4.1. The Effect of the Greenbelt Policy Release on Urban Development
Prior to the analysis of the three individual cities, the regular logistic model that contains city dummy variables was run to identify the city-level effects of development. The maximum likelihood estimator was used to fit the binary logistic regression model. Table 2
shows the estimated coefficients and odds ratios for the model, which contains five variables and two city dummy variables. The results of the model show several development trends in the city-level effects of the GB. First, the previous GB lands show a higher development probability with an odds ratio higher than 1 (1.644) than do the other lands. Therefore, the removal of the GB played a role in making the lands more attractive to develop. Second, urban development in Cheongju (reference city) was more active than in Jinju, but less active than in Chuncheon. However, the effects of the GB on land development in individual cities reveal quite different results from these city-level analyses.
The location of land development before and after the release of the GB restriction in the three cities is presented in Table 3
and Figure 3
. The table and figure provide some important information on the urban development pattern.
There were plenty of developable lands on the inner and outer sides of the GB in Cheongju and Chuncheon cities at the time of the GB removal. The developable lands in the contained lands (between urban core and the GB) were 26.2 km2
in Cheongju and 21.8 km2
in Chuncheon. The existing urbanised areas were 37.1 km2
and 17.6 km2
, respectively, as shown in Table 3
. Even if the GB restrictions were removed, the new development occurred on the inner (G1) and outer sides (G3) of the GB, leaving most of the GB areas (G2) undeveloped in the cases of Cheongju and Chuncheon. In contrast, there was little developable land (11.9 km2
) on the inner side of the GB in the case of Jinju (16.6 km2
of urbanised area) where new development has actively occurred on the previous GB lands.
These different patterns of development before and after GB removal can further be supported by the changes in developed lands and population by area. Overall, Table 3
shows that the growth rate of urbanised areas is faster than that of the population for all three cities. This implies that there has been active non-residential land development such as manufacturing, commercial, and public use in the cities. It shows that each of the three cities reveals distinctive development patterns. In the case of Jinju city, the GB removal leads to urban development through infill. The population has reduced on the outer side of the GB (G3), while that of the inner side and inside the GB (G1) has increased. Urban development is most active in the GB area (G2). The city of Chuncheon reveals a similar pattern of development. However, land development is most active on the outer side of the GB and most of the population increase occurs on the inner side. The urban development of Cheongju city shows the opposite pattern. Even though the GB has been removed, the population inside the GB has decreased, while the population on the outer side has increased. The amount of land development inside the GB is relatively small compared with that in the inner and outer sides. In other words, GB removal has triggered more active new suburban development. During the study period, one additional person was found to bring about 3421 m2
of new land development in Jinju, 1379 m2
in Chuncheon, and only 414.7 m2
in Cheongju. This finding conflicts with the findings of other studies, which maintain that the GB brings about a pattern of leapfrog development. The amount of developable lands inside the contained area may be a crucial factor that determines the pattern of the urban development as well as the effectiveness of the GB policy [14
and Table 5
display the estimated coefficients for the basic DID model (Model-I) containing the three independent variables and Model-II containing the four covariates (two accessibility variables, population change, and land price change). The regression coefficients are as follows.
On the basis of the results of Model-I, some useful findings can be derived. The signs of coefficients are the same across the three cities, though the intensity varies. All variables show expected signs with statistical significance at a 1% significance level except for the DID variable of Cheongju. They are positive for time and DID, and negative for GB. Therefore, the GB removal induces more development inside the GB for Jinju and Chuncheon, but not for Cheongju. The odds ratios (OR) of the DID variables of the three cities also support these development trends. The OR of DID of Jinju is 269.05, and that of Chuncheon is 3.24, implying that the probabilities of development in previous GB lands are 269.05 and 3.24 times higher than in other lands, respectively, by removing the GB restrictions. However, the OR of the DID for Cheongju is not statistically significant.
By adding control variables in Model-II, the explanatory powers of the three models have been highly improved. Since rho (ρ) in Model-II is different from zero for all three city models, the panel estimator is also different from the pooled data. When the control variables were included in Model-II, the signs of the coefficients were not changed, except for the sign of the DID variable in the case of Cheongju. The focus of this study, the DID variable (Time × GB), shows an interesting result. The coefficient of the Jinju DID is positive and significant, that of Chuncheon is positive and not significant, and that of Cheongju is negative and not significant. The negative sign of the DID variable implies that the GB lands were less likely to have been developed than before the removal of the GB restriction. The non-significant DID variable also suggests that the GB removal may not function as a strong incentive for development. The development activities previously outside the GB were more active in Chuncheon and Cheongju than inside the GB while the activities inside the GB were more active than those of other areas in the case of Jinju. These results can be supported further by the ORs of the model. The OR of the DID variable of Jinju is 4.022 while that of Cheongju is less than 1.0 (0.826) and statistically not significant.
Therefore, it is difficult to conclude that there exists a GB policy effect in Cheongju and Chuncheon. The GB policy is valid only in the case of Jinju where developable lands are scarce on the inner side of the GB. The DID coefficient shows that the GB removal accelerates land development in the previously GB boundary. Variations in effects of the GB removal are the results of different demands placed on the GB. There are considerable differences in land development demand for the GB lands among medium-sized cities.
Distance variables have negative signs, implying that higher proximity to the city centre and main road increases development probabilities. The coefficient of the distance to the central business district (CBD) variable of Jinju is not significant, implying that the location of new developments is not clearly distinct by distance. Also, the distance to the main road variable of Chuncheon is positive owing to the mountainous geographical feature. As expected, population and land price variables show positive and significant coefficients.
Although the GB is primarily designed to restrict land development, its removal does not always stimulate land development inside the GB (G2). Land development has increased in all three cities after the GB removal, but the effects of the GB removal are not the same across cities, as the quantity and location patterns of the land development are different among cities. For example, the GB removal apparently contributed to easing development pressures in both G1 and G3, as implied by active development in GB land (G2) in the case of Jinju. Conversely, the GB removal had no significant effects on land development.
The greenbelt policy is still an attractive policy instrument, and many cities around the world maintain the policy or are newly implementing it (Ontario in 2005, and Scotland in 2010). Furthermore, some cities are considering introduction of a GB policy.
Is the greenbelt policy helpful to achieving sustainable development? Or is the greenbelt sustainable? Amati [44
] raised the question of whether the GB policy is a useful tool for managing urban growth in the twenty-first century. The merits of the GB are to preserve green areas around city and amenity values, and to restrict the expansion of built-up areas thereby leading to infill development. However, in many cases, the greenbelt policy leads to a leapfrog development that stimulates lengthier commutes and higher car use, which increases the price of housing/land.
The role of the GB on green area protection can also be challenged. The removal of the GB itself is perceived by environmental groups as a sign of the government’s giving up the protection of green areas around cities. In fact, the greenbelt may not actually be green. It may contain degraded land, little landscape quality, and limited public access [45
]. The GB has an important inter-generational function as a land reservoir for future use as well. Therefore, even though some parts of the GB are not worth protecting for their environmental aspects, the GB’s inter-generational function is valid and should be considered. This may call social attention to the time span of GB policy if development pressure is intense in GB lands. For example, there is a fixed time span (usually 20 years) in UGB policy while the GB is perpetual, once established. How long should the GB be maintained at the expense of the current generation?
Effective land-use regulation contributes to achieving a more efficient form of human settlement, which also improves regional economies [46
]. As in the case of introducing new land-use regulations, the removal of the GB also affects the efficiency of the urban structure. It provides more lands for urban uses near existing urbanised areas. If demand for development is high and developable lands are scarce, then it is very likely to lead to compact development. Conversely, if demand for development is weak, the removal of the GB may have no impact.
The effectiveness of the GB policy may depend on many factors, such as the country-specific political climate concerning more development or preservation [47
], and the amount of available lands for development [14
]. The core question in the GB policy may lie in how to maintain a balance between the demands for development and preservation of the lands [47