Inter American Development Bank (IADB)

Research Department

HOUSING MARKETS IN LATIN AMERICAN AND CARIBBEAN CITIES:

IMPLICATIONS FOR DEVELOPMENT AND MACROECONOMIC STABILITY

THE HOUSING MARKET of

ARGENTINA in the 2000s.

FINAL REPORT1
ONLY FOR COMMENTS

Marcela Cristini
Ramiro Moya
Guillermo Bermúdez2

1
January, 2010. We thank Cesar P. Bouillon, David Dowall, Fernando Cuenin and IADB participants at the
Housing Markets in LAC Seminar for helpful comments and criticisms.
2
Fundación de Investigaciones Económicas Latinoamericanas (FIEL). Corresponding author:
[email protected].

1
 Table of Contents

1. Introduction ................................................................................................................................ 3
2. Urban housing markets: main characteristics............................................................................. 3
2.1. Housing situation in urban areas: general overview........................................................... 5
2.2. Housing in selected urban metropolitan areas ................................................................. 11
2.3. Characterizing the actors of the urban housing market ................................................... 13
2.3.1. Housing informality ................................................................................................... 13
2.3.2. Construction companies and developers (productivity and profitability) .............. 14
2.3.3. Urban dynamics and the housing market: ................................................................ 18
2.3.4. Financial markets (recent evolution, mortgage contract, informal financing,
accessibility) .............................................................................................................................. 20
2.3.5. Government (construction standards, residential regulations and social housing
policies) ........................................................................................... .....................................21
3. Determinants of housing demand and tenure choice. ............................................................. 23
3.1. Estimation of a demand equation ..................................................................................... 24
3.2. Estimate of tenure choices with a multinomial logistic model ......................................... 32
4. Housing market, construction and activity level: stylized facts ................................................ 42
5. Policy recommendations and final remarks. ............................................................................. 44
References..................................................................................................................................... 47
Annex 1............................................................................................. ¡Error! Marcador no definido.
Annex 2: A Sketch to appraise the feasibility of a comprehensive housing program for urban
locations in Argentina ...................................................................... ¡Error! Marcador no definido.

2
 1. Introduction
Argentina is a highly urbanized country. About 70% of the economic activity as measured
by GDP and around 50% of business establishments (manufactures and services) are
concentrated in the seven major metropolitan regions. This spatial distribution of the
population translates into a variety of local housing markets with own characteristics
depending on local growth and development, employment opportunities and geography,
among various factors.

Regarding the housing situation of Argentina, an overall picture shows that its evolution
has been unsatisfactory over the last three decades. Highly cyclical construction activity
affected by disruptions in the macroeconomic scenario and poor design of social housing
policies were two important factors explaining the persistence of housing deficit and the
diminishing quality of life in large urban centers.

The present study analyses the organization and main drivers of the housing market and
relates them to the macroeconomic scenario with the aim to advance a policy agenda for
housing policy reform.

We organize the work in three sections following this introduction. The first section
describes the characteristics of the urban housing markets in detail including their
evolution, the main participants (i.e., households, construction companies, financial
markets and the governments) and the regulatory environment. Section 2 includes a
thorough study of the demand behavior. Section 4 summarizes the operation of market
forces and macroeconomic environment effects on them and elaborates some policy
recommendations.

2. Urban housing markets: main characteristics

Generally speaking, housing construction is considered as a cyclical activity characterized
by fluctuations that replicate the GDP economic cycle in an amplified way. Argentina´s
housing construction activity adjusts to this description, including peaks and troughs of
activity heavily influenced by strong disruptions stemming from significant
macroeconomic crisis. Figure 1 compares the evolution of GDP, construction activity
(private and public) and housing construction since 1993. A salient feature is the rapid
growth achieved after the 2001-02 crisis. (See Total Construction Business Cycle in
Statistical Annex 1, ¡Error! No se encuentra el origen de la referencia.)

3
 Figure 1

Source: Own based on INDEC

Note: Construction aggregates shown here by the Gross Investment in Construction as provided by National
Accounts estimates, are made of four main components: public construction, private construction (includes
housing), industrial construction and construction for the petroleum activity. Unfortunately, separate estimates
for housing construction are not available. One related indicator useful to follow housing construction evolution
(Housing Activity in the chart) is the Synthetic Construction Indicator (ISAC).

Considering a longer perspective (Figure 2) allows illustrating the importance of the 2003-
2010 growth recovery in housing construction: to find a similar activity growth, one has to
go back to the early 80s, suggesting that housing construction has held a very important
role in sustaining the growth path of the 2000s.

4
 Figure 2

Source: Own based on National Accounts-MECON and ISAC. Housing investment corresponds to Fixed Gross
Investment in Housing, Total construction activity and Housing construction activity corresponds to the ISAC

2.1. Housing situation in urban areas: general overview

A brief account of the recent evolution of the housing market in Argentina shows that,
according to Population Census data, between 1991 and 2001 the housing stock increased
19,5% over the period. Very recent (provisional) data from the 2010 Census exhibit a
similar increased of 18,7% of total housing units for this period. This growth was
characterized by more than a proportional increase in houses as compared to apartment
units. New houses and apartments represented 134,000 new units annually. Official
estimates have valuated the housing stock as representing 35% of total stock of capital on
average in the last decade (See Table 1and

5
Table 2).

Table 1

6
 Table 2

Source: Own based on Population and Housing Census

Housing conditions in urban cities are diverse according to the location. Almost 40% of the
quantitative housing deficit3 is located in the province of Buenos Aires and 28% in the
metropolitan area of Greater Buenos Aires (the capital city of Argentina and its
surroundings, including 24 municipalities). More than 53% of the overall deficit
concentrates in the six largest cities (Greater Buenos Aires, Rosario, Santa Fe, Cordoba,
Mendoza and San Miguel de Tucuman).

The qualitative deficit is also a main problem affecting more than 20% of households.
Almost a quarter of this figure is due to housing overcrowding. The localities around
Buenos Aires City and San Miguel of Tucuman have the highest incidence (20%). Buenos
Aires City (the inner city of the largest metropolitan area in the country) has the lowest
incidence with a 7% of qualitative deficit.

Tenancy has also changed over time (see

3
Quantitative and qualitative housing deficits follow the traditional measure in the region (Arriagada Luco
(2003)). According to official estimates of 2003, the housing deficit was at about 3.5 million units, with larger
concentration in the poorest provinces. Size of housing deficit differs according to alternative definitions of
what is considered an ―adequate house unit‖ to cover household´s shelter needs. These definitions have varied
over time and across government levels. Notwithstanding the definition and consequent size of the housing
deficit, one of its characteristics in Argentina has been its persistence over time. Even though the country
growth rate evolved favorable during mid 80s and most of 90s, the housing deficit continued to be an
unsolved problem. This was partly due to poor social housing programs and partly to lack of development in
financial instruments devoted to housing finance.

7
Table 3). Renting a house was a diminishing characteristic of the housing urban market
until very recently when supply for rental dwellings increased noticeably, at least in Greater
Buenos Aires (See Figure 3).

8
 Table 3

Figure 3

Source: Economic Trends

In recent years, there has been observed a trend of households moving to urban
agglomerates exhibiting high employment creation. Since housing supply has not
responded at the same rapid pace, internal migrations resulted in increasing housing
overcrowding (see Figure 4). Moreover, correlations of housing deficit, both quantitative
and qualitative are positive and significant with respect to population density (0.40 to 0.66
depending on deficit definition), employment density (0.3 to 0.51) and poverty (0.73 to
0.77) in urban centers. Between 2007 and mid-2010, estimates based on the Permanent

9
Household Survey confirms a rapid increase in the construction of detached houses (around
416,000 houses) and apartments (258,000 new units). At the lower extreme of the market,
old houses became irrecoverable (113,400) and overcrowding grew in around 75,000
housing units.
Figure 4

Source: Own based in EPH

Affordability analysis: Generally speaking, a housing affordability index is calculated as

the housing cost-to-income ratio. The conventional indicator takes into consideration the
averaged variables and studies the evolution over time or across locations. Alternatively,
the affordability index may also introduce the idea of accessibility to a mortgage loan as in
the case of the US-National Association of Realtors index. In this case, the index measures
whether a typical family with a median income could qualify for a mortgage loan on a
typical house.

In this sub-section we analyze the evolution of housing affordability for Argentina using
alternative indexes based on construction costs and housing market prices per sq meter. We
will postpone the consideration of the relation between income and mortgage loans and
develop it in a following sub-section.

The affordability index will allow us to illustrate two main issues. On the one hand, from
the macroeconomic point of view, the 2001-02 crisis caused a disruption in the economic
regime deeply affecting relative prices. The mega-devaluation of the peso translated into
the increase of prices for tradables relative to non –tradables. In the case of housing costs,
they are composed of tradable and non-tradable goods where the latter exhibit the largest
participation. For instance, in the case of the Housing Construction Cost Index (INDEC),
component weights are the following: labor, 45,6%; construction materials, 46%; other
expenses, 8,4%. The latter two components are indexes made of tradable and mostly non-

10
tradable goods.4 At the same time, family income is often represented through salary, the
―price‖ of labor, a core non-tradable of the economy. As a consequence of the crisis, the
GDP fell 14% between 2001 and 2002. Around 2007, per capita GDP recovered in real
terms to achieve a value similar to the one reported in 1998. Thus for almost a decade, the
average argentine household was poorer than in the period 1993-98, than we are using as a
reference. Accordingly the expected evolution of the affordability index defined as the ratio
between housing construction costs and salaries is of a sharp fall after the crisis with a
tendency towards recovery towards the end of the decade. Besides, increase in employment
in general and in formal occupations in particular, determined an increase of total
household income that also impacted on affordability. The following

4
Increase in construction costs over IV01 y IV09 were: Overall index, 282%; Materials, 272%; labor, 302%
and General expenses, 226%.

11
Figure 5 illustrates this point. The correction of wages by the unemployment rate is a rough
representation to introduce the impact of changes in total household income. Figures
confirmed the recovery of housing affordability on average for the economy.

However, it should be noted that this measure doesn’t include the cost of land. Prices of
land plots are highly variable across the country and between neighborhoods. In this sense,
this affordability index should be considered as a ―frictionless measure‖, without transport
saving costs and other benefits from amenities included in the price of the land.

In fact, on the other hand, at the microeconomic level, dispersion of housing prices due to
quality and location of units and the gap between price and cost as recorded by private
sources (realtors), suggest the segmentation of the market under the current conditions. Due
to the lack of a well developed mortgage loan market, middle income households
participate of the housing market only spasmodically, depending on their availability of
savings. Instead, investors (domestic and foreign) and high income households are active at
the upper end of the housing market demanding expensive units. On the other end, social
housing policy has kept a flow of new units serving the poor families at a pace of 35-50
thousand units per year.

12
 Figure 5

Source: Own based on INDEC and IERIC

The comparison between the episodes of the 90s and 2000s for the housing market is very
relevant. During the 90s, housing construction advanced at a good pace and mortgage loans
widened their scope to reach middle income families. The mix of highly imperfect housing
price indexes as provided by private sources showed prices per sq meter in dollars
relatively lower than the ones prevailing in the 2000s, when those indexes mainly reflected
expensive housing unit prices. Thus, any affordability index based on those prices presents
a distorted story of the market. The use of housing construction costs as the numerator of
the affordability index also presents problems. Housing construction costs may be only
broken into two categories of constructions, detached houses and apartments, both of good
quality. Thus comparison over time is fair but the whole calculation has a bias toward
underestimation of affordability by the average family. Additionally, sources for statistics
on wages also differ due to coverage, definition of labor formality and changes of
composition effects. To give an example of the variability of results according to the
components of the index used in the calculations, see the following

13
Table 4.

14
 Table 4

Source: Own based on INDEC

The most important difference to bear in mind for our analysis is the one corresponding to
formal vs. informal wages, pointing once again to the potential segmentation of the market
in terms of access to loans and repayment capacity over time.

2.2. Housing in selected urban metropolitan areas

At present, 47% of the whole Argentine population lives in only seven large cities
(metropolitan regions). Four out of them are located near the Atlantic coast and exhibit a
potential to become an urban continuum (a megalopolis). Buenos Aires city and its suburbs
(Buenos Aires Metropolitan Region) is a megacity with more than 13 million inhabitants,
being ranked as the 15th largest cities in the world5. The following Table 5 gives an idea of
the relative importance of the selected cities.

Table 5

5
Population as of 2010.

15
Rates of ownership and slum incidence are both higher in larger cities. Being a developing
country, migration and demographic evolution have affected urban housing market
dynamics over time. Episodically, economic swings and changes in preference for location
have determined significant changes in household location that left their imprint in the local
urban development (slums growth during the 2002 crisis, apartment building construction
in the 70s and more recently, the creation of new neighborhoods: the closed ones in the
outskirts of large cities in the 90s and Puerto Madero in Buenos Aires city in the 2000s).
Consequently, housing market conditions vary across cities. On the one hand, construction
costs do not differ much across localities (their different levels depend mainly on the
differences in regional salaries). But on the other, prices per sq meter are quite diverse
depending on demand. Larger cities tend to exhibit higher prices for dwellings with similar
characteristics than smaller cities of the inland. The main reason is that the former
concentrate many factors of attraction (employment, amenities, etc) but have reduced
available space at convenient center distance. Episodes of rapid migration or shifts in
demand located in specific neighborhoods of larger cities provoked an increase in land
prices. This is the recent case of Buenos Aires City and, very likely, of Rosario City. In the
2000s prices by sq meter in Buenos Aires City (average) doubled those of Tucuman and
Mendoza. The latter, in turn, doubled the prices registered in smaller cities such as Jujuy or
Trelew. The ratio of new to old apartment prices also widens when size of cities
diminishes.

As a consequence of differing prices by constructed sq meter and variations in average

income by regions, affordability also changes across cities. In particular, data for the 2000
period suggest that affordability diminishes with urban agglomerates size.

16
Figure 6 shows the affordability measured in terms of monthly salary needed to buy one
housing sq meter in the five largest cities of Argentina. The range goes between 1.1 to 1.8
months. A household living in Buenos Aires City will need 60% more time than one in
Tucumán to acquire a similar dwelling. These differences are larger when smaller cities are
considered. In the case of cities with less than 800,000 inhabitants, market reports indicate
that this affordability measure is 30 to 40% lower, i.e., effort to buy one sq meter of
housing is lower to one monthly salary.

17
 Figure 6

Source: Own based on IERIC. CABA stands for Ciudad Autónoma de Buenos Aires.

Regarding heterogeneity of household characteristics across the country, differences in

income per capita, family size, education of the household head, age of members, etc.,
affect the preferences and accordingly the demand for shelter. Section 2 below will
elaborate on this issue.

2.3. Characterizing the actors of the urban housing market

2.3.1. Housing informality

According to the World Bank (2005), historical high share of ownership has also been
favored by a fairly good system of land and cadastral registration – 90% of the properties in
Greater Buenos Aires have full registration title (even though some of them are illegally
occupied). Still, some 1.6 million households in the country (15.9% of all households) do
not have proper legal titles. This includes ―asentamientos‖, villas,‖ occupiers of private
land in irregular subdivisions, occupiers of undivided rural lands. Illegally occupied lands
are often located in flood–prone or contaminated areas and lack basic infrastructure. There
is ample evidence that the absence of proper titling inhibits families from investing more in
their housing, prevents prospective buyers from obtaining mortgage credit for purchasing
non–titled houses, suppresses the value of houses, and inhibits the development of a proper
housing market (For instance, see Galiani and Schargrodsky (2004 and 2010)).

In order to study the evolution and main characteristics of informality we prepared a

database using several sources (the Census, EPH, ENGH and the Living Condition Survey,
covering only partially over time and in scope the required aspects of informality). (See
Table 6 and also ¡Error! No se encuentra el origen de la referencia. to ¡Error! No se
encuentra el origen de la referencia. in the Annex 1).

18
The Table 6 presents four alternative definitions of informality and the percentages of
formal tenancy along with the incidence of slums by income decile. As shown, informality
is not a dominant problem in Argentina. Slums are located in the periphery of large cities.

Table 6

2.3.2. Construction companies and developers (productivity and profitability)

The construction sector in Argentina comprises a variety of firms of different sizes and
technologies ranging from family enterprises and small builders— highly skilled and less
skilled ones—to larger and more sophisticated companies. The small units still dominate
the sector in terms of number of companies, working in their own projects, providing
services for larger companies or contracting their services to households building their
homes. More recently a handful of high tech expensive dwelling developers have been

19
very active in the market, particularly in the construction of suburban neighborhoods. They
have also developed neighborhoods for foreign and local investors in selected locations in
Buenos Aires, Rosario and Córdoba cities.

Formal companies in activity were around 17,500 in June 2010. The value chain is
composed approximately of 10,500 construction companies, 5,000 contractors and
subcontractors. Sixty percent of them are active in provinces with the largest urban
(Buenos Aires, Córdoba and Santa Fe). The sector has expanded over time (See
Figure 7 below). Small companies dominate the activity. They represent 75% of firms in
number and employ 20% of the workforce (3 workers per company). On the other extreme,
2% of the companies are large ones with more than 100 workers, and they employ around
30% of the total. In the middle, 22% of the companies are medium ones between 10 and
100 workers, explaining half of the employment. The activity is labor intensive and labor
informality used to be very high. In recent years the degree of formality increased along
with employment in the sector (see

20
Figure 8).

Their activities also include other types of construction from public infrastructure to
housing. Large companies mostly concentrate in public infrastructure or commercial
construction. Smaller companies and a bunch of developers specialize in apartment
buildings or suburban closed neighborhoods. Detached houses are built up by independent
constructors (architects, civil engineers, etc).

Figure 7

Note: Small companies renew their inscription in the Construction Register annually, so that quantity of
participants is variable but on average has been increasing over time.

21
 Figure 8

Source: Own based on IERIC

Construction costs keep their relative position related to wholesale prices (tradables) in the
long run.6 Being mostly a basket of non-tradable inputs they were relatively high in the 90s
and adjusted in the early 2000s. Recently they are recovering relatively in a scenario of
increasing prices (consumer price index is growing 25 to 30% in 2010) (See Figure 9).
Returns to construction vary by region but have been growing very fast in Buenos Aires
City, followed by Rosario and Córdoba as shown in Figure 10 through the ratio between
the price of new apartments in sq meters in Buenos Aires and the cost of construction.

The construction activity level increased heavily in the 2000s when the number of permits
grew from 50000 to more than 90000 per year and surface constructed went from 6 to more
than 10 million sq meters for new housing. On top of this, another 20% of built up surface
can be added due expansion of existing dwellings.

Previous studies and our own description of the construction sector have identified only
few constraints to the efficiency of construction companies. Instead, due to the lack of a
fluid mortgage market, financing of new construction has traditionally been linked to the
evolution of financial alternative assets and to the development of financial instruments
such as the ―fideicomiso‖.

In 2004, private analysts estimated that housing production in Argentina averaged 215,000
units per year; 53,000 (40%) were produced by the informal sector without permit or
license (self-construction or informal constructors); 40,000 (18%) were produced by the
public sector; 74,000 were detached houses produced by the formal sector (34%), and

6
Housing in Argentina reflects the Italian and Spanish ethnic backgrounds of the population. Except for
marginal rural dwellings and urban shanty towns, concrete, mortar, and brick are favored as the principal
construction materials

22
48,000 were formally constructed dwelling units (22%). In recent years, this figures
increased particularly by the contribution of the formal sector (see above).

Housing supply appears to be neither bounded by construction input constraints nor is it

heavily influenced by land availability. However spatial differences are of interest and land
availability in highly dense urban areas (Greater Buenos Aires, Greater Rosario and Greater
Córdoba) is an issue under discussion.

Figure 9

Source: Own based on INDEC.

Figure 10

Source: Own based on Toribio Achaval and IERIC.

23
 2.3.3. Urban dynamics and the housing market:

Cities in developing countries are growing at extraordinary rates. In contrast, in Argentina

rapid urbanization took place in the 1940s and 50s crowding cities that had been founded in
early 1800s and had developed in the early 1900s when heavy investment in urban
infrastructure (roads, public transport, public buildings, energy, sanitation, telephones,
hospitals and schools, etc) had taken place. To a great extent the most important cities in
the country still depend on those investments that were over-dimensioned at the time of
their first expansion and now remain at the core of an extended (and sometimes
overcrowded) system. In contrast, investment in housing proceeded spasmodically
depending on various factors such as income growth, business cycles, housing social
programs and migration.

The rate and character of city expansion has been rarely considered as a policy objective
explicitly in Argentina and urban growth has not proceeded orderly enough. Thus,
Argentine large cities are mature structures where many effects of urban dynamics are
present. In this sub-section we will focused on two of them: filtering and gentrification. The
interest in both effects is due to the fact that proper drawing of housing policy should take
them into account. On the one hand filtering may reduce the need for social housing while
gentrification may increase it in certain locations.

Ratcliff (1945) defines filtering down as the process by which the production of new
housing for higher income groups releases used houses to be passed down to successively
lower levels of population down in the income scale until the effect has reached the bottom
of the market. The changing of occupancy is accompanied by a decline in market price
(sales price or rent value). As a consequence, substandard houses tend to be removed from
the stock. Skaburskis (2006) adds that the filtering mechanism should convey a reduction
of rents or sale prices of used units below the economic cost of similar new ones or their
modification.7

Gentrification is the process by which increasing demand for housing grows in inner or
central cities bidding for the land under the ageing buildings to make room for renovation
and replacement.

Both effects have income redistribution consequences through the housing market
operation.

7
Some authors distinguish between neutral and ―welfare improving‖ filtering, where the latter improves the
well-being of low income households. (Baer and Williamson (1988)). It is also usually assumed that the
occupancy of housing by successively lower income groups is concomitant with a deterioration in the quality
of the accommodations. This relationship obtains in a general way, but filtering could take place without any
significant change in quality; and the rate of filtering is not proportionate to the rate of deterioration. Decline
in quality resulting from physical change, obsolescence, and a degeneration of the environment is only one of
several factors that conditions filtering.

24
In this regard, Buenos Aires city probably presents the most extreme situation. It is the
inner city of the largest urban agglomeration and as such it has been subject to various
effects of urban dynamics.

During the 90s the city lost population that moved to gated communities in the suburban
area (sprawling). This may have allowed a ―filtering down‖ process, difficult to identify
due to other changes in the housing market that promote the expansion of demand to new
consumer previously constraint (mortgage market deepened and Buenos Aires city was one
of the first served). In contrast, the 2000s witnessed the opposite movement with
households coming back to the city producing a sort of mild gentrification and pushing
densification in low-cost residential neighborhoods adjacent to traditional costly ones (in
2001-2010, population grew 4% compensating 2/3 of the previous emigration). As already
shown, comparing the evolution of construction costs and final price of new buildings per
square meter suggests that land prices have skyrocketed in this city in the 2000s. Various
factors have been present for this result: demand from real estate investors (domestic and
foreign), rapid income growth in the upper end of the income scale (income concentration),
return of households that had participated of the sprawling in the nineties (for security and
transport cost reasons), etc. Concerning the filtering process over time, we have explored
the response of the relative price between new and old apartments in Buenos Aires city
correlating it with activity level, housing construction, income concentration (Gini) and
rental housing. In all cases the results have shown the expected (positive) sign and
correlations were significant and high. The following figures show the evolution of the
price ratio and the correlation results.

Figure 11

25
 Figure 12 Figure 13

Source: Own based on market information and official figures.

A second piece of information refers to amortization rates implicit in apartment sales.

According to León (2010) those rates varies between 0.2 and 0.7% annually, with an
average of 0.4 which translates in an imputed age over one hundred years for those
dwellings. This suggests a positive excess demand operating in this urban space in 2009,
causing a higher valuation of buildings notwithstanding of their age.

2.3.4. Financial markets (recent evolution, mortgage contract, informal financing,

accessibility)

The Argentine financial market has not fully recovered from the 2001-02 macroeconomic
crisis when the mortgage loan market suffered a disruption due to freezing of evictions and
renegotiation of contracts. Currently, private banks participate financing their clients in the
upper end of the market while public banks act spasmodically offering mortgage loans for
middle income households under more convenient access conditions responding to
government instructions. Credit lines are constrained to the funding of each bank.

The scarcity of mortgage loans partly explains the likelihood of access to home ownership.8
The following chart presents the dramatic evolution of the mortgage loan market after the
2001-02 crisis. Housing accessibility became an issue in Argentina after the 2001 crisis.

8
Anastasi et. al. (2006) studied the availability and uses of financial services across localities in Argentina.
They found that State owned banks have more branches distributed in the territory than private banks. Foreign
banks tend to be located in major urban centers and domestic private entities show greater relative response to
business environment at provincial level. Their econometric analysis shows that there is a link between the
level of banking activity and poverty. One of the main elements conditioning the degree of availability of
banking services is the number of inhabitants: 90% of localities with less than 2,000 people have no supply of
banking services. This percentage falls to 5% in the case of localities with a population exceeding 25,000
inhabitants. Moreover, at the provincial level more than 70% of loans and deposits are concentrated in the
City of Buenos Aires and the province of Buenos Aires. This concentration is partly explained by differences
in the judiciary process efficacy for loan recovery (Cristini et. al. (2002)).

26
The minimum income that was needed to qualify for a mortgage more than tripled in one
year while the average income level of the Argentine households did actually decline. It
was clear that average households were no longer able to afford a mortgage. At the same
time, housing prices recovered over time after a deep fall in dollar terms due to the mega-
devaluation. During 2004 and 2005, average monthly payments to cover a mortgage loan of
AR$ 40,000.- was AR$600 on average. Banks required that monthly payments be equal or
less than 30% of household income, thus monthly household income should amount to
around AR$2,000/1,800. Only the first two deciles of household income distribution
complied with that requirement. The situation has worsened since then due to the incidence
of inflation and the prohibition of indexing loans. Interest rates sometimes around 19% in
nominal terms (they tend to be negative when compared to annual wage increases) and loan
maturities around 12-15 years, determine initial repayment rates that exceed the capacity of
middle income households turning almost impossible access to mortgage.

Figure 14
Stock of Housing Loans
End of Month balances - 2001 constant ARG pesos in Millons and % of GDP-
1994.1 - 2009.12
12,000 4.5

11,000
4.0
Constant ARG pesos of 2001 - in millons-

10,000
3.5
9,000

3.0
8,000

% of GDP
7,000 2.5

6,000
2.0

5,000
1.5
4,000

1.0
3,000

2,000 0.5
Jul-94

Jul-95

Jul-96

Jul-97

Jul-98

Jul-99

Jul-00

Jul-01

Jul-02

Jul-03

Jul-04

Jul-05

Jul-06

Jul-07

Jul-08

Jul-09
Jan-94

Jan-95

Jan-96

Jan-97

Jan-98

Jan-99

Jan-00

Jan-01

Jan-02

Jan-03

Jan-04

Jan-05

Jan-06

Jan-07

Jan-08

Jan-09

Amount (Left Axis) % of GDP (Right Axis)

Source: Own on Central Bank information

2.3.5. Government (construction standards, residential regulations and social housing

policies)

Construction standards and residential regulations have been regarded in the literature on
urban development as a source of over-costs for construction and as a barrier for access to
housing. In Argentina, Goytia et al (2010) provides evidence on the relationship between
land regulation and the (formal/informal) residential tenure condition of households. These
authors collected a nationwide survey of local land use regulation from planning
professionals, covering the municipalities comprised in the large urban agglomerates of
Argentina. A set of indicators were then created allowing the analysis of the regulatory
environment according to some of the main issues (e.g., existence of land use plans;
authorities involved in zoning changes and residential projects approval processes;
existence of building restrictions, infrastructure provision, the presence of access to land

27
regulatory elements, and the cost related to project approvals). Then, using data from the
National Households Survey (2007) and the last available National Census (2001), they
estimated the effect of land regulation on household formal/informal tenure condition. They
found that those municipalities that had incorporated more land planning regulatory
measures into their legal and regulatory frameworks also faced the cost of larger informal
land sectors. They also found negative effects on formality for higher residential approval
costs and tighter regulation (in the form of more authorities involved in authorization).

Social housing policy analysis has been developed at length in Argentina due to this policy
early inception (the National Mortgage Bank was founded in 1882), the historical broad
scope of the housing program (FO.NA.VI) and to the unsatisfactory results over time. Very
briefly, social housing policies have been organized through the National Housing Fund
(FO.NA.VI). This Fund was the most important housing program since mid-70s9. After
more than 30 years in operation FO.NA.VI (now structured under the umbrella of the
National Housing System and minimized in operation due to inception of the Federal
Housing Plans in 2004) was characterized by important inefficiencies that were never
solved. Among them, high administrative costs to manage housing program delivery (an
average of 14% of total disposable funds in early 2000s); the main type of service is to
directly provide houses and their respective financing through highly subsidized mortgage
loans; the construction process was very long, generating high carrying cost and makes the
process sensitive to unanticipated construction costs rises and to potential inflation.

This Program delivery conveyed heavy implicit subsidies. In late 90s, financing was
provided at concessional conditions: repayment of loans involved between 21 and 26 years,
fixed interest rates were well below market standards, between 0% and 6% depending of
the type of program and loan recovery rate was very low: 35%-36% on average.
Altogether, this translated into a de facto subsidization rate of 80% of the housing value,
including debt forbearances. This figure does not include the subsidies embedded in the
provision of land at concessional prices by the local governments.

The beneficiaries of the system were not among the poorest households since no income
ceiling was set (but minimum income was required, to face the - yet largely theoretical-
repayment obligations). The population actually served was at an intermediate level
income. Although the majority of Provinces use a scoring system, based on socio-economic
criteria, lottery and allegedly also political interference played a significant role in the
selection of beneficiaries.

Our previous work has also shown that at the local level there are several factors, probably
related to poor regional institutions that worsen poverty conditions including housing. At
the same time, there has been a smooth but noticeable trend towards decentralization of
households to middle size cities, around 500,000 to 1,000,000 populations. Unfortunately,
Argentine housing programs do not accompany this trend. Organized under a supply
oriented scheme, they prevent households from moving in search of better institutional or

9
Law Nª 19929 of 1972 created FO.NA.VI, and after several modifications it was replaced by Law Nº 21.581
in 1977.

28
labor conditions, reducing flexibility and risking more severe problems like increasing
structural poverty.

Sustained fiscal effort to close the housing deficit has not exceed 0,5% of GDP in
Argentina´s recent history, and in confront to other short run priorities is a difficult
objective to be fulfilled. The new programs (Plan Federal) have combined social housing
provision with labor programs in order to create jobs directly and indirectly through public
housing investment. Spatial distribution of the benefits has been biased by political factors.
(see Cristini (2004) and Cristini and Iaryczower (1997)). Moya et. al. (2010) confirms the
previous diagnosis on FO.NAVI and adds information on the working of the Federal
Housing Plan.

The following Figure 15 shows the evolution of housing units under these programs.

Figure 15

Source: OWN based on MOSP.

3. Determinants of housing demand and tenure choice.

Housing demand is a complex concept. It may refer to housing services stemming from a
group of housing attributes or to the individual housing attributes themselves. It can be
considered narrowly, i.e. the building and facilities in it or more widely, including the
characteristics of the neighborhood. Finally, it may involve not only the use but also the
type of tenure (ownership or renting).

In this section we will develop two aspects of housing demand relevant for our case. First,
we will estimate a demand for housing services including neighborhood characteristics.
Secondly, we will study the determinants of the household´s tenure choices.

29
Demand for housing services will allow for the estimation of price and income elasticities
to be compared with the international and regional results. The analysis of tenure choice
will complement our description of the key variables driving the housing market.

3.1. Estimation of a demand equation

Following Fontela and Gonzalez (2008), Zabel (2004) and Malpezzi (2002) we estimate a
demand equation for housing. The demand is represented by the following function:

qij= q(pj,mij,Ai)

where qij, the physical quantity demanded by the household i located in j, has to be
estimated indirectly as shown below; pj is the house price index; mij is the household
income and Ai is a vector of demographic characteristics such as age, gender and
education. We allow housing prices to be different across markets.

Due to data availability we were restricted to use rental values for our estimations. In
Argentina housing renting amounts to around 12 % of total dwellings and even though the
rental market is small, rents are considered to be reasonably related to housing values (see
below for a discussion on this issue). We developed our estimation in six steps:
1. Permanent income estimates were calculated for each household (renters and
owners).
2. Hedonic equations were estimated for renters by urban agglomerate. Two different
approaches were used: one including permanent income as an independent variable
to approximate the neighborhood characteristics and the other, excluding it.
3. Every household in the sample (owners and renters) was imputed an estimated rent
using the previous hedonic equations in their two versions (with and without
permanent income as an explanatory variable).
4. Relative prices for each of the urban agglomerates were computed based on the
―average‖ dwelling rent of Buenos Aires City.
5. Quantity was estimated as the ratio between rent and ―price‖ for each household by
urban agglomerate.
6. The demand function was estimated and the Heckman correction was applied in
order to correct for potential bias in the sample since owners and tenants are not
randomly selected from the population.

The data:

To undertake this econometric exercise we used the dataset of the National Survey of
Household’s Expenditure and Income (ENGH in Spanish) collected in 1996/1997.10

This national survey sample included households living in urban areas in cities larger than
5,000 inhabitants (according to the National Census of 1991). The sample was designed to
cover 114 cities representing 28 million persons, equivalent to 96% of the urban
10
Unfortunately, a similar survey collected during 2004 and 2005 is still unavailable for the whole country.

30
population. The database informs data at the household level splitting the country in 6
regions and 12 sub-regions. These 12 sub-regions were used as Metropolitan Statistical
Areas (MSA) in our estimation.

The ENGH sample size covered almost 27,000 households. The survey contains
information about:

a) Household head characteristics: age, education level, gender, marital status, activity
condition, type of job, etc.
b) Household attributes: household size, income, level and composition of
expenditures, housing expenses, including payments for domiciliary infrastructure
and maintenance outlays.
c) Housing characteristics: tenure status, number of rooms and bathrooms, connections
to utilities, etc. (See ¡Error! No se encuentra el origen de la referencia. to
¡Error! No se encuentra el origen de la referencia. in the Annex 1 ).
d) Housing quality: outside wall materials, types of floors, roof material, etc.
e) Services in the neighborhood: paved street, street lights, running water, sewerage,
and garbage collection.

The following Table 7 shows the sample distribution by tenure, owners and renters. The
latter represents 12% of total households, 30% in the case of apartments.

Table 7

Sample Size by housing tenure

ENGH - 1996/97
Total Owners Renters Others
Total 26,966 19,210 3,264 4,492
Houses 23,349 17,105 2,176 4,068
Apartments 3,617 2,105 1,088 424
Source: Own estimates based on ENGH Database. See the Statistical Annex for data by regions.

Property values are not available from the survey. However, the ENGH contains rents paid
by home renters and we used them to estimate implicit rents for owners.

Permanent income estimation: Following Zabel (2004) and Fontenla and Gonzalez (2009)
we decomposed income into a permanent and a transitory component. Permanent income
was estimated taking into account household head characteristics: age, educational level,
gender, marital status, employment situation (employed/unemployed) and type of
employment (self-employed, employee). Household location by region was used as a
control variable. Temporary income is defined as the difference between household current
income and permanent income.

Housing hedonic equations for tenants: As expressed by Malpezzi (2002), the method of
hedonic equations is one way expenditures on housing can be decomposed into measurable
prices and quantities, so that rents for different dwellings o for identical dwellings in

31
different places can be predicted and compared. In our model, the hedonic equation is a
regression of rents on housing characteristics for each region. Hn is the vector that
represents housing characteristics (number of bedrooms, surrounding amenities, quality of
construction, etc.) of unit n. Bj is defined as the parameter vector (implicit prices of these
characteristics), which is allowed to vary across markets, for each of the housing
characteristics in Hn. Thus, the rent v of a housing unit n in the city j consumed by
household i is given by the following equation:

vin,j = v(Hn; Bj)

Then, if Hn as well as vin,j are known, it is possible to estimate Bj using a hedonic price
model.

Characteristics included in our model correspond to dwelling structural

(number of rooms, heating and air conditioning, roof quality, etc) and to
characteristics corresponding to access to domiciliary services (natural gas, electricity,
water and sanitation, telephone, etc). Distance to public transportation was also used.
Unfortunately variables reflecting other neighborhood amenities were not available.
list of average characteristics see
Table 8 below and for the details by urban agglomerate (12 sub-regions) see ¡Error! No se
encuentra el origen de la referencia. in the Annex 1.

Table 8
Dwelling and neighborhood characteristics

32
 Source: Own estimates based on ENGH Database.

There are many potential housing characteristics that could be included on the right hand
side of our equation. We tried most of the variables included in the ENGH, but several
conveyed similar information (were highly correlated) or no information at all (every
dwelling in the survey possessed the considered characteristic). Unfortunately, two key
variables missing in the survey were: size of the unit in sq meters and age of the unit. In the
first case, number of rooms and number of household members were used as proxy
variables.

Hedonic regressions were estimated for tenants by agglomerate. A usual criticism when
focusing the estimation on rents stems from the fact that different units may have different
lease terms or contract conditions. In Argentina practice and regulation of the lease contract
make this problem less important since most contracts show similar features: an extension
of two years and a rent net of any utility service payment. As regards property taxes, they
are generally paid directly by the tenant, adding to the rent. This latter fact might be a
source of distortion in our price estimation since property taxes are different across regions
but similar within regions. The parameters of these regressions (the hedonic prices) are
used below to impute rents to the household owners in the sample. This method may be
subject to criticism due to the small proportion of renters in the population and to potential
distortions in the rental market. In fact, to go from rents to house values may entail several
weaknesses such as:

33
 a) Competitive markets are required in order to price attributes through a hedonic
model. Often in developing economies, rents are regulated or threatened to be so.
This problem did not apply in the case of Argentine housing market given that
government intervention was relatively low11 and there were many suppliers.
b) Identical discount rates are assumed across heterogeneous families. If it weren’t the
case, rental prices wouldn’t be linearly related to property values. For instance,
wealthier families might face lower interest rate than the average and implicit
property rent would overestimate the true housing consumption.
c) Also it is possible that families attach different values to house attributes and that
there exist non-observable characteristics. This can be partially solved estimating
the implicit rents with quantile regressions where the hedonic equation is estimated
for each quantile of the conditional income distribution12.

The following

Table 9 shows the rent to income ratio by region in the year of the survey. Percentages are
noticeably similar across the country, suggesting that property value, rents and income
levels, which are different according to the location, vary proportionally between regions.

Table 9

Source: Own estimates based on ENGH Database.

Based on the vector of house rents we estimate a hedonic model as a function of housing
observable characteristics with renters’ data. (See the ¡Error! No se encuentra el origen

11
In spite of its quantitative importance, the State housing program is focused on low-cost shelters for low-
income families.
12
This procedure was followed by Gasparini and Sosa Escudero (2004) to estimate implicit rents in the
Greater area of Buenos Aires.

34
de la referencia. and ¡Error! No se encuentra el origen de la referencia. in the Annex
1). ¡Error! No se encuentra el origen de la referencia. shows the regressions that include
permanent income as an explanatory variable. This inclusion follows the same rationale as
in Fontenla and González (2009) where permanent income is used as a proxy of
neighborhood quality (they used age, educational level and marital status). The exclusion of
this variable alters the results for the hedonic prices, but as we shall see, modifies only
slightly the elasticity estimates. Housing characteristics vary slightly between the two
models (see ¡Error! No se encuentra el origen de la referencia.).

Imputed rents: Using the hedonic equation, we can predict how much a family living owing
its house would be willing to pay for it, thus imputing an implicit rent to home owners. We
present two alternative calculations varying the characteristics included in the hedonic
equation. In the Table 10 below we compare effective rents with the rents imputed by the
model.
Table 10

Both cases exhibit a bias towards rent overestimation when compare to the actual rent
resulting slightly higher in the second case, suggesting the existence of some omitted
variable. Rents imputed to owners exhibit an expected rent/income ratio varying between
22 and 34% across regions, with similar levels to the ones showed for actual tenants. One
important fact to keep in mind for results interpretation refers to the nature of imputed
rents: not being actual out-of-the-pocket costs for shelter they may differ from the owner´s
willingness to pay for housing services. In developing countries owners stay in the property
for very long period and have time to accumulate assets in the form of housing which
imputed rent they may not be able to pay at all times.

Price estimation:

The price index is computed defining a standard unit as H*n which corresponds to the
average characteristics of a housing unit in Buenos Aires City:

Pj = 100 X v(Hn; Bj)/ v(H*n; B1)

35
Where the market for which pj=100 if j=1 (Buenos Aires City).

The following Figure 16 shows the distribution of prices relative to BA City across regions
for the case of rents estimated including permanent income.

Figure 16

We think of each sub-region as a Metropolitan Statistical Area (MSA), i.e. a contiguous

area of relatively high population density, coinciding with distinct urban labor markets.

36
Quantities:

The value of the housing unit n in market j consumed by household i can be expressed as
vin,j =qij*pj; and the quantity consumed can be obtained as:

qij= vin,j/ pj

Therefore, once qij is calculated, it is possible to estimate the demand equation.

Demand equation:

We elaborated two versions for the demand equation. Model I uses the hedonic prices
estimations based on the equation with the permanent income included as explanatory
variable (proxy of neighborhood quality) and Model II doesn’t include it.

Price and income elasticities of the housing demand were estimated with the entire sample
and with the two separated groups, owners and renters. When estimated for the split groups,
we corrected for sample selection using the Heckman (1979) two-step estimator in order to
avoid distorted elasticity values. The results confirm the limited influence of selection bias.
(See Table 11). In the Annex 1 we present the complete results (see the ¡Error! No se
encuentra el origen de la referencia. and ¡Error! No se encuentra el origen de la
referencia.).
Table 11
Housing Demand Estimations
Elasticities 3/
Specification Permanent Price Positive Negative
Income Temporary Temporary Income
Income 4/ 4/
All Households
OLS Model I 1/ Column (1) ¡Error! No (1) 0.534 *** -0.534 *** 0.031 *** -0.010 ***
se encuentra el origen
de la referencia.
OLS Model II 1/ Column (3) ¡Error! No (2) 0.411 *** -0.471 *** 0.036 *** -0.011 ***
se encuentra el origen
de la referencia.
Owners
OLS Model I 1/ Column (1) ¡Error! No (3) 0.517 *** -0.533 *** 0.028 *** -0.005 ***
se encuentra el origen
de la referencia.
OLS Model II 1/ Column (3) ¡Error! No (4) 0.372 *** -0.454 *** 0.033 *** -0.005 **
se encuentra el origen
de la referencia.
Heckman M. 2/ Column (1) ¡Error! No (5) 0.503 *** -0.531 *** 0.028 *** -0.004 ***
se encuentra el origen
de la referencia.
Tenants
OLS Model I 1/ Column (1) ¡Error! No (6) 0.663 *** -0.522 *** 0.055 *** -0.035 ***
se encuentra el origen
de la referencia.

37
OLS Model II 1/ Column (3) ¡Error! No (7) 0.649 *** -0.513 *** 0.054 *** -0.036 ***
se encuentra el origen
de la referencia.
Heckman M. 2/ Column (2) ¡Error! No (8) 0.714 *** -0.508 *** 0.050 *** -0.036 ***
se encuentra el origen
de la referencia.
1/ Corresponds to a log lineal specification. Endogenous variable: natural log of rents. Model I corresponds to the hedonic
price estimations with Permanent Income. Model II corresponds to the equations without Permanent Income.
2/ Regression models with sample selection correction by Heckman (1979) two step procedure. Hedonic price estimations
with Permanent Income.
3/ (***) Statistically significant at the 1%; (**) statistically significant at the 5%; (*) statistically significant at the 10%.
4/ Corresponds to income plus 1, to avoid the indefiniteness of the natural log.

The key results in terms of price and income elasticities are the following:

a. Permanent income elasticities for the sample are in line with those reported in the
literature: from 0.41 to 0.53 according to the model considered (rows (1) or (2)).
b. Tenants are more elastic than owners to changes in the permanent income,
regardless of the specification estimated. In fact, given that owners live in the same
house for long periods once it was bought, they are less responsive to changes in the
permanent income. However, it must be mentioned that Malpezzi and Mayo (1987)
found in a survey for developing countries that income elasticities tend to be lower
for renters than for those owning their home.
c. For the same reasons, elasticities for temporary income changes are lower for
owners than for renters and they are a modest fraction of those calculated for the
permanent income (less than 10%). The response to a transitory fall in the income is
smaller than to an increase. The significant coefficients for the temporary income
shocks suggest that housing consumption smoothing presents obstacles, mainly due
to the deficiencies of the mortgage market.
d. Demand for housing is inelastic to changes in house prices, estimated at around 0.5
with no significant differences between owners and tenants. This result also may be
consistent with a deficient functioning of the mortgage market (i.e. consumption
smoothing is difficult in the housing market) and with the existence of transaction
costs for real estate operations. Fontenla and Gonzalez (2009) also mention this fact
indicating that the lack of good substitutes to owning in México reduces the price
elasticity of demand and many households become owners through self construction
and progressive housing.
e. Corrections for the existence of endogenous sample selections using the Heckman
two-step estimators indicate that, in fact, tenants are different from owners (see the
Annex 1 for complete results and rows (5) and (8) in the Table 11 above). With this
correction included, elasticities estimated for the sample of owners don’t change
significantly. However, in the case of the group of tenants, elasticity of the
permanent income variable becomes slightly larger than the case without the
correction (see rows (6) and (8) for a comparison). That is, renters are even more
responsive to changes in income than owners when differences between them are
taken into account.
f. Regarding other explanatory variables included in the equation (see Annex 1,
¡Error! No se encuentra el origen de la referencia. and ¡Error! No se encuentra

38
 el origen de la referencia.), it can be mentioned that the age variable results
statistically significant and positive in the case of owners suggesting that access and
progressive completion of the house improves with age, probably due to the
changing needs along the cycle of life, the increasing possibility of getting a loan or
accumulating savings for a down payment or, eventually, receiving an inheritance.
This result is opposite to the one found in the Mexican case. When controlling for
selection, the sample of renters shows that at older ages the demand for housing is
lower.
g. Also, if the household head is a male (variable Gender), demand for housing tends
to be lower (similar to México).
h. Finally, if the head of household is married, she diminishes her demand suggesting
some degree of under-consumption. This result may be interpreted as the
consequence of introducing longer term plans when the family has to raise children
(effective or potentially) in an economy where consumption smoothing is difficult
and unemployment fluctuations are important.

3.2. Estimate of tenure choices with a multinomial logistic model

In this section we present an econometric exercise to determine the factors influencing

tenure choices by households. In the standard approach, the election is generally restricted
between ownership or renting. In this case we allow for categories that illustrate household
choices of quality of the house where they are to live.

The following Table 12 summarizes the alternative tenure choices used in our exercise. On
the one hand, being an owner may imply to have a good quality house or a low quality one.
On the other, renters may also choose between two types of houses, an average quality or a
lower one.

Thus, there are 5 different tenure choices: owning and renting a good quality house, owning
and renting a low quality unit or any other remaining options (such as living in a dwelling
provided by the employer or occupying it with the owner´s consent).

Table 12
Tenure Category in the EPH Quality standard
Ownership House and plot ownership All except those included in Owning and Renting a low
Renting Renting quality house.
Owning a low quality House and plot ownership Any of the following:
house - Located near a garbage bump
House only ownership - Located in a flooding area
- Located in a slum
- Plumbing facilities unavailable
Renting a low quality Renting - Lack of electricity
house - Shared bathroom
- Precarious outer wall, roof, floor or ceiling
material
- Overcrowded rooms (e.g. more than 2

39
 persons)

Others The remaining categories All.

Source: Own based in available EPH categories.

The determinants of the tenure choice will be analyzed using a multinomial logistic model.
This model adjusts best to the multiple choices available. There is one equation for each
category to be estimated where each selected category is measured relative to a reference
category (e.g. owning a good unit). Thus, if the first category is used as the reference, then,
for the remaining categories m=2..5, the equation to be estimated is:
K
P (Yi m)
ln m mk
X ik Z mi
P (Yi 1) k 1

Then, there will be 4 equations (m-1) predicting log odds, one for each category relative to
the reference category. Vector X contains all the observable characteristics supposed to
have an effect on the tenure choice. The individual probabilities can be calculated as:

exp( Z mi )
P (Yi m) M

1 exp( Z hi )
h 2

The observable characteristics included in the regression analysis are standard in tenure
choice models (e.g. Chiuri and Jappelli (2003)). In fact, the econometric model posits that
the probability of different tenure choices is a function of the age and squared-age13; level
of education of the household head14; family size measured by the number of children;
gender of household head; current income level; if the head is unemployed; migration
conditions such as recent migrant and origin, domestic or from neighboring countries and,
finally, whether the household head works in the domestic service. As regards the latter,
Goytia et. al. (2010) found that working in the domestic service sector was a significant
determinant of land tenure in Argentina.

We used the dataset of the Permanent Household Survey (EPH in Spanish) corresponding
to the last quarter of 2009. This estimation provides a picture of the structural aspects of the
choice process influenced by the given macroeconomic scenario.

The following Table shows the descriptive statistics of the database. It can be observed that
in the case of good quality houses, 45% of families are owners and 12% are renters over the
total population. In contrast, 25% of total households own and 6% rent a low quality
dwelling.

Life cycle models of tenure choice predict that ownership and home quality rises as
household head ages increase (e.g. Ortaló-Magné and Rady (1998)). From the table
emerges that average age is higher for owners than for renters and also for household heads

13
We omitted cubic age used by the above mentioned authors.
14
It is used as a proxy for the permanent income.

40
living in reasonable quality houses than those living in low quality units. That is, quality
and ownership grow as time passes.

It must also be noticed that income is relatively higher for families living in houses of better
quality but that there aren’t differences, a priori, in household incomes reported by owners
and renters. More years of education are observed in heads renting than owning a house but
this can be explained by the larger proportion of younger among the more educated group.
Nevertheless, again there are differences in the quality related to educated people: as level
of education increases so it does the house quality.

Migrant households are more prone to rent than to own a house. However, migrants from
neighboring countries tend to rent a low quality house whereas domestic migrants (from
other provinces) also rent but higher quality ones. Finally, larger families tend to be owners
but at the cost of lower quality.

Table 13
Basic Statistics

As mig1 but recently

As mig2 but recently

From othe countries

Number of children
years of education

From neighboring

Domestic service
Recent domestic

Unemployed
Ln(income)

countries

migrant
Tenure

female

mig1

mig2

mig3

mig4

mig5
age

Owning Mean
44.7 57.0 0.36 7.95 11.9 0.029 0.031 0.009 0.001 0.002 0.026 0.017 1.0
S.D.
16.2 0.48 0.81 4.1 0.167 0.173 0.095 0.029 0.045 0.158 0.130 1.1
Renting Mean
12.2 41.3 0.40 8.00 13.7 0.032 0.029 0.107 0.007 0.015 0.029 0.021 0.6
S.D.
16.2 0.49 0.73 3.5 0.176 0.167 0.309 0.082 0.120 0.167 0.145 0.9
Owning a Mean
low quality
25.4 49.7 0.31 7.74 9.9 0.055 0.019 0.007 0.002 0.001 0.039 0.044 2.0
S.D.
house 14.7 0.46 0.83 3.8 0.228 0.137 0.083 0.044 0.033 0.193 0.204 1.8
Renting a Mean
low quality
5.4 39.4 0.28 7.72 11.4 0.100 0.038 0.055 0.022 0.015 0.046 0.060 1.7
S.D.
house 12.1 0.45 0.79 3.3 0.300 0.190 0.228 0.146 0.120 0.209 0.238 1.4
Others Mean
12.2 45.9 0.35 7.54 10.4 0.051 0.016 0.017 0.002 0.001 0.040 0.062 1.6
S.D.
16.0 0.48 0.78 3.5 0.220 0.125 0.128 0.045 0.037 0.197 0.242 1.6
Total Mean
100.0 50.9 0.35 7.84 11.4 0.042 0.026 0.024 0.003 0.004 0.032 0.032 1.4
S.D.
16.8 0.48 0.82 4.0 0.202 0.159 0.153 0.056 0.063 0.177 0.177 1.5
Obs. 17,807
Source: Based on the EPH, Fourth Quarter 2009.

Results

41
In

42
Table 14 we present the results from the multinomial logistic model. The Table contains the
exponential coefficients that can be interpreted as odds ratios, the so called relative risk
ratios. They represent the change in the odds of being in the dependent variable category
versus the comparison category associated with a one unit change on the independent
variable, given the other variables in the model are held constant. The reference tenure
choice is ownership of a good quality house, so all the relative risk ratios must be
interpreted as relative to this tenure choice.

43
 Table 14
Multinomial Logistic Regression
With controls by agglomerate
Relative Risk ratios and Standard Deviations
Variable P(Rent)/P(Own) P(Own P(Rent P(Others)/
LQ)/P(Own) LQ)/P(Own) P(Own)
(1) (2) (3) (4)
Age 0.8961 *** 0.9636 *** 0.9301 *** 0.9007 ***
0.0101 0.0122 0.0138 0.0172
Squared Age 1.0005 *** 1.0001 0.9998 1.0005 ***
0.0001 0.0001 0.0001 0.0002
Female 1.4152 *** 0.8177 *** 0.7662 ** 0.8758 *
0.0991 0.0335 0.1023 0.0607
L(income) 1.0681 0.7488 *** 0.6672 *** 0.5302 ***
0.0688 0.0545 0.0210 0.0148
Education 1.0056 0.8702 *** 0.8865 *** 0.8845 ***
0.0247 0.0085 0.0120 0.0134
mig1 1.5363 1.6574 ** 2.8058 ** 1.4897 *
0.5632 0.3295 1.2034 0.3038
mig2 1.4976 *** 1.0268 3.0267 0.7521
0.2173 0.1442 2.2500 0.2405
mig3 4.7195 *** 0.8169 3.8683 *** 1.3536
0.7166 0.1974 1.2905 0.3596
mig4 3.2801 *** 1.7995 4.9486 *** 1.2586
0.3082 1.3437 2.5670 1.3297
mig5 4.0476 1.1330 6.6076 1.4995
4.3995 0.5646 8.1673 1.6162
Unemployed 0.7237 0.9106 0.9837 0.6919 ***
0.2148 0.0867 0.1980 0.0682
Domest Service 1.1541 1.6733 *** 2.2574 *** 2.1176 ***
0.2014 0.2352 0.6125 0.4900
# Children 0.6492 *** 1.4887 *** 1.2471 *** 1.2627 ***
0.0243 0.0306 0.0326 0.0170
Pseudo R-Square 0.1672
Obs. 7,471,367
Log pseudolikelihood = -8,575,637.5
log pseudolikelihood(0)= -10,296,825
Std. Err. adjusted for 32 clusters by agglomerates.

Column (1) presents the relative risk ratios (RRR) for renting a good quality house relative
to owning it. Column (2) displays the RRR for owning a low quality house relative to
owning a good quality home. Similarly, column (3) refers to RRR for renting a low quality
house to owning a good quality one and column (4) to RRR for other choices relative to the
reference tenure choice.

44
For instance, column (1) indicates that as ages increase (coefficient of 0.89), household
heads are expected to become an owner of a good quality house compared to renting a
similar standard dwelling (the coefficient is lower than one which means that the
denominator presents more advantages to be chosen that the numerator).

Following this interpretation, we found that the same behavior is expected regarding the
age of the head relative to owning or renting a low quality house (Columns (2) and (3),
respectively)15. In all the cases, older heads have more chances of owning a good quality
house than any other choices.

But what would be the results for other alternatives, for instance, renting versus owning a
lower quality home? The following Table 15 shows the probabilities for the different tenure
choices along age groups holding the rest of the variables at their sample average. As
expected, until thirty four years old the average household head is more likely to rent a
house, of reasonable quality. However, the next step she would take is buying a house,
although giving up some comfortable features is also a highly probable alternative. At the
age of forty and later, more probably she would buy a house of better quality and the
chances of doing it are increasing as time passes. Renting a good house is more probable
than renting a substandard home except for the elders. Except at early ages, owning a bad
quality house is always more preferable for the average HH than renting any other type of
house. At the average age, about 50.4 years old, there are a 52% of probabilities of owning
a dwelling of reasonable standard, 24% of owning a low quality house and 13% of renting.

Table 15
Probabilities for different age groups
Age of the HH Good quality house Low quality house Others
Owner Renter Owner Renter
20-24 0.095 0.372 0.121 0.049 0.364
25-29 0.141 0.328 0.151 0.051 0.328
30-34 0.204 0.277 0.183 0.052 0.284
35-39 0.284 0.222 0.211 0.050 0.234
40-44 0.371 0.170 0.230 0.046 0.183
45-49 0.465 0.122 0.239 0.040 0.135
50-54 0.552 0.084 0.236 0.033 0.095
55-59 0.630 0.056 0.224 0.026 0.065
60-64 0.697 0.035 0.206 0.020 0.042
65-69 0.750 0.022 0.185 0.015 0.027
70-74 0.796 0.014 0.163 0.011 0.017
75-79 0.832 0.008 0.141 0.008 0.011
80-84 0.860 0.005 0.123 0.006 0.007
85+ 0.896 0.002 0.095 0.004 0.003
Total 0.522 0.096 0.238 0.036 0.108
Note: all the observable characteristics are held at the national sample average.

If the household head is a female, the RRR show that there are more chances of renting
than owning a good quality home (column (1)) compared to males. However, she would
probably own a good unit rather than owning (column (2)) or renting a low standard one

15
Given that the fifth tenure choice is a residual category, we disregard further comments on this results
unless it be necessary.

45
(column (3)). In all the cases, as can be seen in the Figure 17 women choose better
standards for houses relative to men.

Figure 17
PROBABILITIES OF DIFFERENT TENURE CHOICES BY GENDER OF THE HH
Other Variables at their average values
0.60

0.50

Male Female
0.40

0.30

0.20

0.10

0.00
Owner Renter Owner LQ Renter LQ Others

Income levels also help to explain the tenure choices. Table 16 summarizes the exponential
coefficients (or relative risk ratio RRR) of some independent variables for different choices.
The first observation to be noted is that income levels don’t make any difference to explain
between owning or renting when referring to a good quality house (2 to 1 and vice versa).
Also, it barely helps to determine the choices between owning and renting a low quality
unit (3 to 4 and vice versa). That is, household income level is not a good predictor of
choices between renting and owning a house.

Notwithstanding, it helps to determine whether the choice will be between a good or a bad
dwelling (1 to 3, 2 to 4, 2 to 3 and 1 to 4 and vice versa). In fact, renting a comfortable
home is always more probable than living in a low quality one (being renter or owner) as
income grows16. Furthermore, owning a good house is more likely when the family
becomes richer.

Table 16 also shows the RRR for years of education. To be noticed, years of education are
not determinants of renting or owning a house of comparable quality (2 to 1, 3 to 4 and vice
versa). As in the case of income levels, the variable of years of education helps to explain
choices between different dwelling qualities. Again renting a reasonable property is more
probable as education increases than living in substandard homes. The same results are
found comparing owning to other options of lower quality houses.

16
These can be seen in the exponential coefficients that represent the odds ratios. For instance, comparing 2 to
3 the RRR is 1.426 which means that one percent increase in income raises more the probability of choosing 2
than choosing 3 by 42.6%.

46
To be unemployed is not a determinant of any particular type of tenure or quality. Working
in the domestic service increases the probabilities of living in a substandard unit, compared
to buying a higher quality house.

Finally, having more children raises the chances of owning a house compared to renting it
when the quality is the same (uneven alternatives dominate the even ones). Also, it is more
likely to own a low quality house than renting a good one (3 to 2). But if renting, a
substandard unit dominates the election (4 to 2).

Table 16
RRR between tenure choices and independent variables
1 = Owner; 2= Renter; 3= Owner LQ; 4= Renter LQ
Odds comparing Income Years of Unemployed Domestic service Children
alternatives education activity

2 to 3 1.426 ** 1.156 ** 0.795 0.690 0.436 **

2 to 4 1.601 ** 1.134 ** 0.736 0.511 0.521 **
2 to 1 1.068 1.006 0.724 1.154 0.649 **
3 to 2 0.701 ** 0.865 ** 1.258 1.450 2.293 **
3 to 4 1.122 0.982 0.926 0.741 1.194 **
3 to 1 0.749 ** 0.870 ** 0.911 1.673 ** 1.489 **
4 to 2 0.625 ** 0.882 ** 1.359 1.956 1.921 **
4 to 3 0.891 1.019 1.080 1.349 0.838 **
4 to 1 0.667 ** 0.887 ** 0.984 2.257 ** 1.247 **
1 to 2 0.936 0.994 1.382 0.867 1.540 **
1 to 3 1.336 ** 1.149 ** 1.098 0.598 ** 0.672 **
1 to 4 1.499 ** 1.128 ** 1.017 0.443 ** 0.802 **
Notes: RRR is relative risk ratio or e^b = exp(b) = factor change in odds for unit increase in X. We
omit the tenure choice “Others”. Variables: Significant at 1% (**) and 5% (*).

Regarding the migrant condition,

47
Table 17 presents the relative risk ratios for different migrant conditions. Immigrants from
neighboring countries (mig1) are more prone to rent a low quality house (alternative 4) than
to live in a higher standard dwelling. Recently arrived migrants from these countries (mig4)
show higher probabilities of renting to owning whatever the home quality. Also,
immigrants from other non-neighboring countries are more likely to rent a house (mig2)
than owning one of lower quality. However, if they are recently arrived (mig5), they will
rent a substandard dwelling. Finally, migrants from other provinces (mig3) are more likely
to rent a house (even alternatives dominate uneven ones).

48
 Table 17
RRR between tenure choices and migrant condition
1 = Owner; 2= Renter; 3= Owner LQ; 4= Renter LQ

Odds mig1 mig2 mig3 mig4 mig5

comparing
alternatives
2 to 3 0.927 1.459 * 5.777 ** 1.823 3.573
2 to 4 0.548 ** 0.495 1.220 0.663 0.613
2 to 1 1.536 1.498 ** 4.720 ** 3.280 ** 4.048
3 to 2 1.079 0.686 * 0.173 ** 0.549 0.280
3 to 4 0.591 * 0.339 0.211 ** 0.364 ** 0.172 *
3 to 1 1.657 * 1.027 0.817 1.800 1.133
4 to 2 1.826 ** 2.021 0.820 1.509 1.633
4 to 3 1.693 * 2.948 4.735 ** 2.750 ** 5.832 *
4 to 1 2.806 * 3.027 3.868 ** 4.949 ** 6.608
1 to 2 0.651 0.668 ** 0.212 ** 0.305 ** 0.247
1 to 3 0.603 * 0.974 1.224 0.556 0.883
1 to 4 0.356 * 0.330 0.259 ** 0.202 ** 0.151
Notes: RRR is relative risk ratio or e^b = exp(b) = factor change in odds for unit increase in X. We
omit the tenure choice “Others”. Variables: mig1 = migrant from neighboring countries; mig2=migrant
from other countries; mig3= recent migrant from other provinces; mig4= recent migrant from
neighboring countries and mig5= recent migrant from other countries. Significant at 1% (**) and 5%
(*).

As it has been noted before, there is an important dispersion in housing conditions across
cities. For instance, 55.4% of households own a house of reasonable quality in Ushuaia
(Tierra del Fuego) whereas in the other extreme only 29% are in those conditions in Salta;
23.8% are renters in Rio Gallegos, far from the 3% in Santiago del Estero. The same can be
said when analyzing the tenure status of substandard units: 48% of families in Santiago del
Estero own their own house and only 11% in Santa Rosa (La Pampa); 14% rent in Rio
Gallegos but only 1.4% in Santiago del Estero, a city where the stock of house for renting
are of better quality by a large extent than those owned.

Precisely, not only tenure choice varies but also the quality of housing stock is highly
diverse across cities and depending on the tenure. Generally speaking, most of the cities
present a larger proportion of good quality units for renting than those owned: the average
for renting is 69% whereas it is 64% for owned houses. Some agglomerates such as Gran
Tucumán present extreme cases, 81% of rented houses and only 46% of owned units are of
good quality, or Jujuy where the proportions are 47% and 66% respectively17.

17
See ¡Error! No se encuentra el origen de la referencia. in the Annex 1.

49
 Table 18
Rates across agglomerates: 32 main urban cities
2009
Owners Renters Owners LQ Renters LQ Others
Mean (arithmetic) 0.466 0.137 0.240 0.051 0.107
Average (weighted) 0.447 0.122 0.254 0.054 0.122
Median 0.475 0.142 0.235 0.046 0.104
Standard Deviation 0.081 0.056 0.095 0.025 0.036
Coeff. Dispersion 0.181 0.457 0.375 0.454 0.295
Max 0.580 0.238 0.479 0.139 0.177
Min 0.282 0.030 0.100 0.014 0.035
SOURCE: Based on EPH 2009.

What can our multinomial logit model say about these geographical variations? The
following

50
Figure 18 shows the percentage points of actual rates of reasonable quality houses that are
explained by city’s differences in household characteristics relative to the national
average18.

We can observed that almost 20 p.p. of the percentage of good quality houses in the City of
Buenos Aires are explained by the observed household attributes that are different to the
average of the whole sample. In this sense, this figure can be interpreted as the contribution
to the share of reasonable quality dwellings due to the family composition (set of household
attributes) of the agglomerate by itself. Thus, it can be argued that cities with positive
contributions are prone to demand good quality houses because of their family types. In the
case of Buenos Aires City, factors contributing positively to this type of demand are:
incomes, household head’ ages and education levels above the national average. On the
other side, the composition of Formosa’s population shows younger and less educated
household heads, lower incomes and larger family sizes. By the same token, low quality
rates mirror these percentages.

18
Percentage points explained by household’s characteristics relative to the national average are estimated as
the difference between predicted probabilities and the predicted probability for the national average home.

51
 -10.0%
10.0%
15.0%
20.0%

0.0%
5.0%

-5.0%
Ushuaia - Río Grande

-15.0%
-10.0%
20.0%
25.0%

10.0%
15.0%

0.0%
5.0%

-5.0%
Viedma - Carmen de Patagones
Total Formosa
Santiago del Estero - La Banda
Santa Rosa - Toay
Gran Resistencia
Bahía Blanca - Cerri
Posadas
San Luis - El Chorrillo
San Luis - El Chorrillo
Río Cuarto Viedma - Carmen de Patagones
Posadas Gran Catamarca
Corrientes Concordia
Gran Resistencia Corrientes

Neuquén - Plottier Jujuy - Palpalá

Neuquén - Plottier

Owning
Formosa
Rawson - Trelew
Río Gallegos
Gran San Juan
Gran Santa Fe
La Rioja
Rawson - Trelew Ushuaia - Río Grande

Renting
Gran La Plata Partidos del GBA

52
Santiago del Estero - La Banda Gran Tucumán - Tafí Viejo

Concordia Gran Santa Fe

Low Quality
Good Quality

Salta

Figure 19
Figure 18

Gran Catamarca
Bahía Blanca - Cerri
Gran Rosario
San Nicolás - Villa Constitución
Comodoro Rivadavia - Rada Tilly
Santa Rosa - Toay
La Rioja
Total
Jujuy - Palpalá Río Cuarto
Gran Córdoba Gran Mendoza

characteristics relative to the national average

San Nicolás - Villa Constitución Gran Rosario

Gran Paraná Comodoro Rivadavia - Rada Tilly

Predicted rate (Actual household) - Predicted rate (AVG Household)

Gran Paraná

Predicted rate (Actual household) - Predicted rate (AVG Household)

Gran Tucumán - Tafí Viejo
Gran Córdoba
Salta
differential household characteristics relative to the national average
Percentage points of actual rates of good quality houses explained by

Mar del Plata - Batán

Percentage points of tenure rates explained by differential household

Partidos del GBA
Río Gallegos
Mar del Plata - Batán Gran La Plata
Gran Mendoza Ciudad de Buenos Aires
Gran San Juan
Ciudad de Buenos Aires
La Plata or even Rio Grande tend to be locations where renting dominates ownership.
different observed characteristics relative to the sample average. Thus, homes from Buenos

Rio Grande or Viedma due to the family composition. Notice that cities like Rio Gallegos,
Aires City are more prone to own or to rent a house than, for instance, those from Ushuaia-
Finally, the Figure 19 shows the percentage points of actual tenure rates explained by
 4. Housing market, construction and activity level: stylized facts
On the one hand, housing is a durable consumption good but also one of the most important
assets of households. This fact links housing demand to the interest rate and to the
evolution of the financial markets. On the other, construction companies and developers
evaluate market risks taking into account the evolution of alternative financial assets. This
influence may be disproportionate as to distort housing market operation in the case of
extreme macroeconomic volatility and unsecure property rights.

Stylized facts based on the elaboration of the previous sections may be stated for the
Argentine case as follows. Housing supply is subject to cyclical movements and in the long
run, it responds to demand stimulus. It is the nature of this stimulus what distinguishes this
case. When considering potential demand actors by income level, only the higher end of the
income scale is active in the market on permanent basis, though influenced by the business
cycle. For this segment of the market, positive shocks on income translate into increases of
demand for housing due both to a wealth effect and to changes in their investment portfolio
composition. These shifts in income may result from windfall gains such as the increase in
agricultural commodity prices in 2006-2008 or to more permanent economic patterns as
reflected by income concentration since the 90s. Recurrent macro instability may also push
portfolio shifts towards more secure assets, such as new one-bedroom apartments for
renting or for occasional use (complementing suburban homes). Finally, rich families may
pursue new status consumption or changes in preferences (i.e. living in suburban ―gated
communities‖ in the 90s). Instead, demand of medium income households crucially depend
on mortgage loan availability, saving capacity for down payments and affordability issues.
Thus only a portion of them are active in the housing market demanding for new or used
houses. Expansion of demand for new construction depends on credit availability as
exemplified during the 90s in Argentina. In absence of a mortgage loan market this
segment of demand will be constrained to one of the following alternatives: to participate of
the market for used property filtering down from the upper segment, to adapt their homes
through incremental construction or to apply self-construction in their own land plots. Also,
rental housing is an option for this segment under restricted conditions (access requirements
may be stringent and long term rents, uncertain). Very low income households are
generally excluded from the formal housing markets. Generally speaking, housing policies
have been ineffective to promote the improvement of the housing market (but for a very
short period in mid-90s) and in particular, social housing initiatives have been subject to
increasingly poor organization and results over time.

In spite of this apparent unsatisfactory working of the housing market, the housing situation
has not worsened so much as to prompt urgent policy interventions and consequently, the
housing ownership issue was not among the highest priorities in the voter´s scale of
preferences until recently. Evidence shows a relative high share of repairs, reforms and
incremental construction in the activity of the housing market suggesting that progressive
construction is a permanent activity in the market, financed by owners using own savings
or personal loans of small amount. Among the factors favoring the access of middle
income households to formal housing are: the pace of household formation that has been
influenced by lower population growth rate and marriage at older ages; the reduction of
household size allowing for larger inheritance financing down-payment or simply enabling

53
affordability of new houses; and in a smaller portion, the filtering down of houses from
high income deciles to lower income households. Over time, demographic factors shifting
housing demand outwards have also influenced the market: divorce rate increases are
associated with the expansion of one member households; increase in life expectancy
reduced the turnover of the stock. However, the situation may be changing rapidly in large
urban centers as judging from recent land invasions. Several factors have operated to
increase this hazard: the increase of population living in over-crowding conditions in large
urban centers and the higher rents charged for housing accommodation in informal
settlements, among others.

Argentina shows episodes of important disruption of the housing market due to recurrent
macroeconomic crisis. Historically private sector investment in construction19 represented
around 6.0% of GDP. As in most of the countries around the world, this activity has shown
to be closely linked to the overall economic cycle. As a share of the construction sector, the
housing sector did not play an important role leading the activity level, only in a few
exceptional episodes housing construction has been a salient driver of the business cycle.

At the same time, Argentina is one outstanding example of macroeconomic and

institutional instability and housing activity has reflected this fact. Since 1980 the country
went through three scenarios. The first one, during the eighties and until 1989, was
characterized by a high inflation regime, including two hyperinflation episodes (1989-
1990) exhibiting a stagnated per capita income. This period was known as the ―lost
decade‖. During this decade the housing finance was basically publicly served through the
National Mortgage Bank, a state owned entity, and by means of state housing programs,
mainly the Fondo Nacional de la Vivienda (FONAVI).

The second period, approximately since early 1991 until to 2001, was one of very low
inflation rates under the sign of a wide range economic reform conveying to episodes of
economic growth such as the one of 1995-98. During this period, a currency board was set,
the so-called Convertibility Plan, conveying to dollarization of loans and deposits in the
banking system. Also, credit to the private sector grew at double digit rates in real terms
and the mortgage market for housing reached its highest financial depth.

Finally, a third period can be traced since 2002 until now. After a severe macroeconomic
crisis at the end of 2001, the monetary regime changed to one of a floating exchange rate
and a more active monetary policy. On average, inflation rates are higher than in previous
years but lower than in the eighties. Economic activity rebound after the GDP fall of 11%
in 2001, showing a yearly 7% growth rate on average until 2009. The financial system did
not recovered completely since the crisis and the previous disruption of the mortgage
market resulted in a sharp decline of loans to one fourth of the figure of the 90s. State
public programs are more active than in the previous periods and their expenditures exceed
that of the private mortgage flows. The construction sector played a relevant role in growth
recovery and contributed to one third of employment creation.

19
There are not official figures of investment in housing. The closest figures are investment in construction
by the private sector.

54
Prices in the housing market evolved partly reflecting the macroeconomic turbulence and
partly responding to changes in household formation, availability of urbanized land,
affordability issues, etc.

A consequence of a complex number of factors such as the importance of temporary

income shifts in demand, the high transaction costs, demographic factors and the urban
dynamics, is that the own dynamics of the housing market respond to external shocks with
changes larger than in the case of a well-organized market where mortgage loans are in
place. These shocks seem to add variability to the usual construction business cycle.
Besides, the result of a negative shock may be uncertain. For instance, greater macro
volatility accompanied by negative real interest rates or confiscatory measures by the
government (i.e. bank deposits freezing in late 90s or nationalization of pension funds in
the 2000s) might push a portfolio change towards more secure assets including housing.

In this context, a set of policies to promote the improvement of housing market operation
will need to be broad in scope and rich in alternative instruments to assist to a wide range of
different cases.

5. Policy recommendations and final remarks.

This section identifies and discusses policy options for addressing market and regulatory
failures and promoting housing market development in Argentina. We provide a brief
description of the policy ―toolkit‖ adaptable for Argentina. We consider the initial
conditions of the Argentine setting: damaged credibility of mortgage contracts, shallow
financial markets, volatile business cycle and high share of urban poor households.

Policies to improve the housing market operation:

Our diagnosis on the current operation of the housing market in Argentina highlighted
several structural characteristics, some important short-run macro problems and the
recurrent pitfalls of the social housing policies.
As regards the structural characteristics:

A few large metropolitan areas concentrate the major portion of the urban
population.

Urban poverty in large cities is highly associated with persistent housing deficit.

The housing deficit is dominated by qualitative aspects such as deficient

construction and over-crowding of housing units.
Considering now the short run macro problems that could prevent the launching of a
comprehensive housing policy, two are the major aspects to be taken into account.

On the one hand, the financial market has not fully recovered from the consequences of the
2001-02 crisis and the mortgage market has lost its appealing as a financial business due to

55
both the generalized breach of contracts during the crisis that was legitimized through
protective legislation (still in place) and to the lack of long-term funding in the economy.

On the other, the annual inflation rate has fluctuated between 20 and 25% in the last 4
years. However, indexation of credits is forbidden for banks, thus reducing even more the
availability of long-term loans. To extend the term of credit operation over time banks have
to increase the interest rate charged on loans which, in turn, reduces the payment capacity
of potential borrowers.

Finally, concerning the social housing policies, they have been traditionally implemented
as supply side ones and, particularly after the inception of the National Federal Plans, they
have acquired a rather rigid and centralized approach subject to clientelistic problems and
more oriented to solve employment and activity level difficulties than to reduce the housing
deficit.

As pictured, this scenario is not the most favorable one to launch a comprehensive housing
policy unless it is conceived within the context of a wider initiative including
macroeconomic stability aspects.

Moreover, as it was described in the previous sections, one of the main problems of
Argentina has been the persistence of the housing deficit. Recent data on the Population
Census of 2010 illustrate the increase of housing stock up to a total of more than 14 million
units. The 2010-01 increase was around 18,7% (provisional data) which is similar to the
2001-1991 increase of 19,5%. Considering that population growth slowed down from
11,2% to 10,6% comparing the two periods the demographic trend could be operating in
favor of closing the housing deficit. Unfortunately, most of the increase in housing
construction accumulated as ―vacant houses‖ (for holiday or week-end use or for non-
residential use): around 2 million units in both sub-periods, respectively. This fact, along
with the increase of single member households, suggests that housing deficit has evolved
with no significant changes over the last two decades.

We have estimated that this deficit could be eliminated in a 5-8 year period under policy
initiatives oriented to develop the mortgage market and to provide social housing under a
decentralized, demand-side subsidized program (See for the details the Annex 2: A Sketch
to appraise the feasibility of a comprehensive housing program for urban locations in
Argentina). Total investment was calculated for different scenarios and total estimates were
obtained, amounting to 6 to 25% of GDP, depending on the policy coverage.

However, these calculations are subject to numerous caveats and potential criticism.
Among them, we considered hereafter two connected to our calculations on demand
elasticities and tenure choice (Section 3, before).

First, not only demographic changes might influence housing demand evolution but also
changes in the long term income and housing prices will operate to modify the demand for
housing services. For instance, during the 1997-2009, increases in real income of 2,8%
annually would have increased housing service demand in 21% over the period in absence
of changes in other variables affecting the demand. Similarly, the sustained increase of real

56
prices in urban real estate of about 3,5% over the period would have resulted in a decrease
of 28% in demand, ceteris paribus. A total combined reduction effect of around -7% could
have been in operation during the 1997-09 period. This sizable net effect could affect the
development of a program where subsidies were granted without evaluating the housing
price evolution20 (in general, the housing programs include initiatives to keep or reduce
housing costs and prices).

Second, our estimations on evolution of the tenure choice suggested that housing demand is
affected by the evolution of income and other variables over the life cycle. Thus, the
probability of owning a house of good quality grows significantly after the age range of 35-
39 years. This fact suggests that a comprehensive program to eliminate the deficit could
reduce its fiscal cost through assisting the older households in the first place. In our
example, the cost of investment in housing units corresponding to younger households
(under 40 years old) is around 11.8% to 3% of GDP according to the housing solution. To
postpone the access of younger families is equivalent to save 13.8% to 1.2% of the GDP.
These younger households should be induced to participate in alternative programs
including saving incentives until they decide to buy a home or they reach the required
conditions to have full access to the housing program.

In brief, these final remarks have attempted to show the importance of a coordinate social
housing program to eliminate the persistent housing deficit of Argentina and the various
components of the design that are subject to changes in demand characteristics and
household behavior. As a result, sizable reductions in the public cost of the housing
programs might be achieved.

20
Particularly this is remarkably important for demand subsidy schemes which tend to push the demand and
house prices in the market.

57
References

Anastasi, A., E. Blanco, P. Elosegui and M. Sangiácomo (2006). ―Bankarization and

Determinants of Availability of Banking Services in Argentina‖. Working paper of the
Banco Central de la Republica Argentina (BCRA). Diciembre.

Arriagada Luco, C. (2003): ―América Latina: información y herramientas

sociodemográficas para analizar y atender el déficit habitacional‖ Serie Población y
Desarrollo Nº45. Proyecto Regional de Población CELADE-UNFPA, CEPAL, ECLAC,
Chile.

Baer, W. and C. Williamson (1988). The Filtering of Households and Housing Units.
Journal of Planning Literature, Apr 01; 3: 127-152.

Cristini, M. y M. Iaryczower (1997). ―Un análisis económico de las políticas de vivienda

en Argentina: el FONAVI‖. Trabajo presentado en la reunión anual de la Asociación
Argentina de Economía Política, Bahía Blanca.

Cristini, M., Moya R. A. and Powell A. (2002) ―The importance of an effective legal
system for the credit markets: The case of Argentina‖. In M. Pagano (ed.), Defusing
Default: Incentives and Institutions, John Hopkins University Press.

Cristini, M. y R. Moya (2004) Las Instituciones del Financiamiento de la Vivienda en la

Argentina. Departamento de Investigación, Documento de Trabajo 498.

Chiuri, M. and T. Jappelli (2003)."Financial market imperfections and home ownership: A

comparative study". European Economic Review, Elsevier, vol. 47(5), pages 857-875,
October.

Fontela, M. and F. Gonzalez. (2009). ―Housing Demand in Mexico‖. Journal of Housing

Economics 18, 1‐ 12.

Galiani, S.and E. Schargrodsky (2004). "Effects of land titling on child health". Economics
and human Biology, Elsevier, vol. 2(3), pages 353-372, Dec.

________________________( 2010). "Property Rights for the Poor: Effects of Land

Titling". CEDLAS, Working Papers 0103, CEDLAS, Universidad Nacional de La Plata.

Gasparini, L. and W. Sosa Escudero (2004) ―Implicit rents from own-housing and income
distribution: Econometric estimates for greater Buenos Aires‖. CEDLAS, UNLP. Mimeo.

Goytía, C, C. De Mendoza andR. Pasquini (2010) ―Land regulation in the Urban

Agglomerates of Argentina and its Relationship with Households Residential Tenure
Conditions‖ Working Paper Lincoln Institute of Land Policy.

58
Heckman, J. (1979). ―Sample Selection Bias as a Specification Error‖. Econometrica
47(1): 153-161.

León, S. (2010) ―Formaciòn del precio de la oferta de los departamentos en barrios

seleccionados de la Ciudad de Buenos Aires: un análisis hedónico. UADE.

Malpezzi, S. and S. Mayo (1987). "The Demand for Housing in Developing Countries:
Empirical Estimates from Household Data". Economic Development and Cultural Change,
University of Chicago Press, vol. 35(4), pages 687-721, July.

Malpezzi, S (2002). "Hedonic Pricing Models: A Selective and Applied Review".

Wisconsin-Madison CULER working papers 02-05, University of Wisconsin Center for
Urban Land Economic Research.

Moya, R., G. Bermudez and A. Sparacino (2010). ―Programas de estatales de vivienda:

¿Están llegando a los pobres?: El caso argentino‖. Mimeo, IADB, Washington D.C.

Ortalo-Magné, F. and S. Rady (1998). "Housing Market Fluctuations in a Life-Cycle

Economy with Credit Constraints". FMG Discussion Papers dp296, Financial Markets
Group.

Ratcliff, R. U. (1945). ‖Filtering down and the Elimination of Substandard Housing‖. The
Journal of Land & Public Utility Economics, Vol. 21, No. 4 (Nov., 1945), pp. 322- 330

Skaburskis, A. ―Filtering, City Change and the Supply of Low-priced Housing in Canada‖.
Urban Studies, Vol. 43, No. 3, 533–558, March 2006

World Bank (2005) ―Review of Argentina´s Housing sector: Options for affordable housing
policy‖. Policy Research Working Paper. World Bank, Washington D.C..

Zabel, J. (2004). ―The demand for housing services‖. Journal of Housing Economics 13,
16‐35.

59