• 沒有找到結果。

Changing Studies on Households' Employment Structures and Commuting Decisions: The Evidence based on the 1990's and 2000's Data in Taipei Taiwan

N/A
N/A
Protected

Academic year: 2021

Share "Changing Studies on Households' Employment Structures and Commuting Decisions: The Evidence based on the 1990's and 2000's Data in Taipei Taiwan"

Copied!
21
0
0

加載中.... (立即查看全文)

全文

(1)

The 9th Asian Real Estate Society (AsRES)

International Conference

Changing Studies on Households’ Employment Structures and Commuting

Decisions

-- The Evidence based on the 1990’s and 2000’s Data in Taipei, Taiwan

Chen, Shu-Mei* & Chin-Oh Chang**

Abstract

From Year 1990 to 2000, the population characteristics and urban environments in Taipei changed a lot. There are two purposes in this paper. First, we analyze interaction relationships between income, housing consumption and commuting distances under the changing housing markets and traffic conditions. Second, we examine the hypotheses of relative resources and decision-making power, and the family responsibility and obligation, on commuting decisions over time. The empirical study employs the logistic regression model and uses the data from 1990’s and 2000’s Census of Population and Housing to examine 5 hypotheses. The empirical evidences suggest that the economic factors of housing consumptions and income resources play stronger roles than the household responsibility in commuting decision.

Keywords: Household Employment Structure, Commuting Decision, Housing Consumption

* Associate Professor, Department of Real Estate Management, Kun Shan University of Technology, Tainan County, Taiwan.

Address: 949, Ta-Wan Road, Yung- Kang City, Tainan County, Taiwan. E-mail: mayc2110@mail.ksut.edu.tw

Tel: +886-6-2050643 Fax: +886-6-2050578

**Professor, Department of Land Economics, National Chengchi University, Taipei, Taiwan. E-mail:jachang@nccu.edu.tw

Tel: +886-2-29387478 Fax:+886-2-29390251

(2)

1. Introduction

Household’s commuting pattern is significantly interacted with employment structure and housing consumption. The commuting distances represent the relative distances between residential locations and work locations. Household’s employment structure and income imply the commuting costs they are willing and able to pay. Theoretically, the longer acceptable commuting distances for workers usually offer an advantage of avoiding higher housing costs. The commuting costs will be the trade- off for the consumption of housing characteristics based on the classic urban economic model. This study will descript the changing environments of urban and housing markets between Year 1990 and 2000 first. We analysis the differences of households’ employment structures and characteristics, and explore the changing issues on households’ commuting decisions. The results attempt to explain the relation ships between the households’ income, desired housing consumptions and the preference of commuting decisions.

The urban environment and traffic conditions in Taipei City changed a lot between 1990 and 2000. Due to the excess supply of housing market, the vacancy rate increased 81.6% during the 10 years. Households will have more alternative choices of housing while they move. In the same time period, however, the growth rate of households’ disposable income is much more than the increasing rate of housing price. The housing costs will be a less heavy burden for households moving to buy a new house than 10 years ago. On the other hand, the rates for households owning a house increased from 78.5% in 1990 to 82.2% in 2000. In order to buy a house, subject to the expenditure, households might be willing to live closer to the suburban and have longer commutes.

The quantity of manufacture, business and service industries in Taipei City reduced during the same period. The distribution of employment opportunities extended from the CBD in Taipei City to outer ring of Taipei metropolitan. So households might have longer commutes from home to work places. And also in recent years, the MRT systems in Taipei metropolitan begin to work. The new public transport infrastructure might reduce the growing congestion on the streets within metropolitan and promote the convenience for the commuting employees living in suburb. In contrast to 1990, the traffic conditions have been improved and the commuting time also has been saved. The changing traffic conditions of exogenous factors make workers have longer commutes in 2000 than before.

(3)

Table 1 Basic Characteristics in Taipei City

1990 2000

Average Housing Price(NT$/ pin) 182,200 221,000 Average Household Disposable

Income(NT$/year)

600,996 1,237,777

Vacancy Rate(%) 13.3 17.6

Rate of Self- owning Housing (%) 78.5 82.2

No. of Factories 2,669 2,171

No. of Business Companies 144,453 144,136

Indices of Transportation Price 100.00 124.87

Source: 1. The average housing price comes from the Housing Information Quarterly, 2000, Taiwan Real Estate Research Center, National Chengchi University.

2. The others come from the Statistical Abstract of Taipei Municipality, and the Statistical Abstract of Taipei County, 1991 and 2001.

From 1990 to 2000, the socio-economic circumstances in Taiwan changed also. The proportion for the female with higher educational attainment increased. The employment rate and the ratio for female being the primary earner of household both increased also. This study argues that if women’s economic characteristics made greater progress, the female in households would get more decision-making power on commuting decisions. More and more double- income households formed, should women take the responsibility of taking care of family and children like before and commute to work with shorter distances. This study would shed lights on comparing the commuting decisions made by single and double income households between 1990 and 2000.

There are two purposes in this paper. First of all, we would analysis interacted relationships between income, housing consumption and commuting distances under the changing housing markets and traffic conditions. Second, since women earned more income in labor market and even become the primary earner in households during recent years, we would examine the hypotheses of relative resources and decision-making power and the family responsibility and obligation. This study will analysis the changing effects of women’s economic progress on commuting decisions over time.

Since the geographical distribution of employment opportunities are decentralized, the de-concentration of employment challenged the standard urban economic model. The classical consumption equilibrium theory from Alonso(1964) concluded that households will live at the location where marginal decrease of housing cost would be equal to marginal increase of commuting cost subject to income constrains. The simplified assumption in the spatial equilibrium model should be relaxed when the model applied to explain the trade- off relationship in Taipei City. Chen and Chang (2000) analyzed the residential location choice

(4)

and commuting decisions in Taipei City and found that the trade-off relationships between housing cost and commuting cost for the households with moving behavior in 1990, although the assumptions of mono-centric urban and single- worker household type are not consistent with those in Taipei city. This paper will further analysis the commuting decisions of households in 1990 and 2000, and focus on the changing issues because the urban

environment and employment location have been changed over time.

Based on the previous studies, households’ moving decision, housing and commuting decisions are often jointly decided. Singell and Lillydahl (1986) developed a simultaneous model of wage, housing price and commuting cost. Lower earnings and traditional

responsibility of taking care of children made the female have shorter commuting distances in rational economic behavior. Assadian and Ondrich (1993) developed residential location studies from the standard mono-centric models and introduced the second earner into spatial equilibrium model1. They found that housing consumption and location as well as labor supply decision are simultaneously determinate. Baccaïni (1997) focused on the issues of residential and commuting strategies, and concluded that the demographic characteristics, professional skills, life cycle types, housing tenure, housing types are related to the

residential location and commuting distances. Ronwendal and Meijer (2001) reports stated preferences of Dutch workers for combinations of housing, employment, and commuting. The study assumed people dislike commuting, and they adjust their housing and employment situations by migration. Households would find a satisfied combination of housing and employment in a limited urban area. The results suggest that the value of commuting in the model is high in comparison to the wage rate. Some preferences for housing attributes are strong enough for households to make longer commuting. It reveals the trade-off between desired housing characteristics and longer commutes.

This study assumes that workers prefer shorter commutes. The commuting decisions represent relative relationships between residential location and work location. The workers make commuting decisions would take the desired housing preference into consideration. So we could expect that the trade- off between commuting costs and the housing consumption is existed. Workers will accept longer commutes if they want to consume more housing

attributes. And if workers accept longer home-to-work distances, they had better have higher income so that they can pay higher commuting costs. The trade-off between income and commuting cost over time would be studied in this paper also.

Some studies introduced the second earner of household into spatial equilibrium

(5)

model. They found that there are gender differences for husband and wife of double income households to make commuting decisions. Bielby and Bielby(1992) explained the gender differences of the willing for couples to improve a better job opportunity from maximizing the family well-being of a neoclassical model and the relative resources and couple’s decision-making based on social exchange theory. They got conclusions that women are more likely than men to be “tied mover” or “tied stayer” while relocating job or residential location because of her lower earning potential. Another belief of gender- role ideology more often assumed that the husband is the primary earner of a household. Employed women can’t rely on her earning potential as a resource in job relocation and migration decisions. Camstra (1996) posed three hierarchical hypotheses and discussed gender differences of residential moving, remaining job and the sensitivity of commuting distance. They got the similar conclusion that women are more likely to quit their job in double income households if the commuting distance was increased by residential move. The empirical results suggested that women are more sensitive while increasing commuting distance than men. The women in double income households are dual labors for doing their jobs and taking care of family and children. Their home-to-work distances would be shorter than men (White, 1977; Madden, 1981; Singell and Lillydahl, 1986; Assadian and Ondrich, 1993; Freedman and Kern, 1996; Howell and Bronson, 1996).

In the previous studies, we found that employed women have shorter commutes than men because of their lower income and the responsibility of family obligation. It is

reasonable for working women with under school- age children in double income households to make shorter commutation. However, this paper will re- examine the hypothesis of the responsibility of family obligation under the situation that women have better educational attainments and more income resources in 2000, and see if the results could challenge the hypothesis and suggest different conclusions from 1990. In addition, we want to explore ifr the woman has a shorter commuting distance when she make a progress on economic status in double households based on the hypothesis of relative resources and the decision- making power. These commuting issues are worth to be observed further over time.

This paper is organized as follows. The second session presents the empirical

framework and methodology of this study. In session 3, the data is explained. The estimated results are presented and discussed in session 4. Session 5 concludes the paper.

2. Empirical Framework and Methodology

(6)

location and the work location of workers. The home- to- work distance changes when households move to a new house. Households prefer moving to the residential location near the work location for more convenient commutation. Or households prefer to move to a house with more housing services for satisfying their housing demands, but they will face the trade-off between housing consumptions and the commuting costs under the budget constraints. Households pursue the minimized transaction costs of moving, so they will take the housing cost, commuting cost, and wage into considerations and the commuting decision is generated at the same time.

Households consume the desired combined housing commodities, including the characteristics of housing floor area, housing type, tenure and residential location

environment, and pay the relative housing costs. Under the budget constraints, households consume some housing characteristics and the other non-housing commodities including commuting costs. The issues of the trade-off between the housing consumption and the commuting cost are worth to be discussed in details. The worker prefers to find a job with higher wage and shorter commuting distance. From the findings of previous studies, we know the fact that higher- income workers tend to keep the professional job and accept longer commutation. This study describes the environments of residential location, housing supply, traffic conditions, and household income over different time periods and takes them as given. The hypotheses of the trade-off between housing consumption price and the commuting cost, the trade-off between income and commuting cost will be examined. In addition, we can observe the households’ commuting decisions by different employment structures. The dual income households have to consider two work locations comparing to the single- income households and make different commuting decisions with specific behavior aspects. And this paper will further reveal the commuting issues of female workers when they are the primary earners of households. So we divide single- income households into two sub-groups .They are “single- income households with of male” earner and “single- income households with female earner”. The early studies suggest that

household member’s decision- making power is measured by his (her) financial resources. This paper will examine the hypothesis of relative resources and the decision- making power. Husband (wife) with more financial resources and even being the primary earner of household contributes the family income and should bargain to have a shorter commutation. And other studies have focused on the female role in the family labor division and her shorter commutation. This paper will re-examine hypothesis of household’s responsibility of taking care of children, and analysis the changing effects over time after women make

(7)

greater progress on economic status.

This paper will examine 5 hypotheses.

H1: Households live in more expensive residential locations near CBD have shorter commuting distances, and this trend will fade away over time because the employment opportunities in Taipei City decreased. Although the trade-off between housing consumption and commuting costs has been highlighted before, the trade-off relationships have been changed under different urban economic backgrounds.

H2: Households would accept longer home-to- work distances in order to own a house. Tenants pay less housing costs than owners, re-house more frequently, and usually have shorter commuting distances. From 1990 to 2000, the increasing rate of household income is larger than the changed rate of housing price /per unit, the trade- off relationship between housing tenure and commuting cost are expected to be revealed.

H3: The workers with higher income can afford higher commuting costs, and are more likely to accept longer commuting distances. However, the higher marginal income a worker earned, the shorter marginal commuting distance he (she) wants to choose. The worker with higher income still prefers a shorter commutation.

H4: The main provider of a household is more likely to choose a shorter commuting distance than the second earner because of his (her) financial contribution. The economic contribution to family income would help to have stronger bargaining power on commuting decisions.

H5: The household’s responsibility of taking care of children makes woman’s commuting distance shorter. This paper expects the family responsibility is not necessarily stressed on women in 2000, because the gender role ideology is less strong and the social supporting resources to take care of children are much more sufficient than those in 1990.

This paper employs ordinal logit model, which is suitable for the estimation of commuting distance choice. The model is

log[p(y≦j∣X)/1- p(y≦j∣X)]= αj + β2X2 +β3X3 +……+βkXk (1)

where, y is a ranked dependent variable which is generated from sorting each level of commuting distance. If husband (wife) works at home, “y” is set to 0. If husband (wife) commutes to the same location within the family live, “y” is set to 1. If husband (wife) commutes to the adjacent location from where the family live, “y” is set to 2. If husband (wife) commutes to the other location of Taipei City, “y” is set to 3. If husband (wife) commutes to the other cities in Taiwan, “y” is set to 4. Xi is the exogenous explanatory variables. β is the estimated coefficient by the maximum- likelihood function. We use

(8)

logistic procedure of SAS program to estimate the results, including the model fitting information2, coefficient, and the odds ratio3. The positive sign of coefficientβshows a increasing probability to choose shorter commuting distance.

In order to examine the above hypotheses, we divided Taipei’s nuclear households into four groups by employment structures: “double- income households”, “single- income households Ⅰ”with male primary earners, “single- income households Ⅱ”with female primary earners, and “no income households”. This paper chooses the following variables to be explanatory variables of commuting decision models and compares the estimated results of each household group except for no income households.

(1).residential location near CBD:

The residential location where households live, implying the relative housing costs they pay for, is the critical variable in commuting distance choice. Households living in residential locations near CBD have to pay more expensive housing costs and theoretically pay less commuting costs under the implication of standard urban

economic model. This variable can be used to test the trade-off hypothesis between commuting cost and housing cost. If households live near CBD location but commute longer, it implies the employment locations have been decentralized. This variable can be used to test “Hypothesis 1”. If households live in residential locations of Chung-shan, Ta-an, Chung-chen, Sung-Chung-shan, Hsin-i, Ta-tung and Wan-hwa, the dummy variable is set to 1; otherwise, set to 0.

(2).floor area of housing:

Based on the trade- off model of accessibility and space, households pay more commuting costs as a compensation for moving to a suburban house with larger floor areas. The floor area of housing is a continuous variable and used to test the trade- off relationship between housing consumption and commuting costs in commuting decision model.

(3).tenant occupancy:

The choice of renting a house means paying less housing cost than owning a house and often implies a shorter commutation between home and workplace. In contrast, households become owner-occupiers often moving to urban periphery or

2 A test for null hypothesis that the coefficient of each explanatory variable is zero performed by the log

likelihood statistics –2 log L. (-2log L)= -2 Log (Lω/LΩ),Lωis the maximum when maximizing respect to

constant only; LΩis the maximum when maximizing respect to all parameters. It is the same criteria for

assessing model fit as the likelihood ratio ,ρ2= 1-( Log L

Ω/ Log Lω).

3 The odds ratio is the marginal effect of explanatory changes one unit induce the probability changes, which is

(9)

suburban and commute longer. This variable also can be used to test the trade- off between the consumption of housing tenure attribute and commuting cost. If the households rent a house, this dummy variable is set to 1; otherwise, it is set to 0. (4).logarithm of monthly income:

The workers with higher income generally have more affordability to pay high commuting costs. The propensity for working at a great distance from one’s place of residence rises from the least educated to the highly qualified, as well as the level of household income (Baccaïni, 1997). Or workers with higher income have less willing to quit their jobs after changing residential place and commute longer. The estimate results of this variable can examine the hypothesis of the trade-off between income and commuting costs and explain that the partial slope of probability in commuting

decisions changed while each percentage increase in monthly income. (5).with school-age children:

The commuting decision strongly depends on family life- cycle type.

According to the hypothesis of household responsibility, working women have shorter commuting distance if they have to take care of children. This paper further focuses on the impact of the responsibility of taking care of the school-age or younger children. If households with school-age or younger children, this dummy variable is set to 1; otherwise, it is set to 0.

(6).the role of primary earner:

According to the hypothesis of relative resources and decision-making power, the primary earners of households have domain contribution on family financial resources and are expected to get a stronger bargaining power, so they can commute for a shorter distance. The ratios for the female being the primary earners of

households are larger in 2000 than those in 1990. This paper will compare the estimated coefficients of the variable in the double- income household models over time.

(7).Interaction term- the role of main provider by family with school-age children:

If we concern about both hypotheses of relative resources and decision-making power and household responsibility, the interaction of workers being the main provider and double income households with school-age children are expected to make workers have shorter commuting journey. If the worker is the main provider and there is school-age or younger children in family, this interaction term is set to one; otherwise, it is set to 0.

(10)

3. Data

We use the data from “Census of Population and Housing in Taipei, Taiwan” in 1990 and 2000 for empirical statistics. We take the records of self- owned and rental residence only. The houses bought from government housing projects and inherited housing, or vacancy houses are eliminated. When households move to new residential places, they have to consider the minimum costs of housing consumption and commuting. This paper chooses the household samples with moving behavior during recent 5 years to analysis their

commuting decisions after re-housing. This will help to test the trade-off effect between housing cost and commuting cost. And we only choose the households of nuclear family type so that we can observe how the couple’s employment structure and school-age children influence the choice of commuting.

3.1 Some households characteristics among different employment structures in 1990 and 2000

The comparison of individual characteristics among different employment structures in 1990 and 2000 is showed in Table 2. The results suggest that the average floor area of dwellings after re-housing in 2000 is smaller than that in 1990. The ratio of tenant housing in 2000 is larger than that in 1990, especially for the single- income households. Comparing to the household characteristics in 1990, we also find that the ages of couple are both older, the ratios for households with school-age or younger children are larger in 2000. It is possible that people getting married late and having young kids at old age. In addition, the proportion for couple with high educational attainment is higher in 2000, especially for the wives in “double- income households” and the husbands in “singe- income households Ⅰ”. Due to the operation of new public MRT system and the decentralization of employment opportunities, the ratios for husbands commuting to farer work places increased in 2000. However, the ratios for wives with shorter commutation are significantly higher than before. We can find that there are obvious gender differences of commuting distances over time. Generally, the average income of husbands is more than wives, the income for both husband and wife increased over time.

Table 2 The means of households characteristics among different employment structures in 1990 and 2000

(11)

characteristics Double-income HH Single-income HH 1 Single-income HH 2

1990 2000 1990 2000 1990 2000

floor area of housing (pin) 34.2715 29.7156 33.6172 31.8074 32.1587 28.8112 tenant occupancy 0.0785 0.2270 0.0759 0.3253 0.1303 0.3375

Age of husband 38.6239 40.6785 41.2637 41.6686 --

--Husband with high school

education 0.2242 0.2376 0.2895 0.2577 --

--Husband with college

education 0.6421 0.6437 0.3849 0.5649 --

--Log(income of husband) 10.3647 10.8258 10.2595 10.7947 --

--Husband work at home 0.0618 0.0535 0.0864 0.0461 --

--Husband commutes to the

same location 0.4312 0.4509 0.4793 0.4742 --

--Husband commutes to the

adjacent location 0.1905 0.1909 0.1675 0.1823 --

--Husband commutes to the

other location 0.1884 0.1563 0.1487 0.1455 --

--Husband commutes to the

other cities 0.1279 0.1484 0.1179 0.1520 --

--Age of wife 35.3996 37.9139 -- -- 41.9959 43.7886

Wife with high school

education 0.3239 0.3010 -- -- 0.2759 0.3013

Wife with college

education 0.5095 0.5710 -- -- 0.3124 0.3691

Log(income of wife) 9.9976 10.5088 -- -- 9.9155 10.4671

Wife work at home 0.0640 0.0629 -- -- 0.0548 0.0457

Wife commutes to the

same location 0.4759 0.5152 -- -- 0.5203 0.5505

Wife commutes to the

adjacent location 0.2047 0.2104 -- -- 0.1917 0.2256

Wife commutes to the

other location 0.1803 0.1445 -- -- 0.1476 0.1309

Wife commutes to the

other cities 0.0746 0.0670 -- -- 0.0796 0.0473

Household with school-age

children 0.4571 0.5014 0.5090 0.4944 0.3245 0.3975

Numbers of observations 30,328 21,516 34,492 11,114 1,972 634 3.2 The estimation of permanent income

It lacks of income information in the census data. This study uses the sampling data of “Manpower Utilization Survey” to predict the individual permanent income. The data items in this data set are similar to those in the “Census of Population and Housing”, so it is suitable for estimation.

Because we want to analysis the commuting decisions of workers by different employment structures, it is more appropriate to estimate the individual income instead of

(12)

household income. This paper used the relative human capital variables of sex, age, the square term of age, educational level, and marriage status to predict the permanent income of individual worker. The male, older, highly educated, and married persons are expected to have more income resources. We don’t consider the geographical differences for income estimation, since the study scope contains Taipei City only.

We transfer the income into the logarithm form and then estimate the coefficients of the above human capital variables in the income regression models. The predicted income is derived from multiplying the estimated coefficients in income regression models by

individual characteristics in the census data. The predicted income variable will be one of the explanatory variables in commuting decision models. The odds ratios of the logarithm of income in commuting decision models can be explained the percentage change in the odds of choosing shorter commuting distance while one percent increasing in income.

Table 3 The income regression models of workers living in Taipei City Variables

Model 1990 Model 2000

Coefficient t value Coefficient t value

intercept 8.7004** 92.79 8.9387** 61.23

Female -0.3041** -16.7 -0.2555** -10.04

Age 0.0546** 11.47 0.0556** 7.58

Age2 -0.0006** -11.07 -0.0005** -6.09

With high school education 0.1767** 8.11 0.2073** 5.6 With college education 0.4450** 21.51 0.5192** 15.1

Married status 0.1707** 7.81 0.1303** 4.02 2 R 0.2971 0.1873 F value 210.26 112.4 Numbers of observations 2970 2901

** indicates significant at the 1 percent level; * indicates significant at the 5 percent level.

The estimated results of permanent income are shown in Table 3. As a whole, the adjusted R2 ofincome regression model is close to 0.3. Every independent variable of the model is significant. The results reveal that female workers have lower income than the male. An increase in age has a declining increase impact on income. In Model 1990, the coefficient of worker’s age is 0.0546; however, the coefficient of age2 is -0.006. The partial slope for the change in age is composed of the above two terms. The workers with medium or high educational attainment have higher income. Especially the coefficient for the workers in highly educated category is much higher. The married status has a positive

(13)

impact on the income. The estimated results are consistent with our expectations. 4. Empirical Results

Table 4-1 and Table 4-2 show the estimated results of commuting decision models in 1990 and 2000. The statistics support the proportional odds assumption, that is, the partial effect of a dependent variable is invariant to the choice of any commuting distance category. And the interpretation of ordinal logit model is similar to those from binary choice. The value of “–2 log L” suggests the joint significance of all the explanatory variables in this model. The odds ratio4 will be computed to measure the specific effect of an explanatory variable on commuting decision under other things being equal. The results are shown in Table 5.

The empirical evidences show that the commuting decisions among different household employment structures mainly depends on the housing characteristics and income in 1990 and 2000. The estimated coefficient for the variable “residential location near CBD” is significantly positive. It indicates that there is higher probability for those households living in more expensive locations to choose shorter commuting distances. It is consistent with the trade-off between housing consumption and commuting cost. The hypothesis 1 is supported. Comparing the odds ratios for this variable in Table 5, the odds ratios are larger in 1990 than those in 2000 except for the “single-income households 2”. It means that the workers living near CBD have shorter commutation in 1990 than those in 2000 and implies that the

employment opportunities are getting decentralized in 2000. In addition, the odds for the workers in double-income households living near CBD to choose shorter commuting distances are about 1.75 times as high as those living in suburban locations. Such marginal effect on the odds for the “single- income households 1” (with male primary earner) is 1.6 times; and for the “single- income households 2” (with female primary earner) is 1.5 times. Other things being equal, the preference for double- income households living near CBD to choose shorter commutation is stronger than those for the single- income households. The above evidence is different from the previous studies which suggested the double- income households commute longer.

The coefficient estimates for the variable “floor area of housing” is significantly negative. It shows that the workers moving to a larger house have lower probabilities to choose shorter commuting distances. We find the trade-off evidence between the housing

4 The odds ratio is the marginal effect of explanatory changes one unit induce the probability changes, which is

(14)

consumption and commuting cost again. However, such effects for the “single-income households 2” are not significant.

The coefficients estimates for the variable “tenant occupancy” are significantly negative in 1990; but the coefficients are positive in 2000. The workers renting a house in 1990 are more likely to choose longer commuting distances. It implies the tenants have weaker status in housing consumption and commuting. However, the workers buying and owning a house in 2000 have higher probabilities to choose longer commutation than tenants. High

commuting costs are the compensation for the households to own houses. The hypothesis 2 is supported in 2000.

The variable “Log (income)” has a significantly negative impact on the commuting decisions of workers. Workers with higher income have more affordability to pay the commuting costs and are less likely to choose shorter commuting distances. The workers of Taipei City in 2000 have higher income than those in1990, and the marginal effects of “Log (income)” in 2000 are getting smaller. It implies that workers with higher income are not willing to accept higher commuting costs as they did in 1990. The hypothesis 3 is supported. For the example of workers of double- income households in 1990, 1 % increases in

husband’s income decreases the odds of choosing shorter commutation by 90%; however, such effects in 2000 decreases the odds by 74%. And for the wives of double- income households, the effects of decreasing the odds to choose shorter commutation are smaller than those for husbands while each percentage increasing in income. There could be other non-financial factors influencing the commuting decisions of wives.

According to the hypothesis of relative resources and decision- making power, the role of primary earner of households is expected to have stronger impacts on commuting

decisions. The contributions on households’ financial resources would help the workers bargain to have shorter commuting journey. The estimated coefficient of the variable “primary earner” is significantly positive for wives to choose shorter commuting distances; the coefficient is significantly negative for husbands in 1990. There are gender differences for the impacts generated from the role of “primary earner”. However, such effects in 2000 for both husband and wife in double- income households are not significant. The hypothesis 4 is not supported. Although the proportion for wife with the role of “primary earner” of double- income household in 2000 is higher than before, such important role obviously doesn’t make wives choose shorter commuting distances.

(15)

variables Double-income HH

Single-income HH 1 

Single-income HH 2 

Husband Wife Husband Wife

Intercept 1 21.4411 ** 16.9985** 19.6649 ** 17.2099 **

Intercept 2 24.2822 ** 19.9714** 22.4165 ** 20.4809 **

Intercept 3 25.1375 ** 20.9312** 23.2109 ** 21.4200 **

Intercept 4 26.3213 ** 22.4061** 24.2436 ** 22.6707 **

Residential location near CBD 0.6550 ** 0.6130 ** 0.4903 ** 0.2897 ** Floor area of housing 0.0026 ** -0.0014 -0.0048 ** -0.0053 Tenant occupancy -0.3171 ** -0.3506** -0.3645 ** -0.6595 **

Log(income) -2.3862 ** -2.0232** -2.1732 ** -2.0288 **

Household with school-age children 0.2224 ** 0.3869 ** 0.2044 ** 0.2406 **

Primary earner -0.1263 * 0.1875 ** --

--Interaction- Primary earner* with

school-age children 0.0969   -0.1652*  --   --  

Concordant(%) 60.4 60.1 61.0 59.8

-2LogL 84211.468 80247.510 94042.000 4990.763

Numbers of observations 30,320 30,314 34,484 1,960

** indicates significant at the 1 percent level; * indicates significant at the 5 percent level.

The hypothesis of family responsibility suggests that the obligation for wives taking care of younger children makes her choose shorter commuting distances. The estimated coefficients of the variable “Household with school-age children” for the 1990 sample are all significant. The odds of “Household with school-age children” for wives to choose shorter commutes are about 1.27-1.47 times as high as those without school-age children. And such effects are stronger than those for husbands (1.23-1.25 tomes). The hypothesis 5 is supported by the data in 1990. But the variable is only significant for the wives in double- income households in 2000. Such effects are also weaker than those in 1990. The family ties for the wives to take care of school-age children are getting weaker since the social conditions have been changed in 2000.

The interactions of “primary earner and household with school-age children” are almost not significant except that it has significantly negative coefficient for the wives in “double-income households” in 1990. We expected that wives may choose shorter commutation if they are the primary earners and need to take care of school-age or younger children of households. But the interacted effect doesn’t reveal in both 1990 and 2000.

Table 4-2 Determinants of commuting decisions in 2000

variables Double-income HH

Single-income HH 1 

Single-income HH

(16)

Husband Wife Husband Wife Intercept 1 11.5238 ** 7.7824 ** 12.3996 ** 4.9391 Intercept 2 14.5480 ** 10.9253 ** 15.6168 ** 8.4279* Intercept 3 15.4042 ** 11.9708 ** 16.4362 ** 9.5951* Intercept 4 16.3501 ** 13.3099 ** 17.3177 ** 11.0829** Residential location near CBD 0.5644 ** 0.5955 ** 0.4794 ** 0.4372**

Floor area of housing -0.0039 ** -0.0045 ** -0.0042 ** 0.0079 Tenant occupancy 0.4822 ** 0.4633 ** 0.2488 ** 0.3317* Log(income) -1.3680 ** -1.0418 ** -1.4504 ** -0.8186* Household with

school-age children 0.0876 0.2970 ** -0.0049 0.0288

Primary earner -0.0241 -0.0056 --

--Interaction- Primary earner* with

school-age children 0.1102 -0.0369 --

--Concordant(%) 59.2 59.5 58.5 55.9

-2LogL 59063.06 54889.43 29918.97 1523.831

Numbers of

observations 21,516 21,516 11,114   634

** indicates significant at the 1 percent level; * indicates significant at the 5 percent level.

Table 5 propensity to commute a shorter distance by household employment structure

variables

Odds ratio by household employment structure Double-income HH (husband) Double-income HH (wife) single-income HH 1 single-income HH 2 1990 2000 1990 2000 1990 2000 1990 2000

Residential location near CBD 1.925 1.758 1.846 1.814 1.633 1.615 1.336 1.548

Floor area of housing 1.003 0.996 ns 0.995 0.995 0.996 ns ns

Tenant occupancy 0.728 1.620 0.704 1.589 0.695 1.282 0.517 1.393

(17)

Household with school-age

children 1.249 ns 1.472 1.346 1.227 ns 1.272 ns

Primary earner 0.881 ns 1.206 ns -- -- --

--Interaction- Primary earner*

with school-age children ns ns 0.848 ns -- -- --

--5. Conclusion

Comparing the backgrounds over time, the participation rates of female labor force have been increased, and the income resources of female workers have grown up also, and even the ratio for the female being the role of primary earner has been getting higher. The relative issues of employment structure and commuting from home to work are worth to shed some lights on. On the other hand, the geographical distribution of employment opportunities has decentralized from the CBD of Taipei City and the new public MRT systems offer more convenient commuting services in 2000. So that workers might have longer commuting distances than before. This paper analyzed the changing of commuting decisions by different workers of employment structures.

This paper selects the data of workers in nuclear families with re-housing behavior during recent 5 years. Such workers move to new residence places and re-adjust their home- to- work distances. The description statistics show that the percentages for husbands and wives to choose shorter commuting distances in 2000 are lower than those in 1990. It might imply the public transportation services improved and the decentralized trend for distribution of employment opportunities occurred. However, the proportions of commuting shorter distances (commuting to the same location within residential location) for wives in double-income households are larger than those for husbands. We can find some gender differences for the commuting behaviors of workers.

From the empirical results, we can find the evidence of the trade-off between housing consumptions and commuting costs in1990. At that time, the tenants stayed in weaker situation because they had longer commuting distances than the owners. The trade-off effects still exist in 2000; however, the effect for workers living near CBD to choose shorter commutation is getting less strong. The workers in double-income households move to residential location near CBD for more convenient urban services but commute to the work places outside the CBD area and pay more commuting costs. On the other hand, workers earned more income than before, the odds for workers with higher income to choose shorter commuting distances is getting larger in 2000. Further more, the income of female workers increased, so that wife’s income has a stronger impact on commuting decision than the husband’s. We can conclude that the female is more sensitive to the longer commutation

(18)

than men. As the social environment changes, the variable “household with school-age children” has weaker impacts on the commuting decisions for female workers. Taking care of school-age children is still utilized for the commuting decisions of wives in double-income households, but not necessarily for those of female primary earners in single-double-income households. For the female primary earners in single-income households, the income

variable is more important than the variable of taking care of school-age children when making commuting decisions. The empirical evidences show the changes of commuting decisions over time.

This paper draws attention to discuss the effects of economic trade-off relationships, income resources and household responsibility on commuting decisions. Through comparing the estimated results by different groups of employment structures in 1990 and 2000, we find the economic factors of housing consumptions and income resources are significant;

however, the hypothesis of labor division for the female bearing household responsibility is not significantly supported. Economic factors play the stronger role on commuting decisions than the household responsibility. We hope further policy implication will be developed.

References Alonso, William

1964 Location and Land Use: Toward a General Theory of Land Rent. Cambridge, Massachusetts: Harvard University Press.

Assadian, Afsaneh and Jan Ondrich

1993 “Residential Location, Housing Demand and Labour Supply Decisions of One-and Two- Earner Household: The Case of Bogota, Colombia.” Urban Studies 30(1): 73-86.

(19)

Baber, Kristine M. and Katherine R. Allen

1992 Women and Family:Feminist Reconstructions. the Guilford Press. Baccaïni, B

1997 “Commuting and Residential Strategies in the Île-de-France: Individual Behaviour and Spatial Constraints” Environment and Planning A 29: 1801-1829.

Bielby, William T. and Denise D. Bielby

1992 “I Will Follow Him-Family Ties, Gender- Role Beliefs, and Reluctance to Relocate for a Better Job.” American Journal of Sociology 97(5): 1241-67. Blumen, Orna and Kellerman, Aharon

1990 “Gender Differences in Commuting Distance, Residence, and Employment Location: Metropolitan Haifa 1972 and 1983.” Professional Geographer 42(1): 54-71. Camstra, Ronald

1996 “Commuting and Gender in a Lifestyle Perspective.” Urban Studies 33(2): 283-300.

Chang, Chin-Oh, Shu-Mei Chen and Shiawee X. Yang

1998 “Aggregated Needs and the Location Choice of Households in Taipei.” Journal of the Asian Real Estate Society 1 (1): 81-100.

Chang, Chin-Oh and Shu-Mei Chen

1999 “Households Life Cycle and Housing Demand Decision Adjustment in Taipei, Taiwan.” AREUEA/ Asian Real Estate Society International Conference, Hawaii, Maui.

Chen, Shu-Mei and Chin-Oh Chang

2000 “The housing location choice and commuting decision- the comparison of double- income households and single- income households, Taipei,1990.” Journal of Taiwan Society 24: 89-125.(in Chinese)

Freedman, Ora and Clifford R. Kern

1996 “A Model of Workplace and Residence Choice in Two Worker Households.” Regional Science and Urban Economics 27: 241-60.

Hanson, Susan and Geraldine Pratt

1995 Gender, Work, and Space: Routledge Press. Heenan, Deirdre and Anne M. Grey

1997 “Women, Public Housing and Inequality : A Northern Ireland Perspective.” Housing Studies 12(2): 157-171.

(20)

Howell, Frank M. and Deborah Richey Bronson

1996 “The Journey to Work and Gender Inequality in Earnings: a Cross- Validation Study for the United States.” The Sociological Quarterly 37(3): 429-47.

Kristensen, Gustav

1997 “Women’s Economic Progress and the Demand for Housing: Theory, and Empirical Analyses Based on Danish Data.” Urban Studies 34(3): 403-418. Liao, Tim F.

1994 “Interpreting Probability Models: Logit, Probit, and Other Generalized Linear Models.” Pp.07-101 in Sage University Paper Series on Quantitative Applications in the Social Sciences. Thousand Oaks, CA: Sage.

Maddala, G. S.

1983 Limited-Dependent and Qualitative Variables in Econometrics. Cambridge University Press.

Madden, J. F.

1981 “Why Women Work Close to Home.” Urban Studies 18: 191-94. Munro, Moira, and Susan J. Smith

1989 “Gender and Housing:Broading the Debate.” Housing Studies 4(1): 3-17. Rouwendal, Jan and Erik Meijer

2001 “Preferences for Housing, Jobs, and Commuting: A Mixed Logit Analysis” Journal of Regional Science 41(3):475-505.

Singell, Larry D. and Jane H. Lillydahl

1986 “An Empirical Analysis of the Commute to Work Patterns of Males and Females in Two Earner Households.” Urban Studies 23(2): 119-29.

Smith, Susan J.

1990 “Income, Housing Wealth and Gender Inequality.” Urban Studies 27(1): 67-88. Spain, Daphne

1990 “The Effect of Residential Mobility and Household Composition on Housing Quality.” Urban Affairs Quarterly 25(4): 659-683.

White, Michelle J.

1977 “A Model of Residential Location Choice and Commuting by Men and Women Workers.” Journal of Regional Science 17(1): 41-52.

數據

Table 1 Basic Characteristics in Taipei City
Table 3 The income regression models of workers living in Taipei City Variables
Table 4-2 Determinants of commuting decisions in 2000
Table 5 propensity to commute a shorter distance by household employment structure

參考文獻

相關文件

Good Data Structure Needs Proper Accessing Algorithms: get, insert. rule of thumb for speed: often-get

Employment of Foreign Workers, overseas Chinese students shall meet the student status regulated in the Regulations on Study and Counseling Assistance in Taiwan for

In order to apply for a permit to employ Class B Foreign Worker(s), an Employer shall provide reasonable employment terms and register for such employment demands with local

Should an employer find it necessary to continue the employment of the Class A Foreign Worker(s), the employer shall, within four (4) months prior to the expiration of the

Wang, Solving pseudomonotone variational inequalities and pseudocon- vex optimization problems using the projection neural network, IEEE Transactions on Neural Networks 17

Define instead the imaginary.. potential, magnetic field, lattice…) Dirac-BdG Hamiltonian:. with small, and matrix

Monopolies in synchronous distributed systems (Peleg 1998; Peleg

Corollary 13.3. For, if C is simple and lies in D, the function f is analytic at each point interior to and on C; so we apply the Cauchy-Goursat theorem directly. On the other hand,