Research Express@NCKU Volume 13 Issue 6 - April 16, 2010 [ http://research.ncku.edu.tw/re/articles/e/20100416/4.html ]
Does high-speed rail accessibility influence residential
property prices? Hedonic estimates from southern
David Emanuel Andersson
1, Oliver F. Shyr
2,*and Johnson Fu
21Institute of Public Affairs Management, National Sun Yat-sen University
2Department of Urban Planning, National Cheng-Kung University, One University Road, Tainan 701, Taiwan, ROC
Journal of Transport Geography (SSCI), doi:10.1016/j.jtrangeo.2008.10.012
I
ntroductionThe opening of a High Speed Rail line (HSR) in early 2007 drastically reduced travel times along Taiwan's west coast. Investments in transportation infrastructure have often played key roles in the restructuring of urban land use and land price patterns, both in Taiwan and elsewhere. When the Japanese colonial government built Taiwan's first railway network in the early 20th century, it resulted in
restructured urban land use patterns. Regional accessibility and land value maxima shifted from ports and harbors to emerging downtowns around urban stations. Later,
as the first freeways opened in the late 1970s, new towns and high-priced neighborhoods evolved in proximity to freeway interchanges.
According to the Bureau of High Speed Rail, the new rail link is “designed to promote balanced regional
development while minimizing impact on the environment and natural ecology. The total length of Taiwan's HSR line is 345 kilometers. The line has primarily relied on imported technology and hardware from Japan's Shinkansen line, supplemented with a European (TGV and ICE) traffic management system. With an investment cost of approximately US$15 billion, it set a new record as the world's most expensive build-operate-transfer project. The Taipei and Banciao stations are both located within the Taipei metropolitan area and are integrated nodes in the metropolitan commuter rail and rapid transit networks. The Taichung and Zuoying (Kaohsiung) HSR stations can be reached within 15 minutes from downtown railway stations by commuter train or rapid transit. In the smaller metropolitan areas of Taoyuan, Hsinchu, Chiayi, and Tainan, access to HSR is confined to motor vehicles, resulting in minimum access times of between 20 and 40 minutes from downtown locations. New light rail transit and commuter rail projects are being planned for Taichung, Tainan, and Taoyuan. The new projects aim at reducing access times by 10 to 15 minutes1.
station, the area around the new Tainan HSR station at Guiren in Tainan County exhibits low population and land use densities. The database used in this study – which includes all residential property transactions in the Tainan metropolitan area in 2007 – also shows that only 2.7 percent of all sold dwellings were located in the same township as the HSR station, which is even lower than its population share of 2.9 percent.
The aim of this paper is to use hedonic price techniques to estimate the impact of HSR station accessibility on land values in the Tainan metropolitan area. We estimate several functional forms that are all compatible with the underlying economic theory. A comparison of separate estimation results allow us to draw conclusions regarding the robustness of attribute effects. The functional forms include log-linear and Box-Cox transformation. The most general and flexible form of the linear Box-Cox transformation is
X1λ1 = β 1 + β2X2λ2 + … + βiXiλi + … + βkXkλk; (1) λi = λ1, … , λk; where Xiλi = (X iλi – 1) / λi for λi≠ 0; lnXi for λi = 0
Description of the data
The observations on transaction prices and structural characteristics were obtained from the Department of Land Administration of the central government, and cover all 1,550 residential property transactions in 2007 in the Tainan Metropolitan Area. The data on socio-economic neighborhood characteristics are for 2004 and were obtained from the Ministry of Finance. The neighborhoods correspond to districts in Tainan City and townships in the rest of the metropolitan area. The distance measurements are proxies for house-specific access to Tainan's HSR station, the city center, the Tainan Science-based Industrial Park (TSIP), and the nearest freeway interchange, respectively. The distance data correspond to the shortest route for motor vehicles according to a popular GIS program PaPaGo R12 that covers the entire road network of Taiwan.
The hedonic analysis makes use of sales prices rather than rents, since owner-occupied housing accounts for more than 80 percent of Taiwan's housing market. In addition, sales prices reflect expectations of future developments, and should therefore - unlike rents - reflect potential long-term future benefits of new or planned infrastructure investments.
The tested independent variables encompass six structural, five neighborhood and four accessibility attributes. Pair-wise correlation analysis reveals very high positive correlations between neighborhood income and education and relatively high negative correlations between both socio-economic neighborhood variables and the distance-to-CBD variable. The latter correlations imply an income elasticity of demand for accessibility that is on average greater than the income elasticity of demand for space in Tainan's housing market. Correlations between different accessibility variables are all quite low.
“Commercial zones” allow for some residential use on higher floors. For example, downtown residential zones are often used for high-rise apartment blocks with high-value commercial use such as banks and luxury retailing on the first two floors.
Results
Table 1 shows that all structural and neighborhood variables have the expected signs, and most of them are highly significant. The log-linear function explains more of the variance and has greater log likelihood in both its all-variable and reduced versions. All the Box-Cox functions have slightly greater log likelihood than the log-linear function, with the exception of the simple left-hand side model. The transformation parameter of the simple both-hand side model is significantly different from λ=0, which would correspond to the log-linear model. Even so, it is in the same general neighborhood as the log-linear function, and offers identical qualitative conclusions. Each type of functional form has its own advantages and drawbacks: pre-specified functions are simple and easy to interpret, transformations improve the distribution of error terms, but increasingly complex transformation leads to
increasing data specificity and increasing levels of “data mining”.
The estimated accessibility effects point to the essentially monocentric character of the Tainan metropolitan area. The contribution of the distance-to-CBD attribute (which is defined as distance to the old railway station) to residential property prices is similar and highly significant in all eight specifications. The estimates of the log-linear function imply that a halving of the distance is associated with a price increase of about 11 percent. The TSIP is a suburban high technology cluster that is located in the north-eastern part of the metropolitan area; it is a daytime population center for more than 50,000 workers, many of which are high-income managers and engineers. The effect of TSIP accessibility is smaller and less robust: it is insignificant (at one-tailed .05 significance) in two of the functions and is associated with a price-distance elasticity of about -.07.
HSR accessibility is even less robust and significant. Moreover, in two of the functions the parameter estimate takes the wrong (+) sign. Even if we tentatively accept the existence of a price-distance effect, it amounts to no more than a 3 or 4 percent price premium for, say, an otherwise equivalent house at a distance of four instead of eight kilometers from the station. This would correspond to only about half the effect of proximity to any railway station in the Netherlands, regardless of the frequency of trains and network connectivity. In this context it should be mentioned that Tainan HSR station offers a total of 70 departures per day, with connections to all major cities and the transit systems of Taipei and Kaohsiung.
An economic explanation for why there has so far been little in the way of high-speed intercity commuting is expensive HSR fares. By way of example, daily (Monday to Friday) HSR commuting between Tainan and Taichung would – in 2008 – incur a direct ticket cost of NT$23,900 (US$775) per month, which amounts to about 70 percent of the median monthly wage in Taiwan. This could be compared with commuting a similar time distance by train between cities in Western Europe, where a one-month commuter ticket typically amounts to only about 10 percent of the median monthly wage.
other job attributes being equal. The total transportation cost in the HSR case amounts to the sum of the direct cost (NT$23,920) and the subjective time cost (i.e. at least 33 hours of traveling). Though wage rates are higher in Taichung, it is unlikely that a large group of Tainan residents will find jobs in Taichung that pay substantially
more than an additional NT$23,900 per month, if we make the realistic assumption that train commuting is not
their highest-valued leisure activity.
Final remarks
While we have been unable to find any substantial and consistently significant effects of HSR accessibility in the Tainan metropolitan area, this does not necessarily imply that there are no effects on land values. It could be the case that that commercial activities have a higher willingness to pay for HSR station proximity, for example if they engage in frequent business meetings with customers or suppliers from other metropolitan regions. In a well-functioning land market, however, it is the highest bidder for a location that determines the land value, which means that a high valuation in the commercial property market should spill over into the housing market. Under such conditions, only home buyers with an exceptional willingness to pay would be able to outbid commercial property developers in commercially attractive locations.
The land market is however often not “well-functioning” in this sense. There may be substantial transaction and reconstruction costs associated with land use conversions. Moreover, zoning and other land-use regulations may perpetuate the separation of different segments of the real estate market. An additional problem in Taiwan has been that many infrastructural investments have been subject to substantial delays or revisions, which has the effect of impeding the dissemination of reliable information from planners and developers to other market participants, thus delaying the discovery of the highest-valued land uses in various locations.
Finally, the empirical evidence in this study, as well as income statistics from other regions in Taiwan, indicate that the income elasticity of demand for accessibility is greater than the income elasticity of demand for residential space for the modal Taiwanese household. And it is exactly higher-income workers who tend to be the first to take advantage of new commuting opportunities, as a reflection of their pursuit of larger spaces in tranquil upscale neighborhoods. Unlike America, and to some extent Europe, the inner cities of Taiwan are associated with the highest income levels and are perceived as having the best schools.
1. Currently most of the planned light rail transit or commuter rail routes are covered by the free shuttle bus
