The empirical findings are reported in Tables 2 and 3. Table 2 shows the effect of the
distance to MRT station on housing prices within 1.5 km. Model (1) indicates that distance
to MRT stations, the age of the house, size of the house and Taipei are statistically
significant. More importantly, distance has a negative effect on housing prices. The
estimates show that when the distance to MRT station increases by 1 km, the housing
prices decrease by $NT120,300 per ping.
Model (2) demonstrates that when the distance to MRT station decreases by 1 km, the
housing prices in Taipei is $NT81,540 per ping higher than Kaohsiung. It means that the
MRT system in Taipei has a greater impact on housing prices, compare with Kaohsiung.
Model (3) shows the relationship of the distance to MRT stations on housing prices
between Taipei and Kaohsiung. According to the coefficient of distance2, we can
conclude that the relationship between the housing prices and distance is a U shape.
Model (4) is our main model. The estimates indicate that when the distance increases
by 1 km, the housing prices decrease by $NT158,432 per ping in Taipei, and decrease by
$NT97,732 per ping in Kaohsiung. The relationship between housing prices and distance
is U shape as well.
13
Table 2. The Effect of Distance on Housing Prices: Distance to MRT station ≦ 1.5 km)
Independent Variable
Model (1) Model (2) Model (3)
Model (4) Model (5)distance -12.030(4.946)*** -11.496(8.530)*** -35.482(18.845)** -29.065(12.987)* -26.428 (17.022)
distance2 12.485 (12.235)* 10.887(12.558)* 11.273 (12.273)
distance×Taipei -8.154(10.433)** -6.070(10.707)* -15.794(12.656)
distance2 ×Taipei 13.992 (16.566)
age_house -.411(.148)*** -.392(.124)*** -.390(.124)*** -.386(.124)** -.385(.124)***
age_house _missing 6.943(5.205)*** 5.537(4.295)** 6.002(4.320)* 5.950(4.322)* 5.960(4.322)*
size_house -.001(.002)* -.001(.002)* -.001(.002)* -.001(.002)* -.001(.002)*
Taipei 5.474(4.421)*** 12.692(8.425)** 6.243(3.578)* 10.750(8.878)* 17.256(9.246)
building higher than 11 floors -13.756(6.453)** -12.835(5.295)** -12.014(5.340)** -12.027(5.341)** -13.087(5.412)**
building lower than 10 floors -10.233(7.904) -10.743(6.687) -9.316(6.235) -9.421(6.579) -10.743(6.687)
condominium 1.347(6.983)* 1.068(6.004)* 1.458(5.912)* 1.165(5.963)* 1.068(6.004)*
studio apartment -3.573(8.260)* -5.224(7.313)* -3.287(7.156)* -3.754(7.211)* -3.004(7.313)*
14
Notes: 2017 = 0, Single-family house = 0, Kaohsiung = 0.*** Significant at 1% significance level. ** Significant at 5% significance level. * Significant at 10% significance level
Refer to Table 2 (continued)
storefront 12.964(12.431) 8.561(10.350) 9.775(10.303) 9.419(10.324) 9.516(10.205)
commercial building -8.847(16.990) -7.907(9.129) -5.791(8.887) -6.869(9.089) -7.907(9.129)
others -5.754(10.897) -8.448(8.899) -6.990(8.822) -7.127(8.832) -5.844(8.899)
2012 2.521(8.489) 4.239(7.405) 3.318(7.340) 3.606(7.388) 4.239 (7.405)
2013 13.814(6.067)** 12.569(5.144)** 12.304(5.128)** 12.173 (5.135)** 12.569 (5.144)**
2014 11.257(5.607)** 10.233(4.744)** 9.950(4.678)** 9.642(4.719)** 9.033(4.744)**
2015 5.549(5.716) 5.813(4.873) 5.790(4.865) 5.658(4.872) 5.083 (4.873)
2016 2.051(5.584) 2.274(4.742) 2.292(4.741) 2.224(4.743) 2.217 (4.744)
N 1790 1790 1790 1790 1790
F-value 34.061 33.204 32.973 32.448 32.206
R2 .206 .208 .205 .206 .208
15
Model (4) also shows that the impact of MRT system in Taipei on housing prices is
greater than that in Kaohsiung when the distance increases by 1 km. The housing prices
in Taipei decline an additional of $NT60,700 per ping. Due to small sample size, the
coefficients of distance,distance2, distance×Taipei and distance2×Taipei are not
statistically significant in Model (5).
When we look at Model (1), the housing prices drop by $NT4,110 if the age of the
house increase by one year. As Model (2) shows, when the age of house increases one
year, the housing prices decrease by $NT3,920. Model (3) indicates that the housing
prices drop by $NT3,900. Model (4) reveals that the housing prices decrease by $NT3,860
when the age of house increase on year and Model (5) suggests than the housing prices
decrease by $NT3,850 when the age of house increase on year
Moreover, the size of the house has a significant effect on housing prices. All five
models suggest that when the size of the house increases by 10 pings, the housing prices
decrease $NT100. Meanwhile, when we look at Taipei, the estimate infers that the
housing prices in Kaohsiung are less than Taipei.
The housing prices in the year 2013 are higher than in other years. The reason might
be the implementation of the capital gain tax on the stock market in 2012 which redirected
the capital flow to the real estate market and causes the housing prices to rise. However,
16
the government implemented the Combined Housing and Real Estate Tax in 2015;
consequently, the housing prices declined significantly.
Table 3 shows the effect of the distance to MRT station on housing prices when we
limit the sample to those within 3 km from the MRT stations. Model (1) shows that the
distance to the MRT station, the age of the house, size of the house, and location of the
house are all significant determinant of housing prices. Distance to MRT stations has a
negative effect on housing prices, which means that when the distance increases, the
housing prices decrease and this result consists of the previous studies. The results suggest
that when the distance increases by 1 km; the housing prices decrease by $NT50,000/ping
Model (2) shows the different impact of the distance to MRT station on housing prices
in Taipei and Kaohsiung. The result indicates that when the distance increases by 1 km,
the housing prices in Taipei decrease by an additional $NT17,773 per ping, as opposed to
Kaohsiung.
Based on Model (3), Distance2 is statistically significant and the result indicates that
the rate of the decrease in housing prices declines as the distance increases. Model (4)
is our major model. When we have partial housing prices over partial distance, the
estimates indicate that when the distance increases by 1 km, the housing
17
Table 3. The Effect of Distance on Housing Prices: Distance to MRT station ≦ 3 km
Independent Variable
Model (1) Model (2) Model (3)
Model (4) Model (5)distance -5.105(2.278)*** -4.540(3.853)*** -23.087(8.105)*** -23.694(9.070)*** -16.428 (14.672)
distance2 7.719 (3.106)** 7.249(3.142)** 4.496 (5.363)
distance×Taipei -1.7773(4.569)** -1.098(4.595)** -10.123(17.699)
distance2 ×Taipei 4.213 (6.625)
age_house -.298(.105)*** -.295(.106)*** -.318(.106)*** -.320(.106)*** -.318(.107)***
age_house _missing 8.776(3.575)*** 9.523(3.584)** 8.527(3.573)** 8.899(3.629)** 8.344(3.635)**
size_house -.001(.002)* -.001(.002)* -.001(.002)* -.001(.002)* -.001(.002)*
Taipei 10.253(2.891)*** 11.847(5.142)** 8.752(2.960)*** 11.110(5.217)** 13.184(9.685)
building higher than 11 floors -9.532(4.398)** -10.891(4.407)** -9.170(4.396)** -9.414(4.443)** -9.344(4.432)**
building lower than 10 floors -7.877(5.656) -8.045(5.660) -7.731(5.560) -7.822(5.664) -7.992(5.669)
condominium 1.596(5.119)* 1.337(5.112)* 1.849(5.115)* 1.155(5.152)* 1.627(5.139)*
18
Notes: 2017 = 0, Single-family house = 0, Kaohsiung = 0. Standard errors are in parentheses.*** Significant at 1% significance level. ** Significant at 5% significance level. * Significant at 10% significance level
Refer to Table 3 (continued)
studio apartment -2.783(6.257)* -3.480(6.255)* -2.079(6.528)* -2.709(6.296)* -2.617(6.316)*
storefront 10.581(8.591) 10.661(8.660) 10.489(8.528) 10.469(8.569) 10.103(8.615)
commercial building -3.677(7.919) -3.865(7.890) -3.38(7.899) -5.510(7.899) -4.382(8.073)
others -5.406(6.846) -4.657(6.858) -5.472(6.856) -5.472(6.856) -5.340(6.872)
2012 4.477(6.376) 4.533(6.384) 3.841(6.385) 4.533(6.384) 3.699 (6.411)
2013 12.048(4.458)** 12.010(4.462)** 11.537(4.459)*** 12.010 (4.462)*** 11.491 (4.464)***
2014 9.632(4.056)** 9.552(4.906)** 9.513(4.075)** 9.552(4.593)** 9.435 4.098)**
2015 3.276(4.078) 3.238(4.084) 2.839(4.079) 3.238(4.084) 2.722 (4.090)
2016 1.574(4.152) 1.534(4.153) 1.761(4.149) 1.534(4.158) 1.769 (4.156)
N 2244 2244 2244 2244 2244
F-value 34.755 33.953 33.670 33.285 30.397
R2 .272 .276 .277 .275 .279
19
prices decrease by $NT119,467per ping in Taipei and decrease by $NT108,487 per ping
in Kaohsiung. The relationship between the distance to MRT stations and housing prices
is a U shape.
Moreover, the interaction term distance×Taipei is negative. It means that the distance
to MRT station in Taipei has a greater impact than that in Kaohsiung. It is because,
compared with Kaohsiung, Taipei has complete MRT systems; thus, people in Taipei are
more dependent on the MRT system. On the contrary, the MRT system in Kaohsiung is
less complete. Therefore, people would choose to ride a scooter instead of taking MRT.
The coefficients of distance, distance2, distance×Taipei and distance2×Taipei in
Model (5) are not statistically significant. This might be the outcome of the small
sample size. When we look at Model (1), the housing prices drop by $NT2,980 if the age
of house increases by one year. As Model (3) shows, when the age of the house increase
one year, the housing prices decrease by $NT3,180. Model (4) indicates that if the age of
house increases on the year, the housing prices drop by $NT3,200 and model (5) shows
that the housing prices decrease by $NT3,180 when the age of house increase one year.
The size of the house also affects the housing prices. All the models suggest that when
the size of the house increases by 10 pings, the housing prices per ping decrease by
$NT100. In addition, the housing prices in Kaohsiung are lower than in Taipei. The
20
housing prices in the year 2013 are higher than any other years in our sample, which
suggests that the housing market hit the record high in 2013 in Taiwan.