Abstract
This study describes a large longitudinal database of governmental rental market regulations. The regulations are measured using binary variables based on a thorough analysis of the real-time country-specific legislation. Three major restrictive policies are examined — rent control, protection from restriction, and housing rationing. Currently, the database covers 132 countries and states between 1910 and 2020. This allows comparisons of regulation intensity across both time and space. The analysis reveals a surge of all restrictive policies in the first half of the 20th century. However, following World War II, the evolution of policies diverged — while rent control became more flexible or was phased out, tenure security stabilized at a high level or even increased, while housing rationing became used less frequently.In this study, we describe the methodology and data used to construct rental housing market regulation indices. The database is discussed in Konstantin A. Kholodilin (2020). Our aim here is to document the database by providing the list of legal acts and to provide some raw data. In addition, given that the database is constantly evolving, we can present here its most up-to-date version.
Everybody desiring to contribute to the further development of the database is welcome!
In our approach, the quantification of legal acts is carried out in several steps. This section describes the whole algorithm. Its purpose is to make the approach used here as transparent and replicable as possible. The first step consists of exploring the literature that summarizes governmental housing market regulations in the country of interest. In a few select cases, a nice and systematic description of the evolution of such legislation exists. The main sources of such information are the Tenlaw project at the Universitaet Bremen1 for the 28 European Union member states plus Japan, Norway, Serbia, Switzerland, and Turkey; the “Tenancy Law and Procedure in the EU” project of the European University Institute in Florence2 for 13 EU member states plus Switzerland; International Labour Office (1924) for the origins of housing policies in 17 European countries (Austria, Belgium, Czechoslovakia, Denmark, Finland, France, Germany, Great Britain, Hungary, Italy, Netherlands, Norway, Poland, Russia, Sweden, Switzerland, and Yugoslavia); historical and legal studies; as well as preambles of legal acts or parliamentary discussions of law drafts that provide justification of regulations (e.g., Belgium, Portugal, and Romania). Also, the works of John W. Willis are extremely useful, who in the late 1940s made an excellent overview of rent control regulations in many countries across the world (Willis 1947, 1950; J. W. Willis 1948a, 1949b, 1949a). In other cases, the country-specific studies are used in order to identify the relevant legal acts; see Table 1.
Table 1: Country-specific studies on housing regulations
| Country | Study and period under consideration |
|---|---|
| Argentina | Pellejero and Monaiser (2009) 1921–2009, Polotto (2012) 1901–1950, Turnaturi (2013) 1921–2013 |
| Australia | Schneller (2013) 1916-2013 |
| Belgium | Henau (1991) 1914–1940, Bettendorf and Buyst (1997) 1929–1939 |
| Botswana | Frimpong (1989) 1977–1989 |
| Brazil | Bonduki (1994) 1930–1950 |
| Bulgaria | Miletić (2016) 1918–1928 |
| Canada | Thibodeau (2001) 1950–2000, Dillon (1982) 1940–1982, Schuss (2018) 1951–2018 |
| Chile | Hidalgo Dattwyler (2003) 1921–1930 |
| Colombia | Soto Gil (2013) 1946–2013 |
| Costa Rica | Elizondo Calderón (1998) 1920–1928 |
| Cuba | de Peña Mulet (1992) 1902–1985 |
| Ecuador | Novillo Rodas (2006), Bravo Lozano (2015) 1938–2000 |
| Egypt | Attia (2016) 1924–1996 |
| El Salvador | Marmol Navas (2001) 1946–2001 |
| Finland | Malinen (2009) 1917–1922 |
| France | Croizé (2009) 1900–1980 |
| Germany (East) | Meggle (2004) 1945–1990 |
| Germany (West) | Kofner (1997) 1896–1945, Kofner (1999) 1946–1998, Konstantin Arkadievich Kholodilin (2017) 1914–2016 |
| Ghana | Malpezzi, Tipple, and Willis (1990) and Tipple (1988) 1942–1989 |
| Guatemala | Villatoro Illescas (2009) 1932–1992 |
| Hong Kong | Cheung (1975) and Bradbrook (1977) 1921–1977, Choi (2017) 1921–2004 |
| Hungary | Antal (1997) 1946–1997 |
| Iceland | Jóhannsson and Sveinsson (1986) 1939–1986 |
| India | Dey and Dev (2006) 1918–2006 |
| Indonesia | Colombijn (2013) 1918–1966 |
| Israel | Mark (2013) 1948–1954 |
| Japan | Suto (2015a), Suto (2015b), Suto (2016), Suto (2017) 1917–2015, Ono (2017) 1939–1950 |
| Kenya | Noormohamed (1975) and Bonna (1985) 1918–1985 |
| Lebanon | Marot (2018) 1940–2018 |
| Liechtenstein | Nägele (2014) 1812–2014 |
| Lithuania | Kuodys (2012) 1919–1937 |
| Malaysia | Mohit and Sulaiman (2006) 1966–1997 |
| Mexico | Navarro González (1974) 1910–1970, Ramírez Navarro and Ramírez Navarro (2018) 1917–2015 |
| Netherlands | Priemus (1983) 1916–1983 |
| Nigeria | Oni, Ajibola, and Oloyede (2007) 1997 |
| Philippines | Ballesteros (2001) 1946–2001, Ballesteros, Ramos, and Magtibay (2016) 1971–2015 |
| Romania | Tabacu (2006) 1919–2006 |
| Russia | Konstantin A. Kholodilin (2017) and Холодилин (2019) 1917–1922, Kholodilin and Meerovich (2016) 1916–1939 |
| Senegal | Diongue (2018) 1952–2018 |
| South Africa | Maass (2010) 1920–2010 |
| Sweden | Bengtsson (2006) 1907–2006 |
| Switzerland | Rohrbach (2014) 1911–2014 |
| Tanzania | Chembo (2014) 1961–2015 |
| UK | Wilson (2017) 1915–1989 |
| USA | International Labour Office (1925) 1919–1925, J. W. Willis (1948b) 1946–1947, Collins (2016) 1920–2016, Keating (1987) 1969–1985 |
| Venezuela | Lovera De Sola (2015) 1947–2014 |
| Vietnam | Herbelin (2009) 1920–1951 |
| Zambia | Nzonzo (2005) 1940–1994 |
In the second step, a list of relevant legal acts is compiled and the search for their original (not revised) texts is conducted. Since we are interested in the evolution of the housing legislation, we need the “real-time” texts, as formulated at the moment of their enactment. Most frequently, such texts are found in government, or official, gazettes. Fortunately, many of these gazettes are digitized and available as online archives. Hence, it is relatively easy to search for the necessary information. In other cases, laws can be obtained free of charge by contacting the national parliaments (e.g., as is the case for Denmark, Iceland, and Norway). Still other countries charge fees for providing the relevant laws (for example, Bulgaria, Singapore, and Sweden). As in some cases we were unable to locate laws as published in an official gazette, we use drafts of the laws from parliamentary proceedings (e.g., Belgium and Switzerland). In the worst case, answers to questions submitted remotely are not forthcoming (some African, Asian, as well as Latin American and the Caribbean countries) or one must visit a library in the country of interest.
In the third step, the compiled legal acts are summarized. The relevant provisions are identified and recorded. In particular, the following fields are captured: area of application, rent control, tenant protection, housing rationing, and bodies responsible for conflict settling and regulation of the housing sphere. Language barrier is an important challenge at this stage. In many cases, knowledge of the foreign languages (English, French, German, Italian, and Spanish), Russian being his mother tongue, permits the author of the study to understand the legal texts. In other cases, native speakers help decipher these texts (e.g., those in Greek). Otherwise, the author takes advantage of machine translation (the internet service of Google Translator) in order to translate the texts written in the languages he or his colleagues do not speak. Although the quality of modern machine translations is relatively high, there is still a room for mistakes.
In the fourth step, the textual summaries of legal acts are mapped into numeric values. Here, we rely upon the approach of Weber (2017) to code rent laws and tenure security and Konstantin Arkadievich Kholodilin (2017) to code housing rationing. Based on a set of questions contained in Table 2, binary variables are constructed that equal one, if regulation is more stringent, and zero, otherwise: \[ I^k_{jt}= \begin{cases} 1,& \text{if restriction $j$ of type $k$ is present in period $t$}\\ 0, & \text{otherwise} \end{cases} \] where \(k\) is a regulation type (rent laws, tenure security, or housing rationing) and \(t\) is the date on which the law containing such provision is enacted. Thus, each binary variable represents an answer to a question characterizing a particular aspect of the corresponding regulation. If the answer is “yes”, then the binary variable takes value 1, if the answer is “no”, then the variable takes the value zero. Typically, the positive answer corresponds to more limitations from the standpoint of landlords. Below, the coding is described in more detail.
Rent control. In his dissertation, Weber (2017) defines six binary variables: Real rent freeze (the rents are not allowed to grow faster than inflation), Nominal rent freeze (the rents are frozen in nominal terms), Rent level control (some government body, arbitration council, or court fixes the rent level at the beginning of new contracts), Intertenancy decontrol (if the rent control ceases with a change of tenant), Other specific rent decontrol (certain types of dwellings or settlements are no longer subject to the rent control), and Specific rent recontrol (certain types of dwellings or settlements are subject to more stringent controls).
Tenure security. This protection is summarized using four binary variables. The binary variables Eviction protection during term or period and Eviction protection at the end of term or period take the value one, if, in order to evict a tenant during the contract term or at the end of it, the landlord is required to present justified reasons. The Minimum duration variable equals one, if the contract duration must be at least two years, while the Short-term tenancies’ variable is 1, if letting dwellings for a period of less than one year is prohibited.
Housing rationing. This policy is approximated with eight binary variables. Registration of housing equals one, if landlords are obliged to register vacant, or all available, premises. The binary variable Protection of housing is one, if it is prohibited to use dwellings for non-residential purposes, merge or demolish them, or to convert rental dwellings into condominiums. The variable Creation of housing space equals one, if the state prescribes the use of all available space for housing purposes, e.g., through the reconstruction or conversion of non-residential premises or through the subdivision of big dwellings into smaller ones. The dummy variable Requisition equals one, if requisition with subsequent compulsory letting of the vacant dwellings is conducted. Restriction of freedom to move is one, if residential mobility is restricted: for example, if access to areas with an acute housing shortage is closed to all persons, who are neither “indispensable” for these areas nor residing there on a permanent basis. Conservation of social composition is one, if a balanced social composition of population in particular urban areas is protected through interdiction to upgrade the dwellings to a state considered being above the standard level. The variable Housing consumption norms equals one, if restrictions on the amount of housing that might be used by tenants are imposed. The dummy Nationalization of housing takes the value one, if the state nationalizes housing stock, and zero, if no nationalization or privatization occurs. Unlike requisition, nationalization means the loss of property rights for the owner and no compensation for property taken. Table 2 defines the individual indices (binary variable) making up the three regulation types.
Table 2: Composition and definition of regulation indices
For each regulation type \(k\), a composite index is computed as a simple average of binary variables: \[I^k_{t}= \frac{1}{N_k}\sum_{j=1}^{N_k} I^k_{jt} \] where \(k=\{\mbox{Rent laws},\mbox{Tenure security}, \mbox{Housing rationing}\}\) and \(N_k\) is the number of binary indices. For example, for rent laws the composite index is based on six binary indices. When all of them are equal zero, their simple average is equal to zero too implying that no limitations on rent setting are present. The more such limitations the closer the index to 1. The composite index equal to 1 means the highest intensity of regulations of type \(k\).
The binary and composite indices are constructed for a large balanced panel of 132 countries or subnational regions (111 nations) covering the period 1910–2020, see Table 3 and Figure 1. The choice of countries is dictated by the availability of legal acts. The best coverage exists for Europe, Asia, and Latin America. In Africa, mainly former French and Portuguese colonies are covered, since it was relatively easy to locate the historical legal acts for them. For North American countries, coding is complicated by the fact that housing regulations there are created at the regional level, including not only states or provinces, but also cities. All in all, our data set covers the majority of the world population, given that the population of countries for which rental market regulation indices are constructed makes up 80.6% of the total world population as of 2010.
Table 3: List of countries, for which regulation indices are constructed