Greener Journal of Social Sciences Vol. 9(2), pp. 39-44, 2019 ISSN: 2276-7800 Copyright ©2019, the copyright of this article is retained by the
author(s) DOI Link: https://doi.org/10.15580/GJSS.2019.2.082219159 http://gjournals.org/GJSC |
|
Factors Affecting House Ownership in
Nigeria; A Probit and Heteroscedastic
Probit Model Approach
Fasakin I.J.1*; Olanrewaju O.O.2;
Umeokeke N.I.3
Department of Agricultural
Economics, University of Ibadan, Nigeria
ARTICLE INFO |
ABSTRACT |
Article No.: 082219159 Type: Research DOI:
10.15580/GJSS.2019.2.082219159 |
With a rapid rate of
urbanization in Nigeria, almost half of the population already lives in
cities thereby putting great pressure on the country’s urban settlement.
Fundamental to this problem is the decision to become a homeowner which is
an issue of financial and social decision. Thus, this study investigates the
nexus between house ownership and socio economic variables of the household
head. The study makes use of nation-wide cross-sectional data of the
2015/2016 General Household Survey (GHS) conducted by the Nigeria Bureau of
Statistics. Results show that the main determinants of house ownership in
Nigeria are: Occupation, household size, education level, marital status,
income and age of household head. The study, therefore, suggests that
policies that are targeted at maintaining moderate or lean household size
and promoting non-farming occupation along gender-oriented development
policies should be advocated and promoted by policy makers, since these
variables are considered critical to increasing house ownership among
households in Nigeria. |
Submitted: 22/08/2019 Accepted: 25/08/2019 Published: |
|
*Corresponding Author Fasakin
I.J. E-mail: idowufasakin2010@
gmail.com |
|
Keywords: |
|
|
|
1.0. INTRODUCTION
Nigeria is rapidly becoming urbanised with almost half of
the population already living in cities and this is anticipated to increase to
75 percent by 2050 (CAHF, 2016 and Burgoyne, 2008), the increase in urban settlement is directly
proportional to the demand for housing. The World Bank estimated Nigeria’s population to be 200 million,
accounting for approximately 47% of West Africa’s population, with 64% of the
population living below the poverty line (World Bank, 2018). Poverty remains
highest in rural areas, remote communities and among female headed households. The
decision to become a homeowner seems to be an important financial and social
decision, and individual choices regarding whether to own a house or to rent are
important consumer behaviours (Çağlayan,
2012 and Silver,
1988). The housing census conducted last in Nigeria counted 28, 197, 085 households in 2006, of which
51% of these are flats, 9% in semi-detached houses, 14% rented a room in
someone’s house, and 1% lived in informal houses. 83% of Nigerians owned the
houses in which they lived, while 11.04% rented 3.01% occupied rents free, 2.01%
owned but had not yet fully paid off, and 0.52% squatted (NBS, 2006 and CAHF,
2016).
Household
Heads characteristics has been identified as the most important factors that
determine home ownership in some empirical studies(Çağlayan,
2012, Campbell
and Cocco, (2007), Constant et al. (2006), Case et al.,
2005;Boehm and Schlottmann, 2004 and Deng et al., (2003), and it was anticipated
that their studies determined that the characteristics of the Household Head
such as gender, income level, household size, age, marital status, and educational
background are influential in the decision to buy a house. Other studies
indicated that income level has both a direct and indirect impact on home
ownership. Çağlayan, (2012) indicated that besides the demographic characteristics of
the head of the household, education, employment, and income are also
influential factors in regard to home ownership, with the choice to live in
either rural or urban areas also having a significant impact. In studies
examining home ownership status, qualitative preference models e.g. Çağlayan, 2012,
Guriset al., (2011), Capeau
et al., (2003), Li (1977) are
commonly used among others.
Hence,
this study focuses on the heteroscedasticity problem in Probit
model by aiming at determining the factors affecting the probability of owning
a house in Nigeria. Data from the 2015/2016 General Household Survey (GHS)
conducted by the Nigeria Bureau of Statistics was used for the study. Heteroscedasticity
can cause problems such as incorrect standard errors, biased and inconsistent
parameters, which are ignored in many studies. This study is unique in this
methodology because the determinants of home ownership using the heteroskedastic
Probit model as well as the standard Probit model will
be examined. In the empirical literature, there have been some studies
conducted which have applied the Heteroscedastic Probit models, such as Alvarez and Brehm
(1995, 1998), Busch and Reinhart (1999), Krutz
(2005), Litchfield et al. (2011) Çağlayan, 2012,
among others.
This
paper is organized as follows: The following section includes the introduction.
Section 2 presents both the methodology (standard Probit and Heteroscedastic Probit
models). The data and variables are introduced in Section 3. Section 4 reports
the estimation results. The final section presents the conclusion
2.0. METHODOLOGY
2.1. Heteroscedastic Probit
Model
The Heteroscedastic Probit model is a
generalization of the Probit model because it allows
the scale of the inverse link function to vary from observation to observation
as a function of the independent variables. For a Probit model, the binary Probit model is based on the assumption that a latent
variable is linearly related to the observed
..., (1)
The standard probit model assumes
that the error distribution of the latent model has a unit variance. The
heteroskedastic probit model relaxes this assumption, and allows the error
variance to depend on some of the predictors in the regression model.
Heteroscedastic Probit
model is a generalization of the Probit model; Let;
j = 1…, N, be a binary outcome variable taking on the value 0 or 1. In the
probit model, the probability that
takes on the value 1 is modelled as a
nonlinear function of a linear combination of the k independent variables
= (
,
)
Pr(=1)=
(
)
..., (2)
in which () is the cumulative
distribution function (CDF) of a standard normal random variable, that is,a
normally distributed (Gaussian) random variable with mean 0 and variance 1. The
linear combination of the independent variables,
, is
commonly called the index function, or index.
Heteroscedastic Probit generalizes the Probit
model by generalizing to a normal CDF with a variance that is no longer fixed at 1 but
can vary as a function of the independent variables. Heteroscedasticity Probit
models the variance as a multiplicative function of these m variables
= (
) ..., (3)
Following
Harvey
(1976),
=
(4)
Thus the probability
of success as a function of all the independent variables is
Pr(=1)=
(
}
..., (5)
From equation (5)
above, it is clear that, unlike the index,
no constant term can be present in
if
the model is to be identifiable.
Suppose
that the binary outcomes are generated by thresholding an unobserved
random variable, w, which is normally distributed with mean
and
variance 1 such that
=
..., (6)
This process gives
the Probit model;
=
) ..., (7)
Now
suppose that the unobserved are heteroskedastic
with variance,
=
..., (8)
Relaxing the heteroscedastic assumption
of the Probit model in this manner yields our
multiplicative heteroscedastic Probit model;
} ..., (9)
The log-likelihood function
for the heteroscedastic Probit model is;
..., (10)
Where S is the set of all
observations j such that and
denotes the optional weights.
is
maximized for the model (Blevins and Khan, 2013, Greene, 2012).
For
this study, household head owned house is an indicator function of ownership of
house as follows (Glewwe, 1991);
(11)
Y is the dependent variable;
ownership of house, it takes the value of 1 or 0. (1= if the household head
owned house and 0= otherwise)
= Gender of the household
head (1 if male; 0, if otherwise)
= Age of household head (Years)
= Education household head
(Years)
= Household size
= Occupation of household head (1 if farming; 0, if otherwise)
= Marital Status
(1=married, 0=otherwise)
= Monthly income household
head
= Error term
3.0. Data Source and Sampling Technique
Data for
this study was extracted from the third wave (2015/2016) of the General
Household Survey Data (GHS) conducted by the Living Standard Measure (LSM) and
the Nigeria Bureau of Statistics. The data set had a household
response status of 99.72% (4987 out of 5000 respondents interviewed). Farm
households constituted 65% of the respondents, amounting to about 3000
respondents in year 2010, 2012 and 2015 respectively (NBS, 2016).
The
GHS panel survey used a two-stage probability sampling. The primary sampling
Unit (psu) were enumeration Areas (EA). These were
selected based on probability proportional to size (Pps)
of the total EAs in each state and FCT and the total households listed in those
EAs. A total of 500EAs were selected using this method. Households were
randomly using the systematic selection of ten (10) households per EA. This
involved obtaining the total number of households listed in a particular EA, and
then calculating a sampling interval (S.I) by dividing the total households by
ten (10). The next step was to generate a random start ‘r’ from the table of
random numbers which stands as the first selection. Consecutive selection of
households was obtained by adding the sampling interval to the random start. In
all, 500 clusters/EAs and 5000 households were interviewed. These samples were
proportionally selected in all the states such that different states had
different sample sizes. However, the selection covers all the Local Government
Areas and all the states in Nigeria, the urban and rural areas were also
included in the sample. Most of the explanatory
variables used in the study are presented in table 1 below;
Table 1:
Variables |
Abbreviations |
Descriptions |
House
ownership*** Income |
OWNH INC |
Own
house=1, rented=0 Nigeria
Naira |
Age of
Household Head |
AGE |
Year |
Education
|
EDUC |
Primary=0,
Secondary=1 and Tertiary=2 |
Occupation
of Hh Head |
OCCUP |
Agriculture=0,
Non-agriculture=1 |
Marital
Status |
MSTAT |
If
married =1, otherwise (single=0) |
Gender |
GEN |
Male=1,
female=0 |
Household
Size |
HHSZ |
Number
of people in the households |
Note: *** Dependent variable
(House ownership)
4. 0. RESULTS AND DISCUSSION
Table 2 below presents the results
of the Probit regression and heteroscedatic model used
to investigate the determinants of house ownership in Nigeria. House ownership
which takes the value of one (1) if a household head owns the house and zero
(0) if otherwise formed the dependent variable of the model. The marginal
effect estimates of the explanatory variables are also presented alongside of
the Probit and heteroscedastic Probit model on Table 2
respectively. The essence of presenting the two models is to facilitate
comparison and based on the statistical tests conducted, chose the model that
best explains the determinants of house ownership in the study area. Given the
drawbacks, as highlighted in section 2.1 and the statistical tests conducted, the
Probit model becomes more admissible for the analysis
of the determinants of house ownership in Nigeria. This position is
corroborated by the likelihood ratio chi-square value of 416.61 of the model
with a p-value of 0.0000 which reveals that the Probit
model as a whole is statistically significant at 1% level, and thus implies a
good fit.
From Table
2, going by the Probit model, Age of the respondents
has a positive effect on house ownership status and is statistically significant
at 1%, and thus implies that the older an household is, the probability of
owing a house. Nonetheless, while the coefficient of the model could be
admissible, the marginal effect estimate of the model becomes more useful in
providing direction of causality, and the extent to which this variable
influences house ownership status. Therefore, the marginal effect estimates of
Age of the household head reveals that a unit increase in Age increases the
likelihood of owning a house by 0.1% and is statistically significant at 1%.
Relatively, the marginal effect estimates of Age for the heteroscedastic Probit
appears to be superior to the Probit model, as seen in
its estimates (0.2%). This may be attributed to correction of variance in the
skewed distribution of age and house ownership status that the model proffers.
Gender of
the respondents is statistically significant at 1% level in explaining house
ownership in Nigeria. The Probit model has a marginal effect of -0.069, and by
implication means that gender reduces the likelihood of owning a house by 6.9%.
This is a better estimate in providing direction of causality compared to the
marginal effect estimate of the Heteroscedatic Probit model which shows that
gender reduces the likelihood of owning a house in the study area by 6.6%.
Further to this, it could further imply that being a male increase the
probability of owning a house in Nigeria by 6.6% and vice-versa. Relatively to
marginal effect estimates of other explanatory variables in the model, this as
well indicates that gender is critical to determining ownership of house in
Nigeria. Marital status has a positive relationship with house ownership
status, and it is statistically significant at 1% level. The marginal effect
estimates of the Probit model shows that marital status increases the
likelihood of owning a house by 0.8%, and thus implies that married respondents
are more likely to own a house in Nigeria. The heteroscedatic Probit model was also at a close call with a marginal effect
estimate of 0.007, implying that marital status reduces the likelihood of being
an house owner by 0.7%.
With
reference to occupation which is operationalized as farming and non-farming across
the respondents, it reveals that occupation positively influences house
ownership status. Table 2 jointly shows that being into non-farming activities
increases the probability of owning a house in the study area. Also, the result
of the marginal effect estimates of the Probit model show that a change from
farming occupation increases the probability of owning a house by 5.8%. This
result is highly comparable to the heteroscedastic model. Education level has a
negative relationship with ownership of house status. It is statistically
significant at 1% level. The marginal effect estimate of the Probit
model reveals that education level reduces the probability of owning a house by
2.8%. This is the same estimate as obtained from the heteroscedastic Probit model.
Household
size is estimated to negatively influence house ownership status in Nigeria.
The marginal effect of the Probit model reveals that
household size reduces the likelihood of owning a house by about 0.3%, and thus
implies that the smaller the household size is, the higher the probability of
being a house owner. This may be connected to the huge financial implication of
maintaining large households, which must have eaten deep into the income of the
household which could have being alternatively used in owning a house. This
marginal effect estimates obtained from the Probit model appears superior to
the heteroscedatic model, as it does not only present a more superior
confidence interval, the estimates are also slightly better. Surprisingly, against
a-priori expectation, income negatively influences house ownership in Nigeria.
Both the Probit and heteroscedatic model reveals that as income of the
household head increases, the likelihood of owning a house increases. This
could be connected with the fact that house ownership increases with age, as
earlier revealed from marginal estimates of age, a variable whose increase is
always associated being economically active. The older the household, the farer
he is from the agile and economic active cadre. Consequently, this category of
household head is associated with low income.
Table 2: Determinants of House Ownership in Nigeria
Variables |
Probit Model |
Heteroscedasticity Probit
Model |
|
|||
|
|
dy/dx |
|
dy/dx |
|
|
Age |
0.014*** (0.001) |
0.0012*** (0.000) |
0.015*** (0.005) |
0.002*** (0.000) |
|
|
Gender (Male=1; 0
otherwise) |
0.603*** (0.091) |
-0.069*** (0.011) |
-0.602 --- |
-0.066*** 0.011 |
|
|
Marital Status (Married=1;
0 otherwise) |
0.069*** (0.020) |
0.008*** (0.002) |
0.069 --- |
0.007*** (0.003) |
|
|
Occupation
(Agriculture=0; 1 otherwise) |
0.463*** (0.106) |
0.053*** (0.013) |
0.463** (0.185) |
0.051** (0.022) |
|
|
Education Level |
-0.228*** (0.028) |
-0.026*** (0.003) |
-0.228 --- |
-0.026*** (0.005) |
|
|
Household Size |
-0.024*** (0.008) |
-0.003*** (0.001) |
-0.023 --- |
-0.0023*** (0.001) |
|
|
Income |
-0.000*** (1.52e-06) |
-1.46e-06*** (0.000) |
-0.00002* (5.66e-06) |
1.44e-06 (0.0000) |
|
|
Constant |
-0.618*** (0.150) |
|
-0.618*** (0.200) |
|
|
|
No of Observation |
3585 |
|
|
|
|
|
LR Chi2 |
416.61*** |
|
|
|
|
|
Pseudo-R-Squared |
0.1100 |
|
|
|
|
|
Homoskedasticity
(LM Test) |
233.18*** |
|
|
|
|
|
Wald |
|
|
237.19*** |
|
|
|
Log- likelihood |
1685.9006 |
|
-1685.9006 |
|
||
Ln Sigma Square |
|
|
57.88*** |
|
||
Source: Authors
Computation, 2019 ***, **, * Sig at 1%, 5% and 10% respectively. Numbers in
parenthesis are Standard Errors (SE)
5.0. SUMMARY
AND CONCLUSION
Most empirical
studies are usually analyzed using Probit models among other methods, and are
often done without testing the normality or/and the heteroscedasticity
properties of the models.
The biasness and
inconsistent of the standard maximum likelihood estimators is a major problem
in Probit model, if the disturbances are abnormal or heteroscedastic.
Hence,
we used heteroscedastic
probit model and standard Probit
model to examine the determinants of home ownership in Nigeria. The Wald statistics (237.19***) indicates that the explanatory
variables in a model are significant,
and they contribute meaningful to the result of the model. The high value of
Sigma Square (57.88***) shows that the variance of the error term spread out
very from the mean and highly significant. The nexus between socio-economic variables
and house ownership status in Nigeria was explored in this study using Probit regression model and its variants. The Probit model
was used to statistically identify the determinants of house ownership. The Probit
model shows that age, gender, marital status, occupation, household size and
income are the main determinants of house ownership in Nigeria. While in the
Heteroscedasticity model, thesame variables were identified as the determinants
of house ownership in Nigeria. Arising from the findings of this study, it is
therefore suggested that policies that are targeted at maintaining modest or
lean household sizes should be encouraged. Also, there should be promotion of non-farming
occupations by policy makers in the country, since these variables are
considered critical to increasing house ownership among households in Nigeria.
REFERENCES
Alvarez, R. M., Brehm, J. 1995. American Ambivalence Towards Abortion
Policy: Development of a Heteroskedastic Probit model of competing values, American
Journal of Political Science, 39(4), 1055-1082.
Alvarez, R. M., Brehm, J. 1998. Speaking in Two voices; American
Equivocation about the Internal Revenue Service, American Journal of
Political Science, 42(2), 418-452.
Boehm, T.P. and Schlottman, A.M. 2004. The Dynamic of Race Income and Home ownership;
Journal of Urban Economics, 55(1): 113-130.
Busch, M.L.,
Reinhart, E. 1999. Industrial Location and Protection: The political and
Economic Geography of US Nontariff Barriers, American Journal of Political
Science, 43(4): 1028-1050.
Çağlayan,
E. 2012.
Heteroscedastic Probit Model: An
Application of Home Ownership in
Turkey. Regional And Sectoral Economic
Studies Vol. 12-2 (2012)
Campbell,
J.Y. and Cocco, J.F. 2007. How Do House Prices
Affect Consumption? Evidence From Micro Data. Journal
of Monetary Economics, 54(3): 591-621.
Capeau, B., Decoster, A., Vermeluen, F. 2003.
Home ownership and the Life Cycle: an
ordered Logit Approach, Centre for Economic Studies,
Discussion Paper Series, 03.08, http://econ.kuleuven.be/ces/discussionpapers/default.htm
Centre for Affordable
Housing Finance in Africa (CAFH), 2016. Understanding Nigeria's Housing
Finance Market. http://www.auhf.co.za/conference/housingandafricasgrowthagenda/.
The32nd annual conference of the
African Union for Housing Finance took place in Abuja, Nigeria, in September
2016,
Constant, A., Gataullina, L and Zimmermann, K.F. 2006 Gender, Ethnic
Identity and Work, IZA Discussion Papers, 2420
Deng, Y., Ross, S.L.,
Wachter, S.M. 2003. Racial Differences in
homeownership: the effect of residential location, 33(5): 517-556.
Glewwe, Paul, 1991. "Investigating the determinants of household welfare in Cote d'Ivoire," Journal of Development Economics,
Elsevier, 35(2): 307-337.
Greene, W. H. 2012.
Econometric Analysis; 7th ed. Upper Saddle River, NJ: Prentice Hall.
Guris, S., Caglayan, E., Un, T. 2011.
Estimating of Probability of homeownership in Rural and Urban Areas: Logit, Probit and Gompit Model, European Journal of Social Sciences,
21(3): 405-411.
Krutz,
G.S. 2005.
Issues and Institutions: Winnowing in the US Congress, American Journal of
Political Science, 49(2): 436-477.
Li, M. 1977. A Logit Model of Homeownership, Econometrica,
45(5): 1081-1097.
Litchfield J., Reilly,
B., Veneziani, M. 2011 An
Analysis of Life Satisfaction in Albania: A Heteroscedastic Ordered Probit
Model Approach, Journal of Economic Behavior & Organization, doi:
0.1016/j.jebo.2010.12.020
Nigeria Bureau of
Statistics 2015 http://www.nigerianstat.gov.ng official website to Nigeria’s
Federal Bureau of Statistics
NBS, 2016
Nigeria Bureau of Statistics, Sector Statistics Retrieved from:
http://www.nigerianstat.gov.ng/
Silver, S.D. 1988
Interdependencies in Social and Economic Decision Making: A conditional Logit model of the Joint Homeownership-Mobility Decision, Journal
of Consumer Research, 15(2):234-242.
World Bank 2016 World Bank publication on
quantitative indicators on business regulations and the protection of property
rights that can be compared across 189 economies over time available at www.doingbusiness.org the doing business indicators.
Cite this Article: Fasakin, IJ; Olanrewaju, OO; Umeokeke, NI (2019). Factors Affecting House Ownership in
Nigeria; A Probit and Heteroscedastic
Probit Model Approach. Greener Journal of
Social Sciences, 9(2): 39-44, https://doi.org/10.15580/GJSS.2019.2.082219159. |