P- value Prob>Chi Squared
6.6 ANALYSIS OF MULTINOMIAL LOGIT MODEL RESULTS
162
significant as indicated by the F-statistics and its significance levels. Similar to medium residential density area, most of the variables are highly significant statistically.
What can be inferred from the results thus far is that most of the human capital variables are highly correlated with income variable. This therefore justified their exclusion from multinomial logit model estimations because using them together may lead to spurious results and consequently generate wrong policy diagnosis and recommendations.
163
increase preferences and probabilities for flats, duplexes and single household houses by 7.24, 4.87 and 3.23 times respectively over multi-household houses. The probabilities, however, decreased by 0.02 and 0.85 times for squatters‘ settlements and a room in the main building relative to multi-household houses. These results are more sensible given the fact that a condition of higher and better living is always preferred to lesser one. Single household houses are not statistically different from multi-household houses should there be an increase in the housing price. This situation can be explained by the fact that the take-home pay of the household head may not be enough to secure single-household houses in Lagos State even if income increases. The odds of a household head choosing a room in the main building and squatters‘ settlement decrease by 0.85 and 0.002times relative to multi-household houses where there are increases in the household income.
Household size variable also appears significant only for a flat in a block of flats relative to the base category. As the size of household increases, the odds of choosing to live or rent a flat relative to multi-household houses increases by 1.082 times. This can be interpreted to mean that as the size of household increases, a spacious and conducive house is more likely to be preferred. The condition of choosing to live in single household houses, room in a main building and squatters‘ settlements may not actually be better alternatives.
The reasons for this may be likened to prohibitive nature of house rents for single household houses and non-conduciveness of squatters‘ settlements. Also choosing to rent a duplex may not be an option because the cost of renting a duplex is mostly out of reach because of excessive rent that is associated with it.
Gender criterion as depicted on the Table is not an important determinant of residential housing choice as suggested by the level of non significance of z-statistics. Thus, being a female-headed household is not statistically different from that of men on issues relating to making residential housing choice. Though, being a female would increase the probabilities or preferences for single-household house, flats, duplexes while decreasing the likelihood for choosing squatters‘ settlement and a room in the main building but they are insignificant.
Age is another important demographic variable influencing residential housing choice in the housing literature. As revealed from the Table, household head whose age is above 65 years is more likely to choose a flat in a block of flats and duplex as his residential homes than the household head whose age falls within 15 to 65years relative to multi-household houses. This result can simply be attributed to the decision the elderly are likely to take since they are no longer economically productive. Thus, they may simply require a quiet, healthy
164
and conducive environment to live. It is also interesting to note that household head whose age is over 65 years is more likely than those within the age bracket, 15 to 65 years to choose a squatters‘ settlement to multi-household houses though at different level of significance to that of living or renting a flat. The odd of a household head (the elderly) increases more for flats (4.233times) than for both duplex (3.157times) and squatters‘ settlement (3.776times) but all compare to bench-mark category. This may be applicable to the elderly who could not afford to rent either a flat or a duplex in Lagos. They may simply prefer to provide their own informal and illegal housing structures devoid of hustling and bustling which characterises city life. This is more prevalent in Lagos metropolis where all sort of illegal structures are erected all over available spaces. Also, household head within age bracket 15-65 and above 60 years are not significantly different in their likelihood of choosing to live in either single household houses or room in a main building as against multi-household houses.
Religion is also an important factor when making residential housing choice as noted in the housing literature. This is in fact supported by several empirical studies eg Farley et al (1997). Our result also provides additional evidence to this. For example, household head whose religion is islam are less likely than the christians to choose a flat in a block of flats relative to multi-household houses. The odd of a muslim household head making a choice of a flat relative to multi-household houses decreases by 0.754 times than christian household.
Also household heads with other religious faith are more likely to choose a duplex than the christian in relation to multi-household houses. Similar argument is also hold for a muslim in the choice of duplex as a residence. In addition, household heads which are from other religious sects are more likely than the christians to make a choice of a room in a main building relative to multi-household houses. The odd of a head of household from other religious sects choosing a room in the main building increases by 1.654times relative to multi-household houses than household head that is a christian. A simple inference that can be drawn from this is that religious doctrines and ideology may affect people‘s ways and styles of life greatly.
Ethnicity also plays an important role in determining the choice of residence in the housing literature. This factor shows no significant effect on the choice of single household house. Household heads who are Ibo by tribe are more likely than Yoruba household head to choose flats relative to multi-household as depicted by the z-statistics value on the Table.
Whereas household heads from other tribes other than Hausa and Ibo are less likely than the Yorubas to choose a duplex as their residence relative to multi-household houses. Ibo people also are more likely to make a choice of residence in favour of a room in the main building
165
than Yoruba as against multi-household houses. This is quite common in the Lagos metropolis for the reason that has to do with the fact that majority of the people actually migrated to Lagos metropolis for business purposes. They mostly prefer to build expensive houses in their home towns than Lagos which they view as business centres where they come to make money back home. Ibos also, are more likely than the Yoruba to choose squatters‘
settlement relative to multi-household houses whereas household headed by Hausa are more likely than Yoruba to demand for squatters‘ settlements relative to multi-household houses.
Table. 6.22: ESTIMATED MULTINOMIAL LOGIT MODEL RESULTS FOR THE FULL SAMPLE SIZE
Independent variables
Dependent variable : Residential Housing Choicea Single
Household House
Flats in a block of flats
Duplex Room in the main Building
Squatters’
settlement Hedprice 0.004
(-0.79)
4.581***
(9.48)
3.501***
(8.04)
0.476***
(-8.12)
0.332***
(-3.90) Logincm 3.233***
(3.77)
7.242***
(6.55)
4.873***
(4.26)
0.85***
(-3.05)
0.02***
(-4.32)
Hsize 1.062
(1.27)
1.082**
(2.35)
0.984 (-0.32)
1.019 (0.86)
0.964 (-0.71) Genderb 0.986
(-0.07)
1.094 (0.57)
1.208 (0.95)
0.881 (-1.24)
0.867 (-0.55) Age_abv65c 0.467
(-0.73)
4.233***
(3.37)
3.157**
(2.22)
1.284 (0.78)
3.776**
(2.20) Rel_musf 0.866
(-0.72)
0.754*
(-1.69)
0.510**
(-2.92)
1.090 (0.93)
1.121 (0.45) Rel_oth 1.219
(0.48)
2.200**
(2.97)
3.714***
(4.16)
1.654**
(2.50)
0.865 (-0.27) Ethn_Haug 0.629
(-0.45)
1.026 (0.04)
0.550 (-0.57)
1.099 (0.25)
3.920**
(2.51) Ethn_Ibo 1.209
(0.89)
0.641**
(2.23)
0.846 (-0.74)
1.094**
(2.20)
1.402**
(2.13) Ethn_oth 1.052
(0.13)
0.980 (-0.08)
0.324***
(-3.25)
0.951 (-0.26)
1.282 (0.52)
Sample Size=4433, LR chi2(90)=1070.87, Prob>chi2=0.0000, Pseudo R2=0.0933, Log Likelihood=-5203.6228 Note: The variables‘ coefficients are odd ratios, which are referred to as relative risk ratios(rrr) in Stata program The figures in brackets are z-statistics, with (***) representing 1 percent level of significance (**);5 percent level of significance; and (*) 10 percent level of significance.
The following are the base categories for each of the power alphabets: a=Multi-household houses=Male c=Age bracket 15_65 d=Christianity and e=Yoruba.
Source: Estimations based on Lagos State Housing Survey data, 2006.
Housing price which is a composite price of housing characteristics plays a very critical role in determining the choice of residential housing in high residential density areas of Lagos state as depicted on Table 6.22. The hedonic housing price is statistically significant at a 1%
166
level of significance for both the flats and a room in a main building while that of squatters‘
settlement is significant at 5% level of significance. Household heads are2.629 times more likely to choose flats relative to multi-household houses while household heads are 0.709 and 0.810 times less likely to choose room in the main building and squatters‘ settlements relative to multi-household houses. For flats however, an increase in the housing price will likely increase the probability of choosing flats relative to multi-household houses. Contrary arguments hold for a room in a main building in the sense that an increase in price of houses will likely reduce the choice of demanding for a room in a main building relative to the reference group category. This same argument goes for squatters‘ settlement. It is therefore logically plausible to argue that it would be desirable for any rational person to have taken this kind of decisions since quality that is attached to flats is much more than that of multi-household houses. This may at the same time mean demanding less for a room in a main building and squatters‘ settlement because of the low and poor housing quality that may be associated with them.
Household income is significant for a flat in a block of flats, duplex and squatters settlements at both 5% and 10 % respectively. This simply suggests that at higher levels of income, household heads in high residential density areas will likely choose to live in a flat or duplex than living in a multi-household house. This variable is also significant for squatters‘
settlements. By implication, this means they are statistically different from that of multi-household houses because multi-households will likely demand less for squatters‘ settlements at higher house rents relative to multi-household houses. For the duplex as a choice of residence, inadequate household income level within the density areas may be enough reason coupled with prohibitive housing rents to reduce the likelihood of choosing it. This outcome goes to show that with increasing income level, the probability is high for the household heads to demand for less and low housing quality but rather demand for an improved housing condition.
The size of household also plays a prominent role in determining a person‘s choice of residence out of a feasible affordable alternative choice sets . From the table, it is discernable that an increase in the size of the household may likely limit the choice of duplex relative to multi-household houses. This may be explained by the fact that majority of people with a large number of family are commonly found living in multi-household houses in high residential density areas in Lagos. In actual fact this is more prevalent in this residential neighbourhood. Thus it is not a surprising to find a family of seven or eight living in a small-sized multi-household house in the Lagos, metropolis. It is equally not surprising to observe
167
that the odd of a household decreases by 0.572 times relative to multi-household houses. This is not unexpected given the nature of the people residing in such area.
The issue of gender in the determination of residential housing choice is also very critical as can be observed from the Table. A female-headed household have a higher likelihood than male in choosing a single household house as against the choice of multi-household houses. This contrast sharply with what will occur if it involves a room in a main building. What is observed from the Table is that a female-headed household is less likely than a male counterpart in choosing a room in a main building relative to multi-household houses. Thus from the Table ,it is evidently clear that while the odd of female-headed household increases by 2.676 times for a single-household house and decreases by 0.676 times for a room in the main building. This outcome simply suggests that a female-headed household has a very strong preference for better and qualitative houses than male counterpart in relation to occupying or choosing multi-household houses.
Age remains a significant factor in making a residential housing choice concerning duplex and squatters‘ settlement in high residential density areas. A household head whose age is 65 years and above will more likely than those within the age bracket 15 to 65years to choose duplex and squatters‘ settlement as their residential housing choice. What this implies is that at the age above 65 years, there is high likelihood that household head will choose to live in either a duplex or squatters‘ settlement depending on the financial status of the individual involved. If it were to be a wealthy or high net-worth individual, a duplex may be a preferred option most especially at retirement age. Squatters‘ settlement may provide an option if the person is not financially buoyant. The reasons for this are not far-fetched given the fact that the odd of a household head whose age is above 65 years increases by 4.623 times for a duplex and 3.272 times for squatters‘ settlement relative to multi-household houses than a household head within the age bracket 15 to 65years. At age 65 and above, the assumption is that people may no longer be economically productive therefore, would require a more lonely and spacious housing environment to live than those that are still within their economically productive years.
As for the religion, people with different faiths are more likely than christians choose to demand for single household house, flats and a room in the main building. This corroborates some beliefs that muslims and christians care less about material things. The odd of a household head choosing flats (3.851 times) increases more than that of both single-household houses (5.125 times) and a room in the main building(1.451 times) for single-household head with other religion than christian household heads in relation to multi-household houses.
168
Interesting results also emerged from ethnic variable as one of the determinants of residential housing choice. Here, an Ibo will more likely than the Yoruba choose and live in a single-household house, flats, a room in the main building and squatters‘ settlements relative to multi-household houses. This is significant at both 1% and 10% level of significance respectively. The odd of an Ibo person choosing a squatters‘ settlement is highest by 3.384 times than Yoruba person relative to multi-household houses while the odd for the same Ibo person increases by 2.870 and 2.331 times for both flats and a room in the main building.
Table 6.23: ESTIMATED MULTINOMIAL LOGIT MODEL RESULTS FOR HIGH RESIDENTIAL DENSITY AREAS
Independent variables
Dependent variable : Residential Housing Choicea Single
Household House
Flats in a block of flats
Duplex Room in the main Building
Squatters’
settlement Hedprice 0.008
(-0.32)
2.629***
(3.63)
1.344 (0.02)
0.709***
(-6.00)
0.810**
(-2.60)
Logincm 3.251
(1.37)
2.682**
(2.49)
5.709**
(2.96)
1.207 (1.03)
0.392**
(-2.91)
Hsize 1.022
(0.13)
1.096 (1.27)
0.572**
(-2.80)
1.015 (0.46)
1.012 (0.16) Genderb 2.676*
(1.65)
0.902 (-0.29)
2.615 (1.55)
0.676**
(-2.53)
2.816 (-0.57) Age_abv65c 1.96e-16
(-0.00)
2.994 (0.98)
4.623***
(3.48)
1.609 (0.97)
3.272**
(2.57) Rel_musf 2.641
(1.38)
0.860 (-0.45)
0.795 (-0.31)
1.171 (1.17)
0.768 (-0.73) Rel_oth 5.125*
(1.75)
3.851***
(3.34)
1.131 (0.14)
1.451*
(1.65)
1.325 (0.53) Ethn_Haug 4.335
(1.28)
3.595 (1.54)
2.168 (0.57)
0.907 (-0.21)
3.22e-15 (-0.00) Ethn_Ibo 3.565*
(1.73)
2.870***
(3.66)
1.520 (0.57)
2.331***
(5.98)
3.384***
(4.11)
Sample Size=1753, LR chi2(85)=1070.87, Prob>chi2=0.0000, Pseudo R2=0.0879, Log Likelihood=-1675.6973 Note: The variables‘ coefficients are odd ratios, which are referred to as relative risk ratios(rrr) in Stata program The figures in brackets are z-statistics, with (***) representing 1 percent level of significance (**);5 percent level of significance; and (*) 10 percent level of significance.
The following are the base categories for each of the power alphabets: a=Multi-household houses,b=Male c=Age bracket 15_65 d=Christianity and e=Yoruba.
Source: Estimations based on Lagos State Housing Survey data, 2006.
MEDIUM
Similar to earlier results, hedonic housing price was still found to be significant for a single household house, flats, duplex, a room and squatters‘ settlement respectively. Hence, housing price is found to be significant across all the residential housing choice. What is obvious from this result is that at a higher housing price, preferences are more likely to be shifted in favour
169
of flats, duplex and squatters‘ settlement relative to multi-household houses. This result appears reasonable in the sense that a rational thinking household head would prefer living in a flat and duplex should there be increases in the housing price than staying in a multi-household house which is devoid of privacy that are inherent in a qualitative house. This result contrasts sharply with what is obtainable in the case of a single household house and a room in a block of flats. Given these residential housing choice, increasing house rents are less likely to make a household demand for them relative to multi-household houses. This by implication suggests that preferences are likely to be shifted in favour of multi-household houses than demanding for a single household house and a room in a main building. It is surprising however to note that household head places more and a higher premium on squatter settlement over multi-household houses. Table 6.6.3 depicts that the odds of a household choosing flats, duplex and squatters‘ settlement increase by 5.334, 5.151 and 6.553 times relative to multi-household houses while the odd of a household choosing a single-household house and a room in the main building decrease by 0.535 and 0.421 times relative to multi-household houses. All these occur at different levels of significance.
Household income in the medium residential density areas is only significant for flats and duplex at the highest level of significance. A simple inference that can be drawn from this result is that household head with increased income will likely demand for flats or duplex as their choice of residence relative to multi-household houses as their z-statistics values are significant at 1% level of significance. But for other residential housing choice, household head position is not significantly different from that of multi-household houses. Thus, for the increased household income, the odds of a household choosing either flats or duplex increase significantly by 3.261 and 2.419 times relative to multi-household houses.
Household size appears significant for both flats and duplex at 5% and 10% level of significance respectively. What this suggests is that the higher the size of the household family, the more likely the preference for flats and duplex would be relative to multi-household houses. This result goes contrary to what was obtained in the case of high residential density areas where the odd of a household head choosing a duplex decreases by 0.572 times relative to multi-household houses.
In the case of gender dummies, it is observed that there was no significant difference between being a male or female in the choice of residence as clearly depicted on the Table.
For age dummies, it is only those within the age bracket of 65years and above that are likely to demand for flats as their choice of residential homes as against multi-household houses.
The assumption here is that at that age, a lonelier and private place of residence will be
170
preferable by the elderly. Also the odd of a household head of 65 years and above choosing squatters‘ settlements decreases by 0.577 times relative to multi-household houses than those household head within the age bracket of 15 to 65years.
Religion is also a very critical factor in the determination of residential choice in the medium residential density areas but to a lesser degree. Other religions relative to being a christian appears significant in choosing either a flat or duplex. Comparatively, muslims are less likely than christians to choose a duplex relative to multi-household houses.
Tribal sentiments also appear significant over half of the residential housing alternatives which include flats, a room in a main building and squatters‘ settlement respectively. What this suggests is that Ibos are likely than Yorubas to live in these type of residential houses as against staying in a multi-household houses.
Table 6.24: ESTIMATED MULTINOMIAL LOGIT MODEL RESULTS FOR MEDIUM RESIDENTIAL DENSITY AREAS
Independent variables
Dependent variable : Residential Housing Choicea Single
Household House
Flats in a block of flats
Duplex Room in the main Building
Squatters’
settlement Hedprice 0.535**
(-2.06)
5.334***
(3.49)
5.151***
(3.92)
0.421*
(-1.75)
6.553**
(2.40)
Logincm 0.671
(-1.23)
3.261***
(5.14)
2.419***
(3.15)
1.046 (0.21)
0.427 (-1.33)
Hsize 1.068
(1.21)
1.122**
(2.76)
1.108*
(1.96)
1.050 (1.30)
0.920 (-0.79) Genderb 0.840
(-0.65)
1.147 (0.70)
1.298 (1.16)
1.068 (0.38)
0.395 (-1.19) Age_abv65c 0.680
(-0.35)
5.362**
(2.84)
2.336 (1.21)
1.983 (1.18)
0.577***
(-5.03) Rel_musf 0.831
(-0.74)
0.875 (-0.62)
0.553**
(-2.20)
0.943 (-0.33)
1.357 (0.53) Rel_oth 0.778
(-0.74)
1.756**
(2.50)
1.542*
(1.67)
1.205 (0.86)
0.940 (-0.10) Ethn_Ibo 2.267
(-0.08)
2.267***
(4.73)
0.669 (-1.57)
3.898***
(4.02)
2.738**
(2.13)
Sample Size=1745, LR chi2(80)=406.60, Prob>chi2=0.0000, Pseudo R2=0.0772, Log Likelihood=-2428.8723 Note: The variables‘ coefficients are odd ratios, which are referred to as relative risk ratios(rrr) in Stata program The figures in brackets are z-statistics, with (***) representing 1 percent level of significance (**);5 percent level of significance; and (*) 10 percent level of significance.
The following are the base categories for each of the power alphabets: a=Multi-household houses=Male c=Age bracket 15_65 d=No education; e= Unemployed person; f=Christianity and g=Yoruba.
Source: Estimations based on Lagos State Housing Survey data, 2006.