Stylized facts on housing vacancies

Dans le document Essays on location choice: agglomeration, amenities and housing (Page 103-109)

housing vacancies in the suburbs

3.2 Stylized facts on housing vacancies

Data on housing come from the French national statistical institute (INSEE) which pro-vides detailed information on the stock of housing for all French municipalities. Dwellings are classified as being either occupied – including main residences and second homes – or vacant. According to INSEE, a dwelling is considered as vacant when “it is for sell or proposed to let with no buyer or renter, when it is awaiting occupation, pending succes-sion settlements, kept by an employer for future use by one of their employees, or kept vacant without any specific attribution”. Primary data are collected along with the French population census. Information is provided annually from 2006 to 2011, with additional data for the years 1999, 1990, 1982, 1975 and 1968. A vacant dwelling corresponds to a unit (house or flat) that is empty at the date of the census. Second homes – used for weekends or holidays – and occasional accommodations are not considered as vacant.

The geographical units of analysis are the French municipalities. As there are more than 36’000 municipalities in France, data are fairly disaggregated at the geographical

level5. For each municipality, I compute an annual housing vacancy rate, defined as the share of vacant dwellings in the total stock of housing.

3.2.1 Size and evolution of vacancies

The extent of housing vacancy is large in France. There were about 2.5 millions vacant dwellings in mainland France in 2011, among which 327’000 are concentrated in the region Ile-de-France (the metropolitan area of Paris). This number has to be compared with the few 67’818 vacant units that could be found in the five boroughs of New York City the same year6. In 2011, 7.4% of existing dwellings were indeed unoccupied in France. This ratio is relatively high compared to other countries. For instance, it reaches 3.3% in Great Britain and an extremely low rate of 1% in Switzerland (Cheshire et al.,2014; Thalmann, 2012). According to the most recent data provided by the U.S. Census Bureau (for the last quarter of 2014), the vacancy rate in the U.S. reaches 7% for rental dwellings against 2% for home-owner housing.

Figure 3.1 describes the evolution of the vacancy rate in France. In addition, Table 3.1 provides numerical information on the stock of housing, for France and the Paris area from 1968 to 2011. The ratio of vacant to total dwellings fluctuates over time. Indeed, it experienced a strong increase between the two first censuses, both in the country and the capital region. This increase is partially explained by a substantial rise in housing supply that occurred between 1968 and 1975, as shown by the average annual growth rate of housing stock that exceeds the one of occupied dwellings.

Figure 3.1: Evolution of housing vacancy rates (1968-2011)

(a) France (b) Paris area

5In addition, all French municipalities with more than 10’000 inhabitants – and many between 5’000 and 10’000 – are broken down into districts, calledIRIS. The latter broadly correspond to census tracks in the U.S. Information on housing is also provided at this geographical scale.

6New York City Rent Guidelines Board 2014 Housing Supply Report.

Table3.1:Stockofhousingbyoccupancystatus 19681975198219901999200620072008200920102011 France Vacancyrate6,6%7,7%7,6%7,2%7,0%6,4%6,5%6,7%7,0%7,2%7,4% Totalunits1844921287239562654029151316123198532371327393309733450 -2,1%1,7%1,3%1,0%1,2%1,2%1,2%1,1%1,1%1,1% Occupied1723619653221362463227106295962990430196304573071730984 -1,9%1,7%1,3%1,1%1,3%1,0%1,0%0,9%0,9%0,9% Vacant12131634182019092046201620812174228223802465 -4,4%1,6%0,6%0,8%-0,2%3,2%4,5%4,9%4,3%3,6% Parisarea Vacancyrate4,6%6,7%6,9%6,5%8,1%6,2%6,1%6,1%6,0%6,0%6,0% Totalunits35864098442347465083530953405378541754535486 1,9%1,1%0,9%0,8%0,6%0,6%0,7%0,7%0,7%0,6% Occupied34223823411644374673498050145053509151265158 -1,6%1,1%0,9%0,6%0,9%0,7%0,8%0,8%0,7%0,6% Vacant164275307309409329325325326327328 -7,7%1,6%0,1%3,2%-3,1%-1,1%0,1%0,2%0,2%0,6% Total,occupiedandvacantunitsareexpressedinthousands.Averageannualgrowthrateinitalics.

After 1975, the evolution of owner-occupancy rates in the Paris area differ from the rest of the country. In mainland France, the vacancy rate has progressively decreased over the period 1975 - 2006. This fall has been followed by an increasing trend during the last census years (from 2006 to 2011). In contrast, vacancy rates remained quite constant in the Paris area over the last 35 years, apart from a strong peak in 1999. The latter is also explained by a rise in the supply of housing that was not followed by a proportionate increase in occupancy. Besides, the most recent years in Paris are characterized by a moderate increase in housing supply, associated with a proportionate increase in the number of occupied dwellings. As a result, the vacancy rate remained quite stable.

3.2.2 Distribution of vacant dwellings in the Paris area

There are 1280 municipalities in the Paris area, to which we should add the city of Paris. Figure 3.2 provides a map of these municipalities and their respective vacancy rate for the year 2011. There are important disparities in housing vacancies within the Paris metropolitan area. Vacancy rates indeed range from 0 to 25% in 20117. Moran’s coefficient that measures the spatial autocorrelation in vacancies across contiguous municipalities is quite low at the value of 0.12. This coefficient is close to a spatial random distribution8.

Figure 3.2: Housing vacancy rates in the Paris area

In addition, Table 3.2 shows that this relatively large variation in vacancy rates ob-served in the dataset during the period 1968 - 2011, is almost equally explained by a between (variation across municipalities) and a within component (variation over time within municipalities).

In that respect, the central city of Paris is a striking example of a strong within variation. The left hand-side of Figure 3.3 shows how the ratio of vacant to existing

7An extreme value of 58% might be found which corresponds to a small municipality of 30 inhabitants in 2011 which experienced a wave of desertification. Such outliers will be excluded in the empirical analysis.

8A Moran’s coefficient of 0 indicates spatial randomness. A coefficient of 1 represents perfect correla-tion (if municipalities with high vacancy rates agglomerates, and vice versa). A coefficient of -1 is perfect dispersion, when each municipality that has a high vacancy rate is surrounded by jurisdictions with low vacancy rates.

Table 3.2: Between and within variations of vacancies in Parisian municipalities Std. Dev. Min Max Observations

Overall .035 0 .603 N = 14079

Between .023 0 .410 n = 1280

Within .027 -.283 .554 Years = 10.999

dwellings in the city of Paris has evolved over time. This ratio indeed rose from a very low rate of 3.3% in 1968 to 7.3% in 2011, and reached a staggering value of 10.3% in 1999. The right panel of Figure 3.3 describes the evolution of vacancies in the rest of the metropolitan area – excluding the central city of Paris. This graph shows that the evolution of housing vacancy varies from the city center to the suburbs.

Figure 3.3: Evolution of housing vacancy rates in the Paris area (1968-2011)

(a) Municipality of Paris (b) Suburban areas

These different trends might be explained by changes in urban policy and housing supply across space. In that respect, Table 3.3 describes how the stock of housing has evolved in all districts (French d´epartements) of the Paris area. This table shows that distant locations have indeed experienced a strong increase in the supply of housing from 1968 to 2011 compared to proximate areas or to the central city of Paris. This change in housing supply – that might contribute to explain the existence of high vacancy rates in the suburbs – will therefore be taken into account in the empirical analysis.

3.2.3 Characteristics of vacant dwellings

Table 3.4 compares the characteristics of housing goods based on their occupancy status in 2011. In France, 57% of existing dwellings are houses while 43% are apartments. These shares respectively reach 27% and 73% in the Paris area. Table 3.4 shows that, in both cases, the largest share of vacant units are apartments. Apartments indeed represent 51.5% of vacant units in France, and a staggering 80% in Ile-de-France. The size of dwellings (expressed in terms of number of rooms) also statistically differ across the two samples (vacant vs occupied units). In France, around 60% of vacant dwellings have 3 rooms or less, while they only represent 40% of total inventory. The relative prevalence of small vacant dwellings is even stronger in Ile-de-France. In the Paris area, three fourth

Table 3.3: Growth rates of housing supply in the Paris metropolitan area (1968 - 2011) 1968–1975 1975–1982 1982–1990 1990–1999 1999–2006 2006–2011

City of Paris 1.4 3.3 1.9 1.4 0.9 1.6

Petite couronne (proximate locations)

Hauts-de-Seine 9.2 5.1 5.0 6.9 6.1 2.3

Seine-Saint-Denis 15.1 6.2 5.2 5.2 4.1 3.1

Val-de-Marne 17.9 4.1 6.0 7.7 5.0 3.4

Grande couronne (distant locations)

Seine-et-Marne 26.3 18.9 17.8 15.7 8.7 7.2

Yvelines 32.2 16.5 12.0 9.7 4.9 3.5

Essonne 38.4 12.5 13.1 11.6 6.1 4.6

Val-d’Oise 27.3 14.5 13.5 10.9 5.5 4.3

of vacant dwellings have fewer than 3 rooms, compared to 57% of occupied dwellings.

This table also shows that unoccupied dwellings tend to be older than their occupied counterparts. Recent dwellings (that have less then 20 years) represent only 12% of vacant units in France (11% in Ile-de-France) while they account for almost 20% of total available dwellings (15% in the Paris area). To summarize, the characteristics of dwellings differ depending on the occupancy status. Vacant dwellings tend to be apartments, located in older buildings with a fewer number of rooms, than their occupied counterparts.

Table 3.4: Characteristics of housing units (in %)

France Paris area

Occupied Vacant All Occupied Vacant All Housing type

Houses 57.5 48.5 56.8 27.8 19.7 27.3

Apartments 42.5 51.5 43.2 72.2 80.3 72.7

Housing size

1 room 5.8 12.8 6.3 10.9 25.2 11.8

2 rooms 13.3 20.6 13.9 20.7 28.6 21.2

3 rooms 21.6 25.9 21.9 25.6 22.3 25.4

4 rooms 25.1 21.1 24.8 21.0 13.1 20.5

5 rooms 18.7 11.0 18.1 12.3 5.8 11.9

6 rooms or more 15.5 8.6 15.0 9.5 4.9 9.3


Before 1946 27.7 45.1 29.0 29.1 42.6 29.9

1946 - 1990 52.4 42.7 51.7 56.0 46.4 55.4

1991 - 2008 19.9 12.2 19.3 14.9 11.1 14.6

Dans le document Essays on location choice: agglomeration, amenities and housing (Page 103-109)