101
102
Table 5.1: DESCRIPTION OF HOUSING CHOICE VARIABLES
Variables Definitions and Measurements
Thedprice Transformed hedonic price Hhincome Household income(naira value)
Hhsize Household size (in numbers)
Age
15-65 1 if household head age falls within 15-65 and 0 if otherwise 65 and above 1 if household head age falls within 65 and above and 0 if
otherwise Gender Dummy
Male 1 if household head is male and 0 if otherwise Female 1 if household head is female and 0 if otherwise Educational
Qualification Dummy
Edunone 1 If household head has no formal education and 0 if otherwise Edupry 1 If household head has primary education and 0 if otherwise Edusec 1 If household head has secondary education and 0 if otherwise Edutert 1 If household head has tertiary education and 0 if otherwise Eduvotec 1 If household head has either vocational or technical education
and 0 if otherwise Occupational Status
Dummy
Occu_unemply 1 if household head is unemployed and 0 if otherwise Occu_pub 1 If household head is a public servant and 0 if otherwise
Occu_priv 1 if household head is a private-salaried worker and 0 if otherwise Occu_selfemply 1 if household head is self-employed and 0 if otherwise
Occu_stua 1 if household head is a student or apprentice Ethnicity Dummy
Ethn_yoruba 1 if household head is a Yoruba and 0 if otherwise Ethn_hausa 1 if household head is a Hausa and 0 if otherwise Ethn_Ibo 1 if household head is a Ibo and 0 if otherwise
Prior to estimation of the equation 16, both Hp and M will first of all, be estimated. The rationale behind this is that in the literature, housing price and quantity in equation (16) are not directly observable but, observed jointly either as rent paid or as the owners‘ estimates of housing value. It has also been observed that for developing countries, the information is scarce and studies either omit the price term from the demand equation (Jimenez and Keare, 1984) or derive price estimates in indirect ways . The production function approach was applied to Korean data by Follain et al. (1982) and the hedonic approach by Ingram (1981) to Colombian data and by Grootaert and Dubois (1988) to Ivory Coast cities. Various solutions have been suggested in the literature because of this problem. The most appealing and often used approach is to estimate housing price either as actual or imputed rent or owner‘s value of the house (see Mayo, 1981; Malpezzi, Mayo and Gross 1981, Arimah, 1992 and Phillip,
103
2006). Housing price is a variable which poses a serious challenge to all housing researchers.
Hence, different methods have been adopted to proxy it, in the literature. The most favoured approach in the literature has been that of Hedonic pricing.
The basic premise of the hedonic pricing method is that the price of a marketed good is related to its characteristics or the services it provides. The characteristic of a given dwelling unit or housing structure includes among others the property characteristics that affect selling prices, such as lot size, number and size of rooms and number of bathrooms;
locations of residential properties; accessibility characteristics that affect prices, such as distance to work and shopping centres, availability of public transportation and the neighborhood characteristics that affect selling prices, such as property taxes, crime rates and quality of schools. In capturing the housing price (P) in equation (16), we specify hedonic price function of the general form as follows.
0 ( )
p
i i j i
LogH X ……… (17) Where Hp is
hedonic prices, Xi is a vector of characteristics of the house traits (characteristics) like structural, neighbourhood and locational traits. This classical hedonic price model reflects a relationship between housing prices and traits. The housing traits can be classified into three categories: structural traits denoted by S; neighbourhood traits denoted by N; and locational traits denoted by L. Above equation (17) can be explicitly rewritten as:
( , , )
Hp H S N L --- (18)
The structural traits consist of roofing materials, walling materials, flooring materials, lighting types and water sources while the neighbourhood traits as identified in the survey are waste disposal methods, security services and pollution and locational traits also are distance to workplace, children schools, public transports, hospitals and water supply. In the light of the above, the empirical model of hedonic pricing was specified as follows:
0 1 2 3 4
( p) _ ij _ ij _ ij _ ij
Log H Roofing Mat Walling Mat Flooring Mat Lighting Typ 5Toilet_ facij 6Water soc_ ij 7Waste disp_ ij 8Securityij 9Pollutionij 10Distance emply_ ij 11Distance chdsch_ ij 12Distance_pubtranij
13Distance_pubtranij 14Distance hosp_ ij 15Distance watssp_ ijij-(19) Each of the explanatory variables is further sub-divided into different levels with each carrying zero and one value as dummy variables. The 1- 15, are the coefficients of the parameters to be estimated. The details of the variable description are shown on Table 5.2.
below.
104
Table 5.2: DESCRIPTION OF HEDONIC PRICING METHODOLOGY VARIABLES
Variables Definitions and Measurements
Structural Characteristics Roofing materials
types
Corrugated_roof 1 if the house is roofed with corrugated roofing sheet and 0 if otherwise
Cement_roof 1 if the house is roofed with cement and 0 if otherwise Tile_roof 1 if the house is roofed with tiles and 0 if otherwise Asbestos 1 if the house is roofed with asbestos and 0 if otherwise Wooden_roof 1 if the house is roofed with wooden roof and 0 if otherwise Thatched_roof 1 if the house is roofed with thatched roof and 0 if otherwise Mud_bricks 1 if the house is roofed with mud bricks and 0 if otherwise Walling materials
types
Mud_wall 1 if the house is walled with mud wall and 0 if otherwise
Burnt_wall 1 if the house is walled with burnt bricks wall and 0 if otherwise Cement_wall 1 if the house is walled with cement wall and 0 if otherwise Wooden_wall 1 if the house is walled with wooden wall and 0 if otherwise Corrugated_wall 1 if the house is walled with corrugated wall and 0 if otherwise Cardboard_wall 1 if the house is walled with cardboard wall and 0 if otherwise Flooring materials
types
Earth_mud_floor 1 if the house is floored with earth mud and 0 if otherwise Wood_tile_floor 1 if the house is floored with wood/tile and 0 if otherwise Plank_floor 1 if the house is floored with plank and 0 if otherwise Concrete_floor 1 if the house is floored with concrete and 0 if otherwise Dirt_straw_floor 1 if the house is floored with dirt/straw and 0 if otherwise Lighting Source types
PHCN 1 if Power Holding Company of Nigeria supplies the light and 0 if otherwise
Generator 1 if the lighting comes from generator and 0 if otherwise Candle 1 if the lighting comes from candle and 0 if otherwise Battery 1 if the lighting comes from battery and 0 if otherwise Gas 1 if the lighting comes from gas and 0 if otherwise
Kerosene 1 if the lighting comes from kerosene /paraffin and 0 if otherwise Wood_coal 1 if the lighting comes from wood/coal and 0 if otherwise
Toileting facilities
Flushpipe 1 if the toilet facility is flush to piped sewer and 0 if otherwise Flush_septic 1 if the toilet facility is flush to septic tank and 0 if otherwise Flush_pit 1 if the toilet facility is flush to pit and 0 if otherwise
Composting 1 if the toilet facility is composting and 0 if otherwise
VIP_pit 1 if the toilet facility is pit latrine with slab and 0 if otherwise Covered_pit 1 if the toilet facility is covered pit and 0 if otherwise
Uncovered_pit 1 if the toilet facility is uncovered pit and 0 if otherwise Hanging 1 if the toilet facility is hanging type and 0 if otherwise Pail/bucket 1 if the toilet facility is by pail/bucket and 0 if otherwise No_toilet 1 if there is no toilet facility and 0 if otherwise
105 Water source types
Pipebor_water 1 if water source is from pipe borne water and 0 if otherwise Public_water 1 if water source is from public tap and 0 if otherwise Borehole 1 if water source is from borehole and 0 if otherwise Well_water 1 if water source is from thewell and 0 if otherwise
SSvendor_water 1 if water source is from small scale vendor and 0 if otherwise Tanker_truck 1 if water source is from tanker truck and 0 if otherwise Other_water 1 if water source is from other water sources other than those
earlier mentioned and 0 if otherwise
NEIGHBOURHOOD CHARACTERISTICS Waste Disposal
Methods
PSP 1 if wastes are being collected by the government through private sector participation and 0 if otherwise
Dump_ground 1 if wastes are dumped in unauthorised places and 0 if otherwise Truck_push 1 if wastes are being collected by the truck pushers and 0 if
otherwise
Comp_dump 1 if wastes are dumped within the house compound and 0 if otherwise
Other_dump 1 if wastes are dumped through other methods and 0 if otherwise Security services
Com_pol 1 if security services are provided by the community police e.g like vigilante group, maid-guards etc and 0 if otherwise
Govt_pol 1 if security services are provided by the government police and 0 if otherwise.
Pollution
Littering 1 if pollution is mainly in form of littering and 0 if otherwise Public_urine 1 if pollution is mainly in form of urinating in the public places and
0 if otherwise
Poor_traffic 1 if pollution is in form of poor traffic and 0 if otherwise Illegal_trad 1 if pollution is in form of illegal trading and 0 if otherwise.
LOCATIONAL CHARACTERISTICS Distance to
employment
Distemployd0_14 1 if distance to household head place of employment takes between 0-14 minutes
Distemployd15_29 1 if distance to household head place of employment takes between 15-29 minutes
Distemployd30_44 1 if distance to household head place of employment takes between 30-44 minutes
Distemployd45_59 1 if distance to household head place of employment takes between 45-60 minutes
Distemployd60_abv 1 if distance to household head place of employment takes between 60-above minutes
Distance to children school
Distschdsch0_14 1 if distance of household head to children schools takes between 0-14 minutes
Distschdsch15_29 1 if distance of household head to children schools takes between
106 15-29 minutes
Distschdsch30_44 1 if distance of household head to children school takes between 30-44 minutes
Distschdsch45_59 1 if distance of household head to children schools takes between 45-59 minutes
Distschdsch60_abv 1 if distance of household head to children schools takes between 60-above minutes
Distance to public transport
Distpubtrans0_14 1 if distance of household head to public transport takes 0_14 minutes
Distpubtrans15_29 1 if distance of household head to public transport takes 15_29 minutes
Distpubtrans30_44 1 if distance of household head to public transport takes 30_44 minutes
Distpubtrans45_59 1 if distance of household head to public transport takes 45_59 minutes
Distpubtrans60_abv 1 if distance of household head to public transport takes 60_above minutes
Distance to hospital
Disthosp0_14 1 if distance of household head to the hospital takes 0_14minutes Disthosp15_29 1 if distance of household head to the hospital takes 15_29minutes Disthosp30_44 1 if distance of household head to the hospital takes 30_44minutes Disthosp45_59 1 if distance of household head to the hospital takes 45_59minutes Disthosp60_abv 1 if distance of household head to the hospital takes 60_above
minutes Distance to market
Distmkt0_14 1 if distance of household head to marketplace takes 0_14minutes Distmkt15_29 1 if distance of household head to marketplace takes 15_29minutes Distmkt30_44 1 if distance of household head to marketplace takes 30_44minutes Distmkt45_59 1 if distance of household head to marketplace takes 45_59minutes Distmkt60_abv 1 if distance of household head to marketplace takes 60_above
minutes Distance to water
supply
Distwat0_14 1 if distance from household head house to water supply takes between 0_14 minutes
Distwat15_29 1 if distance from household head house to water supply takes between 15_29 minutes
Distwat30_44 1 if distance from household head house to water supply takes between 30_44 minutes
Distwat45_59 1 if distance from household head house to water supply takes between 45_59 minutes
Distwat60_abv 1 if distance from household head house to water supply takes between 60_above minutes
However, the estimation of the implicit prices of the hedonic prices can be done by regressing market values of house pricesHp, measured as rents, as a function of various
107
housing attributes such as what we have in equation (19). The next stage in the implementation of this model is the choice of the functional form for the hedonic prices. The issue that has generated debates in the housing literature borders on the choice of functional form specification. In fact, quite a large number of studies on residential choice impose a prior restriction on their model before estimation. This may lead to model misspecification, erroneous and inconsistent estimates. In specifying a model, it is pertinent to derive information from the data itself (data-based model specification) instead of imposing an untested restriction in advance. Hence, we employ Box-Cox transformation which is a flexible form of obtaining an accurate functional form. In addition, since we do not have any prior notions about the shape of the hedonic functions, we estimate alternative forms of Box-Cox transformations. We estimate the general Box-Box-Cox functional forms given below:
( )
0 1
1
( ) i 0.5 i j
k p
i ij i j
i i j
H Z X X X
---(20) whereHp( )X ( ) [(Hp( ))X ( ) 1)]/ ---(21) and
( ) ( )
( 1) /
i i
X X i---(22) Where λ is a parameter used to transform housing characteristics to do Box-Cox transformation and τ is transformation parameter for rent (Hp). Nonlinear methods are used to find optimal values of transformation parameters. The thesis employs a Box-Cox transformation to transform the specification in equation (19) and the Box-Cox transformation generated the maximum likelihood estimates of the parameters according to
( )p 1 k 1
k
X
H
---(23)Where N(0,2) and , ( , ). The dependent variable Hp is transformed by the parameter , and each of the independent variables Xk is transformed by the same parameter λ. The transformed variables must be strictly positive to be defined for all values of and λ.
Thus, variables that have negative values or contain zeros such as dummy variables are not transformed.
Since Box-Cox transformation embeds several standard functional forms, estimating
and λ allows us to test these functional forms without imposing them a priori on the data.
108
In particular, when =λ=1, then equation (19) becomes linear. When =λ→0, the transformed elements of equation (19) become log-linear. Finally, when =λ=-1, the transformed elements of the regression become the multiplicative inverse specification.
Another benefit to be derived from using the Box-Cox transformation is that it makes the residuals more closely normal and less heteroskedastic.
In addition, M can further be calculated since researchers are divided on which measure to be used in capturing appropriate household income. Different measures have been applied in the literature but the most favoured measure is to regress current income on life cycle variables. This is achieved by running a household income regression using household demographic characteristics. The fitted value of the regression provides permanent income while the temporary income is calculated as the residual. The main justification for using income equation specified in equation 24 is whether there exists multicollinearity between income and human capital variables. The study adapted the version of Mincerian human capital equation. This is expressed as follows:
0
ij k ij ij
LnM H --- (24)
Where LnMij is natural logarithms of monthly household income of an individual in residential density area j, Hij are human capital and other background characteristics of household i in residential density area j, εij is the error term of zero mean and constant variance. 0,and k,are parameters to be estimated. If the variables are highly collinear, then human capital variables like education and occupational status will not enter into estimable models but if otherwise, all the variables will enter into the final estimable equation.
A Priori Expectation of Model Parameters
Among the explanatory variables considered in the equation 16 as factors influencing the residential housing choice are housing prices, household income, household size, age, gender and ethnicity.
The important role of housing price (housing rents in the case of rented houses or house value in the case of owner-occupier houses) in the determination of residential housing choice is well established in the housing literature. Apart from the fact that the strong surge in housing prices can place affordable housing beyond the reach of many demographic groups, it is also possible for such increase in housing price to influence the choice of residence type
109
and location. For instance, one may prefer a duplex to multi-household houses but if house rents were to increase substantially, such preference may change from renting a duplex to multi-household houses. Housing price (hedonic price) is a composite price in the sense that it represents a schedule of prices that a household faces. Thus, what one values in houses vary from one individual to another. While some place higher premiums on accessibility to workplace, others may value proximity to market place or central business districts (CBDs) and some place recognition on the presence of housing physical fittings than anything else.
Since all these cannot be explicitly traded in the market place, it must have been captured into prices, but at higher prices individual preferences may be distorted. Hence, the impact of housing price on residential housing choice is negative.
Household income is another vital explanatory factor in the residential housing determination. Housing literature clearly distinguishes permanent and temporary or current income but in studies of durable consumer purchases, permanent income has been shown to be the relevant variable in consumers‘ housing decision (Friedman,1957; Mayo,1981 and ; Malpezzi and Mayo,1987). This hinges primarily on the permanent-income hypothesis which states that in well-functioning capital markets, a household‘s consumption of durable goods is determined by the permanent income, which takes cognisance of the flow of income over a long time. This is because current income usually contains transitory components which bring about fluctuations in the flow of income over a given period. This in part resulted to a downward bias in estimates obtained via current income. (Follain etal,1980; Jimenez and Keare,1984 and ; Shefer,1990). The income data collected through the survey was based on average total income of the household per month. Thus we could not differentiate between permanent and current income for our analysis. More importantly, income usually impacts on the choice of residence positively, that is income is predicted to be positively related to residential housing choice.
The size of household could also determine greatly the residential housing choice behaviour of an individual household. The higher the size of the household, the greater would be the need to demand for spacious houses and hence influence the choice of residential housing . A family of two that hitherto had been staying in a room apartment may want to demand for two or three bedroom apartment once the family size increase to three or four.
Thus, the size of the family has a positive relationship with choice of residence. Households with children and relatively high incomes tend to live in suburbs because of the need for larger houses, larger lots and good schools ( Filion et al.1999).
110
Age has been found in many studies to have a strong influence on residential tenure choice decisions. Age captures the experience as well as the stage in the life cycle of the household. Life-course model of residential mobility suggests that different age groups with different household characteristics have different desires for their residences and that these preferences change over the life course (Wenning,1995 and Morrow-Jones,2005). Rowles (1993) noted that older people want to live close to their children and still be independent in their residential location decisions. What this suggests is that age as a factor determining choice of residence could either exert a positive or negative impact depending on the stages in the life-cycle of the individual household. Empirical evidence from Tiwari (2000) also showed that if age of household head increases by 1%, the market share of single family and multi-family ownership houses increases.
Gender is also an important factor in determining residential housing choice in the literature. It has been observed that females usually prefer and cherish their privacy than their male counterparts. Hence they prefer houses like flats, single-household houses to multi-household houses and squatters‘ settlement. Also, women or females are seen to prefer renting houses close to their workplaces and markets than their male counterparts. The issues relating to proximity to workplaces and markets are believed to have been captured under hedonic prices. What this suggests is that irrespective of the amount charged on house rents, females will always prefer all these features prior to their choice of houses whereas male counterparts hardly pay attention to these features in relation to the female counterparts.
The relationship between the level of education and residential housing choice is positive in many studies. The direction of the relationship tends to vary across different residential types and location. For example, it is expected that somebody with tertiary education tends to prefer flat, duplex and single-household houses to multi-household houses, room in the main building and squatters‘ settlements relative to someone with primary or no education. Thus, demand for different residential housing types tend to vary significantly across different level of educational attainments. The higher the level of education one attains, the higher the level of sophistication. Generally speaking, education tends to exert positive influence on both the demand for housing and choice of residence. Thus, the level of schooling determines greatly the type and choice of residential location of an individual household.
Another variable that was included in the estimated model is the occupational status of the household head. Occupation often measures the social status of the head of the household thereby indicating that household in which the head is employed in a white-collar
111
job are more likely to consume greater quantities of housing attributes. The results are consistent with those obtained by Blomquist and Worley (1981, 1982), and Witte, Sumka and Erekson (1979). The effect of occupation on housing consumption is generally positive but this tends to vary depending on the type and nature of occupation. For instance, public salaried workers tend to live in rented apartments than both self-employed and private-salaried workers. The choice made of residential housing also varies from one occupation to another.
Religion may be a factor in the emergence of residential concentrations as people who share cultural backgrounds including religion seek to live near each other or are attracted by services provided by religious organisations. In addition, religion may be used as a dimension in the identification of residential concentrations of people.
People of the same ethnic group are more alike, while people of different ethnic groups within the same racial group may be quite different. This suggests that individuals are drawn more to people of their own ethnic group rather than to people of other ethnic groups in the same racial group. In other words, people choose residences based on proximity to co-ethnics rather than other co-racial, but since all co-co-ethnics are of the same race, both ethnicity- and race-based concentration results. For instance, social capital theories suggest that ethnicity and race form important social and economic networks, leading people to gravitate towards others in the same group and ultimately resulting in geographic concentration by race and ethnicity. Thus, people will move to a neighbourhood or a place where there is a large population of coethnics.