Article 5 – Inequality of real estate wealth in early modern Stockholm, 1730–1850
3. Theory and earlier research
The aims of this thesis relate to several fields in economic history. The research literature that it has to position itself against is relatively vast. Each article incorporates a discussion of relevant theory and earlier research. But some overarching themes are important for the thesis as a whole and will therefore be discussed in this section.
Real estate prices and economic instability
A major theme in the first three articles of this thesis is the development of real estate prices and periods of price turbulence. Behind this lies a theoretical insight that house prices may have a severe impact on the rest of the economy.
Therefore, it is important to dwell on this supposed connection.
Real estate prices affect the real economy mainly through two mechanisms:
1) construction activity tends to rise if house prices increase faster than production costs and, 2) an improvement in the household sector’s balance sheet can produce wealth effects that increase aggregate demand. Both mechanisms work in the opposite direction if prices fall.16
The balance sheet mechanisms concern people’s wealth portfolios and are mediated through financial markets. Real estate is used as loan securities, and rising prices consequently increase the opportunities to borrow money. Loans can then be used for residential investments or to buy other goods. Therefore, rising real estate prices are together with growing private indebtedness associated with higher consumption. Reversely, falling prices have a negative impact on aggregate consumption as wealth effects are canceled. A more pessimistic expectation of future economic development makes households less inclined to both consume and invest.17
However, this effect is dependent on the financial institutional setting that regulates the access to credits. Given that housing is a major part of household wealth and an often used collateral for loans, house prices strongly impact the macroeconomy and financial markets.18 If mortgages are increasing in number and volume, households become more exposed to price fluctuations. The asset
16 Eurostat, Handbook on Residential Property Prices Indices (RPPIs).
17 Girouard, ‘Housing and Mortgage Markets: An OECD Perspective’; Goodhart and Hofmann, House Prices and the Macroeconomy; Hilbers, Lei, and Zacho, ‘Real Estate Developments and Financial Sector Soundness’; Miles, Housing Financial Markets and the Wider Economy; Swartz, ‘Finansiella Kriser Och Depressioner’.
18 Goodhart and Hofmann, House Prices and the Macroeconomy.
18
value of houses has become decisive for the balance sheets of a growing number of households, and so have repayment conditions and rent levels.19
The importance of the balance sheet position of investors, households, and enterprises has been most famously theorized by Minsky. His insights have been employed by various researchers to analyze house prices and related financial risks.20
According to Minsky, enterprises in times of economic growth are rewarded with higher profit when they take on higher debt. This creates an incentive for them to increase their debt level to compete with other actors on the market. As long as the business cycle is good, this leads to greater prosperity. But as increased debt levels are a means for increasing profit, there is a tendency toward greater speculation inherent in the economy. This may eventually lead to the buildup of a speculative bubble.
The thesis of Minsky is that stability and growth create the incentives for bubbles, thus creating the preconditions for future instability. If the business cycle turns and the expectation of the future becomes pessimistic, this will negatively affect inflated asset prices. And falling asset prices will lead to bad balance sheet positions for those who have taken on large debts to buy assets.
As the worsening balance-sheet positions harm investments and demand, through the mechanisms described above, this might cause a downward spiral.
Falling prices strengthen depressive tendencies in the real economy, while depressive tendencies in the real economy put even more downward pressure on prices.21
Minsky’s financial instability hypothesis is not explicitly tested in any of the papers in this thesis. It does however provide a theoretical ground from which we can understand why real estate prices are important to study to assess the wider economic history. Now, the nexus between financial markets, the construction sector, and the real estate market has undergone many changes over the last 300 years. These must be taken into consideration to assess the meaning of historical price changes.
Stockholm’s real estate market from a historical perspective
Stockholm was a city in economic and demographic stagnation during large parts of the period 1750-1850. The causes for this were numerous. The city’s importance as a central town for trading diminished as the Baltic trade lost
19 Lapavitsas, Profiting without Producing. How Finance Capital Exploits Us All;
Dymski, ‘Bank Lending and the Subprime Crisis’.
20 See discussion in Kincaid, ‘Marx after Minsky: Capital Surplus and the Current Crisis’.
21 Minsky, ‘Uncertainty and the Institutional Structure of Capitalist Economies’;
Minsky, Stabilizing an Unstable Economy.
19
importance when the Atlantic trade grew.22 At the same time, its textile industry was to a large extent out-rivaled by other Swedish towns, mainly Norrköping. Lars Magnusson notes that between 1816 and 1845, textile production declined by 60 percent in Stockholm, while it increased by 460 percent in Norrköping.23
The stagnation is visible in figure 1, which depicts the population of Stockholm from 1730 to 2020. Even more telling is that the population level during this period was upheld by immigration. Due to the high death rates, the city was not able to reproduce the number of inhabitants by itself.24 In this period, few new houses were built. The total number of houses in 1766 was 4 462, and by 1782 the number had fallen to 4 137. The number of houses and plots in 1828 was 4 526.25
During this period, the construction of new houses was mainly paid for by rich merchants and industrialists. They were often built to inhabit the owner and his family, and maybe a business and/or tenants.26 The building contractors had no modern financial market to turn to. The private credit market was the major credit market, and personal relations were probably more important than banks or other financial institutes for access to money.27
With the industrialization of 1850, Stockholm experienced economic and demographic growth.28 Among many fundamental changes, the guild system was abolished in 1846. That allowed anyone to set up a construction company.
These changes had a huge impact on the housing sector, as smaller entrepreneurs set up construction businesses to profit from the growing housing demand. They often lacked the capital for investments. Instead, they relied on credits – mainly from smaller financial institutions, private loans, and suppliers of raw materials and labor – to finance their projects.29
22 Söderberg, Jonsson, and Persson, A Stagnating Metropolis. The Economy and Demography of Stockholm, 1750-1850.
23 Magnusson, Sveriges ekonomiska historia, 216; For a discussion on Norrköpings success and Stockholm failure when it comes to textile production, see for example Nyberg, Kommersiell kompetens och industrialisering; Schön, Från hantverk till fabriksindustri.
24 Stockholms utredningskontor, ‘Befolkningen i Stockholm 1252-2005’; See also discussion in Söderberg, Jonsson, and Persson, A Stagnating Metropolis. The Economy and Demography of Stockholm, 1750-1850.
25 Kungliga Statistiska Centralbyrån, Statistisk tidskrift.
26 Perlinge, Bubblan som sprack. Byggboomen i Stockholm 1896-1908;
Hammarström, ‘Urban Growth and Building Fluctuations. Stockholm 1860-1920’.
27 Nyberg, ‘The Early Modern Financial System and the Informal Credit Market’, 18.
28 Ahlberg, Stockholms befolkningsutveckling efter 1850; Cederqvist, Arbetare i strejk. Studier rörande arbetarnas politiska mobilisering under industrialismens genombrott. Stockholm 1850-1909; Gustafson, Industrialismens storstad. Studier rörande Stockholms sociala, ekonomiska och demografiska struktur 1860-1910.
29 Perlinge, Bubblan som sprack. Byggboomen i Stockholm 1896-1908; Gejvall, 1800-talets stockholmsbostad, 39ff.
20
Figure 1: Population of Stockholm, 1730–2020.
Source: Utrednings- och statistikkontoret (2005) and Statistiska centralbyrån.
At the same time, the financial sector underwent major changes. Between 1824 and 1846, several banking acts were passed that together marked a transition from a mercantilist legislation toward a more liberal regulation.30 The changes accelerated, eventually leading to what has been characterized as a financial revolution, taking place somewhere between 1850 and 1870. An important part of this revolution was the liberalization of the banking system, with the banking act of 1864 as an institutional milestone. 31 Before, banks had to have a new charter accepted by the Crown every ten years. The new act standardized this procedure and turned charters prolongation into a formality.
Moreover, the act stipulated the rights of holders of bank liabilities and an absolute right to redeem these at the bank of issue. Whit the new banking act, financial services rapidly increased in Sweden.32 Between 1860 and 1900, the volumes on the Swedish credit market almost doubled every tenth year.33
To finance the urban construction sector, mortgage institutes had been used least since 1865. By the turn of the century, mortgage institutes financed between 15 and 20 percent of investments in urban buildings, insurance companies had the same share, commercial and the national bank held around
30 Andersen, The Evolution of Nordic Finance, 144f.
31 Ögren, ‘The Swedish Financial Revolution: An In-Dept Study’, 9f.
32 Ögren, 10f.
33 Petersson, ‘Cooperation, Interbank Markets and Bank Industry Networks: The Growth and Characteristics of Swedish Bank Lending, 1860–1910’, 64.
21
20 percent, and savings banks around 25 percent. During and after the international financial crisis in 1907, interest rates grew. To ensure the availability of cheap credits for urban housing, Stadshypotekskassan was founded by the state in 1909.34
From 1850 onwards, the construction activity in Stockholm was driven by the increase in economic activity in the wake of the industrialization process and the dramatic population increase that accompanied it. This building activity had a clear cyclical pattern which correlated with the general economic activity and the availability of credits.35 Economic turmoil could quickly turn into a wave of bankruptcies as the small building companies were dependent on a constant flow of credits to maintain production.36 Since Sweden was dependent on capital import during the early industrialization, the economy was also sensitive to international financial turmoil.37
After experiencing several financial crises, the state took on a more active role from the 1930s and onwards. This was, of course, partly due to the democratization of Sweden and the beginning of the political dominance of the Social-democratic labor movement. From early on, housing was targeted as an important sector to dampen business cycles. The central bank, Riksbanken, was used to transfer financial risk away from the construction enterprises to the state in exchange for them building cheap rental apartments and cooperative houses (bostadsrätter).38
Even if this housing policy was unique for Sweden, financial regulation was a global phenomenon. In most high-income countries, these regulations were dismantled mainly during the 1980s, which many theorists describe as an epochal change in the history of capitalism. Financial incentives had an increasing impact on real production.39 In Sweden, this process more or less coincided with the abandoning of a more interventionist housing policy.40 One
34 Larsson, ‘The State and the Financial System: Regulation and Regime Change around 190’, 172.
35 Hammarström, ‘Urban Growth and Building Fluctuations. Stockholm 1860-1920’.
36 Hammarström.
37 Kock, Kreditmarknad och räntepolitik 1924-1958; Edvinsson, ‘Major Recessions in Sweden 1850–2000’.
38 Larsson and Söderberg, Finance and the Welfare State Banking Development and Regulatory Principles in Sweden, 1900–2015; Larsson, Staten Och Kapitalet;
Englund, ‘Bostadsfinansieringen och penningpolitiken’; Jonung, ‘Kreditregleringens uppgång och fall’.
39 Krippner, Capitalizing on Crisis. The Political Origins of the Rise of Finance;
Stockhammer, ‘Financialization and the Global Economy’; Wray, ‘The Rise and Fall of Money Manager Capitalism’.
40 Strömberg, ‘Sweden’; Stenfors, ‘Financialisation and the Financial and Economic Crises: The Case of Sweden’; Stenfors, ‘Swedish Financialisation: “Nordic Noir” or
“Safe Haven”?’; Jacobson, .... ‘.... Och Mödan Gav Sin Lön. Om Bostadspolitik Och Bostadskooperation i Stockholm 1870-1930’.
22
of the conspicuous results of this in both Sweden and on a global level is the rise of household indebtedness based on mortgages.41
The above analysis of the trajectory of Stockholm’s real estate market in relation to construction activity and financial markets is obviously simplified.
Still, it points toward two periods under which the interrelation between the real estate market and the financial sector was more important: first from around 1850 to 1930, second from 1980 until today. During both these periods, financial deregulations created access to private credit, that found its way to the real estate market. Given the financial instability hypothesis outlined in the previous section, one might expect that price turmoil would have been more regularly occurring in these periods and possibly had a worse real economic effect.
Dominating approaches to constructing real estate price indices
Previous sections have outlined the theoretical and historical relationship between real estate prices and the real economy. A take-away from that is the importance of real estate price development. In recent years, the number of studies attempting to assess the long-run trajectory of real estate prices has increased. Still, only a few studies exist that cover periods before the twentieth century and they typically employ similar methods.
As a major part of this thesis consists of constructing new price indices, it is interesting to see how many other such studies there are. But, to be able to assess them, an understanding of the most common methods used. These methods will be briefly described in this section.
Three dominating approaches exists for real estate price index construction:
hedonic regression, repeated sales (RS) analysis, and sales appraisals ratios (SPAR).
Hedonic regressions include relevant indicators of the qualitative characteristics of the properties sold and estimate a regression model. The RS method uses information about the price changes between sales of individual properties. This requires the assumption of constant quality of individual properties between each sales pair; it is often controlled for by gathering information about renovations. The SPAR method combines sales data with taxation records for each property included in the dataset. Appraisals of taxation values between taxation years are supposed to control for qualitative changes in the properties sold.
41 Waldenström, ‘The National Wealth of Sweden, 1810-2014’; Poppe, Collard, and Jakobsen, ‘What Has Debt Got to Do with It?’; Karacimen, ‘Dynamics behind the Rise of Household Debt in Advanced Capitalist Countries: An Overview’; Turk,
‘Housing Price and Household Debt Interactions in Sweden’.
23
Table 1 summarizes the specific kind of data each method requires. They all use some kind of data that reflects the market price of properties. Normally, this can be derived from some kind of transaction record archived by a state or city institution.
It is not in theory necessary to identify individual properties in hedonic regressions, which is the case when using RS analysis or SPAR. Hedonic regressions always require numerous other variables that together describes the qualitative characteristics of the sold item. This might include area, heating, location and so on. What characteristics to include is not predetermined. When constructing historical series, data availability often sets the limits. No other variables are needed for the RS or SPAR methods.
Except for transaction records, hedonic regressions do not require other types of sources. In practice, however, one might have to obtain information that is not part of the transaction records in order to be able to do a regression.
In those cases, other sources might be necessary. In theory, the RS method also only requires a market price and property ID. But in practical analysis one often needs to control for possible qualitative changes in the real estate stock in-between sales. In those cases, other sources might be necessary. The SPAR method requires appraisal values. They are sometimes included in transaction records, but most often have to be collected from another archive.
Table 1: Source requirements of different methods.
Method Market
Long run real estate price series are uncommon. Most series start around 1960 or later. Series that stretches back into the 1800s are rare. This section provides a brief overview over existing long-run indices. The time span covered for
24
different cities are summarized in Figure 2, with their linetype indicating which method is used.
As can be seen, only Stockholm’s index also covers the Medieval age. It was constructed by Franzén and Söderberg and covers the years 1300–1600.
Due to the character of their source material, they were not able to use any of the three methods listed above. They instead rely on weighted median values, divided into stone houses and wooden houses.42
Other studies on Swedish cities includes real estate indices for Gothenburg for the years 1875 to 1957;43 Stockholm between 1818 to 187544 and 1875 to 1957.45 In all three cases, a SPAR method were used.46
Karagedikli and Tunçer have constructed a price index for Ottoman empire Edirne, 1720–1814. They estimated a hedonic regression with numerous variables that gives information about house quality, neighborhood and geographical location.47
One of the most famous indices is Shiller’s index for the US 1890–2014.48 His index is a combination of different indices constructed with different methods. It has been criticized for applying partly outdated methods.49 Recently, Jonathan D. Rose has presented a new repeated sales index for Baltimore that cover the years 1880–1953.50
Deeter, Duffy and Quinn (2016) have estimated real estate prices in Dublin 1708–1949. They use a hedonic regression that provides information about location, if a garage or garden was included and if the transaction included numerous properties.51 Ronan Lyons employs a similar method for Dublin 1900–2015 but adds the size of sold dwellings to his model.52
One of the first studies that attempted a repeated sales method was conducted by Gaston Duon. He constructed a house price index for Paris 1840–1944 already in 1946. Duon’s results have been used by Friggit. This index has been linked to two other Paris indices. Already in 1894,
42 Franzén and Söderberg, ‘Hus, gårdar och gatubodar. Fastighetspriser i Stockholm och Arboga 1300–1600’.
43 Bohlin, ‘A Price Index for Residential Property in Göteborg, 1875-2010’.
44 Edvinsson, Eriksson, and Ingman, ‘A Real Estate Price Index for Stockholm, Sweden 1818–2018’.
45 Edvinsson, Blöndal, and Söderberg, ‘A Price Index for Residential Property in Stockholm, 1875–2012’.
46 For comparison, all studies also constructed an index with a repeated sales method.
47 Karagedikli and Tunçer, ‘House Prices in the Ottoman Empire’.
48 Shiller, Irrational Exuberance.
49 Lyons, ‘Measuring House Prices in the Long Run: Insights from Dublin,1900-2015’, 5.
50 Rose, ‘Reassessing the Magnitude of Housing Price Declines and the Use of Leverage in the Depressions of the 1890s and 1930s’.
51 Deeter, Duffy, and Quinn, ‘Dublin House Prices’.
52 Lyons, ‘Measuring House Prices in the Long Run: Insights from Dublin,1900-2015’.
25
Georgesd’Avenel presented a real estate price index for the centre of Paris 1200–1800. Duon built on that material to in 1946 present an index for Paris 1625–1850. d'Avenel’s series builds on transaction prices averaged over 25-years periods, and he did not control for changes in the housing stock. Duon presented average prices calculated over 10 to 50 years for the years 1625 to 1790. Between 1790 and 1850, he used the repeated sales method to calculate an index for 10 years periods.53 Paris stand out for its long indices. But an annual index building on modern econometric methods exists only after year 1840. The price changes 1200–1840 should be used with caution.
For Beijing, Raff, Wachter and Yan have constructed an index for the period 1644–1840. They use a hedonic regression with information about number of rooms, geographical location, number of court yards, if the house had a well and building material.54
Figure 2: Cities for which long run real estate price indices exists.
Figure 2: Cities for which long run real estate price indices exists.