# RECORD 69288e963b8055b19e0fd077000b14eb34270ae1
# TITLE: THE EFFECTS OF TRANSACTION HOTSPOTS AND FLIPPING HOTSPOTS ON HOUSING PRICES
# YEAR: 2024.0  VENUE: International Journal of Strategic Property Management  DOI: 10.3846/ijspm.2024.20899
# AUTHORS: Bor-Ming Hsieh; Chih-Yuan Yang
# PAGES: 15  CANDIDATE TABLE PAGES (1-based): [8, 9, 10, 13, 14, 15]
# ABSTRACT: In contrast to much of the literature that focuses on the issue of spatial dependence in housing price research, this study addresses the spatial aggregation of housing transactions and analyzes the effects of transaction hotspots and short-term flipping hotspots on housing prices by using real housing transaction data in Taipei City, Taiwan. The empirical results show that after controlling for the effects of spatial dependence and individual housing attributes, the impact of transaction hotspot areas on housing prices is significantly negative, while the impact of flipping hotspot areas on housing prices is significantly positive. The results verify that the key to driving up housing prices lies in flipping activities. Furthermore, the results of the spatial quantile regression model show that low-priced residential properties are more sensitive to the spatial concentration of housing transactions and flipping transactions in the housing market. Our results have implications for the government’s policy intending to control hot trade volumes to cool skyrocketing housing prices in a booming housing market. It is suggested that the government should pay attention to restraining short-term flipping activities in the housing market rather than setting constrains on housing transactions.


===== PAGE 1 (score 11, numeric-paren 0, table-words 0, var-words 11) =====

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), 
which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Copyright © 2024 The Author(s). Published by Vilnius Gediminas Technical University
ISSN: 1648-715X / eISSN: 1648-9179
INTERNATIONAL JOURNAL 
OF STRATEGIC PROPERTY 
MANAGEMENT
INTERNATIONAL JOURNAL 
OF STRATEGIC PROPERTY 
MANAGEMENT
2024
Volume 28
Issue 1
Pages 1–15
https://doi.org/10.3846/ijspm.2024.20899
THE EFFECTS OF TRANSACTION HOTSPOTS AND FLIPPING HOTSPOTS ON 
HOUSING PRICES
Bor-Ming HSIEH1, Chih-Yuan YANG2,*
1 Department of Land Management and Development, Chang Jung Christian University, Tainan, Taiwan
2 Department of Civic Education and Leadership, National Taiwan Normal University, Taipei, Taiwan
Article History:  
Abstract. In contrast to much of the literature that focuses on the issue of spatial dependence in housing 
price research, this study addresses the spatial aggregation of housing transactions and analyzes the effects 
of transaction hotspots and short-term flipping hotspots on housing prices by using real housing transac-
tion data in Taipei City, Taiwan. The empirical results show that after controlling for the effects of spatial 
dependence and individual housing attributes, the impact of transaction hotspot areas on housing prices is 
significantly negative, while the impact of flipping hotspot areas on housing prices is significantly positive. 
The results verify that the key to driving up housing prices lies in flipping activities. Furthermore, the results 
of the spatial quantile regression model show that low-priced residential properties are more sensitive to the 
spatial concentration of housing transactions and flipping transactions in the housing market. Our results have 
implications for the government’s policy intending to control hot trade volumes to cool skyrocketing housing 
prices in a booming housing market. It is suggested that the government should pay attention to restraining 
short-term flipping activities in the housing market rather than setting constrains on housing transactions.
	■received 31 August 2023
	■accepted 26 December 2023
Keywords: housing prices, flipping, spatial aggregation, housing volume, hotspot analysis.
 * Corresponding author. E-mail: cyy@ntnu.edu.tw
1. Introduction
Housing transactions and housing prices have usually 
been the most important issues in housing market studies. 
Previous studies have indicated that housing transactions 
and housing prices interact (Rothenberg et al., 1991). In 
many cases, housing transactions can serve as an indicator 
of whether the housing market is hot or not. A traditional 
expectation based on this viewpoint suggests that the vol-
ume of housing transactions predicts housing prices; for 
example, a rapid increase in housing transactions induces 
a rise in housing prices (Hua et al., 2001; Yiu et al., 2009). 
A boom in housing transactions and housing prices in an 
area often indicates that the housing market is prospering; 
hence, more house buyers are attracted, including owner-
occupiers and investors, and as a result, this drives hous-
ing prices up. However, it is difficult to distinguish rigid 
residential demand from speculative investing demand 
from observing housing transactions. By analyzing the fre-
quency of housing transactions in a certain period of time, 
it can differentiate buyers’ demand for owner-occupation 
or short-term speculation in the housing market (Depken 
et al., 2009).
In the housing market, the behavior of short-term buy-
ing and reselling properties with a profit can be seen as 
speculation or called flipping (LaCour-Little & Yang, 2023). 
Bayer et al. (2020) pointed out that investors who repeat-
edly buy and resell properties within two years are seen 
as flippers. Most flippers choose the location of property 
and time of investment and resell it when the housing 
price rises promptly (Yilmaz, 2014). Previous studies have 
indicated that short-term flipping in the housing market 
causes a boom in housing transactions and thus rapidly 
driving housing prices up, which is not conducive to the 
sound development of the housing market (Leung & Tse, 
2017; Li et al., 2023). However, fewer studies have con-
cerned the spatial content of housing flipping and exam-
ined its impact on housing prices. 
Regarding the spatial concept of housing, adjoining 
housing units share similar environmental conditions, 
such as neighborhoods, parks, and open space. Hous-
ing prices of adjoining housing units are affected by each 
other, which is explained as the spatial dependence of 
house prices (Basu & Thibodeau, 1998; Case et al., 2004). 
In addition to the spatial dependence of housing prices, 
the concentration of traded residential properties also 
has spatial impacts (ripple effects) on the prices of ad-
jacent properties. As stated earlier, previous studies have 
indicated that a rapid increase in housing transactions 
tends to drive up housing prices. In this sense, a boom in 


===== PAGE 8 (score 54, numeric-paren 29, table-words 5, var-words 7) =====

8
B.-M. Hsieh, C.-Y. Yang. The effects of transaction hotspots and flipping hotspots on housing prices
accounting for 50.3% of the total repeated-sale samples.9 
This figure illustrates the gravity of the issue of flipping 
in Taipei’s housing market. By using a hot spot analysis 
approach to calculate the Getis-Ord Gi* statistic for each 
flipping sample in 50 × 50 m grid areas in the city, the hot-
spot areas of housing flipping are identified as red areas 
in Figure 2.
As shown in Figure 2, the red areas account for ap-
proximately 50% of areas with repeated sale samples and 
account for 1% of total housing transaction areas in the 
city, representing the most significant and high concentra-
tion of flipping transactions in the housing market over 
the past decade. The transaction hotspot areas and flip-
ping hotspot areas do not exactly overlap. Compared to 
the hotspot areas of housing transactions (the blue ar-
eas), the hotspot areas of housing flipping are more likely 
to be concentrated in western areas of the city, such as 
Zhongshan District, Datong District, Zhongzeng District, 
and Wanhua District, where they are old city center areas 
in Taipei City. Furthermore, most flipping hotspot areas are 
located in relatively lower-priced areas. This implies that 
lower-priced properties are more likely to be flipped than 
high-priced properties in Taipei City.
9	 Of the 62,140 transactions in our sample, 3,302 occurred within 
2 years. The proportion of 5.31% is larger than the correspond-
ing ratio (3.13% = 3,133/100,076) in the paper of Li et al. (2023) 
which using the dataset of residential property transactions in 
Clark County, Nevada in the USA for the period 2003–2013.
4.2. Results of OLS model and spatial lag 
model
Regarding the housing price model, this study employs 
the spatial lag model to improve the spatial dependence 
of housing prices.10 The results of the OLS and the spa-
tial autoregression models are shown in Table 2. The full 
model results are shown in the Appendix. The results show 
that there is a significant and positive coefficient of spatial 
autocorrelation, suggesting that spatial autocorrelations 
exist significantly among housing prices. Thus, using the 
spatial lag regression can alleviate the problem of spatial 
dependence among housing prices and therefore improve 
the model goodness-of-fit. As shown in Table 2, the good-
ness-of-fit of the spatial lag model is better than that of 
the OLS model, while the R-square of the spatial lag model 
is slightly higher than that of the OLS model.11
In both models, the directions and coefficients of the 
controlling variables, such as district dummy variables, are 
as expected and consistent with prior studies. An increase 
of one square meter in floor area increases housing prices 
by 0.8%, and an increase of one bathroom rises housing 
prices by approximately 6%. In contrast, an increase of one
10	The Moran’s I for housing transaction prices is 0.1741, with a 
z-score of 859.99, indicating a significant positive spatial cor-
relation among housing transaction prices.
11	As expected, the results show the housing prices in all 11 dis-
tricts are significantly higher than in the Wanhua District. In ad-
dition, housing prices decreased compared to 2021, indicating 
a steady increase of housing prices over the past decade.
Table 2. Results of OLS model and spatial autoregression models
Model
OLS model
Spatial lag model
Spatial error model
Variable
Coeff.
t value 
Coeff.
z-statistic 
Coeff.
z-statistic 
Constant
16.149
1394.93***
16.184
1426.16***
16.224
1350.06***
Area
0.008
315.66***
0.008
316.60***
0.007
179.92***
Floor
–0.0004
–1.05 
–0.0001
–0.37 
–0.0008
–2.33** 
TStory
–0.0003
–1.01 
–0.0004
–1.33 
–0.0009
–2.69***
Bathroom
0.059
24.31***
0.060
25.31***
0.061
25.42***
Age
–0.011
–118.36***
–0.011
–121.44***
–0.011
–114.10***
Dist. MRT
–0.011
–40.04***
–0.011
–39.90***
–0.011
–37.84***
Dist. CC
–0.004
–32.22***
–0.003
–27.49***
–0.003
–31.16***
THA
–0.102
–34.74***
–0.099
–34.56***
–0.101
–34.42***
FHA
0.014
1.08 
0.022
1.70* 
0.020
1.55 
THA × FHA
0.042
1.68* 
0.050
2.02** 
0.032
1.30 
Administrative districts
Yes
Yes
Yes
Trading years × quarters
Yes
Yes
Yes
Spatial autocorr. coeff.
 –
 
0.196
52.50***
0.056
22.32***
(Pseudo) R-square
0.853
0.859
0.855
F value
6215.69***
–
–
AIC
14781.55
14678.08
14737.21
Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. The number of observations is 62,140. The results of controlling dummies 
of the administrative district and trading year and quarter are provided in the Appendix.11


===== PAGE 9 (score 88, numeric-paren 54, table-words 5, var-words 11) =====

International Journal of Strategic Property Management, 2024, 28(1), 1–15
9
year in dwelling age decreases housing prices by approxi-
mately 1.1%. The distance to the nearest MRT station and 
the distance to the city center all have negative effects on 
housing prices, which indicates that the closer to the MRT 
station and to the city center, the higher the housing prices.
The coefficient of THA is –0.099 and significant in the 
spatial lag model, indicating that housing prices in trans-
action hotspot areas decrease 9.9% compared to those 
not in hotspot areas. This is probably because of the price 
comparison effect; that is, when there are many properties 
for sale in the market, the competitive effect for on-sale 
properties usually makes transaction prices down. Such re-
sults are compatible with the findings of Deng et al. (2022). 
Properties in the areas of many transactions have more 
sufficient information and more competitors and thus re-
duce their transaction prices.
Furthermore, the coefficient of FHA in the spatial lag 
model is 0.022, denoting that housing prices increase up 
to 2.2% in hotspot areas of flipping activities. These re-
sults are in line with the viewpoints of short-term flipping 
speculation in the housing market (Leung & Tse, 2017; 
LaCour-Little & Yang, 2023). House flipping increases 
housing prices and harms sound markets.12 The sum-
marizing effect of both THA and FHA on housing prices 
is  –0.027 (–0.099+0.022+0.050), denoting that housing 
prices decrease by 2.7% in both hotspot areas of housing 
12	See article titled “How Tales of ‘Flippers’ Led to a Housing Bub-
ble” by Robert Shiller, at Economist’s View on May 18, 2017, 
available from https://economistsview.typepad.com/economists-
view/2017/05/how-tales-of-flippers-led-to-a-housing-bubble.html
transactions and hotspot areas of flipping activities. Since 
properties in the FHA drive up neighboring prices, the 
negative effect of the interactive THA and FHA on hous-
ing prices is less than that on THA only. In short, the above 
results suggest that a high concentration of flipping activi-
ties in the housing market, rather than a high concentra-
tion of normal housing transactions, is an important driv-
ing force of rising housing prices in the housing market 
of Taipei City.
4.3. Results of the spatial quantile regression 
model
This subsection discusses the results of the spatial quantile 
regression model, which estimates the effects of transac-
tion hotspot areas and flipping hotspot areas on various 
quantiles of housing prices with the consideration of spatial 
autocorrelation. The results are summarized in Table 3. The 
full model results are also shown in the Appendix. The trad-
ing floor (Floor) and total story (TStory) have positive effects 
on properties in middle to high housing price levels, show-
ing that the influence of being located on a high floor and 
high-rise buildings on housing prices increases with a rise 
in housing price quantile levels. The number of bathrooms 
(Bathroom) has a greater influence on properties at lower 
housing price levels than at higher price levels.
Table 3. Results of the spatial quantile regression model
Model
Low
……………………………….. Housing prices ………………………………..
High
0.1 q
0.25 q
0.5 q
0.75 q
0.9 q
Variable
Coeff.
t value
Coeff.
t value
Coeff.
t value
Coeff.
t value
Coeff.
t value
Constant
15.916 582.66***
16.067 991.19***
16.155 1303.56***
16.236 1297.69***
16.334 
1150.88***
Area
0.008 
131.06***
0.008 
230.96***
0.008 
305.09***
0.008 
300.55***
0.008 
265.16***
Floor
–0.002 
–2.15** 
0.001 
1.51 
0.001 
3.91***
0.002 
5.064***
0.002 
3.27***
TStory
–0.004 
–5.40***
–0.001 
–3.01*** 
0.001 
3.31***
0.003 
7.32***
0.003 
6.37***
Bathroom
0.069 
12.16***
0.061 
17.98***
0.055 
21.54***
0.054 
20.79***
0.049 
16.63***
Age
–0.012 
–55.95***
–0.012 –88.20***
–0.011 –105.21***
–0.010 –100.58***
–0.010 
–86.66***
Dist. MRT
–0.012 
–17.80***
–0.013 –33.05***
–0.011 
–38.23***
–0.010 
–32.26***
–0.009 
–25.01***
Dist. CC
–0.002 
–9.48***
–0.003 –16.04***
–0.003 
–26.20***
–0.004 
–29.40***
–0.004 
–26.04***
THA
–0.122 
–17.62***
–0.124 –30.11***
–0.097 
–30.80***
–0.061 
–19.22***
–0.038 
–10.64***
FHA
0.069 
2.24** 
0.050 
2.74***
0.014 
1.69* 
–0.018 
–1.26 
–0.024 
–1.51 
THA × FHA
–0.023 
–0.40 
0.037 
1.07 
0.056 
2.10** 
0.089 
3.30***
0.045 
1.45 
Administrative districts
Yes 
Yes 
Yes 
Yes 
Yes 
Trading years × quarters
Yes 
Yes 
Yes 
Yes 
Yes 
Spatial autocorr. coeff.
0.163
18.14***
0.185 
34.70***
0.200 
48.77***
0.214 
51.82***
0.217 
46.49***
(Pseudo) R-square
0.577
0.606
0.641
0.671
0.693
Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. The number of observations is 62,140. The results of controlling dummies 
of administrative district and trading year and quarter are provided in the Appendix.


===== PAGE 10 (score 15, numeric-paren 0, table-words 5, var-words 3) =====

10
B.-M. Hsieh, C.-Y. Yang. The effects of transaction hotspots and flipping hotspots on housing prices
The spatial quantile regression allows us to assess 
the effects of THA at specific quantiles of the depend-
ent variable. Thus, we estimate the effects of THA in both 
low-priced (10th quantile) and high-priced (90th quantile) 
areas. The results consistently show negative and signifi-
cant coefficients for THA across all quantiles, indicating 
that THA remains a significant negative factor not only in 
low-priced areas but also in high-priced areas. In Table 3, 
for example, in the 0.1 q column (representing the 10th 
percentile of Taipei City housing prices), the coefficient for 
THA is –0.122 (t = –17.62; p-value < 0.01). Similarly, in the 
0.25  q column (representing the 25th percentile of Tai-
pei City housing prices), the coefficient for THA is –0.124 
(t = –30.11; p-value < 0.01). These results demonstrate that 
transaction hotspots are not synonymous with low-priced 
areas, as THA remains significantly negative in low-priced 
regions even after controlling for administrative districts 
and distance-related spatial variables.
Since this study addresses the influences of transac-
tion hotspots and flipping hotspots on housing prices, 
the coefficients of THA and FHA in different quantiles 
are depicted in Figure 3. It is interesting to observe that 
the coefficients of THA are –0.122, –0.124, –0.097, –0.061, 
and –0.038 at quantiles of 0.1 q, 0.25 q, 0.5 q, 0.75 q, and 
0.9 q, respectively.13 That is, these transaction hotspot ar-
eas have greater negative impacts on properties at lower 
price levels than at middle to high price levels. In other 
words, prices in lower-priced properties are more likely to 
be reduced where they are located in areas with a high 
concentration of housing transactions.
Furthermore, the coefficients of FHA are 0.069, 0.050, 
0.014,  –0.018, and  –0.024 at quantiles of 0.1  q, 0.25  q, 
0.5 q, 0.75 q, and 0.9 q, respectively. This indicates that 
flipping activities have greater positive effects on prop-
erties at lower price levels than at middle to high price 
levels. The lower-priced properties are more likely to be 
flipped and thus raise their prices in Taipei’s housing 
market. In addition, the coefficients in the 0.75 and 0.9 
quantiles are insignificant, indicating that the influences 
of flipping activities are insignificant and weak in higher 
priced properties. The effects of properties in both THA 
and FHA are significant on housing prices in the 0.5 and 
0.75 quantiles, respectively. The interactive effect is greater 
in the 0.75 quantile than in the 0.5 quantile. These results 
indicate that the interactive effect of THA and FHA have 
greater and significant influences on middle- to high-
priced houses in Taipei’s housing market. Simply stated, 
with a rise in housing price level, the negative effects of 
transaction hotspots and the positive effects of flipping 
hotspots on housing prices decrease. Low-priced residen-
tial properties are more sensitive to the spatial concentra-
tion of housing transactions and flipping transactions in 
the housing market of Taipei City. 
5. Conclusions
In contrast to much of the literature that focuses on the 
issue of spatial dependence in housing price research, this 
study addresses the spatial aggregation of housing trans-
actions and analyzes the effects of transaction hotspots 
and short-term flipping hotspots on housing prices by us-
ing real housing transaction data in Taipei City, Taiwan. The 
empirical results show that after controlling for the effects 
of spatial dependence and individual housing attributes, 
the impact of transaction hotspot areas on housing prices 
is significantly negative, while the impact of flipping hot-
spot areas on housing prices is significantly positive. The 
results verify that the key to driving up housing prices lies 
in flipping activities. Furthermore, the results of the spatial 
quantile regression model show that low-priced residential 
properties are more sensitive to the spatial concentration 
13	As shown in Figure 3, the trend of the coefficients of THA is 
rising. However, the coefficients of THA are negative, so the ris-
ing trend means the negative influences of transaction hotspot 
areas on housing prices are decreased.
Figure 3. The coefficients of THA, FHA, and THA × FHA in 
different quantiles


===== PAGE 11 (score 9, numeric-paren 0, table-words 0, var-words 8) =====

International Journal of Strategic Property Management, 2024, 28(1), 1–15
11
of housing transactions and flipping transactions in the 
housing market. 
Hot spot analysis has been widely used to analyze the 
spatial correlation of housing prices in the housing market, 
but less research has addressed the effect of “hotspot are-
as” of two types of housing transactions on housing prices. 
This study provides another perspective to analyze the re-
lationships between trade volumes and prices in housing 
market research. The results off two important implications 
for government. First, areas of hot transactions will not 
drive housing prices up, but hot flipping areas will. The 
government should not be overly concerned about wheth-
er the housing market is hot or not but should pay atten-
tion to short-term flipping (repeated transactions within 
2 years). In addition, the areas with low-priced houses are 
easily influenced by spatial aggregation of trading volume. 
Flipping detection and support measures should focus on 
areas with low housing prices, where they are more sen-
sitively influenced by flipping transactions. Corresponding 
policies such as selective credit controls should refer to 
transaction (flipping) hotspot areas and set the optimal 
applicable areas of administrative districts.
Future research could explore more about the poten-
tial limitations of this study. First, this paper considers the 
hotspot effect through quantitative aggregation. However, 
if there is a significant urban‒rural disparity in certain re-
gions, areas with fewer residences are less likely to ex-
perience the emergence of hotspots. For instance, spot 
A mainly comprises low-density mansions, so there are 
few transactions, while spot B has many high-rise build-
ings and many transactions. However, the turnover rate 
of spot A, i.e., the ratio between the number of transac-
tions relative to the number of housing units, is higher 
than that in spot B. It is an important issue to compare 
the nuanced difference between the quantity and the ra-
tio concept. Fortunately, the urban characteristics of Taipei 
City, which has a relatively small geographical area and 
more balanced development, may mitigate such concerns. 
Second, the endogeneity of flipping activities is also an im-
portant consideration. For example, in a search-theoretic 
model of house flipping, as proposed by Leung and Tse 
(2017), flipping activities are considered endogenous. The 
relationships among housing prices, transactions, and flip-
ping may exhibit multidirectional influences. Finally, it is 
worth noting that the transaction and flipping hotspots 
may change over time. Identifying the time-varying trans-
action and flipping hotspots could provide more practical 
implications. For future research, employing space-time 
permutation scan statistic methods (such as Kulldorff & 
Nagarwalla, 1995; Kulldorff et al., 2005) to analyze changes 
in clusters over time would be particularly suitable.
Acknowledgements 
We would like to thank the Editors and three anonymous 
referees for providing highly constructive comments. The 
authors acknowledge the funding support received from 
the Chin-Oh Chang’s Inheritance Award, Chinese Society of 
Housing Studies. Special appreciation is extended to Pro-
fessor Chin-Oh Chang for his valuable and insightful com-
ments on this paper. Chih-Yuan Yang extends thanks to the 
National Science and Technology Council of Taiwan for fi-
nancial support through grant (NSTC 112-2410-H-003-131).
References 
Anacker, K. B., & Schintler, L. A. (2015). Flip that house: Visualising 
and analysing potential real estate property flipping transac-
tions in a cold local housing market in the United States. Inter-
national Journal of Housing Policy, 15(3), 285–303. 
https://doi.org/10.1080/14616718.2015.1051401
Anselin, L. (1988). Spatial econometrics: Methods and models. Klu-
wer Academic. https://doi.org/10.1007/978-94-015-7799-1
Anselin, L. (1992). Spatial data analysis with GIS: An introduction 
to application in the social sciences. National Center for Geo-
graphic Information and Analysis.
Anselin, L., & Bera, A. K. (1998). Spatial dependence of linear re-
gression models with an introduction to spatial econometrics. 
In A. Ullah & D. E. A. Giles (Eds.), Handbook of applied economic 
statistics (pp. 237–289). Marcel Dekker.
Basu, S., & Thibodeau, T. (1998). Analysis of spatial autocorrelation 
in house prices. Journal of Real Estate Finance and Economics, 
17(1), 61–85. https://doi.org/10.1023/A:1007703229507
Bayer, P., Geissler, C., Mangum, K., & Roberts, J. W. (2020). Specu-
lators and middlemen: The strategy and performance of inves-
tors in the housing market. The Review of Financial Studies, 
33(11), 5212–5247. https://doi.org/10.1093/rfs/hhaa042
Bilal, A., & Rossi-Hansberg, E. (2021). Location as an asset. Econo-
metrica, 89(5), 2459–2495. https://doi.org/10.3982/ECTA16699 
Bø, E. E. (2018). Housing match quality and demand: What can we 
learn from comparing buyer characteristics? Journal of Housing 
Economics, 41, 184–199. 
https://doi.org/10.1016/j.jhe.2018.06.007
Bourassa, S., Cantoni, E., & Hoesli, M. (2007). Spatial dependence, 
housing submarkets, and house price prediction. Journal of 
Real Estate Finance and Economics, 35, 143–160. 
https://doi.org/10.1007/s11146-007-9036-8
Bourassa, S., Cantoni, E., & Hoesli, M. (2010). Predicting house 
prices with spatial dependence: A comparison of alternative 
methods. Journal of Real Estate Research, 32(2), 139–159. 
https://doi.org/10.1080/10835547.2010.12091276
Carrillo, P., & Pope, J. (2012). Are homes hot or cold potatoes? The 
distribution of marketing time in the housing market. Regional 
Science and Urban Economics, 42(1–2), 189–197. 
https://doi.org/10.1016/j.regsciurbeco.2011.08.010
Case, B., Clapp, J., Dubin, R., & Rodriguez, M. (2004). Modeling spa-
tial and temporal housing price patterns: A comparison of four 
models. Journal of Real Estate Finance and Economics, 29(2), 
167–191. https://doi.org/10.1023/B:REAL.0000035309.60607.53
Chakravorty, S. (1995). Identifying crime clusters: The spatial prin-
ciples. Middle States Geographer, 28, 53–58.
Chang, C. O., & Chen, S. M. (2018). Dilemma of housing demand 
in Taiwan. International Real Estate Review, 21(3), 397–418. 
https://doi.org/10.53383/100267
Chang, C. O., & Hsieh, B. M. (2018). Changes in housing policy, 
housing wellbeing and housing justice in Taiwan. In R. L. H. Chu 
& S.-K. Ha (Eds.), Housing policy, wellbeing and social develop-
ment in Asia (pp. 88–105). Routledge. 
https://doi.org/10.1201/9781315460055-6


===== PAGE 13 (score 163, numeric-paren 131, table-words 1, var-words 3) =====

International Journal of Strategic Property Management, 2024, 28(1), 1–15
13
Appendix
Table A1. Complete results of OLS model and spatial autoregression models
Model
OLS model
Spatial lag model
Spatial error model
Variable
Coeff.
t value 
Coeff.
Variable
Coeff.
Variable
Constant
16.149
1394.93***
16.184
1426.16***
16.224
1350.06***
Area
0.008
315.66***
0.008
316.60***
0.007
179.92***
Floor
–0.0004
–1.05 
–0.0001
–0.37 
–0.0008
–2.33** 
TStory
–0.0003
–1.01 
–0.0004
–1.33 
–0.0009
–2.69***
Bathroom
0.059
24.31***
0.060
25.31***
0.061
25.42***
Age
–0.011
–118.36***
–0.011
–121.44***
–0.011
–114.10***
Dist. MRT
–0.011
–40.04***
–0.011
–39.90***
–0.011
–37.84***
Dist. CC
–0.004
–32.22***
–0.003
–27.49***
–0.003
–31.16***
THA
–0.102
–34.74***
–0.099
–34.56***
–0.101
–34.42***
FHA
0.014
1.08 
0.022
1.70* 
0.020
1.55
THA × FHA
0.042
1.68* 
0.050
2.02** 
0.032
1.30
Administrative districts
Zhongzheng
0.250
37.19***
0.174
25.94***
0.232
34.38***
Datong
0.036
5.26***
0.025
3.70***
0.034
4.83***
Zhongshan
0.119
22.63***
0.077
14.82***
0.102
19.26***
Songshan
0.300
46.97***
0.253
40.03***
0.284
44.32***
Da’an
0.395
58.62***
0.275
39.49***
0.361
52.55***
Xinyi
0.287
44.97***
0.209
32.59***
0.265
41.20***
Shilin
0.298
45.58***
0.235
36.03***
0.283
43.23***
Beitou
0.149
17.49***
0.102
12.14***
0.144
16.93***
Neihu
0.193
35.74***
0.133
24.65***
0.187
34.63***
Nangan
0.251
37.74***
0.160
23.70***
0.249
37.47***
Wensha
0.032
5.86***
0.016
3.06***
0.036
6.62***
Trading years × quarters
2012Q3
–0.259
–24.86***
–0.260
–25.51***
–0.247
–23.76***
2012Q4
–0.243
–25.68***
–0.249
–26.90***
–0.231
–24.45***
2013Q1
–0.186
–18.83***
–0.190
–19.73***
–0.176
–17.95***
2013Q2
–0.130
–14.11***
–0.135
–15.00***
–0.123
–13.36***
2013Q3
–0.112
–11.56***
–0.118
–12.50***
–0.106
–11.04***
2013Q4
–0.093
–9.87***
–0.096
–10.40***
–0.087
–9.25***
2014Q1
–0.075
–7.49***
–0.077
–7.87***
–0.071
–7.09***
2014Q2
–0.131
–13.41***
–0.137
–14.37***
–0.122
–12.55***
2014Q3
–0.087
–8.60***
–0.092
–9.26***
–0.080
–7.88***
2014Q4
–0.089
–9.10***
–0.090
–9.35***
–0.086
–8.83***
2015Q1
–0.073
–6.71***
–0.074
–6.91***
–0.070
–6.46***
2015Q2
–0.059
–5.68***
–0.066
–6.54***
–0.054
–5.24***
2015Q3
–0.066
–6.16***
–0.068
–6.48***
–0.058
–5.43***
2015Q4
–0.095
–9.33***
–0.095
–9.54***
–0.087
–8.64***
2016Q1
–0.107
–8.92***
–0.109
–9.37***
–0.099
–8.36***
2016Q2
–0.064
–6.42***
–0.068
–6.93***
–0.058
–5.79***
2016Q3
–0.102
–9.38***
–0.104
–9.76***
–0.096
–8.83***
2016Q4
–0.107
–9.75***
–0.106
–9.83***
–0.100
–9.07***
2017Q1
–0.109
–9.48***
–0.107
–9.52***
–0.101
–8.81***
2017Q2
–0.093
–8.47***
–0.091
–8.50***
–0.084
–7.70***
2017Q3
–0.115
–10.39***
–0.116
–10.71***
–0.106
–9.59***
2017Q4
–0.130
–11.95***
–0.130
–12.19***
–0.121
–11.16***
2018Q1
–0.111
–9.63***
–0.108
–9.56***
–0.102
–8.93***
2018Q2
–0.123
–11.44***
–0.121
–11.55***
–0.115
–10.72***


===== PAGE 14 (score 178, numeric-paren 139, table-words 2, var-words 4) =====

14
B.-M. Hsieh, C.-Y. Yang. The effects of transaction hotspots and flipping hotspots on housing prices
Model
OLS model
Spatial lag model
Spatial error model
2018Q3
–0.104
–9.39***
–0.103
–9.47***
–0.097
–8.79***
2018Q4
–0.073
–6.65***
–0.075
–7.00***
–0.065
–5.98***
2019Q1
–0.086
–7.64***
–0.090
–8.15***
–0.079
–7.01***
2019Q2
–0.091
–8.62***
–0.092
–8.98***
–0.085
–8.10***
2019Q3
–0.082
–7.62***
–0.085
–8.03***
–0.077
–7.22***
2019Q4
–0.074
–7.03***
–0.075
–7.28***
–0.070
–6.71***
2020Q1
–0.071
–6.28***
–0.072
–6.54***
–0.065
–5.80***
2020Q2
–0.053
–5.06***
–0.054
–5.23***
–0.047
–4.50***
2020Q3
–0.036
–3.48***
–0.041
–4.13***
–0.032
–3.05***
2020Q4
–0.034
–3.28***
–0.039
–3.81***
–0.031
–2.99***
2021Q1
–0.019
–1.82**
–0.022
–2.09**
–0.018
–1.71*
2021Q2
–0.014
–1.30
–0.018
–1.65*
–0.013
–1.20
2021Q3
–0.014
–1.25
–0.017
–1.61
–0.012
–1.11
Spatial autocorr. coeff.
–
 
0.196
52.50***
0.056
22.32***
(Pseudo) R-square
0.852
0.859
0.855
F value
6215.69***
–
AIC 
14781.55
14678.08
14737.21
Robust lag
–
683.37***
Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.
Table A2. Complete results of the spatial quantile regression model
Model
Low
……………………………….. Housing prices ………………………………..
High
0.1 q
0.25 q
0.5 q
0.75 q
0.9 q
Variable
Coeff.
t value
Coeff.
t value
Coeff.
t value
Coeff.
t value
Coeff.
t value
Constant
15.916 
582.66***
16.067 
991.19***
16.155 1303.56***
16.236 1297.69***
16.334 1150.88***
Area
0.008 
131.06***
0.008 
230.96***
0.008 
305.09***
0.008 
300.55***
0.008 
265.16***
Floor
–0.002 
–2.15** 
0.001 
1.51 
0.001 
3.91***
0.002 
5.064***
0.002 
3.27***
TStory
–0.004 
–5.40***
–0.001 
–3.01*** 
0.001 
3.31***
0.003 
7.32***
0.003 
6.37***
Bathroom
0.069 
12.16***
0.061 
17.98***
0.055 
21.54***
0.054 
20.79***
0.049 
16.63***
Age
–0.012 
–55.95***
–0.012 
–88.20***
–0.011 –105.21***
–0.010 –100.58***
–0.010 
–86.66***
Dist. MRT
–0.012 
–17.80***
–0.013 
–33.05***
–0.011 
–38.23***
–0.010 
–32.26***
–0.009 
–25.01***
Dist. CC
–0.002 
–9.48***
–0.003 
–16.04***
–0.003 
–26.20***
–0.004 
–29.40***
–0.004 
–26.04***
THA
–0.122 
–17.62***
–0.124 
–30.11***
–0.097 
–30.80***
–0.061 
–19.22***
–0.038 
–10.64***
FHA
0.069 
2.24** 
0.050 
2.74***
0.014 
1.69* 
–0.018 
–1.26 
–0.024 
–1.51 
THA × FHA
–0.023 
–0.40 
0.037 
1.07 
0.056 
2.10** 
0.089 
3.30***
0.045 
1.45 
Administrative districts
Zhongzheng
0.227
14.01***
0.113
11.76***
0.130
17.66***
0.190
25.65***
0.226
26.85***
Datong
0.017
1.65* 
–0.020
–2.11** 
0.015
2.02** 
0.038
5.06***
0.054
6.34***
Zhongshan
0.010
0.83 
–0.016
–2.18** 
0.078
13.67***
0.130
22.51***
0.170
26.03***
Songshan
0.280
18.37***
0.209
23.20***
0.227
32.81***
0.251
36.07***
0.270
34.13***
Da’an
0.282
16.76***
0.233
23.41***
0.253
33.15***
0.294
38.27***
0.326
37.37***
Xinyi
0.233
15.08***
0.159
17.36***
0.189
26.95***
0.229
32.37***
0.252
31.44***
Shilin
0.236
15.03***
0.170
18.28***
0.213
29.98***
0.272
37.88***
0.308
37.82***
Beitou
0.051
2.52** 
0.013
1.08 
0.102
11.11***
0.172
18.66***
0.199
18.95***
Neihu
0.186
14.31***
0.080
10.36***
0.101
17.17***
0.141
23.62***
0.163
24.13***
Nangan
0.236
14.52***
0.129
13.31***
0.143
19.40***
0.141
18.97***
0.118
14.04***
Wensha
0.070
5.46***
–0.027
–3.49***
–0.007
–1.20 
0.017
2.85***
0.023
3.39***
End of Table A1


===== PAGE 15 (score 220, numeric-paren 179, table-words 1, var-words 1) =====

International Journal of Strategic Property Management, 2024, 28(1), 1–15
15
Model
Low
……………………………….. Housing prices ………………………………..
High
0.1 q
0.25 q
0.5 q
0.75 q
0.9 q
Trading years × quarters
2012Q3
–0.323
–13.15***
–0.269
–18.48***
–0.254
–22.80***
–0.220
–19.58***
–0.206
–16.19***
2012Q4
–0.294
–13.19***
–0.269
–20.35***
–0.243
–24.05***
–0.206
–20.19***
–0.200
–17.32***
2013Q1
–0.258
–11.09***
–0.220
–15.99***
–0.175
–16.64***
–0.155
–18.30***
–0.145
–12.03***
2013Q2
–0.185
–8.54***
–0.144
–11.17***
–0.120
–12.20***
–0.108
–10.83***
–0.108
–9.56***
2013Q3
–0.170
–7.46***
–0.133
–9.84***
–0.109
–10.61***
–0.087
–8.34***
–0.097
–8.19***
2013Q4
–0.144
–6.53***
–0.109
–8.33***
–0.086
–8.60***
–0.073
–7.16***
–0.073
–6.33***
2014Q1
–0.118
–5.01***
–0.086
–6.16***
–0.072
–6.75***
–0.061
–5.67***
–0.053
–4.30***
2014Q2
–0.325
–14.12***
–0.136
–9.93***
–0.091
–8.66***
–0.065
–6.13***
–0.059
–4.90***
2014Q3
–0.109
–4.55***
–0.103
–7.30***
–0.093
–8.63***
–0.062
–5.66***
–0.078
–6.29***
2014Q4
–0.111
–4.79***
–0.088
–6.38***
–0.068
–6.45***
–0.066
–6.26***
–0.069
–5.78***
2015Q1
–0.138
–5.37***
–0.107
–7.05***
–0.067
–11.89***
–0.041
–3.50***
–0.040
–3.06***
2015Q2
–0.098
–4.00***
–0.083
–5.71***
–0.054
–4.70***
–0.041
–3.65***
–0.058
–4.60***
2015Q3
–0.093
–3.68***
–0.076
–5.07***
–0.070
–6.36***
–0.066
–5.72***
–0.071
–5.39***
2015Q4
–0.132
–5.53***
–0.112
–7.88***
–0.103
–9.50***
–0.082
–7.44***
–0.081
–6.52***
2016Q1
–0.128
–4.54***
–0.117
–6.99***
–0.097
–7.63***
–0.096
–11.62***
–0.112
–7.68***
2016Q2
–0.096
–4.09***
–0.084
–6.00***
–0.077
–7.23***
–0.057
–5.26***
–0.057
–4.69***
2016Q3
–0.136
–5.29***
–0.125
–8.21***
–0.112
–9.65***
–0.082
–7.00***
–0.089
–6.68***
2016Q4
–0.129
–4.97***
–0.124
–8.06***
–0.116
–9.83***
–0.085
–7.17***
–0.086
–6.37***
2017Q1
–0.139
–5.12***
–0.122
–7.56***
–0.115
–9.31***
–0.105
–8.42***
–0.110
–7.82***
2017Q2
–0.109
–4.24***
–0.107
–7.02***
–0.104
–8.91***
–0.094
–7.96***
–0.084
–6.31***
2017Q3
–0.148
–5.66***
–0.120
–7.71***
–0.122
–10.32***
–0.115
–9.63***
–0.120
–8.86***
2017Q4
–0.158
–6.17***
–0.138
–9.05***
–0.135
–11.61***
–0.144
–12.26***
–0.122
–9.20***
2018Q1
–0.129
–4.78***
–0.109
–6.77***
–0.118
–9.60***
–0.111
–8.97***
–0.107
–7.58***
2018Q2
–0.153
–6.06***
–0.116
–7.75***
–0.123
–10.71***
–0.119
–10.31***
0.122
–9.32***
2018Q3
–0.114
–4.37***
–0.116
–7.46***
–0.112
–9.43***
–0.099
–8.31***
–0.089
–6.65***
2018Q4
–0.102
–3.96***
–0.093
–6.09***
–0.086
–7.34***
–0.059
–4.98***
–0.073
–5.45***
2019Q1
–0.086
–3.23***
–0.095
–6.01***
–0.097
–8.02***
–0.095
–7.83***
–0.108
–7.83***
2019Q2
–0.093
–3.73***
–0.077
–5.23***
–0.095
–8.42***
–0.098
–8.64***
–0.111
–8.66***
2019Q3
–0.100
–3.93***
–0.096
–6.35***
–0.093
–8.09***
–0.070
–6.04***
–0.087
–6.57***
2019Q4
–0.090
–3.63***
–0.070
–4.77***
–0.084
–7.50***
–0.067
–5.89***
–0.082
–6.34***
2020Q1
–0.069
–2.61***
–0.076
–4.83***
–0.077
–6.35***
–0.073
–5.96***
–0.092
–6.67***
2020Q2
–0.049
–1.97** 
–0.054
–3.72***
–0.065
–5.80***
–0.060
–5.28***
–0.070
–5.48***
2020Q3
–0.039
–1.69* 
–0.036
–2.53** 
–0.052
–4.72***
–0.051
–4.60***
–0.067
–5.33***
2020Q4
–0.036
–1.48 
–0.041
–2.88***
–0.035
–3.19***
–0.038
–3.46***
–0.047
–3.77***
2021Q1
–0.033
–1.31 
–0.020
–1.35 
–0.023
–2.04** 
–0.015
–1.34 
–0.026
–1.99** 
2021Q2
–0.029
–1.14 
–0.022
–1.43 
–0.022
–1.84* 
–0.014
–1.16 
–0.030
–2.21** 
2021Q3
–0.017
–0.67 
–0.015
–0.97 
–0.018
–1.55 
–0.016
–1.39 
–0.036
–2.67***
Spatial autocorr. coeff.
0.163
18.14***
0.185 
34.70***
0.200 
48.77***
0.214 
51.82***
0.217 
46.49***
(Pseudo) R-square
0.577
0.606
0.641
0.671
0.693
End of Table A2