500 Years of Urban Rents, Housing Quality and Affordability in Europe

Transcription

1 500 Years of Urban Rents, Housing Quality and Affordability in Europe Piet Eichholtz Maastricht University The Netherlands maastrichtuniversity.nl Matthijs Korevaar Maastricht University The Netherlands maastrichtuniversity.nl Thies Lindenthal 1 University of Cambridge United Kingdom htl24@cam.co.uk July 14, 2017 Abstract Housing affordability remains a concern in modern cities. To study its long term trajectory, we first construct a large database of rent observations and use these to create indices of market rents dating back to 1500 for seven cities: Amsterdam, Antwerp, Bruges, Brussels, Ghent, London, and Paris. We relate rent developments to wages and the prices of other consumption goods, partly employing newly created price and wage indices. We decompose median rents in a market and quality component, in order to measure long-run changes in (implied) housing quality. Real rents in European cities have developed similarly in the long run, but reflect differences in local economic and political conditions in the shorter run. While long-run growth in real rents has been limited, housing quality has risen considerably. At the same time, wage growth has outpaced growth in market rents, implying that affordability has improved significantly, particularly in the 20 th century. Our results suggest that most of the increases in housing expenditures relative to income can be attributed to increasing housing quality, rather than rising real market rents. Keywords: Economic history, Housing markets, Housing affordability 1 Lindenthal is the corresponding author: University of Cambridge, 19 Silver Street, Cambridge CB3 9EP, United Kingdom. Acknowledgements: We thank David le Bris, Gregory Clark, Marc Francke, Katharina Knoll, Arthur van Riel, Yves Segers, Ronan Tallec, and participants to the 2016 AREUEA International Conference, the European Macrohistory Workshop at the University of Bonn, the RCA Research Seminar at the University of Amsterdam, the Penn-Wharton China Conference and the 2017 World Congress of Cliometrics for their helpful comments and/or for generously providing their data. Jeanine van den Bosch, Sjoke Merk and Paul Bouts are thanked for their research assistance to collect different parts of the dataset. Korevaar has been financed by the Netherlands Organisation for Scientific Research under the Research Talent scheme. 1

2 1. Introduction Increasing house prices and rents raise concerns regarding housing affordability, especially in cities. Falling house prices during the financial crisis reduced policy attention to housing affordability, but nowadays the topic is firmly back on the agenda. For example, London s mayor Sadiq Khan recently handed over the keys to the first tenants living under the new London Living Rent scheme 2, which aims to improve affordability in the city. Similar, and equally debated, measures are in place in Paris ( encadrement des loyers ) and various large German cities ( Mietpreisbremse ). Such concerns are not new: explicit government interference in the housing market started around 100 years ago, when many countries announced rent controls following the housing shortages caused by World War I. Despite the importance of housing affordability, and the long history of housing policy, it is still notoriously difficult to define and measure it. First of all, affordability measures are often tied to housing standards, which are likely to change over time. Second, little information is available on how housing costs and standards have evolved over time. The paper has two main goals. First, we construct continuous annual rental price indices from 1500 to the present for seven European cities: Amsterdam, London, Paris and the Belgian cities Antwerp, Bruges, Ghent and Brussels. Second, we combine these rent indices with estimates of wages and consumer prices, based on existing and newly-constructed indices, to investigate urban housing quality and affordability in the long run. We focus on urban rents, because these tend to be the main target of affordability measures, and we have selected these particular cities based on data availability and their historical significance in Western Europe. Also, they were located in relative proximity of each other, such that we can look at interdependencies in rents and affordability. At the same time, these cities had very different fortunes and developments, some Bruges, Antwerp growing quickly initially before slowing down, while others London, Paris have continued growing into the 21 st century. To our knowledge, the dataset presented in this study is among the largest historical real estate datasets constructed to date, complementary to the recent work of Knoll, Schularick & Steger (2017) who collect and combine long-term house price indices for 2 Mayor Hands over keys of first London Living Rent Homes. May 3, 2017: 2

3 a variety of cities. This paper devotes a lot of attention to the collection and classification of all data sources, and to the construction of the corresponding rental indices. We have, at the time of writing, collected more than 235,000 individual rent observations based on archival sources and (digitalization of) existing sources, which allows us to estimate constant-quality indices using a unified repeated-measures method. We subsequently use these indices to trace developments in (implied) housing quality and affordability over time. We believe a more long-run perspective on housing quality and affordability is warranted for two reasons. First, urban rental markets, in particular in Europe, have been under some form of government regulation since the early 20 th century, often due to concerns over affordability. Our long term study thus allows observing housing rents and affordability for a period of 400 years without distortions caused by regulatory bodies, as well as the past century when such interventions likely played an important role. We will show that housing affordability, in terms of the ratio of quality-controlled rents to wages, has improved significantly in each of the studied cities, in particular during the 20th century. Second, our long-term perspective allows to observe gradual shifts in housing quality over time. A house is a collection of attributes, and for the housing stock as a whole, the variety and quantity of these attributes varies gradually over time. In the early centuries covered in our study, the only attribute of a typical dwelling was space, and little of that. Gradually, housing space per capita increased, and attributes like heating, water and plumbing, access to sewers, and electricity became standard features of urban dwellings. We shall see, using a newly developed index of implied housing quality, that housing quality has indeed changed fundamentally, and started expanding as early as the 16 th century. A short-term view would not have picked up the better part these momentous changes. More importantly, it seems that increases in housing expenditures are predominantly the result of rising housing quality, rather than rising market prices. It is important to state that there is no consensus in the literature as to the exact meaning of the term housing affordability, and the resulting definitions therefore vary strongly. On one extreme, housing affordability is considered largely as an 3

4 income issue. For example, in the residual income approach, which is favoured in the overview of Stone (2006), affordability issues arise if insufficient income is left after paying housing costs. The main problem with such an approach is that it is tied to housing or income standards, which vary over time. On the other extreme, affordability issues are considered to arise if house or rent prices exceed their fundamental values, independent of income. In this case constraints on housing supply, for example through zoning regulations, have been found to be a major driver of expensive housing (Glaeser & Gyourko 2002, Quigley & Raphael 2005). Most approaches combine measures of house prices and rents with estimates of income. This is also the approach taken in this paper. We view affordability as a matter of housing costs relative to wages and consumption prices. However, we need to be careful in defining housing costs. For example, simple cost-to-income ratios are still widely used to assess housing affordability, while it is well known that they do not control for the quality of housing nor take account of differing tastes for housing amenities (Linneman & Megbolugbe, 1992). Hence, our main analysis will investigate developments in the rental prices of a constant-quality housing unit, and compare those to developments in wages. Our approach is similar to Gyourko & Linneman (1993) and Gyourko & Tracy (1999), who study developments in housing affordability for owner-occupied housing in the United States from the 1960s until the 1990s. Few studies look at developments in housing affordability over longer periods of time. An exception is Quigley & Raphael (2004), who apply various measures to trace affordability in the United States from 1960 until the early 2000s. They find that affordability has worsened in particular for low-income renters, although they also point out that part of this trend might be due improvements in housing quality. While our long-term perspective on housing affordability is a new phenomenon, studies tracing the historical cost of housing are abundant. An early study is Eichholtz (1997), who created the Herengracht house price index for Amsterdam, covering the period. More recently, studies have appeared for many different regions 4

5 and time periods around the world 3. In light of this study, the papers of Carmona, Lampe & Rosès (2016) and Clark (2002) are particularly worth mentioning. The former studies housing rents and prices in early 20 th century Spain and, as in this paper, attempts to relate this to measures of housing affordability. The study of Clark (2002) for England, whose primary data are also used in this study, gives significant attention to changes in housing quality over time. Similar to our study, he estimates implied changes in housing quality. Except for the smaller sample, the main difference between his study and our approach is the fact he combines urban and rural rents, and applies a different estimation methodology. While all these studies have uncovered lots of useful primary rental data on particular cities and eras, a consistent overview and comparison across cities and countries updated to current times is lacking. This makes it hard to employ the currently available urban housing rent indices to study housing affordability in the long run, which is the raison d être for this study. For house prices, such an overview has been presented by Knoll, Schularick & Steger (2017). They first provide a comprehensive overview of the literature on historic housing market studies, and they use the increased availability of historic house price studies to put together annual house price indices as diligently as possible, covering 14 different advanced economies from 1870 to However, contrary to this paper, they do not collect primary from archives or existing studies. To our knowledge, the only attempt to compile rent indices across countries has been the work of Hoffmann et al. (2002), as part of a broader study on real inequality. This study does not differentiate between urban and rural rents though, and more importantly, as in Knoll et al. (2017), they combine indices constructed with different methods that sometimes do and sometimes do not control for quality. As the authors acknowledge, and this paper will show, this has a dramatic impact on the underlying rent indices. We have therefore made great efforts to create quality-controlled indices that are as comparable as possible, by using one and the same construction technique for each of them. In that respect, our paper can deliver an important contribution to historical cost-of-living studies. So far, few cost-of-living 3 For housing rents, recent works include Eichholtz, Straetmans & Theebe (2012), Cvrcek (2013), Drelichman & Agudo (2014) and Kholodilin (2016). For house prices, e.g. Raff, Wachter & Yan (2013), Nicholas & Scherbina (2013) and Lyons (2015) 5

6 studies have been able to include housing rents, and even fewer have been able to use quality-controlled indices. Our paper does not only expand the spectrum of qualitycontrolled indices, it also suggests that the share of housing costs in the urban household budget might have been higher and more variable than assumed in wellknown cost-of-living studies, such as Allen (2001). Our quality-controlled rental indices provide interesting insights in the development of rents over time. In real terms, urban rental prices have grown little in the long run: in most Belgian cities, market rents even seem to have stayed roughly constant since In the other cities, rents have grown significantly, but with relatively low annualized (geometric) real growth rate at around 0.25% per year in each city. At the same time, we also observe large differences in rents across cities in the short to medium run. Periods of growth seem to correlate with periods of economic expansion: Antwerp for example experiences its heyday in the early 16 th century, and reaches exactly in that period its highest level of real rents. In combination with the consumer price and wage information, the dataset presented in this paper provides a solid basis for further research in economic history and urban economics. The remainder of this paper is organized as follows. We first introduce and discuss the data and data sources, after which we explain the index estimation methods and present the resulting indices. Subsequently, we will discuss our methods to analyse urban housing affordability and quality over the very long run, and provide the results of this analysis. We end the paper with a short summary and some conclusions. 2. Data sources To provide a long-term overview of house rental markets, we have collected as much primary rental data for each of these cities. We have compiled, and often digitalized, rental data from dozens of historical and contemporary studies, while at the same time hand-collecting archival data when necessary. This has resulted in the collection of about 235,000 individual rent observations, coming from Amsterdam, London, Paris, and the four Belgian cities 4. When primary data was not available, mostly since the 4 We are still hand-collecting additional primary data for Paris ( ) and London ( ). These will be incorporated in a next version of this paper. 6

7 20 th century, we employed secondary sources, as much as possible selecting sources that control for housing quality. An overview of our rental sources is presented in Tables 1 and 2. A more detailed discussion of the rental sources can be found in Appendix A. Table 1 gives information as to whether the rental sources can produce a representative index. Coverage reflects whether the index or data cover just our city of interest or the whole country, whereas type corresponds to the nature of the source: primary or secondary. The column headed sample representativeness reflects the extent to which the sample consists of dwellings that are representative for the overall housing market in the city or country. The next column, constant quality market rents (CQMR), classifies whether the data correspond to quality-controlled market rents. On top of the four Belgian cities already mentioned, Table 1 also includes information for Liège and Leuven, which we will use to create a rent index for Belgium as a whole. Table 2 lists the number of rental observations and properties we have currently collected for each city. We have only listed both the total number of observations collected and the number of observations that were used to construct the repeat-rent rent indices. As rents are the primary interest of this study, and we only create new consumer price and wage indices for Belgium ( ), we discuss the data and methodology for our wage and consumer price indices only in Appendix B and C. --- Tables 1 and Virtually all our primary sources, and most secondary sources before the end of World War I, originate from ledgers of various social institutions, mostly churches, monasteries, hospitals or orphanages. These institutions had considerable housing portfolios, mostly resulting from bequests or donations over time, and used the rental cash flows of these homes to finance their activities. They were the precursors of the modern-day institutional investors, and kept extensive archival records of their accounts, of which many have survived the test of time. Such institutions were present in most large cities in Western Europe, and they all operated on a fairly similar basis. 7

8 Although it might seem counterintuitive to use rents from properties owned by such social institutions to estimate market prices of rental housing, there is an overwhelming amount of evidence showing these rents were representative for the market. First of all, despite the social nature of these institutions, their real estate portfolios were not used to provide low-cost housing to the poor, and in each city there was considerable variety in the homes being leased, varying from sober cottages to splurge mansions. Second, most institutions relied almost exclusively on rental streams to finance their core activities and could therefore not permit themselves to ask below-market rents. 5 It is also unlikely that these institutions could engage in rentseeking through a dominant local market position. In all the cities we study, there were multiple institutions renting out a housing portfolio, as well as considerable numbers of private landlords. For example, about 25% of data from Ghent stem from private landlords, while in most cities our dataset contains around 20 different social institutions that rent out property. The representativeness of the rents paid to social institutions is confirmed by the data. Le Roy Ladurie & Couperie (1970) collected about 12,000 rental observations from privately owned Parisian properties for 23 years and compared these with the level of rents from properties owned by social institutions. Overall, they found no notable differences in the level and the development of rents. For early 19 th century England, this was confirmed through a comparison of rents from charities with average rents identified in tax records (Feinstein 1988, Clark 2002). While institutional rents might be representative for the market, the homes in the sample can still be biased towards specific neighbourhoods or classes of homes. Note that this is only problematic if the prices of homes in different classes or neighbourhoods evolve differently: the repeated-measures methodology only requires percentage changes in the observed rents to be representative for the market, and not the level of rent itself. We have found little evidence that homes in different classes 5 This is highlighted in two rare complete annual accounts of the Hopital-Saint Jacques in Paris for 1636 and 1648, where rents provided respectively 83% and 96% of total income (Le Roy Ladurie & Couperie, 1970). 8

9 evolved differently prior to the 20 th century, and our archival data collection as well as the collection from the historical studies we use, has been targeted to collect rental observations from a wide variety of housing market segments and neighbourhoods. To investigate in more detail the locational spread and the status of the neighbourhoods, we have recovered the locations of all properties in our Amsterdam sample. We select Amsterdam for this purpose as we can compare the locations of the properties to the home values of all properties existing in Amsterdam in 1832 (see Lindentahl, Eichholtz & Geltner, 2017). For the large majority of properties, we were not only able to locate the street, but also the (almost) exact location of the property on the street. Figure 1 plots the locations of the properties in our sample from 1700 to 1850, a period during which the city did not expand, next to a map of the property values across Amsterdam in As can be seen, the properties are very well spread throughout the city and not limited to neighbourhoods with mostly cheaper homes. Also during the 16 th and 17 th centuries (not reported) the properties are evenly distributed throughout the expanding city of Amsterdam. We see a similarly even distribution for Antwerp, and historians have confirmed this pattern in other cities Figure Rent Index Estimation The current literature on the estimation of rent indices has relied on hedonic models, repeated-measures models or a combination of both. In this paper, we solely rely on repeated-measures regression. Hedonic models control explicitly for quality characteristics, but since we have very little quality information on the homes in the sample, this approach is not viable. The repeated-measures methodology controls for quality by focusing on changes in rents on the same property, under the assumption that the quality of the property remains constant over time. While initially the methodology was used for the construction of house price indices, it recently has also 6 Plots are also available for Ghent (Van Ryssel, 1967) 9

10 been applied for housing rents (Ambrose, Coulson & Yoshida 2015 and Eichholtz, Straetmans & Theebe 2012). Even though all historical studies compiled have discarded rental observations if there was an indication a home was rebuilt, renovated or significantly affected in some other way, it is likely that the data do not completely satisfy the assumption of constant quality. First of all, we cannot account for the effect of aging on the property as we do not know the year in which a property was built. Second, minor changes to the property might not have been registered. Property changes are likely to have been common: evidence from Lindenthal, Eichholtz & Geltner (2017) shows that between 1832 and 2015, 79 percent of lots in Amsterdam were involved in land assembly. However, we believe that the errors are likely rather small. First, homes seem to have been well maintained. For some periods, the institutional archives contain books that register all costs that were made for each home. In general most money was spent on repairs, in order to avoid the property from depreciating. If renovations occurred, they were often very large, such that we register the home as new observations. Second, various Parisian arrests (see Felibien, 1725) make note that tenants were often responsible for the upkeep of the property, and could be held liable when they failed to do so: various records make indeed note of payments for repairs. In order to estimate a repeated-measures index that is a reflection of market circumstances, it is important to classify which observations can be treated as repeated observations. A rent observation is only representative for the market in the year a new rental contract is signed. Unfortunately, all rent data from the Belgian cities, and part of the Paris and Amsterdam data, do reflect rent payments rather than rent contracts. Most historical studies have therefore included all observations, while Eichholtz, Straetmans & Theebe (2012) have argued only observations where the rent is being revised should be included: when the rent is revised one is sure a new contract has been signed. We apply the latter approach to deal with parts of the data for which we do not have contractual data, with the exception of the data from the study of Henau (1991) for the Belgian cities from : the data indicate that rents were set annually, as in the majority of cases the rent changes every year. The main 10

11 disadvantage of this approach is that it wastes otherwise useful data: analysis of Parisian data, for which 80% of data come from rent contracts, reveals that in 30% of the contractual rent was the same as in the previous contract. However, this data loss is minor relative to the number of observations that does not correspond to a rental contract, and therefore is rightfully removed. For the cities for which we have contractual data, only observations on homes for which we have a single contract are dropped; these do not constitute repeats. Additionally, as in Clark (2002), we drop observations for London with a contract length of more than 21 years, which generally correspond to ground rents rather than regular leases. For both Paris and London, data prior to 1500 is not included in the index estimation. For London, less than a third of observations remain, whereas for Paris, which mostly based on contractual data, 80% of the observations remain in the sample. We subsequently apply the repeated-observation methodology to the selected data to construct new repeat-rent indices for each city. While for most cities such indices were not yet available, Eichholtz, Straetmans & Theebe (2012) already constructed a repeated-rent index for Amsterdam from , which we improve by applying our more efficient estimation method and a larger dataset. Clark (2002) has already estimated an index for London from that controls for the quality of the properties by only including repeated observations, but he applies a different index estimation methodology. More importantly, his index is at a 10-year frequency, whereas we construct annual indices. The basic repeated measures methodology from Bailey et al. (1963) starts with the simple observation that the natural logarithm of the price on any asset, in this case the log rental price! " on a particular home i, can be represented as the sum of three components:! "# = % " + ' # + ( "# 11

12 The first term! " reflects the underlying value, and therefore quality, of the home: the key assumption is that this does not change over time. The second term,! ", is the value of the log rental price index, while the third term! "# reflects price noise and is assumed to be distributed as!(0, % & ). Hence, for any periods! " and! ", the difference between log prices on home i can be written as:! ",$% -! ",$' = ) $% -) $' + + ",$% - + ",$' If generalized for all time periods, this yields a regression equation, where D refers to a set of dummy variables that take on the value of 1 if! =! # and -1 if! =! #, and! "# equals the difference in the two error terms: -! "#$ -! "#& = ( # ) #," +, "#, 0 = 1,, 3 #./ OLS regression of the ratio of log rental prices on this set of dummies produces predicted values " # for each year, which correspond to the growth rate of the index relative to the base year. Exponentiation of these estimates and multiplication by 100 yields the index values. The basic model has several shortcomings, and in our case the most severe are small sample problems: for most cities, in particular London and Bruges, the annual number of observations is fairly low. In such cases noise in the individual rent prices can have a large impact on the estimation of the indices. The literature has proposed several adaptations of the original model in order to improve the signal-to-noise ratio. Probably the most notable of these is the study of Goetzmann (1992), which proposes a Bayesian ridge estimator, and shows this estimator outperforms conventional models. In this paper, we apply the methodology of Francke (2010). There are three main advantages of Francke (2010) relative to other procedures. First of all, being a generalization of Goetzmann s method, it allows for very general model specifications, which can easily be compared using likelihood criteria. Second, this set of methods allows to compute price indices even if no price information is available in a given 12

13 year, as is sometimes the case in our sample. Third, the proposed one stage maximum likelihood procedures results in more efficient estimation compared to other methods. In Francke (2010), the betas are not treated as fixed unknown parameters that can be estimated using least squares, as in the standard case, but rather as a stochastic process, more specifically a local linear trend model:! "#$ =! " + ( " + ) ", ) " ~,(0, / )! "#$ =! " + ( ", ( " ~ +(0,. / 0 1 ) The dependence between the betas is based on several components: the stochastic trend! ", which is defined by its initial value! " and the signal-to-noise ratio " #, and the signal-to-noise ratio " #. If! " = 0 for all t, the stochastic trend vanishes and the betas simply follow a random walk. The signal-to-noise ratio play an important role in this model: if the ratio is high, the variance of the error terms of the index are high, and hence the dependence between the betas will be small. Correspondingly, the model will reduce to the standard case with fixed unknown parameters as the variance of the error terms approaches infinity. However, if the signal-to-noise ratio is low, the dependence between the! " s will be strong, resulting in a smoothening of the index compared to the standard case. To estimate the index, Francke (2010) proposes an empirical Bayes procedure, which we have applied here. Conditional on the variance parameters " #, " # and! " estimates of the annual coefficients and! " can be obtained using generalized least squares. The variance parameters are subsequently estimated by maximum likelihood. For more details, see Francke (2010). 7 We have applied the Akaike Information Criterion and the Bayesian Information Criterion for each city to select the appropriate model specification. For all cities, both information criteria selected the random walk model. 7 Note that we have not included the parameters, additionally discussed in Francke (2010), that capture dependence of the returns and their variance on the holding periods, as is generally observed in house price transactions. However, in rental markets such considerations do not seem to play a role. 13

14 4. Estimating implied housing quality While we have argued that the use of simple rent indices (mean or median indices) is problematic, as they do not control for quality, we should not completely disregard them. Because the indices are for the majority of years based on primary data, we can construct quality-controlled indices and simple rent indices from the same set of data. Since the only difference between the two is that the former controls for quality and the latter does not, the difference between the two is an estimate of implied housing quality. If representative, such an index reflects the monetary value of quality improvements over time. These quality improvements have two dimensions: improvements in the quantity of space consumed over time, and changes in the quality of a given space due to construction improvements, such as plumbing and insulation. Quality improvements external to the property, such as changes in pollution or local infrastructure, are not taken into account: these are implicitly included in the market prices. The main problem is that the construction of such quality indices requires another, much stronger assumption: at any point in time, the average level of rents in the sample should be representative for the city under investigation. This assumption is likely violated in some cases, as explained in the data section. To make matters worse, it is in general difficult to assess quantitatively the severity of this bias, as in most cases no other data is available. If rental data from the institutions in the sample can be considered a random draw from the available set of rental homes in a city, then through the Law of Large Numbers the mean of such a sample will converge to its true value as the sample size increases. Hence, if we can create a sufficiently large representative sample, it might be possible to construct a representative quality index. Such a sample can clearly be created when pooling the data together for the Belgian cities. This yields a dataset of approximately 160,000 rents: as our interest is now the median rent paid, rather than quality-controlled market rents, we keep all observations in the sample. On average, 14

15 we have 360 observations per year with a minimum of 75 observations in Individual observations are converted into a single currency, the Belgian franc. In order to eliminate large short-run fluctuations in median rents, we take median rents for ten year periods for each city, with the starting period moved forward one year annually. In addition, we exclude data after 1920 as the sample is not updated anymore with new homes. Median rents for Bruges are interpolated from 1795 until 1815, given lack of data. Weights for each city in the total Belgian quality index are based on interpolated population numbers from various sources (Deprez 1957, Wyffels 1958, Van Ryssel 1967, Boumans & Craeybeckx 1974, Segers 1999). The total quality index subsequently follows by indexing the difference between the 10- year moving average repeat-rent and median rent index, and applying the weights. In a similar fashion, but excluding the city weighting, we compute a quality index for Amsterdam and Paris. However, given the lower number of observations, we use 20 rather than 10 year periods to compute the index. The Amsterdam index is based on about 200 observations per year, with a minimum of 46 observations in Unsufficient data is available after 1916, such that we do not construct a median rent index after this period. In addition, since the mid-19 th and early 20 th century, the homes in the sample are dominated by a single institution, the Burgerweeshuis, which owned homes that on average fell in the cheaper segment of the market. Hence, the median rent index cannot be considered representative in this period. However, we will use the rent index values from Van Riel (forthcoming), which is mostly based on non-quality controlled rents, from 1830 until 1913 to assess the severity of this bias. The Paris index is based on a much lower number of observations, on average only 30 per year, but it must be accounted for that these correspond to contracts rather than rent payments. Given the median contract length of 6 years, the sample can be considered comparable to that of Amsterdam. No quality index is constructed for London, as too few observations are available to produce a representative simple rent index. Next to the large amount of irregular long-term leases, the London rents often refer to multiple dwellings, such that we can 15

16 only compute the average rent if the number of dwellings is known. In the end, this reduces the median number of observations to just two per year. Note again that the repeated-rent index is unlikely to control fully for changes in dwelling quality, which implies that our housing quality index might still misestimate the true increase in housing quality. In addition, despite our efforts to construct a representative quality index, the estimates can still contain significant noise in the short run due to large sample composition changes. However, in the long run the indices should give a reasonable estimate of implied quality changes. 5. Rents in the Long Run Figure 2 reports the total real rent indices for Amsterdam, Paris, London and Belgium, with the latter index based on population-weighted average from six cities (until 1940). Rents are deflated using consumer price indices 8. The indices for Paris, London and Amsterdam are based on national data starting in from 1960, 1940 and 1981 respectively. Estimates of the variance parameters from the random walk model are reported in table Figure 2, Table The first and most striking conclusion from the figure is that rental prices have shown little growth in the long run. The real rent levels currently observed in the Belgian cities are similar to those observed around For Paris, London and Amsterdam, the implied annual (geometric) growth rate of real rents is around 0.25% in each city over the whole sample period, while in the Belgian cities real rents do not seem to have increased at all. In addition, for most time since the 17 th century, the long run developments across these rental markets have been strongly correlated. This suggests these cities have had close economic connections in the last 400 years, as well as that benefits from geographic diversification, for example for very long-term rental housing 8 Tabulated indices are available upon request. 16

17 investors like sovereign wealth funds, might be smaller than previously thought (Eichholtz et al., 1995). The result that long-term growth in rental prices is limited confirms scarce existing evidence for house prices. Eichholtz (1997) finds, based on a repeat-sales index, little long-term price appreciations for the homes on the Herengracht, Amsterdam s most expensive canal. Raff, Wachter & Yan (2013) find that price appreciation has been limited in Beijing when comparing the price level of to that of 2004; although house prices have exploded recently. While house prices differ from rental prices in the short to medium run, Ambrose, Eichholtz & Lindenthal (2013) show based on data form Amsterdam that the long-run developments in house prices and rents have been similar. The long-run stability of rental prices also shines new light on existing studies. For example, Hoffman et al. (2002) note that rising housing rents contributed to rising real inequality prior to 1900, most notably in pre-revolutionary Paris, as the poor had to rent their homes. However, after controlling for quality, rents in Paris remain stable in real terms. In all cities, most growth in real rental costs seems to have occurred in the 19 th century, most likely because of increasing urbanization and economic growth. Before the 18 th century, there are no significant increases in rents in the long run. During the 16 th century, rents even decline significantly in London and, more notably, in the Belgian cities. After 1900, real rental costs have only increased very significantly in London, which should not be very surprising as London is currently the most expensive European city in terms of housing costs. However, it is remarkable that real rental costs have not increased strongly in the other cities. Part of this might be due to increased government interference in the market: it is exactly at the start of the 20 th century that governments start intervening in rental markets, worried by high rental prices and, more importantly, the corresponding low housing standards. Despite the similar and limited long-run growth in the indices, significant differences exist in short- and medium-term rent movements. Although we cannot cover each of these movements in detail due to space considerations, we note that they often 17

18 coincide with large political and/or economic shifts that significantly impact urban housing markets. We will shortly discuss the most significant example: the period at the end of the 16 th century. As noted previously, Belgian rents are at the start of the 16 th century at relatively high levels, but decline significantly throughout the century. Closer inspection of nominal rent developments in Belgium, Figure 3, reveals that this decrease in real rents is mostly caused by a strong increase in consumer prices: throughout the course of the 16 th century nominal rents increase significantly, in particular in Antwerp. Until the 1560s, Antwerp is the only Belgian city that does not face a decline in real rents. This is not surprising: at that time, Antwerp is one of the leading cities in the European economy. Historical records make note of shortages in housing supply: the population of Antwerp more than doubles from 1500 to 1570, while around 6000 new homes are built intra-muros (Marnef, 1996). While living within the city walls is an amenity, the wall itself is also an explicit zoning regulation. In that perspective, the strong rise in Antwerp rent prices relative to the other cities can be seen as a confirmation of the results of Glaeser & Gyourko (2002) and Quigley & Raphael (2005). The preeminence of Antwerp, and to a lesser extent that of the other two Belgian cities, ends in Antwerp is captured by the Spanish, and the Dutch began a naval blockade of the river Scheldt, preventing Antwerp its access to the sea. This created a deep economic slump, and housing rents declined dramatically. At the same time, those in Amsterdam rose significantly, as Amsterdam took over Antwerp s pre-eminence as Europe s main trading city (see Figure 2 and 3). Only when the Scheldt was gradually reopened at the end of the 18 th century did Antwerp s housing rents recover. --- Figure Although Parisian rents do not significantly increase or decrease throughout the 16 th century, there is a very remarkable decline and resurgence of rents in the city at the end of the century. The ongoing Wars of Religion seem to be the main culprit for these dramatic swings in house rents, most notably the Siege of Paris, which took place in The Siege had devastating effects on the city, causing the starvation of 18

19 approximately a quarter of the Paris population. The situation was so exceptional, that the Parlement de Paris, the most important court, released twice an arrest that mandated tenants to pay only a fraction of their contractual rent amount. The first arrest was released on the 15 th of April 1589, on the diminishment of Parisian rents (De Thou, 1734). Our primary data reveal that this law reduced rents on all existing contracts by 25%. On the 8 th of January 1592, a second arrest was published which mandated tenants to pay only 25% or 50% of their contractual rent, depending on whether the contract was signed before or after August 1590 (Felibien, 1725). Although the Parlement de Paris was a court, and not a law-making authority, these arrests can be considered the very first recorded instances of explicit interference in the rental market. From the mid-1590s, rents start to recover quickly to their old levels. It is interesting that short-term socioeconomic developments seem more strongly reflected in housing markets than in income. Figure 4 plots the level of real wages in each of the studied cities from 1500 to the present. As can be seen, wages show little variation across cities, and do not clearly reveal any of the events described in the previous section. This suggests that from a historical point of view, long time series from the housing market provide a more direct perspective on the turbulent history of cities than time series on wages. It is important to note that part of this difference might be attributed to the fact that most historical wage series are based on day wages, without knowledge of the working year. New evidence indicates that the number of days worked, on which little to no information is available, might have varied significantly over time (Humphries & Weisdorf, 2016). To investigate the extent to which this could have affected our sample, we aggregated and digitized data from Verlinden ( ) on annual wages and the number of weeks worked for 175 compositors of the Plantin Press in Antwerp, from Although the annual wages varied strongly, likely related to the Press its productivity, developments per decade were closely correlated with constructions worker s day wages. However, the median number of weeks worked for full-time employees (defined as working at least 35 weeks), increased from 47 weeks in the late 16 th century to 52 weeks in the early 18 th century. As a result, total wage growth was about 10% higher for the compositors 19

20 of the Plantin Press. Thus, the use of day wages might have slightly underestimated total wage growth during this period. --- Figure There are considerable differences in the volatility of rental growth over time and across cities. Most of these differences across cities can be attributed to the underlying data. For example, London rents show a much smaller volatility of nominal rents, since the index is heavily smoothed due to the low signal-to-noise ratio (see table 3). Belgian real rents are much more volatile, since its deflator, the CPI, does not have to rely on interpolation for missing observations. On the other hand, data on clothing, which had in most countries a very stable price and made up for about 10% of household expenditures, was not available. The increasing stability of consumer prices over time accounts for the significant decline in the volatility of real rental growth from the 16 th towards the early 20 th century. Because of technological progress, food production became less reliant on external circumstances, thereby stabilizing prices. However, it is during the first part of the 20 th century that we observe the most remarkable volatility in the real rental cost of housing, which is depicted in more detail in figure 5. Following the sustained rise of real market rents in the 19 th and early 20 th century, and large urban housing shortages caused by World War I, governments started to interfere directly in the price of rental housing. Housing was extremely scarce during this period: records from Amsterdam show that in most months less than 10 dwellings were vacant in the whole city. To address affordability concerns, all countries in the sample start to adopt rent controls from (the end of) World War I onwards. It is remarkable how similar governments in the four countries studied here have acted during this turbulent period. There are two reasons for this pattern. First of all, most rent regulation focused on the nominal level of rents, rather than stabilizing rents in real terms. As can be seen in Figure 5, the countries that experience the highest inflation in the first decades of the 20 th century experience the largest fall in real rents. The reverse holds for the deflationary period right after the end of the First World War. Second, it seems that at approximately the same time governments 20

21 realized that the combination of frozen rents and high inflation left little incentives for landlords to invest, harming the supply of rental housing. Thus, in each country rents were slowly deregulated from the late twenties onwards, such that the level of real rents could catch up again. Exactly the same process happens again during and after World War II: real rents initially decline significantly, but catch up as soon as rent controls are abolished or weakened. After this turbulent period, most countries start to introduce more sophisticated policies (see Arnott, 1995, for an overview), which has likely had a dampening effect on real rent volatility. --- Figure Housing affordability and quality in the long run Government intervention creates a key distinction between rental markets before and after World War I. It is difficult to identify the exact effects of these measures, since housing policy is endogenous. However, it is possible to analyse to what extent markets could provide affordable housing at stable prices before and after governments started intervening. The simplest measure of housing affordability expresses the market level of rents relative to income, which we have indexed and reported in Figure 6. Note again that the affordability index does not correspond to actual expenditure shares, since the index reflects market rents rather than actual rents paid. --- Figure The pattern emerging from the figure is striking. Before World War I, affordability does not significantly decline or improve in the long run, although it does vary over time and across cities in the shorter run due to changes in the level of market rents. 9 Thus, the concern that free markets would cause rental housing to get less and less affordable in the long run seems invalid, at least for these four centuries. It also confirms the hypothesis of Hoffman et al. (2002) that, until the 19 th century, housing 9 Paris forms a small exception to this pattern, with a slight worsening of affordability. However, it must be noted that most of the worsening in affordability occurs in the mid 19 th century, when we rely on a rent index that does not control for quality and hence likely overstates increases in rental costs. 21

22 is probably the one major consumer price that may have kept pace with the wages of labour, when quality is held constant. After World War I, affordability improves significantly with wage growth outpacing rent growth by about 1.5% per year, although this trend is temporarily disrupted in the thirties, possibly due to the abolition of rent controls and the depression. Even though it is impossible to causally identify the effects of government interference, it is remarkable that rental housing affordability started improving at about the same time when governments began interfering directly in the level of rents. In more recent decades, improvements in affordability have halted, with affordability actually worsening in most cities. With that in mind, it is not very surprising that discussions about housing affordability have regained prominence recently. While we have sketched a relatively positive story about housing affordability, with affordability improving significantly throughout the 20 th century, our view is at odds with evidence from actual rental expenditures. Table 4 reports actual estimated expenditure shares on rents for Antwerp (prior to 1800). For some homes, our sources listed the profession of the tenant. In case this was a mason or carpenter, we could pair this to the day wages of these professions, and obtain an estimate of the expenditure share. First, the table reveals that the income share in Antwerp, by far the most expensive city in Belgium in the 16 th century, is currently close to its level in Second, it shines new light on existing studies on the historical cost of living. Many cost-of-living studies have not been able to include rent prices. They generally assumed that housing costs were relatively small and stable, with estimates ranging from 5 to 10 percent of the income (e.g. Allen 2001, Ozmucur & Pamuk 2002 and Allen et al. 2011). The Antwerp data indicate that urban expenditures shares could vary significantly over time, and were likely much higher in periods of strong economic expansion. Third, rent-to-income shares seem to have risen considerably in recent decades, with expenditure shares in the range of 20 to 30 percent. To put this in historical perspective, various budget studies for Belgium reveal that in the 19 th century around 10 percent of income was spent on rent. In 1909, this had risen to an average of approximately 12 percent (Avondts & Scholliers, 1977). This high recent expenditure-to-income ratio is also visible in the other countries we study. 22

23 --- Table To understand the contradiction between expenditure shares and our affordability estimates, it is essential to look at measures of housing quality. Figure 7 reports the index of implied housing quality for Belgium, Amsterdam, Paris and the Clark index for England. Compared to the changes in real housing rents, increases in housing outpace growth in real rental costs in each city. Hence, these quality indices suggest that the main reason why tenants have been spending more and more of their income on housing over time is not because of rising market prices, but mostly due to tenants living in higher-quality homes with both more space and better facilities. Although we do not have sufficient data to create a quality index for London, it is possible that this pattern does not apply for London in this period: growth in London real rents has outpaced the implied quality change for England from Clark (2002). --- Figure Although our last quality index ends in the early 20 th century, it is likely that growth in housing quality has continued to outpace growth in real rents for several decades, as affordability continued to improve at least until the 1970s. For the United States, this process has been documented in detail by Gordon & Van Goethem (2005) and Quigley & Raphael (2005). Similar processes occurred in Europe. For example, Dutch housing policy in the first decades after World War II was both aimed to keep housing affordable as well as to improve housing quality by promoting renovations and subsidizing new construction. Correspondingly, Berghauser Pont & Haupt (2010) document that housing space per capita in Amsterdam still quadrupled in the 20 th century. Before we attempt to explain reasons for the strong observed increases in housing quality, it is important to make three more remarks. First of all, the index is an estimate of implied increases in housing quality, rather than indexing actual quality characteristics such as the number of people per square meter, or the improvement of 23

24 amenities such as plumbing, heating and insulation. And while we believe our index corrects appropriately for large interventions in housing quality since these would take place as major renovations, for which we control we are not able to account directly for smaller and more gradual improvements in housing quality or the effect of aging on properties. Second, we believe the quality index is in the shorter term still sensitive to changes in the composition of the sample. One should thus be careful in interpreting the short-term volatility of housing quality. Last, the implied quality index reported in Clark (2002) for England and Wales does not seem to differ much from our quality indices, despite the much lower number of observations and the fact that this index combines rural and urban rents. Hence, this suggests that in the long run the improvements in housing quality were highly correlated across cities and countries. While we do not possess hedonic data which could directly identify the sources of these quality upgrades, we have various hypotheses that might explain the trajectory of our quality indices. Prior to the 19 th century, housing quality increases strongest in Paris. We hypothesize that a significant part of these increases is due to improvements in housing space. Estimates reported in Hillairet (1963) indicated that the population density of the city reduced from 640 people per hectare in the late 14 th century to just 180 people per hectare in 1789: part of this increase in space per capita was likely in housing. Although not as dramatic, comparable reasons might explain the strong quality improvements in Amsterdam and the Belgian cities during the 16 th and 17 th century. In the Belgian case, in particular Antwerp and to a lesser extent Ghent and Bruges, population expanded significantly in the early 16 th century and went hand-in-hand with a growing economic importance of these cities. New homes were constructed to accommodate for the rising population, but when housing construction only occurred intra-muros, housing supply would be constrained significantly, which happened in Antwerp in the mid-16 th century (Marnef, 1996). The comparatively high rent prices in this period, combined with a more densely populated city, might explain the decline in housing quality observed in the mid-16 th century. Similarly, the large 24

25 decline in population after the capture by the Spanish in these cities might explain why housing quality recovers strongly again at the turn of the 17 th century: rent prices are lower, and space per inhabitant increases, as many inhabitants leave the Belgian cities, most notably to Amsterdam. However, Amsterdam responds to the increased population through a large coordinated expansion of the city: the four stage Uitleg that occurred from the late 16 th century until the mid-17 th century. The size of the city expands significantly, and despite the large increases in population and construction, building plots remain available. In addition, the increased number of wealthy citizens causes more high-quality homes to be constructed: it is exactly in this time periods that Amsterdam s famous canal ring is built. We do not believe that increases in housing space can explain all quality improvements over this period. Technological advances in construction rates likely played an important role as well. During the 16 th and 17 th century, wooden and clay homes are gradually replaced by stone homes, and roofs are constructed using tiles rather than thatch. Van Ryssel (1967) has attempted to quantify the magnitude of these changes for the case of Ghent. Prior to 1612, stone homes were only a possibility for the wealthiest citizens, but after 1612 citizens can receive a government subsidy to construct a more durable stone façade. As a result, at the end of the 17 th century the large majority of homes is built from stone. The change from thatch to tiles likely took place earlier, as already in the 15 th century the government required that all homes should be constructed using tile homes. However, this process likely took until the late 16 th century to complete, as city fires continued to occur frequently. Housing quality improves significantly again in the late 19 th and early 20 th century, most notably in Belgium. It is very well possible that introduction of plumbing, sewerage, electricity and other new technology in this period improved housing quality substantially. We are more careful though in interpreting this quality change, as there seem to be significant differences between the Belgian cities and Amsterdam that might be driven by the underlying data. In this period, the Belgian index most relies on data from Antwerp and Brussels, the largest Belgian cities, and in both cities seemingly expensive properties are added to the sample. At the same time, the Amsterdam index starts to rely on relatively cheaper properties. Hence, when using 25

26 average rents in The Netherlands as proxy for Amsterdam rents, the difference between the two reduces. Part of the difference might also be attributed to decreases in housing space in Amsterdam: census data from the city reveals that the number of people per house increased from 10 in the 1850s to almost 14 at the end of the 19 th century Conclusion In this paper, we have presented a long-term overview of the rental market in Western Europe, relying both on existing indices and newly constructed indices on rents, housing quality, wages and consumer prices. For the first time, it is possible to trace the rental trajectories for various European cities all the way from 1500 to the present on a continuous annual basis. The combination of the wage, rent and quality indices tells a compelling story. Until the 19 th century, growth in real market rents was close to zero or even negative. However, housing quality gradually expanded, and tenants started to spend more of their income on housing. From the 19 th century onwards, real rents start to increase in most cities, while keeping up with the pace of real wages. Importantly, direct government interference in the rent level did not exist in the first four centuries we study, and the interplay of market forces seems to have done its work in stabilizing long-term real rent levels relative to wages. When wages started to outpace growth in rents during the first 75 years of the 20 th century, possibly with the support of intervening governments, households could expand their housing consumption (and expenditure shares on housing) even further to the levels currently observed. We believe these results have important implications for housing policy, in particular since urban housing affordability is considered a major policy issue around the world. Our findings show that to understand housing affordability, it is essential to look at income, quality-controlled market rent indices and measures of housing quality; 10 Note that this measure is different from the number of families / persons per unit of housing: homes in Amsterdam were very often split in several parts (an upper home, a lower home, back home etc.) which were rented out separately. In our dataset, we have registered these rents separately as well. 26

27 excluding any of these components can result in a markedly different interpretation of the data. Unfortunately, these considerations are currently missing in the literature. Contemporary quality-controlled rent indices are barely available (e.g. Ambrose, Coulson & Yoshida, 2015), and our study is one of the first to produce long-run estimates of the developments in housing quality. Hence, an essential first step for future academic work, statistics offices or other parties is to create contemporary and comparable indices of market rents and housing quality. The relevance of this study goes beyond housing affordability. For investors, our results reveal that in in the long-run real rents have grown little, and have developed similarly in each of the studied cities. Thus, the long-run benefits of regional diversification in residential real estate are likely smaller than previously thought. At the same time, there can be substantial heterogeneity in short-run returns across cities, as local economic and political conditions can have dramatic impacts on the rental market. For economic historians, our rent indices and estimates of expenditure shares and housing quality shine new light on the historical standards of living: housing quality seems to have improved very significantly prior to the 20 th century, and expenditure shares on housing could vary significantly over time. Last, the presented indices provide a valuable data source for economists, for example to be employed in future work on the dynamics of housing markets. 27

28 6. References Algemene Directie Statistiek (2016). CPI vanaf Retrieved May 10, 2016: Allen, R. C. (2001). The great divergence in European wages and prices from the Middle Ages to the First World War. Explorations in Economic History, 38(4), Allen, R. C., Bassino, J. P., Ma, D., Murata, C., & Van Zanden, J. L. (2011). Wages, prices, and living standards in China, : in comparison with Europe, Japan, and India. The Economic History Review, 64(s1), Ambrose, B. W., Coulson, N. E., & Yoshida, J. (2015). The repeat rent index. Review of Economics and Statistics, 97(5), Nationaal Instituut voor de Statistiek ( ). Annuaire statistique de la Belgique (et du Congo Belge). Years Arnott, R. (1995). Time for revisionism on rent control?. The Journal of Economic Perspectives, 9(1), Avondts, G. (1973). Een Brusselse huishuurindex ( ), barometer voor de economische conjunctuur in het Ancien Regmie. Annalen van de Faculteit Economische Sociale Politieke Wetenschappen. VUB, Brussel. Avondts, G., & Scholliers, P. (1977). De Gentse textielarbeiders in de 19e en 20e eeuw (Vol. 3). Vrije Universiteit Brussel-Dienst Uitgaven. Ambrose, B. W., Eichholtz, P., & Lindenthal, T. (2013). House prices and fundamentals: 355 years of evidence. Journal of Money, Credit and Banking, 45(2 3), Bailey, M. J., Muth, R. F., & Nourse, H. O. (1963). A regression method for real estate price index construction. Journal of the American Statistical Association, 58(304), Bank of England (2017). A millennium of macroeconomic data. Retrieved June 24 from: Banque National de Belgique ( ). Belgische Economische Statistieken ( ). Tijdschrift voor documentatie en voorlichting. Years , , Baulant, M. (1971). Le salaire des ouvriers du bâtiment à Paris de 1400 à In Annales. Histoire, Sciences Sociales (Vol. 26, No. 2, pp ). EHESS. Berghauser-Pont, M. Y., & Haupt, P. (2010). Spacematrix: space, density and urban form. Rotterdam: NAi Publishers. 28

29 Boumans, R., & Craeybeckx, J. (1974). Het bevolkingscijfer van Antwerpen in het derde kwart der XVIe eeuw. Tijdschrift voor geschiedenis, 60, Carmona, J., Lampe, M., & Rosés, J. (2016). Housing affordability during the urban transition in Spain. The Economic History Review. Cassiers, I. (1980). Une statistique des salaires horaires dans l'industrie belge, Recherches Économiques de Louvain/Louvain Economic Review, CBS Central Bureau of Statistics (2017a) Historische reeksen. Retrieved June 24, 2017: statline.cbs.nl CBS Central Bureau of Statistics (2017b) Prijzen: Consumentenprijsindex, Retrieved June 24, 2017: statline.cbs.nl CBS Central Bureau of Statistics (2017c) Lonen, CAO-lonen en arbeidskosten. Retrieved June 11, 2017: statline.cbs.nl CGEDD (2016). Le prix de l immobilier d habitation sur le long terme. Retrieved March 11, 2016: n a1048.html Clark, G. (2002). Shelter From The Storm: Housing And The Industrial Revolution, Journal of Economic History 62, Clark, G. (2005). The condition of the working class in England, Journal of Political Economy, 113(6), Cvrcek, T. (2013). Wages, prices, and living standards in the Habsburg empire, The Journal of Economic History, 73(01), Daumard, A. (1965). Maisons de Paris et propriétaires parisiens au XIXe siècle, Editions Cujas. Deprez, P. (1957). Het Gentse bevolkingscijfer in de tweede helft van de achttiende eeuw. Handelingen der Maatschappij voor Geschiedenis en Oudheidkunde te Gent, 11(1). De Thou, J. A. (1734). Histoire universelle depuis 1543 jusqu'en 1607 (Vol. 12). De Vries, J., & Van der Woude, A. (1997). The first modern economy: Success, failure, and perseverance of the Dutch economy, Cambridge University Press. Drelichman, M., & Agudo, D. G. (2014). Housing and the cost of living in early modern Toledo. Explorations in Economic History, 54, Durand, Y. (1966). Recherches sur les salaires des maçons à Paris au XVIIIe siècle. Revue d'histoire économique et sociale, 44(4),

30 Eichholtz, P.M.A., Hoesli, M., MacGregor, B.D., Nanthakumaran, N. (1995). Real estate portfolio diversification by property type and region. Journal of Property Finance 6, Eichholtz, P. (1997). A long run house price index: The Herengracht index, Real Estate Economics, 25(2), Eichholtz, P., Straetmans, S., & Theebe, M. (2012). The Amsterdam rent index: The housing market and the economy, Journal of Housing Economics, 21(4), Feinstein, C. H., & Pollard, S. (1988). Studies in Capital Formation in the United Kingdom Oxford University Press. Félibien, M. (1725). Histoire de la ville de Paris (Vol. 1). chez Guillaume Desprez, Imprimeur [et] Libraire du Roi. FOD Werkgelegenheid, Arbeid en Sociaal Overleg (2017). Indexcijfer van de conventionele lonen, arbeiders. Retrieved June 21, 2017: Francke, M. K. (2010). Repeat sales index for thin markets. The Journal of Real Estate Finance and Economics, 41(1), Glaeser, E. L., & Gyourko, J. (2002). The impact of building restrictions on housing affordability. NBER Working paper Goetzmann, W. N. (1992). The accuracy of real estate indices: Repeat sale estimators. The Journal of Real Estate Finance and Economics, 5(1), Gordon, R. J., & Van Goethem, T. (2005). A Century of Housing Shelter Prices: Is There a Downward Bias in the CPI? (No. w11776). National Bureau of Economic Research. Gyourko, J., & Linneman, P. (1993). Affordability of the American dream: An examination of the last 30 years. Journal of Housing Research, 4(1), 39. Gyourko, J., & Tracy, J. (1999). A Look at Real Housing Prices and Incomes: Some Implications for Housing Affordability and Quality. Federal Reserve Bank of New York Economic Policy Review, 5(3), 63. Henau, A. (1991). De Belgische huishuren gedurende het interbellum (Vol. 1)}. Katholieke Universiteit Leuven, Centrum voor Economische Studiën. Hillairet, J. (1963). Dictionnaire historique des rues de Paris. Paris: Editions de minuit. Hoffman, P. T., Jacks, D. S., Levin, P. A., & Lindert, P. H. (2002). Real inequality in Europe since Journal of Economic History, 62(02),

31 Holmans, A.E. (2005). Historical Statistics of Housing in Britain. Cambridge Centre for Housing and Planning Research, University of Cambridge, Cambridge. Humphries, J., & Weisdorf, J. (2016). Unreal Wages? A New Empirical Foundation for the Study of Living Standards and Economic Growth in England, (No. 310). Competitive Advantage in the Global Economy (CAGE). INSEE (2017). Macroeconomic database: prices and price indices. Retrieved May 20, 2017: Karagedikli, G., & Tunçer, C. (2016). The people next door. Housing and neighbourhood in Ottoman Edirne, Kingdom of The Netherlands ( ). Staatsblad van het Koninkrijk der Nederlanden. Kholodilin, K. A. (2016). War, housing rents, and free market: Berlin's rental housing during World War I. European Review of Economic History. Knoll, K., Schularick, M., & Steger, T. M. (2017). No price like home: Global house prices, American Economic Review. 107(2): Le Roy Ladurie, E., & Couperie, P. (1970, July). Le mouvement des loyers parisiens de la fin du Moyen Age au XVIIIe siècle. In Annales. Histoire, Sciences Sociales (Vol. 25, No. 4, pp ). EHESS. Lyons, R. (2015). Measuring House Prices in the Long Run: Insights from Dublin, Linneman, P. D., & Megbolugbe, I. F. (1992). Housing affordability: Myth or reality?. Urban studies, 29(3-4), Lesger, C. (1986). Huur en conjunctuur: de woningmarkt in Amsterdam, (Vol. 10). Historisch Seminarium van de Universiteit van Amsterdam. Lindenthal, T., Eichholtz, P., & Geltner, D. (2017) Land Assembly in Amsterdam, Regional Science and Urban Economics Marnef, G. (1996). Antwerpen in de tijd van de Reformatie. Ondergronds protestantisme in een handelsmetropool 1550, Marnata, F. (1961). Les loyers des bourgeois de Paris: A. Colin, Fondation nationale des sciences politiques. Michotte, F. (1937). L'évolution des prix de détail en Belgique de 1830 à Bulletin de l'institut des Sciences Économiques, 8(3), Monin, H., & Lazard, G. (1920). Sommier des Biens nationaux de la Ville de Paris. 31

32 Nicholas, T., & Scherbina, A. (2013). Real estate prices during the roaring twenties and the great depression. Real Estate Economics, 41(2), ONS (2016). RPI historical datasets 1947 to Retrieved May 28, 2016: k/ons/guide-method/user-guidance/prices/cpi-and-rpi/cpi-and-rpi-basket-of-goodsand-services/index.html ONS (2017). Experimental Index of Private Housing Rental Prices. Retrieved June 24, 2017: privatehousingrentalprices/previousreleases Özmucur, S., & Pamuk, S. (2002). Real wages and standards of living in the Ottoman empire, The Journal of Economic History, 62(02), Parker, H. (1957). Manpower A Study of War-time Policy and Administration. Her Majesty s Stationary Office & Longmans, Green and Co., London. Peeters, M. (1939). L'évolution des salaires en Belgique de 1831 à 1913.Bulletin de l'institut de Recherches Economiques, 10(4), Quigley, J. M., & Raphael, S. (2004). Is housing unaffordable? Why isn't it more affordable?. The Journal of Economic Perspectives, 18(1), Quigley, J. M., & Raphael, S. (2005). Regulation and the high cost of housing in California. The American Economic Review, 95(2), Raff, D., Wachter, S., & Yan, S. (2013). Real estate prices in Beijing, 1644 to Explorations in Economic History, 50(3), Rougerie, J. (1968). Remarques sur l'histoire des salaires à Paris au XIXe siècle. Le mouvement social, (63), Samy, L. (2015). Indices of House Prices and Rent Prices of Residential Property in London, (No. 134), Discussion Papers in Economic and Social History. Oxford. Scholliers, E. (1960). De levensstandaard in de XVe en XVIe eeuw te Antwerpen: Loonarbeid en honger. De Sikkel. Schrage, P., Nijhof, E., & Wielsma, P. (1989). Inkomensontwikkeling van werkenden en werklozen in Nederland, '. Tijdschrift voor Sociale Geschiedenis, 15, Segers, Y. (1999). Huishuren in België, Reconstructie en analyse van een nationale huurprijsindex. Tijdschrift voor Sociale Geschiedenis, (2), Singer-Kérel, J. (1961). Le coût de la vie à Paris de 1840 à Paris: A. Colin. 32

33 Smits, J. P. H., Horlings, E., & Zanden, J. L. (2000). Dutch GNP and its components, (p. 115). Groningen: Groningen Growth and Development Centre. Stone, M. E. (2006). What is housing affordability? The case for the residual income approach. Housing policy debate, 17(1), Van den Eeckhout, P., & Scholliers, P. (1979). De Brusselse huishuren: Vrije Universiteit Brussel. Van der Wee, H. (1963). Growth of the Antwerp Market and the European Economy, 14th to 16th Centuries, 3 vols. The Hague, (6). Van Riel, A. (forthcoming). Prices and economic development in the Netherlands, ; Markets institutions and policy restraints. Universiteit Utrecht. Van Ryssel, D. (1967). De Gentse huishuren tussen 1500 en 1795: bijdrage tot de kennis van de konjunktuur van de stad. Pro Civitate, Van Zanden, J. L. (2008). Prices and Wages and the Cost of Living in the Western Part of the Netherlands, International Institute of Social History List of Datafiles of Historical Prices and Wages.. Amsterdam. Available online at: Verlinden, C. (Ed.). ( ). Dokumenten voor de geschiedenis van prijzen en lonen in Vlaanderen en Brabant: Documents pour l'histoire des prix et des salaires en Flandre et en Brabant. Vol I.-Vol IV., 125 e, e, 153 e and 156 e editie. De Tempel, Brugge. Wyffels, A. (1958). De omvang en de evolutie van het Brugse bevolkingscijfer in de XVIIde en de XVIIIde eeuw. Revue belge de philologie et d'histoire, 36(4), Archived sources Archives de l Assistance Publique Hopitaux de Paris Fosseyeux archives, sous-series 782. [various documents relating rents] Registres des loyers des maisons et des maisons vendues. 782 FOSS and 782 FOSS Fosseyeux archives, sous-series 7M. [various documents relating to hospital finances] Registres de caisse et de comptabilité générale. 7M FOSS 1-35 Archives Nationales, Paris, site Pierrefitte-sur-Seine EHESS CRH archives. [transcribed rental contracts] Enquete sur les loyers parisiens XVe XVIIe siècle. Inventory , boxes 1-7 CRH archives, fonds Emmanuel le Roy Ladurie. [scattered rent documents] Enquete sur le batiment. Inventory , box

34 Stadsarchief Amsterdam Burgerweeshuis, oud archief [ledgers] Huur- en kasboeken. Archief 367.C, inv. nr. 100, and 320. Burgerweeshuis, nieuw archief. [ledgers] Verhuurregisters. Archief 367.A, inv. nr. 114, 135, 136 and 136A. Archief van de Verenigde Doopsgezinde Gemeente en rechtsvoorgangers. [ledgers and rent contracts] Huurboeken, kasboeken en huurcontracten. Archief 1120, inv. nr , , Archief van het Rooms-Katholiek Jongensweeshuis [ledgers and rent contracts] Huurboeken, kasboeken en huurcontracten. Archief 191, inv. nr Archief van de Remonstrantse Gemeente [rent contracts] Huurcedulen en ingekomen brieven betreffende huur. Archief 632, inv. nr. 432 Archief van de Maatschappij Nederland tot Exploitatie van Onroerende Goederen N.V. [rent books] Huurboeken en huurstaten. Archief 613, inv. nr ,

35 4. List of Tables and Figures Figure 1A: location of rental properties in Amsterdam Figure 1B: 1832 value of all lots in Amsterdam Notes: The upper figure shows the location of each of the homes in the Amsterdam sample between , with each home indicated with a red dot. Multiple rental units on the same location (e.g. rooms, or ground and upper floor rented separately) are indicated as one point. The plot below shows the 1832 value of all lots, from Lindenthal, Eichholtz & Geltner (2017). The dark red areas correspond to the most valuable lots, whereas the dark blue correspond to the lowest values. 35

36 Figure 2: Real rent indices, Notes: The figure displays rental indices for three major European cities and the aggregated Belgian index, in log scale. The rents are indexed with 1900 as base year. Missing values are interpolated linearly. Figure 3: Nominal rent indices, Belgium, Notes: The figure displays the rental indices for the four Belgian cities, in nominal values (log scale). The base year of the index is

37 Figure 3: Real Wages, Notes: The above plot shows the combined development in real wages in Europe. Data switch in the first half of the 20 th century to national indices. Missing values are interpolated linearly. Figure 5: Real rent and consumer price indices, Notes: The transparent upper lines represent the consumer price indices in each of the cities, with 1913 chosen as base year. The lower lines represent the real rental indices, indexed at 50 in The grey lines correspond to the start and end of WWI and WWII. 37

38 Figure 6: Indexed market rent-to-wage ratio s, Figure 7: Implied housing quality index Notes: The indices have been smoothed using moving averages. The lighter blue line corresponds to the implied quality changes in Amsterdam when using the average Dutch rent index from Van Riel (forthcoming). For comparison, we have plotted the decadal quality index of Clark (2002) for England and Wales. 38

39 39