates in Hong Kong are low. (Arguably, this is linked to high property prices, as the government redistributes money via the property market, rather than the taxation system.) Secondly, Hong Kong also has great income and wealth disparity. The top 10% of all households account for nearly 40% of total income. The 90th percentile household income is around HK$70,000 per month (around US$9,000), compared with a median household income of around HK$21,000 (around $2,700). In reality, only the highest income households can afford to buy private housing. Finally, Hong Kong does not have capital gains tax or inheritance tax, making it easier to accumulate and pass on wealth.
The simplest measure of affordability is the price-to-income ratio. I have calculated the ratio since the early 1990s, for what I have assumed is a typical purchaser, buying a typical Hong Kong flat; dividing the price of a 50m2 in the New Territories (calculated using average price data for different classes of flat produced by the Rating and Valuation Department) by the 75% percentile household income. The New Territories has accounted for almost all of Hong Kong’s population growth in the last few decades; its population has almost trebled in the last 30 years and the majority of Hong Kong’s residents now live there. Realistically, the average new buyer is likely to be buying in the New Territories rather than on Hong Kong Island, where average prices are more than 50% higher; and whilst a 50m2 flat might be small for London or New York, in Hong Kong this is a normal size. I have used the estimated 75th percentile household income rather than the median income to calculate the ratio, as it is more representative of the income level of potential property buyers.
The price-to-income ratio for the period from 1993 is shown in Chart 12. The ratio is now more than 8x, similar to the previous peak in 1997, and around 50% above the average ratio of 5.3x over the entire period. The ratio is certainly high by historic standards. Nevertheless, whilst it would be hard to argue that affordability is not stretched, the position is perhaps not as extreme as some other sources might suggest. I will consider affordability in terms of the actual cost of ownership in my next post.
In this post, I have looked in detail at historic price trends for the Hong Kong property market, including prices relative to their long-term trendline (GDP per capita), rents and household income. All three give a consistent answer, suggesting residential property prices are currently around 50 to 60% above trend levels. This consistency is perhaps not too surprising, however, as the various metrics used to scale prices are to a greater or lesser degree correlated.
A number of conclusions can be drawn. As might be expected, Hong Kong property prices are highly cyclical and historically have diverged from long-term trend levels for substantial periods. However, in itself, the fact that prices are above trend is not conclusive evidence that property is overvalued, even if prices do eventually revert to the trend level. This would also be consistent with low expected future returns, resulting from current low real interest rates. Nonetheless, whilst recognising the limited number of previous cycles for which data is available, prices are now at levels, from which historically they have rapidly fallen. In that sense at least, Hong Kong property is a bubble. To reach a different conclusion, requires an answer to the question: why is this time different?