Regression Result

Table 2 is the result of multi-nomial logit model of tenure choice with owner without loan as the reference group. It is suggested that household income, head age, head in public sector, and extended family are negatively associated with the probability of renter while household size, income earner, male headed, urbanization, two generation with income parent headed are positively associated with renter (relative to owner without loan). For owners with loan, household income, intact couple, higher educated head are positively correlated with the

propensity to own with loan than that without. Household size and extended family are negatively correlated with. Regarding head’s age, since linear and quadratic terms are both significant, we

9

try to graph the age ownership pattern and the result is shown in graph 1.

Table 2 Multi-nomial logit of tenure choice

Coefficient p-value Coefficient p-value Household characteristics

household income (hhy_tot) -1.494 0.000** 0.410 0.000**

Household size (hhsize) 0.076 0.035** -0.127 0.000**

Income earners (tot_ernr) 0.211 0.008** 0.045 0.291

Head characteristics

Head sex (hd_sex) 0.264 0.016** 0.022 0.751

Head age (hd_age) 0.021 0.360 0.048 0.002**

Head age squared (hd_age^2) -0.001 0.000** -0.001 0.000**

Spouse exist (sp_exist) 0.101 0.411 0.304 0.000**

Head in public sector (public) -0.432 0.010** 0.107 0.105 Head education-primary (hd_edelem)

Head education-junior high (hd_edjrhi) -0.026 0.821 0.068 0.344 Head education-senior high (hd_edhigh) 0.163 0.122 0.218 0.000**

Head education-college ( hd_edcoll) -0.226 0.142 0.312 0.000**

Housing location

urban 1.742 0.000** 0.863 0.000**

suburban 1.277 0.000** 0.906 0.000**

Demographic characteristics

Extended with child head (gen_2y) -1.407 0.000** -0.906 0.000**

Extended with parent head (gen_2o) -0.431 0.054* -0.339 0.002**

Two generation income with child head (inc_2y) 0.181 0.413 -0.178 0.133 Two generation income with parent head (inc_2o) 0.379 0.023** 0.052 0.588

Intercept 16.944 0.000 -7.197 0.000

Note: owner without loan as the reference group

Renter Owner w/ loan

Given the results of Table 2, it is needed to test the hypothesis of IIA to prove the validity of multi-nomial logit model. The result shows that IIA hypothesis is not fully supported by the results.

Table 3 IIA tests

renter vs. owner w/o loan renter vs. owner w/ loan owner w/ v.s. w/o loan

-16.970 1208.030 -21.110

p-value 0.312 0.000 0.267

conclusion not rejected rejected not rejected

χ2

10

Life-cycle effect

-1.00 -0.80 -0.60 -0.40 -0.20 0.00 0.20 0.40 0.60 0.80

20 23 26 29 32 35 38 41 44 47 50 53 56 59

We then turn to nested logit model. As shown in Table 4, we divide the decision to be two layers: the lower level for loan demand and the upper level, tenure choice. For the lower level, income level, head in public sector and educational level are positively associated with loan tendency of the owners while household size, income earner and head age are the reverse. For tenure choice, urbanization and two generation with income are negatively associated with ownership while extended family are positively associated with.

11

Table 4 Nested logit model of tenure choice and loan choice

Lower level: with loan = 1; without loan = 0 coeff p-value Household characteristics

household income (hhy_tot) 3.710 0.000

Household size (hhsize) -0.745 0.002

Income earners (tot_ernr) -0.635 0.000

Head sex (hd_sex) 0.336 0.264

Head age (hd_age) -0.172 0.000

Head in public sector (public) 0.726 0.039

Head education-junior high (hd_edjrhi) 0.779 0.038 Head education-senior high (hd_edhigh) 1.171 0.002

Head education-college ( hd_edcoll) 1.420 0.003

Head education-primary (hd_edelem)

Obs 10,831

Upper level: renter = 0; owner = 1 coeff p-value

Head characteristics

Spouse exist (sp_exist) 0.042 0.727

urban -1.683 0.000

suburban -1.116 0.000

rural

Extended family with child head (gen_2y) 1.241 0.000 Extended family with parent head (gen_2o) 0.384 0.083 one generation family

Two generation income with child head (inc_2y) -0.171 0.436 Two generation income with parent head (inc_2o) -0.374 0.023 one generation with income

Obs 11,952

12

Table 5 Saving rates of different types of household

saving rate sample size Renter

Total sample 0.162 1,121

One generation Income 0.143 927 Two gen. Income Young head 0.256 48 Two gen. income old head 0.254 145 Homeowner with loan

saving not includes mortgage

Total sample 0.102 2,855

One generation Income 0.080 2,315 Two gen. Income Young head 0.196 149 Two gen. income old head 0.196 388 Homeowner with loan

saving includes mortgage (forced saving)

Total sample 0.251 2,855

One generation Income 0.237 2,295 Two gen. Income Young head 0.310 147 Two gen. income old head 0.311 385 Homeowner without loan

Total sample 0.187 7,976

One generation Income 0.158 6,099 Two gen. Income Young head 0.272 650 Two gen. income old head 0.286 1,204

Following the analysis of tenure choice, next question is saving rate of different types of housing ownership. For renter and homeowner without loan, the saving rates are similar.

Homeowners with loan tend to save less. However, if we add the loan payment (defined as forced saving) back to saving, the saving rate will be the highest among three types of household.

Table 6 Houeshold saving function with the consideration of Heckman correction

Explainedvar: log Y - log C coeff p-value coeff p-value coeff p-value coeff p-value coeff p-value coeff p-value Household size -0.160 0.000** -0.244 0.000** -0.135 0.000** -0.171 0.000** -0.130 0.000** -0.177 0.000**

Household size squared 0.010 0.003** 0.017 0.000** 0.008 0.002** 0.010 0.000** 0.004 0.020** 0.010 0.000**

Permanent income 0.252 0.000** 0.460 0.000** 0.363 0.000** 0.309 0.000** 0.292 0.000** 0.334 0.000**

head sex 0.039 0.094* 0.017 0.607 -0.028 0.074* -0.003 0.905 0.039 0.000** 0.021 0.097*

head age -0.003 0.008** 0.002 0.247 0.002 0.049** 0.004 0.000** 0.001 0.221 0.002 0.000**

head edu junior high 0.045 0.096* -0.027 0.497 -0.025 0.151 0.043 0.159 0.018 0.130 0.030 0.032**

head edu high school 0.012 0.593 -0.001 0.984 -0.035 0.012** 0.002 0.929 -0.010 0.322 -0.010 0.395 head edu college -0.108 0.002** -0.115 0.023** -0.054 0.001** -0.033 0.274 -0.019 0.172 -0.053 0.001**

housing land space -0.003 0.000** 0.000 0.742 0.001 0.039** -0.001 0.123 0.000 0.589 0.000 0.423

housing indoor space -0.005 0.009** 0.001 0.684 0.000 0.370 0.001 0.234 0.000 0.239 0.000 0.160

one floor 0.187 0.001** 0.078 0.325 0.025 0.699 0.165 0.033** 0.147 0.000** 0.039 0.129

2-3 floor 0.161 0.000** 0.059 0.279 -0.029 0.244 0.020 0.506 0.066 0.000** -0.009 0.669

4-5 floor 0.088 0.002** 0.057 0.217 -0.047 0.021** 0.014 0.631 0.018 0.215 -0.048 0.029*

intercept -3.251 0.000 -5.592 0.000 -4.617 0.000 -3.455 0.000 -3.616 0.000 -3.771 0.000

Obs. No.

mills lambda 0.321 0.000 -0.008 0.862 0.058 0.051 -0.168 0.000 0.128 0.000 -0.113 0.000

2,137 718 4,977 3,349

Renter 1 gen Renter 2 gen. Owner w/ loan 1 gen Owner w/ loan 2 gen Owner w/o loan 1 gen Owner w/o loan 2 gen

865 256

Table 6 shows that household size is not favorable toward saving though the effects are decreasing along with size. The marginal propensity to save out of permanent income (MPS) is the highest for renters of two generation and then owners with loan of one generation. These two types of households tend to be under large pressure to save more either for home buying or loan payment. That renters of one generation tend to save the least may reflects the fact that the disadvantaged group has lower priority for housing purchase.

However, it is worth to note that the MPS of one generation renter is as high as 25%. Male headed household tend to save more than female headed one except owner with loan one generation household. It is suggested that female headed one generation household (may be single person family) tend to save more than the male headed counterpart. Higher educated heads consistently save less may be associated with consumption preference. Housing space is negatively associated with renter’s saving but positively associated with that of owner with loan.

Reference

Bourassa, S. (2000). Ethnicity Endogeneity, and Housing Tenure Choice. Journal of Real Estate Finance and Economics , pp. 323-341.

Bourassa, S. (1995). The Impacts of Borrowing Constraints on Home-Ownership in Australia. Urban Studies , pp. 1167-1173.

Chu-Chia Lin, a. C.-L.-J. (n.d.). Life Cycle, Mortgage Payment, and Forced Savings. INTERNATIONAL REAL ESTATE REVIEW , pp. 109-141.

Chu-Chia Lin, a. Y.-F. (2003). Housing Prices,Mortgage Payments and Savings Behavior in Taiwan: A Time Series Analysis. Asian Economic Journal , pp. 407-425.

Deaton A., a. C. (1998). Aging and Inquality in Income and Health. Amercian Economic Review , pp.

248-263.

Deaton A., a. C. (1995). Saving, Inquality and Aging：An East Asia Perspective. Asia Pacific Economic Review , pp. 7-19.

Fumo, H. (1986). Analysis of Household Saving: Past, Persent, and Future. Japanese Economic Review , pp.

21-33.

Hsueh, L.-M. (2000). The Relationship between Housing Price, Tenure Choice and Saving Behavior in Taiwan. INTERNATIONAL REAL ESTATE REVIEW , pp. 11-33.

Lin, C.C., 1993. The relationship between rents and prices of owner-occupied housing in Taiwan. J. Real Estate Finan. Econ. 6, 25-54.

Lin, C.C., Yung, C.S., Chang, C.O., 1996. Housing price index in Taipei. J. Housing Stud. 4, 1-30 (in Chinese).

Lin C.C., Lai, Y.F., 2003. Housing price, mortgage payments and savings behavior in Taiwan: a time series analysis. Asian Econ. J. 17, 407-425.

McFadden, D. (1974). Conditional Logit Analysis of Qualitative Choice Behavior. Frontier in Econometrics , pp. 105-142.

Phang, S.-Y. (2004). House prices and aggregrate consumption: do they move together? Evidence from Singapore. Journal of Housing Economics , pp. 101-119.

Tachibanaki, T. a. (1988). Household Saving, Life Insurance, and Public Pension. Bunken Journal , pp.

23-58.

Tachibanaki, T. (1994). Housing and Saving in Japan. Housing Markets in the United States and Japan , pp.

161-190.

Tse, C.Y., Leung, C.K.Y., 2002. Increasing wealth and increasing instability: the role of collateral. Rev. Inter.

Econ. 10, 45-52.