Introduction

Since at least 1940, there was a remarkable increase in the size of homes and a significant improvement in the overall availability of housing worldwide. For instance, the average floor space per person in Germany more than doubled from 18.4 m\(^2\) in 1956 to 47.5 m\(^2\) in 2023.1 Improved income levels welfare enabled construction of larger homes. The baby boom of the 1950s and 1960s also apparently contributed to this process, as larger families required more spacious dwellings.

However, a turning point emerged in the late 1990s — early 2000s, marking the beginning of the reversal of this trend. This transformation is not a phenomenon that occurred in just one country. Similar developments were observed in various European countries, as well as in Japan and the United States. Three factors appear to contribute to this shift. First, private households are becoming smaller and smaller, and smaller households require smaller homes. Second, rising housing prices make larger dwellings less affordable both during recurrent housing price booms and in the long run. Third, some municipalities are adopting bans on the construction of single-family homes that are typically larger than the dwellings in multi-family houses.

Evidence of diminishing size of new housing

First, I will examine the data on the size of new dwellings. Depending on the national statistics, the data are either on building permits, housing starts, or completions of dwellings. These indicators are closely related: an increase in the average size of approved dwellings means an increase in the average size of completed dwellings a few years later.

The figure below shows the evolution of the average size of new dwellings in selected countries. The choice of these countries is dictated mainly by the availability of the long time series on the floor spaces of newly built dwellings.

Figure 1: Average size of newly built housing in selected countries

Figure 1: Average size of newly built housing in selected countries

Sources: Statbel, Destatis, Statistics Denmark, INSEE, Central Bureau of Statistics of Irland, ISTAT, Statistics Bureau of Japan, SSB, Statistics Poland, CSU RSFSR, and Rosstat

Note: Dashed vertical lines indicate the year when the peak in new housing size was reached.

Even among developed countries, there are significant differences in dwelling size. Due to the large proportion of single-family houses in the total housing stock and in the newly built housing, the dwelling sizes in Belgium and Norway are substantially larger than in other countries in the sample. In Japan, on the other hand, where the urban landscape is dominated by high-rise buildings, newly built dwellings are much smaller. This is because apartments in multi-family buildings are generally significantly smaller than single-family homes.

Despite these differences in size, a common trend can be observed in all countries shown in the figure above. The average size of newly built dwellings increased over several decades. At some point, however, a peak was reached and the size began to decline. The exact timing of this peak varies from country to country. In Belgium, Japan, and Norway, for example, it was reached before the turn of the century. Germany, France, Poland, and Russia experienced this shift between 2002 and 2005, while Denmark and Ireland followed a similar pattern after the Great recession of 2008–2009. In Italy, the average living space of new dwellings began to decline in 2019.

Possible explanations of declining home size

Four factors — both on supply and demand side — have potentially contributed to the phenomenon of decreasing average size of the newly built dwellings.

First, the current demographic transition leads to shrinking household sizes and smaller families require less living space (Ellsworth-Krebs 2020; Snell 2017). In Germany, for example, the average household size fell from 2.9 to 2 persons between 1961 and 2021, while the proportion of single-person households rose from 21% to 41% over the same period. In major cities like Berlin, Hamburg, and Munich, this figure oscillates around 50%. By contrast, in 2021, the share of 1-room dwellings in the total housing stock was just 3.5%, while that of 1- and 2-room dwellings was about 13%.2 In Berlin, the proportion of smaller dwellings is somewhat higher (4.9% for 1-room dwellings, and 23.2% for 1- and 2-room dwellings), but is still well below the proportion of single-person households.3 Figure below compares the distribution of dwellings by the number of rooms (including kitchens) to the distribution of households by the number of persons in Germany and in Berlin.

Figure 2: Mismatch between household and housing stock in Germany

Figure 2: Mismatch between household and housing stock in Germany

Sources: Destatis, Berlin-Brandenburg Statistik, and own calculations

As seen, while the majority of dwellings has four rooms or more, most households are comprised of 1 or 2 persons. This testifies a large mismatch between the housing stock and the households. The reason for this discrepancy is that the housing stock changes only slowly, while demographic processes are faster. This is because buildings remain in place for several generations and are rarely replaced. For example, in 2022, the proportion of homes built before 2000 in Germany was over 87%.4 In developed countries, the structure of the housing stock therefore corresponds to the demographic structure of the mid-20th century. However, when the number of smaller private households increases to such an extent that their needs can no longer be met by the existing housing stock, the construction industry begins to respond by building more smaller dwellings.

Second, since 2010, housing prices and rents rose a lot, thus, making it increasingly challenging for families to afford larger dwellings. Hence, many families must downgrade the size of housing they are going to buy or rent. Many are forced to give up their dream to live in a single-family house and switch to dwellings in multi-family houses (MFH) (Lee, Myers, and Park 2000).

Third, the rising housing prices incentivize investors to build multi-family houses instead of single-family houses, since the former allow them obtaining higher returns on their investment than the latter: the same building plot can be used to build a handful of single-family houses or a large building with dozens of flats. This means that the increased land costs are spread across several apartments, reducing the proportion of land costs in the total construction costs per apartment. Indeed, there is a notable transformation in the composition of newly constructed housing, with a greater emphasis on multi-family dwellings. In Germany, for instance, between 2001 and 2022, the proportion of dwellings in new multi-family residential buildings (buildings with 3 and more dwellings) compared to the total number of newly built dwellings in residential buildings increased from 34% to 58%. Typically, flats within multi-family houses are smaller than single-family homes. Figure below depicts the share of dwellings in newly built multi-family houses (residential buildings with three dwellings or more) in the total number of newly completed residential buildings.

Figure 3: Multi-family buildings crowd out the single-family houses

Figure 3: Multi-family buildings crowd out the single-family houses

Sources: Destatis and own calculations

The percentage of newly constructed residential units in multi-family buildings exhibits a remarkably consistent cyclical trend, with each cycle lasting approximately 8 to 15 years. These cycles bear a striking resemblance to housing booms: during the periods of growth in these booms, the proportion of multi-family constructions rises, while in the downturn phases, it declines (see Figure below).

Figure 4: Housing booms in Germany: real house prices

Figure 4: Housing booms in Germany: real house prices

Source: Destatis, OECD, and Macrohistory

The cycle in the real estate market is illustrated here using the ratio of purchase price to rent. This ratio is often used as an indicator of speculative real estate price bubbles. This is because during booms, the purchase price can deviate significantly from its fundamental value, whereas the rental price remains close to this value. When purchase prices rise, more properties are bought in order to sell them at a higher price that is expected in the future. In such periods, purchase prices rise faster than rents, which increases the purchase price-to-rent ratio. The correlation between the proportion of apartments in multi-family buildings and the purchase price-to-rent ratio is not perfect, but it is clearly discernible. As a rule, the increase in the proportion of apartments in multi-family buildings follows the increase in the purchase price-to-rent ratio with a delay. Due to the comparatively long construction period, builders cannot react immediately to price changes.

Similar relationship between housing price-to-rent ratio and number of dwellings in multi-family houses can be observed for Sweden between 1938 and 2022 (see Figure below).

Figure 5: Fluctuations of multi-family dwellings share in Sweden

Figure 5: Fluctuations of multi-family dwellings share in Sweden

Sources: SCB, OECD, Macrohistory, and own calculations

The relation is not perfect. However, three housing market booms since 1950 are accompanied by surges in the construction of multi-family housing.

Not only is the proportion of apartments in multi-family buildings increasing, but multi-family buildings are also becoming larger and the apartments in them smaller. Figure below shows the change in the average living area per dwelling for houses with 1, 2, and 3 and more dwellings (left panel) and the average number of dwellings per multi-family house in Germany between 1993 and 2024. Moreover, similar indicators are shown for West Germany (dashed lines), since for it data are available starting from 1980.

Figure 6: Sizes of newly built dwellings in Germany by building types

Figure 6: Sizes of newly built dwellings in Germany by building types

Sources: Destatis and own calculations.

Between 2004 and 2024, the average number of apartments per multi-family house rose from around 7.5 to more than 10.5. Until 2015, the average living space per newly built apartment increased. Since 2015, however, the floor space of single-family and multi-family houses has developed differently. The trend of increasing dwelling sizes in single-family houses continues, contributing to the overall growth in house dimensions. Conversely, it is the size of dwellings in multi-family houses that has driven the recent decline in newly built housing sizes. Interestingly, the data for West Germany indicate that the dwelling size seems to follow a cyclical pattern similarly to the proportion of dwellings in multi-family houses.

A similar trend is observed in the United States. The figure below shows how the average size of completed dwellings has changed in multi-family and single-family houses.

Figure 7: Average size of newly built housing in the USA

Figure 7: Average size of newly built housing in the USA

Sources: US Census Bureau and own calculations

As can be seen, the United States experienced a decline in the size of newly constructed dwellings in multi-family homes beginning in 2008, which coincided with the start of the Great Recession. In contrast, the size of single-family dwellings continued to increase until 2015, after which it began to decrease.

Fourth, governmental land-use regulations can play a role. Thus, at the beginning of the 2020s, some municipalities in Germany (for example, in Hamburg, Munster, and Wiesbaden) have begun to ban the construction of single-family homes.5 On the one hand, such houses consume too much of the space that is scarce, especially in large cities. On the other hand, they are considered environmentally unfriendly due to their higher energy consumption and the fact that they presumably lead to more soil sealing than the multi-family buildings. Therefore, the environmentally concerned politicians are fighting against this form of housing in urban areas.

Discussion

The above evidence shows that the discrepancy between the demand and supply of housing is widening. Households are becoming smaller and demanding smaller dwellings, yet the existing housing stock is mainly made up of large dwellings. This is largely a result of decisions made decades ago when the demographic situation was radically different. For example, in 2018, three-fourths of the housing stock in Germany were dwellings constructed 30 years or more ago. The shortage of affordable housing that many societies face today is partly related to this discrepancy. This issue can be solved by changing the structure of newly built housing, which is already happening, as well as by rebuilding existing housing to adjust it to the current population structure.

The observed reduction in floor space in newly built housing also has positive implications for reduced energy consumption. Research shows that living space is the main determinant of residential energy consumption (Huebner et al. 2015; Huebner and Shipworth 2017).

Due to the low construction rates observed in developed countries, it will perhaps take decades before the smaller sizes of newly built housing noticeably affect the average floor space of existing homes. However, it is important to note that we are already in the midst of this process.

References

Ellsworth-Krebs, Katherine. 2020. “Implications of Declining Household Sizes and Expectations of Home Comfort for Domestic Energy Demand.” Nature Energy 5 (1): 20–25.
Huebner, Gesche M, Ian Hamilton, Zaid Chalabi, David Shipworth, and Tadj Oreszczyn. 2015. “Explaining Domestic Energy Consumption — the Comparative Contribution of Building Factors, Socio-Demographics, Behaviours and Attitudes.” Applied Energy 159: 589–600.
Huebner, Gesche M, and David Shipworth. 2017. “All about Size? — the Potential of Downsizing in Reducing Energy Demand.” Applied Energy 186: 226–33.
Kholodilin, Konstantin A., and Dirk Ulbricht. 2014. “Mietpreisbremse: Wohnungsmarktregulierung bringt mehr Schaden als Nutzen.” DIW Wochenbericht 81 (15): 319–27.
Lee, Seong Woo, Dowell Myers, and Heon Soo Park. 2000. “An Econometric Model of Homeownership: Single-Family and Multifamily Housing Option.” Environment and Planning A 32 (11): 1959–76.
Snell, Keith DM. 2017. “The Rise of Living Alone and Loneliness in History.” Social History 42 (1): 2–28.

  1. See the data in Kholodilin and Ulbricht (2014) and those of the German Statistical Office.↩︎

  2. See Destatis “Wohnungsbestand in Deutschland”; https://www.destatis.de/DE/Themen/Gesellschaft-Umwelt/Wohnen/Tabellen/liste-wohnungsbestand.html#115202.↩︎

  3. See Statistik Berlin Brandenburg “Wohnungen nach der Anzahl der Räume: Anteil in %”; https://www.statistik-berlin-brandenburg.de/f-i-1-j.↩︎

  4. Ergebnisse des Zensus 2022 — Gebäude- und Wohnungszählung.↩︎

  5. See, for instance, Jana Werner (2021) “In Hamburg ist ein Traum linker Ideologen wahr geworden” Welt (https://www.welt.de/regionales/hamburg/article225612769/Verbot-von-Einfamilienhaeusern-In-Hamburg-ist-ein-Traum-linker-Ideologen-wahr-geworden.html), Oliver Bock (2023) “Dem Häuslebauer geht der Boden aus” Frankfurter Allgemeine Zeitung (https://www.faz.net/aktuell/rhein-main/frankfurt/neubau-wiesbaden-will-keine-einfamilienhaeuser-mehr-zulassen-18913524.html), and Amy Walker (2023) “Klimaschutz: In diesen Gemeinden dürfen keine Einfamilienhäuser mehr gebaut werden” Merkur (https://www.merkur.de/wirtschaft/in-diesen-gemeinden-duerfen-keine-einfamilienhaeuser-mehr-gebaut-werden-klima-92302511.html).↩︎