Houses are typically the largest component of household wealth, the key collateral for bank lending and play a central role for long-run trends in wealth-to-income ratios and the size of the financial sector (Piketty and Zucman, 2014; Jordà et al., 2014). (Page 5)
Based on extensive historical research, we present house price indices for 14 advanced economies since 1870. (Page 5)
Houses are heterogenous assets and when combining data from a variety of sources great care is needed to construct plausible long - run indices that account for quality improvements, shifts in the composition of the type of houses and their location (Page 5)
For the construction of the long-run database, we were able to build in part on the existing work of economic and financial historians such as Eichholtz (1994) for the Netherlands and Eitrheim and Erlandsen (2004) for Norway. (Page 5)
In addition to house price data, we have also assembled, for the first time, corresponding long-run data for construction costs, farmland prices as well as expenditures on housing services. (Page 5)
We go one step further and study the driving forces of this hockey-stick pattern of house prices. Houses are bundles of the structure and the underlying land. An accounting decomposition of house price dynamics into replacement costs of the structure and land prices demonstrates that rising land prices hold the key to understanding the upward trend in global house prices. (Page 6)
Our decomposition suggests that about 80 percent of the increase in house prices between 1950 and 2012 can be attributed to land prices. (Page 6)
. Our emphasis is on the different dynamics in land supply before and after the middle of the 20th century. From the 19th to the early 20th century the transport revolution – mostly the construction of the railway network, but also the introduction of steam shipping and cars – led to a massive and well-documented drop in transport costs, often referred to as the transportation revolution (Jacks and Pendakur, 2010; Taylor, 1951). An important effect of the transport revolution was to substantially augment the supply of economically usable land. (Page 6)
We show that this land-augmenting decline in transport costs subsides in the second half of the 20th century so that land increasingly became a fixed factor. (Page 6)
Ricardo (1817) argued that, in the long run, economic growth disproportionatly profits landlords as the owners of the fixed factor. As land is highly unequally distributed across the population, market economies therefore produce ever rising levels of inequality (Page 7)
The key result of the decomposition is that land price dynamics hold the key to understanding the observed long-run house price dynamics. The sixth section discusses, empirically and theoretically, explanations for the observed trajectory of land prices. We show (i) how the sharp drop of transportation costs during the late 19th and early 20th century expanded land supply and capped prices; and (ii) that this factor not only faded in the second half of the 20th, but coincided with rising expenditures shares for housing services as well as growing restrictions on land which pushed up prices. (Page 7)
Whereas rural indices may underestimate house price appreciation, urban indices may be upwardly biased. Second, house prices can either refer to new or existing homes, or a mix of both. Price indices that cover only newly constructed properties may underestimate overall property price appreciation if new construction tends to be located in areas where supply is more elastic (Case and Wachter, 2005). (Page 8)
Australian residential real estate prices are available from 1870 to 2012 (Figure 1). They cover the principal Australian cities. The index that we use is computed on the basis of two series for Melbourne from 1870 to 1899 (Stapledon, 2012b; Butlin, 1964) and an aggregate index for six Australian state capitals (Adelaide, Brisbane, Hobart, Melbourne, Perth, and Sydney) from 1900 to 2002 (Stapledon, 2012b). (Page 11)
First, house prices in advanced economies increased in real terms since the 1870s, although there is considerable cross-country heterogeneity. Second, the time path of this trend follows a hockey-stick pattern: real house prices remained broadly stable from the late 19th-century to the mid-20th century and increased strongly since then. Third, we demonstrate that urban and rural house prices display similar long-run trends. We also present a number of additional test and consistency checks to corroborate these stylized facts. (Page 20)
On average, house prices in advanced economies have risen threefold since 1900, equivalent to an average annual real rate of growth of a little more than 1 percent. (Page 20)
That is to say, house prices have risen significantly over the past 140 years relative to the consumer prices but have lagged income growth in most countries. We will return to this point later. (Page 20)
Our data show that house prices remained constant until World War I, fell in the interwar period and began a long lasting recovery after World War II. On average, it took until the 1960s for real house prices to recover their pre-World War I levels. (Page 21)
This shape is all the more surprising since income growth much more stable over time. (Page 23)
House prices remain, by and large, stable before World War I despite rising per capita incomes. Relative to income, house prices decline until the mid-20th century. After World War II, the elasticity of house prices with respect to income growth was close to or even greater than 1. Finally, in the past two decades preceding the 2008 global financial crisis, real house price growth outpaced income growth by a substantial margin. (Page 23)
The second striking fact is that, as in the case of house prices, the path of farmland prices also follows a hockey-stick pattern. Prior to World War II, farmland prices were, by and large, stationary. Yet for the second half of the 20th century, there is a clear upward trend with real farmland prices rising on average by about 2 percent per annum. Farmland surpassed house prices. (Page 25)
As the quality of homes has risen notably over the past 140 years, the long-run trends could be upwardly biased if the quality improvement of houses is understated. (Page 26)
A house is a bundle of the structure and the underlying land. The replacement price of the structure is a function of construction costs. If the price of the house rises faster than the cost of building a structure of similar size and quality, the underlying land gains in value (Davis and Heathcote, 2007; Davis and Palumbo, 2007). (Page 29)
We then introduce a stylized model of the housing market in order to study the role of replacement costs and land prices as drivers of the increase in house prices over the past 140 years. The result is straightforward: higher land prices, not construction costs, are responsible for the rise in house prices in the second half of the 20th century. Real land prices remained, by and large, constant in the majority of countries between 1870 and the 1960s, but rose strongly in the following decades. (Page 29)
Figure 24 shows average construction costs side by side with house prices.7 It can be seen from Figure 24 that construction costs, by and large, moved sideways until World War II. Construction costs before World War II were likely held down by technological advances such as the invention of steel frame which allowed for the construction of taller buildings. (Page 30)
Primary historical data for the long-run evolution of residential land prices are extremely scarce. We were able to locate price information on residential land prices for six economies, mainly for the post-World War II era. The series are displayed in Figure 25. The figures show a substantial increase of residential land prices in recent decades, but the sample is clearly small. (Page 32)
As noted above, the trajectory of land prices in the second half of the 20th century is not as puzzling from the perspective of a standard neoclassical model. With continuous economic growth, the value of land could be expected to grow. However, two additional factors might have contributed to an even starker increase of land prices. (Page 41)
First, empirical data show that the mean housing expenditure share remained nearly constant in the pre-World War II period (average annual growth rate: 0.06 percent), whereas it grew by an average annual growth rate of 1.1 percent after World War II. (Page 41)
The intuition is simple. As the production of housing services relies more heavily on land – the land cost share in production is higher – compared to the manufacturing sector, aggregate demand for land rises when the expenditure share for housing services rises. With fixed land supply, the land price increases. A back-of-the-envelope calculation on the basis of the model yields the following results. (Page 42)
We show that after a long period of stagnation from 1870 to the mid-20th century, house prices rose strongly in real terms during the second half of the 20th century, albeit with considerable cross-country heterogeneity. These patterns in the data cannot be explained with quality improvements or composition shifts in the index. Moreover, urban and rural house prices have risen in lockstep in recent decades and farmland prices have also increased. (Page 43)
First, we demonstrated how the transport revolution in the late 19th and early 20th century led to a substantial drop in transport costs, which triggered an increase of land supply. This decline in transport costs petered out in the second half of the 20th century, so that land increasingly behaved like a fixed factor. (Page 44)