Has tourism driven house prices in Germany? Time-varying evidence since 1870

Abstract

Using a historical data set and recent advances in non-parametric time series modelling, we investigate the nexus between tourism flows and house prices in Germany over nearly 150 years. We use time-varying non-parametric techniques given that historical data tend to exhibit abrupt changes and other forms of non-linearities. Our findings show evidence of a time-varying effect of tourism flows on house prices, although with mixed effects. The pre-World War II time-varying estimates of tourism show both positive and negative effects on house prices. While changes in tourism flows contribute to increasing housing prices over the post-1950 period, this is short-lived, and the effect declines until the mid-1990s. However, we find a positive and significant relationship after 2000, where the impact of tourism on house prices becomes more pronounced in recent years.

Keywords

Germany house prices tourism tourist arrivals

Introduction

Tourism is one of the most important contributors to economic development in many countries. Given that tourism creates jobs and facilitates infrastructure development and tourism-related income generation activities, it has long been considered as an essential generator of economic activity (Balli and Tsui, 2016; Wu, 2019). A large body of literature has, thus, examined the relationship between tourism and economic growth (see Brida et al., 2016: for a review). To further inform policy on tourism development, a substantial body of literature has also examined the factors that influence tourism growth (see, e.g. Arana and León, 2008; Awaworyi Churchill and Nuhu, 2018; Ceron and Dubois, 2005; Crouch, 1995; Crouch, 1994; Fourie and Santana-Gallego, 2011; Naudé and Saayman, 2005; Qiu and Zhang, 1995).

Despite the benefits of tourism, concerns about the potential negative externalities of tourism, especially on house prices, have recently heightened. For instance, in some major cities around the world, there have been protests about the undesirable effects of tourism on house prices and the housing market in general (see, e.g. Daily Mail, 2018; Henley, 2020, López Díaz, 2017; Milano et al., 2018). The interest in understanding the impact of tourism on house prices has, thus, gained traction among policymakers and researchers (Biagi et al., 2012, 2015, 2016; Paramati and Roca, 2019; Wu, 2019).

What is the impact of tourism on house prices? The theoretical literature suggests that the effects of tourism on house prices could be negative or positive (Biagi et al., 2012). Empirically, relatively few studies have examined the impact of tourism on house prices, and two groups of studies mostly dominate this literature. The first group has primarily focused on the impact of tourism-related facilities on house prices in tourist destination or on prices of tourism accommodation (see, e.g. Benson et al., 1998; Conroy and Milosch, 2011; Do and Grudnitski, 1995; Fleischer and Tchetchik, 2005; Le Goffe, 2000; Rush and Bruggink, 2000). The second group of studies examines how the demand for holiday homes have influenced the house princes in tourist destinations inflation (see, e.g. Gallent and Tewdwr-Jones, 2001; Marjavaara and Müller, 2007). Very few studies examine the impact of tourism on house prices in general as opposed to prices of holiday or tourism accommodations (Biagi et al., 2015, 2016; Paramati and Roca, 2019; Wu, 2019). Although these studies provided useful evidence on the tourism-housing market dynamics, they are limited because they are often based on specific case studies that focus on a city or neighbourhood or focus on the role of specific amenities (Biagi et al., 2015). Evidence from this literature is also drawn from either cross-sectional data or panel/time series data that are based on a relatively short period. This makes it unlikely to observe the evolution of tourism over time and how it has influenced house prices.

We examine the impact of tourism on house prices in Germany using annual data covering the period 1870–2016. In doing so, we advance the literature in at least two important ways. The first is that by using a data set that spans almost 150 years, our study focuses on a period which is longer than any of the previous studies in this line of inquiry. The use of historical data is important given that it allows us to understand how the evolution of tourism over time, in a country that has strong historical roots in tourism, has influenced house prices. The advantage of using such a historical data set is that we are able to examine the time-varying nature of the relationship between tourism and house prices, thus, capturing the considerable variation in tourism growth and house prices over time. Second, unlike previous studies, our methodological approach establishes the inherent non-linearities in the variables under investigation, given that we make use of historical time series data. We apply the time-varying coefficient (TVC) model (Cai, 2007; Orbe et al., 2000, 2005, 2006; Phillips, 2001; Park and Phillips, 2001) to determine the evolution of tourism changes on house prices using the non-parametric local liner method, which has been utilised in the non-parametric literature given its valuable properties such as efficiency, adaptation of boundary effects and bias reduction (Awaworyi Churchill et al., 2019; Chen et al., 2018b; Hailemariam et al., 2019; Ivanovski et al., 2020). The TVC has the advantage of flexibility and robustness to functional form and misspecification (Cai, 2007; Orbe et al., 2000, 2005, 2006).

We situate our study in Germany, given the unique historical characteristics of German tourism and the trends in house prices. Germany is known for numerous natural and cultural resources and has a well-established infrastructure for both business and leisure tourism, which evolved since the nineteenth century (Semmens, 2005). Germany is also Europe’s largest economy and one of the region’s most competitive tourist destinations (Dustmann et al., 2014; Vismara et al., 2012). Importantly, rising house prices is a concern for policymakers in Germany. House prices in Germany have been found to deviate significantly from the Organisation for Economic Co-operation and Development (OECD) average and long-term projections (OECD, 2017). German policymakers are, thus, increasingly concerned about the implications of house prices and rent growing faster relative to income and are exploring various avenues to address the ensuing challenges of housing affordability, well-being, and inclusive growth (OECD, 2018). The focus on Germany is therefore appropriate and contributes to ongoing policy debate.

Our results suggest that the effect of tourism on house prices is time-varying. Pre-World War II, we find mixed evidence on the impact of tourism with both positive and negative effects on house prices at different times. Tourism contributed to soaring house prices in the early 1950s for a short period, after which it led to a persistent decline in house prices until the mid-1990s. We find a positive and significant relationship after 2000 with more pronounced effects of tourism on house prices in recent years.

The remainder of the article is structured as follows. The next section presents a brief overview of the literature. The third section discusses the empirical methods and data used. The fourth section presents the results from preliminary data analysis and trends. The fifth section presents the time-varying estimates while the sixth section concludes.

Brief overview of related literature

Discussions in the existing literature suggest that tourism and tourism-related activities can directly or indirectly influence housing markets.¹ The broader literature on the determinants of house prices has shown that house prices depend on location, structural characteristics, and a set of external factors including neighbourhood characteristics that influence suppliers’ and consumers’ decisions to sell and buy (Biagi et al., 2012; Cheshire and Sheppard, 1998). Thus, tourism can affect house prices either by directly influencing demand and supply of housing or by shaping structural and neighbourhood characteristics that influence the dynamics of demand and supply in the housing market.

Tourists create extra demand for housing and increase competition for land use, thus, directly influencing land and property prices (Tsui et al., 2016a). Indirectly, tourism also affects house prices via externalities emerging from tourism-related activities and economic growth, which contributes to higher standards of living and with that, higher prices for properties. Tourism tends to facilitate local infrastructure development including airport infrastructure, which could negatively or positively influence house prices (Bilotkach, 2015; Chalermpong, 2010; Chen et al., 2018a; Cidell, 2015).

Purpose-built and natural tourist attractions also tend to add value to properties within their vicinities, thus, increasing house prices (Cebula, 2009; Voicu and Been, 2008), while others negatively influence property values (Lin et al., 2013). The literature here demonstrates that willingness to pay for houses is higher in areas with purpose-built and natural attractions (Biagi et al., 2012). However, purpose-built or natural amenities may also be a source of negative social and environmental externalities that can be detrimental to neighbourhood appeal, and consequently, a decline in house prices (Biagi and Detotto, 2014; Recher and Rubil, 2020).

Empirically, the majority of the studies that have examined the tourism–house price relationship focus on the effects of tourism on prices of tourism-related accommodation. These studies tend to apply the hedonic price method to measure the impact of amenities on tourism accommodation. Studies in this tradition examine the effects of location facilities on the price of holiday cottages, hotels and tourism accommodation in general. In doing so, the bulk of studies focus on specific locations or neighbourhoods within a country or examine the role of specific tourism-related amenities (Biagi et al., 2016). The findings from this literature generally suggest that purpose-built and natural tourist attractions and amenities tend to put an upward pressure on price in most locations (Espinet et al., 2003; Fleischer and Tchetchik, 2005; Hamilton, 2007; Nelson, 2010; Taylor and Smith, 2000), while in other locations this can drive prices down (Le Goffe, 2000; Vanslembrouck et al., 2005).

A related strand of literature focuses on holiday homes. The growing demand for holiday homes by tourists has been found to influence the dynamics of local housing markets. In major tourist destinations, tourists and tourism-related workers who relocate for employment opportunities represent external demand in direct competition for land and housing with local residents (Biagi et al., 2012). Research has shown that the additional demand for holiday homes puts upward pressure on house prices and causes house price inflation (see, e.g. Gallent and Tewdwr-Jones, 2001; Marjavaara and Müller, 2007). The literature demonstrates differential effects on house prices depending on preferences for location and amenities, the length of occupancy of tourists, the type of housing demand and preferred leisure activities, among others (Gill, 2000; Muller et al., 2008; Sautter and Leisen, 1999). There is also evidence that tourism impacts housing rent (Schäfer and Hirsch, 2017) and results in a decline in housing affordability (Gu and Ryan, 2008).

Our study also relates to the literature that has examined the impact of airport activity on various aspects of the housing market (see, e.g. Chalermpong, 2010; Cohen and Coughlin, 2008; Matos et al., 2013; Pope, 2008; Tsui et al., 2016b).² Chalermpong (2010) examines the effects of airport noise on property values in Bangkok, while Cohen and Coughlin (2008) examine a similar relationship in Atlanta. Pope (2008) examines the impact of airport noise disclosure on the property values, while Matos et al. (2013) compare the effect of airport noise on property prices around seven airports in the United Kingdom. Tsui et al. (2016b) show that airport traffic volume influences house prices in New Zealand. Our study also relates to those that examine the impact of rail stations (including location and rail station activities, among others) on property values (see Debrezion et al., 2007, for a review). Our study relates to these studies, given that an increase in tourist arrivals can be linked to increased airport and rail activities. However, our focus is on tourist arrivals rather than airport noise, transport infrastructure or airport and rail station activities.

The closest studies in the literature to ours are the very few studies that have examined the impact of tourism on house prices in general as opposed to prices of holiday or tourism accommodations (Biagi et al., 2015, 2016; Paramati and Roca, 2019; Wu, 2019). Using data for a panel of Italian cities, Biagi et al. (2015) found a positive association between tourism and house prices. Biagi et al. (2016) adopted a regime-based model to examine a similar relationship and find that tourism activity positively influences house prices in some Italian cities but either decreased or had no effect in others. Using cross-country aggregate data for 20 OECD countries, Paramati and Roca (2019) have shown that tourism is positively associated with house prices.

Overall, the empirical literature on tourism and housing market dynamics are mostly microeconomic studies that have modelled tourism amenities as a function of prices of different tourism accommodations. Other studies have examined how the growing demand for holiday homes by tourists influence house prices or rents. While another group of studies, which are closely related to ours, focus on the impact of tourism on house aggregate house prices either focusing on sub-national units within a single country (see, e.g. Biagi et al., 2015, 2016) or cross-country macro indicators (Paramati and Roca, 2019). We build on these last group of studies by presenting evidence on the time-varying nature of the relationship between tourism and house prices in a macroeconomic context.

Methodology and data

TVC framework

The primary objective of this study is to estimate the time-varying impact of tourism on house prices by utilising a non-parametric model. In particular, we make use of the TVC model in which the estimated coefficients change over time. The appeal of this modelling approach is the flexibility and robustness to functional form and misspecification and continues to receive significant research interest (Cai, 2007; Orbe et al., 2000, 2005, 2006; Park and Phillips, 2001; Phillips, 2001) since its initial development by Robinson (1989).

The empirical specification is given by:

H P_{t} = x_{t}^{⊤} γ_{t} + u_{t}, t = 1, ., T,

where $H P_{t}$ is the dependent variable, real house prices and constructed as nominal house prices over the consumer price index (CPI). $x_{t} = {(x_{1 t}, x_{2 t}, \dots, x_{d t})}^{⊤}$ , where $d = 1, \dots, 6$ and contains the vector of independent variables, which includes our main variable of interest, tourist arrivals ( $T O U R$ ) and the control variables real gross domestic product (GDP) per capita ( $R G D P$ ), the interest rate ( $I R$ ), inflation ( $I N F$ ), mortgage loans-to-GDP ( $M L$ ) and non-mortgage loans-to-GDP ( $N M L$ ) ratios. $γ_{t} = {(γ_{1 t}, γ_{2 t}, \dots, γ_{d t})}^{⊤}$ is the vector of estimated coefficients and u_t is the stationary error term. Following Robinson (1989), Cai et al. (2009), Li et al. (2011) and Chen et al. (2018b), we assume that:

γ_{t, j} = γ_{j} (τ_{t}), τ_{t} = \frac{t}{n}, t = 1, \dots, n .

where n is the sample size and $γ_{j}$ is the TVC function of the independent variables to be estimated such that $j = 1, \dots 6$ . This type of specification allows for the non-parametric estimation of the coefficient functions via sampling of data points at each point in the interval [0,1]. Let $x_{t} = {(1, H P_{t - 1}, T O U R_{t}, R G D P_{t}, I R_{t}, I N F_{t}, M L_{t}, N M L_{t})}^{⊤}$ and replace the coefficient functions with those defined in equation (2), the non-parametric model is given by:

H P_{t} = γ_{0} (τ_{t}) + γ_{1} (τ_{t}) T O U R_{t} + γ_{2} (τ_{t}) R G D P_{t} + γ_{3} (τ_{t}) I R_{t} + γ_{4} (τ_{t}) I N F_{t} + γ_{5} (τ_{t}) M L_{t} + γ_{6} (τ_{t}) N M L_{t} + ε_{t} .

To estimate the coefficient functions, we use the local linear (LL) technique, which has been utilised in the non-parametric literature given its valuable properties such as efficiency, adaptation of boundary effects and bias reduction (Chen et al., 2018b). Under the assumption that each $γ_{j} (\cdot)$ has a continuous second-order derivative over the interval [0,1], $γ_{j} (τ_{t})$ is approximated by a linear function through the first-order Taylor expansion:

γ_{j} (τ_{t}) ≃ p_{j} + q_{j} (τ_{t} - τ)

where $p_{j} = γ_{j} (τ)$ and $q_{j} = γ_{j}^{'} (τ)$ , which is its first-order derivative of $γ_{j} (τ)$ . Accordingly, equation (3) can be locally approximated by:

H P_{t} {\tilde{x}}_{t}^{⊤} λ (τ) + ε_{t},

where ${\tilde{x}}_{t}^{⊤} = {[x_{t}^{⊤}, x_{t}^{⊤} (τ_{t} - τ)]}^{⊤}$ and $λ (τ) = {\{λ^{⊤} (τ), {[λ^{'} (τ)]}^{⊤}\}}^{⊤}$ . Based on equation (4), we can specify the locally weighted sum of squares by:

\sum_{t = 1}^{n} {[H P_{t} - {\tilde{x}}_{t}^{⊤} λ]}^{2} R_{h} (τ_{t} - τ),

where $R_{h} (u) = \frac{1}{h} R (\frac{u}{h})$ , $R (\cdot)$ is a kernel function and $h = h_{n} > 0$ is the bandwidth satisfying the condition that $h \to \infty$ and $n h \to \infty$ as $n \to \infty$ . Note that h determines the amount of smoothing used in the LL estimation. Consequently, $\hat{λ} (τ)$ is found by minimising equation (6) with respect to $λ$ and the expression of $\hat{λ} (τ)$ is given by:

\hat{λ} (τ) = {[\begin{matrix} L_{n 0} (τ) & L_{n 1}^{⊤} (τ) \\ L_{n 1} (τ) & L_{n 2} (τ) \end{matrix}]}^{- 1} [\begin{matrix} J_{n 0} (τ) \\ J_{n 1} (τ) \end{matrix}]

with

L_{n k} (τ) = \frac{1}{n} \sum_{t = 1}^{n} x_{t} x_{t}^{⊤} {(τ_{t} - τ)}^{k} R_{h} (τ_{t} - τ), k = 0, 1, 2,

J_{n k} (τ) = \frac{1}{n} \sum_{t = 1}^{n} x_{t} (H P_{t}) {(τ_{t} - τ)}^{k} R_{h} (τ_{t} - τ), k = 0, 1.

The LL estimate of $γ (τ) = {[γ_{0} (τ_{t}), \dots, γ_{7} (τ_{t})]}^{⊤}$ is $\hat{γ} (τ)$ which consists of the first four elements of $\hat{λ} (τ)$ , and the LL estimates of the derivative $γ^{'} (τ),$ which consists of the last four elements of $\hat{λ} (τ)$ ³. Moreover, the LL technique may be sensitive to the bandwidth, h, choice and is essential for empirical specifications. We follow Chen et al. (2018b) in selecting the cross-validation (CV) optimal bandwidth for time series applications (see Härdle and Vieu, 1992). Let ${\hat{γ}}_{- t} (τ_{t})$ denote the LL estimated value of $γ (τ_{t})$ by using the sample $\{H P_{s} : s \neq t\} .$ The optimal bandwidth, $h_{o p t}$ , is given by minimising the following CV average predictive squared-error criterion:

min_{h_{o p t}} C V (h) = \frac{1}{n} \sum_{t = 1}^{n} {[H P_{t} - x_{t}^{⊤} {\hat{γ}}_{- t} (τ_{t})]}^{2} .

Data

One of the novel features of this article is that it examines the effect of tourism on house prices in Germany over the many decades, which stretches back to the 1870s. The historical time series data is obtained from several sources. Real house prices, real GDP per capita, interest rate, inflation rate, mortgage loans-to-GDP and non-mortgage loans-to-GDP are taken from the Jordà–Schularick–Taylor Macrohistory Database for the period 1870–2016.⁴ Data on tourist arrivals from 1870 to 1994 are taken from Rahlf (2016).⁵ We update tourist arrival data for the years 1995–2016 from the World Bank’s Development Indicators database. Accordingly, our time series data set consists of annual observations for the period 1870–2016. Table A1 in the Online Appendix details the description and data sources of the variables used.

Standard empirical analyses of time series data utilise parametric techniques that impose linearity assumptions regarding the data generating process and involves satisfying the basic estimation assumptions such as normality, homoscedasticity and no autocorrelation. These assumptions of linearity often lead to using techniques that exhibit low power, despite the fact they may be non-linear. Table 1 presents the descriptive statistics of the time series variables which exhibit volatility in $H P$ and $T O U R$ (as well as in the control variables). The Jarque–Berra (JB) test rejects the null hypothesis of normality at the 5% level of significance for all variables, while the Ljung–Box Q (LB(Q)) test shows that there is evidence of autocorrelation up to the 12th order in all variables. Furthermore, Figure 1 illustrates the time series plots of $H P$ and $T O U R$ since 1870, which shows both series exhibit similar trending behaviour over time.

Table 1.

Descriptive statistics (log-level).

	$H P$	$T O U R$	$R D G P$	$I R$	$I N F$	$M L$	$N M L$
Mean	3.97	13.47	3.41	1.57	−6.41	−1.60	−1.15
Maximum	4.70	17.38	4.76	2.34	5.06	−0.64	−0.44
Minimum	0.37	7.97	2.38	−2.40	−25.18	−4.52	−3.27
Standard deviation	0.62	2.99	0.77	0.52	13.87	0.80	0.55
JB	369.81***	14.589***	15.62***	3533.01***	24.90***	96.04***	27.41***
LB(Q)	722.28***	1357.8***	1308.5***	311.70***	1409.9***	556.43***	737.37***

Note: JB: Jarque–Berra; LB(Q): Ljung–Box Q.

*** denotes a rejection of the null hypothesis of independence if the probability < 0.05.

Figure 1.

Plot of real house prices (1990 = 100) and tourist arrivals (in logs).

Preliminary analysis

Unit root and time-varying cointegration test

We begin the discussion of the empirical findings with preliminary testing procedures for our time series variables. Conventional analysis of time series data requires testing for stationarity in the variables to determine the order of integration. First, we begin with traditional unit root testing by conducting the augmented dickey–fuller (ADF) unit root test. Second, given the long-run nature of our data, global events such as the world wars and various economic shocks may have caused structural shifts in the variables. Since traditional unit root tests, such as the ADF test, are unable to capture significant structural shifts in the data series, we undertake unit root testing that accommodates for structural breaks. Clemente et al. (1998) develop a test for non-stationarity in the presence of double structural breaks in the series and based on the framework of the additive outlier (AO) and innovative outlier (IO).

Table 2 reports the unit root test results for each variable. Under the ADF test, which includes a constant and a time trend, we find that all variables are non-stationary in levels. Upon first differencing, we find that all variables are stationary (expect for $I N F$ ). However, since the data series spans almost 150 years, it is likely that there will be structural shifts. Thus, Table 2 also reports the Clemente et al. (1998) unit root test with the null hypothesis of non-stationarity and allowing for two up to structural breaks. The results indicate that the null hypothesis of non-stationarity of the series cannot be rejected at the conventional level of confidence. Therefore, all variables exhibit a unit root with at least two structural breaks, where the majority of these breaks occur during the Great Depression of the 1920s and both world wars. Also, $T O U R$ experienced two structural breaks before World War I (1909) and just after World War II (1949).

Table 2.

ADF and Clemente et al. (1998) unit root test.

	ADF test		Clemente et al. (1998) test
	Level	First differencing	Coefficient	t-stat.	TB1	TB2	Model
$H P$	−2.911	−9.914***	−0.079	−1.924	1920	1962	AO
$T O U R$	−2.317	−11.542***	−0.097	−4.718	1909	1949	IO
$R D G P$	−2.725	−8.221***	−0.038	−4.574	1931	1944	IO
$I R$	4.200	−9.631***	−0.009	−0.147	1915	2000	AO
$I N F$	−1.589	0.713	−0.033	−0.317	1920	1924	AO
$M L$	−2.992	−6.351***	−0.091	−5.229	1939	1948	IO
$N M L$	−2.956	−6.551***	−0.140	−3.003	1925	1962	AO

Note: AO: additive outlier; IO: innovative outlier; TB: trend breaks. A constant and time trend is included in the ADF unit root test, but the time trend is removed in the first difference equation. The optimal lag order is selected based on the AIC in the ADF test equation. The critical value for the Clemente et al. (1998) unit root test is −5.490 for two structural breaks.

*** Significance at the 1% level.

Given the non-stationarity of the variables, we procced with examining the cointegrating relationship between $H P$ and $T O U R$ .⁶ However, traditional cointegration testing, such as Johansen (1991), does not account for structural breaks in the time series and assumes the cointegrating vector is constant over time. As a result, we employ the Bierens and Martins (2010) cointegration technique. Under this framework of a time-varying vector error correction (VECM) model, in which the cointegrating relationship varies smoothly over time, the framework proposes a likelihood ratio statistic to test for time-invariant cointegration. Chebyshev time polynomials are employed in expressing the cointegrating vectors, and the maximum likelihood approach is employed in the estimation. The null hypothesis of time-invariant cointegration can thus be tested via the likelihood ratio test statistic.

The χ² distribution with $m k r$ degrees of freedom describes the asymptotic distribution of the test statistic, where $m$ is the degree of the Chebychev time polynomial, k is the number of series in the VECM (two in our analysis: $H P$ and $T O U R$ ) and r in the number of cointegrating vectors. The results are presented in Table 3, with the degree of Chebyshev time polynomial m, ranging from 1 to 4. The results reveal a time-varying cointegrating relationship; thus, the evidence ( $m = 1$ ) provides support on the rejection of the null hypothesis that the long-run coefficients of the series have been stable over the sample period. The trace statistics results obtained from the Bierens and Martins (2010) time-varying cointegration analysis are illustrated in Figure 2 and show that a long-run relationship between $H P$ and $T O U R$ is present.

Table 3.

Bierens and Martins (2010) test for time-varying cointegration.

Dependent variable: $H P$	Chebyshev time polynomials	Test statistic	5% CV	p-Value
$T O U R$	$m = 1$	1.14	5.99	0.564
	$m = 2$	5.24	9.49	0.263
	$m = 3$	17.00	12.59	0.009
	$m = 4$	31.86	15.51	0.000

Note: CV: critical value; $m :$ Chebyshev time polynomials. We report the results for m up to four and conclude that time variation is observed unless at least one m fails to reject the null hypothesis.

Figure 2.

Scaled trace statistic of cointegration parameters.

In sum, the findings from the Clemente et al. (1998) unit root test and Bierens and Martins (2010) time-varying cointegration test provide suggestive evidence that the relationship between $H P$ and $T O U R$ is characterised by abrupt changes and display time-varying characteristics. Consequently, non-linearities may be inherently present within the variables and thus proceed to investigate the relationships between $H P$ and $T O U R$ non-linearly.

Non-linear analysis

Given the findings from the previous section, we test for the existence of a non-linear relationship. Since we observe the presence of time breaks in the variables, we apply the Broock et al. (1996) BDS test. The results from the BDS test are reported in Table 4 and confirm the variables exhibit non-linear dependencies. Based on the test statistics (across the embedded dimensions), the null hypothesis of independent and identically distributed residuals is rejected. Accordingly, this result strongly confirms non-linearity in the time series variables.

Table 4.

BDS non-linearity test.

	Embedding dimension = M
	$M = 2$	$M = 3$	$M = 4$	$M = 5$	$M = 6$
$H P$	0.180***	0.302***	0.383***	0.434***	0.465***
$T O U R$	0.193***	0.325***	0.415***	0.473***	0.511***
$R D G P$	0.195***	0.328***	0.418***	0.479***	0.520***
$I R$	−0.008***	−0.019***	−0.019***	−0.019***	−0.019**
$I N F$	0.207***	0.350***	0.449***	0.516***	0.562***
$M L$	0.179***	0.298***	0.374***	0.420***	0.446***
$N M L$	0.181***	0.302***	0.383***	0.437***	0.469***

Note: BDS: Brock, Dechert and Scheinkman.

* The rejection of the residuals null at 10%.

** The rejection of the residuals null at 5%.

*** The rejection of the residuals null at 1%.

Next, we apply the Kapetanios et al. (2003) (KSS2003) and Kapetanios and Shin (2008) (KSS2008) non-linear unit root to determine the integration of the variables. The detection of the existence of non-stationarity against non-linear processes is a distinct characteristic of these tests. Given the long span of the data, the use of the non-linear unit root tests is desirable, since the data is characterised by the existence of structural changes and time-varying volatility. The lag length is based on fixed lag and AIC conditions.

Table 5 reports the findings from the KSS2003 and KSS2008 non-linear unit root tests and reveals that the null hypothesis of a unit root in the series is rejected for $H P$ , $M L$ and $N M L$ . The evidence suggests that these variables follow a stationary process. In contrast, the results for $T O U R$ , $R G D P$ , $I R$ and $I N F$ suggest that these variables follow a non-stationary process. Thus, mixed evidence exists with regards to the statistical properties of the series over time. While some variables exhibit non-stationarity, others follow a stationary non-linear exponential smooth transition autoregressive (ESTAR) process.

Table 5.

Kapetanios et al. (2003) and Kapetanios and Shin (2008) non-linear unit root test.

	KSS2003		KSS2008
$H P$	−8.604***	−8.731***	−8.807***	−8.934***
$T O U R$	−1.099	−1.067	0.862	0.715
$R D G P$	0.076	−0.074	0.797	0.635
$I R$	10.055	14.376	8.315	9.272
$I N F$	−0.895	−0.898	−0.063	−0.022
$M L$	−9.126***	−9.126***	−9.784***	−9.784***
$N M L$	−6.457***	−6.845***	−5.734***	−6.165***
Criteria	Fixed	AIC	Fixed	AIC

Note: The null hypothesis stipulates that the series is non-stationary. The alternative hypothesis states that a series follows a stationary non-linear ESTAR model. The lag length selection method is shown in the criteria column.

*** Significance at the 1% level.

Based on the mixed evidence on stationarity, we consider whether a cointegrating relationship is present using the non-linear autoregressive distributed lag (NARDL) model. Note that the NARDL model may also be applied regardless of whether the variables are (non-linearly) stationary, non-stationary, or a combination of both. The NARDL technique is also useful as it simultaneously models asymmetries in both the pattern of dynamic adjustment and the long-run relationship. Thus, the negative and positive long-run relationships between the variables of interest are disentangled, and sudden shocks in the series are captured.

The results from the NARDL model are reported in Table 6. Based on the F-test, the results reveal the presence of a long-run asymmetric relationship between $H P$ and $T O U R$ . This finding implies that there are differences in the effect of positive and negative changes in $T O U R$ . The approach also reveals a significant and negative long-run effect such that $H P$ decreases with a decrease in $T O U R$ . The long-run positive effect is statistically insignificant, but the magnitude of the coefficients confirms that $H P$ falls more with negative changes in $T O U R$ . Also, the long-run asymmetric relationship is observed for all other variables except for $N M L$ , which has a more direct relationship with $H P$ . The long-run positive effects are significant for $R G D P$ , $I R$ and $M L$ while the long-run negative effects are significant at 5% level of confidence.

Table 6.

Shin et al. (2014) non-linear ARDL cointegration test.

	Long-run effect [ $+$ ]			Long-run effect [ $-$ ]
	Coefficient	F-stat.	$P > F$	Coefficient	F-stat.	$P > F$
$T O U R$	−0.157	2.508	0.126	−0.489	5.894	0.023
$R D G P$	3.591	18.36	0.000	0.186	0.029	0.865
$I R$	−2.39	11.27	0.003	0.448	1.12	0.300
$I N F$	0.137	1.709	0.203	−2.886	3.77	0.064
$M L$	−0.572	4.559	0.043	−0.933	1.642	0.212
$N M L$	1.97	3.163	0.087	−0.351	0.134	0.717
	Long-run asymmetry		Short-run asymmetry
	F-stat.	$P > F$	F-stat.	$P > F$
$T O U R$	10.74	0.003	6.011	0.022
$R D G P$	9.741	0.005	6.287	0.019
$I R$	25.26	0.000	6.581	0.017
$I N F$	3.147	0.088	4.775	0.038
$M L$	4.242	0.050	5.341	0.029
$N M L$	0.982	0.331	8.066	0.009

Note: NARDL: non-linear autoregressive distributed lag. NARDL technique assumes that changes in the $H P$ can be obtained by decomposing positive and negative unit changes in $T O U R$ and control variables around a threshold of zero.

Since the results provide evidence for an asymmetric relationship, it becomes necessary to provide more precise estimates of how these changes in $T O U R$ induce different movements in $H P$ .

LL TVC results

The findings from the previous section provide evidence of a time-varying relationship and that non-linearities are inherent within the variables. As a result, we now turn to our main analysis by estimating the effects of $T O U R$ on $H P$ using the TVC specification coupled with the non-parametric LL estimation method, as described in the third section.

The TVC LL estimates are presented in Figure 3 and accompanied by their confidence intervals (90%). In relation to the main variable of interest, we find that $T O U R$ has a time-varying effect on HP. In particular, the results show that the TVC of $T O U R$ has a positive and negative impact over the sample period. Intuitively, a negative (positive) time-varying relationship indicates that a decline (rise) in tourism is associated with a decline (rise) in real house prices. For instance, from 1870 to 1920, $T O U R$ and $H P$ are negatively related, albeit statistically insignificant. However, the negative relationship becomes more pronounced and significant from 1920 to 1935. The $H P$ elasticity of $T O U R$ reaches its lowest point across the entire sample period, where the estimated coefficient shows that a 1% increase in $T O U R$ is associated with 1.2% decreases in $H P$ . This finding may reflect the hyperinflation crisis that occurred in Weimar, Germany, in 1923 and continued until the Great Depression of late 1920s. During this period, the exchange rate skyrocketed, and German corporations found it impossible to do business or trade abroad. This also affected the middle class who relied on investments, savings or incomes from pensions and rents. Besides, this hyperinflation period significantly impacted housing prices (Silverman, 1970), since renters gained at the expense of property owners because rent ceilings did not keep pace with the general level of prices.

Figure 3.

LL TVC estimates (levels). Notes: Dark grey lines denote the 90% confidence interval levels. Black line denotes TVC estimates. TVC: time-varying coefficient; LL: local linear.

From 1935 to 1940, the TVC of $T O U R$ increases, although remaining negatively related to $H P$ . Spode (1997) found that less than 15% of German adults took annual vacation trips prior to the outbreak of WWII, and international visitors also significantly declined. The relationship over the following decade (i.e. between 1940 and 1950) is indistinguishable from zero where $T O U R$ does not affect $H P$ – again reflecting the turbulent years of World War II and early years following the post-WWII recovery. Also, the German currency reform, which took place towards the end of the 1940s, impacted on tourism flows and caused severe food rationing, transport bottlenecks and a lack of domestic economic incentives to rent rooms to tourists (Kopper, 2009).

However, the post-1950 period witnessed a surge in the TVC of $T O U R$ on $H P$ which may reflect the boom in economic recovery of the time period. However, the positive effect of $T O U R$ on $H P$ is short-lived and once again begins to decline between 1960 and reaches another low point in the mid-1990s. Nevertheless, it once again begins to increase, and after 2000, an increasing impact of $T O U R$ is observed. This effect becomes more pronounced in recent years and is statistically significant.

Regarding the control variables, for most of the periods under investigation, $R G D P$ is rising with rising $H P,$ and the effect is statistically significant. The impact of $I R$ is negative and statistically significant over the early parts of the sample period, although it remains insignificant and mostly positive across the many decades. $I N F$ has a mixed effect over the period of study with positive and negative TVC values. Both $M L$ and $N M L$ are statistically insignificant in determining $H P$ for the period of study. Similar results are obtained when the growth rate of the variables is used (see Figure A1 in the Online Appendix).

Conclusion

This study examined the impact of tourism flows on house prices in Germany since the first globalisation boom of the 1870s. We move beyond traditional parametric techniques and adopt non-parametric methods that track how the tourism–house prices relationship has evolved over time. Our main analysis considers a time-varying non-parametric framework which is more suitable for modelling a dynamic relationship between tourism flows and house prices due to structural breaks caused by global shocks such as the world wars and financial crises, changes in policies, reforms, regulations as well as continued financial development over the many decades.

The initial testing procedures indicate our variables are influenced by abrupt changes due to various economic shocks over the decades and display time-varying characteristics. Based on this evidence, we proceed to determine the non-linear behaviour in the data. We find evidence of an asymmetric relationship between tourism flows and house prices, where negative long-run effects of tourism flows dominate the positive long-run effects for house prices. The main results show a time-varying relationship between tourism and house prices. Tourism has a negative impact on the TVCs on house prices over the early part of the sample period. This negative relationship peaks in the build up to World War II before stabilising during this war period. However, the post-1950 period witnessed a positive relationship between tourism and house prices, albeit for a brief period, after which the effect reverts to a negative relationship up until the mid-1990s. Finally, a second positive relationship if identified between tourism and house prices after 2000 and becomes more pronounced in recent years.

Understanding the dynamic relationship between tourism and house prices is of importance to policymakers. The time-varying relationship between tourism and house prices suggests that other underlying factors mediate the relationship between tourism and house prices and can explain the differential effects over time. Thus, policymakers need to understand the mechanisms through which tourism transmits to house prices to formulate relevant policies. The strong positive effect of tourism on house prices in recent years (i.e. post-2000) could be linked with other externalities, causing an increase in house prices together with tourism growth. While tourism contributes to domestic economic growth, employment and service export (OECD, 2016), many OECD nations, such as Germany, face significant issues of housing affordability coupled with greater tourist flows each year. Policymakers need to find a balance between maintaining growth in the tourism industry and develop regulatory frameworks surrounding real estate properties and accommodation facilities for tourists.

While empirically exploring the mechanisms or channels through which tourism transmits to house prices is beyond the scope of this study, future research can explore these dynamics. There is a general lack of empirical studies on the potential channels through which tourism influences house prices, although these channels can provide meaningful direction for policy. Future research could also focus on other major tourist destination countries in assessing how tourism flows affect house prices or even consider a panel of countries using non-parametric techniques to investigate the time-varying relationship. Several studies have also shown that income inequality may be an important factor that affects tourism–house price nexus (Alam and Paramati, 2016; Paramati and Roca, 2019; Raza and Shah, 2017). Thus, future studies may consider the moderating role of income inequality and tourism on house prices using time-varying non-parametric techniques. Furthermore, while this study considers the effect of tourism on house prices at the country-level, future studies may also consider a within-country (or territorial-level) analysis. An advantage of this type of analysis is that it can reveal the dynamic impact according to the type of region as well as the type of tourism affecting house prices. This, however, does depend on data availability at the regional level.

Supplemental material

Supplemental Material, sj-docx-1-teu-10.1177_13548166211008832 - Has tourism driven house prices in Germany? Time-varying evidence since 1870

Supplemental Material, sj-docx-1-teu-10.1177_13548166211008832 for Has tourism driven house prices in Germany? Time-varying evidence since 1870 by Sefa Awaworyi Churchill, John Inekwe and Kris Ivanovski in Tourism Economics

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Sefa Awaworyi Churchill

Supplemental material

Supplemental material for this article is available online.

Notes

References

Alam

Paramati

(2016) The impact of tourism on income inequality in developing economies: does Kuznets curve hypothesis exist? Annals of Tourism Research 61: 111–126.

Arana

León

(2008) The impact of terrorism on tourism demand. Annals of Tourism Research 35(2): 299–315.

Awaworyi Churchill

Nuhu

(2018) The impact of ethnic diversity on tourism revenue and tourist arrivals. Tourism Economics 24(6): 753–761.

Awaworyi Churchill

Inekwe

Smyth

, et al. (2019) R&D intensity and carbon emissions in the G7: 1870–2014. Energy Economics 80: 30–37.

Balli

Tsui

WHK

(2016) Tourism demand spillovers between Australia and New Zealand: evidence from the partner countries. Journal of Travel Research 55(6): 804–812.

Benson

Hansen

Schwartz

, et al. (1998) Pricing residential amenities: the value of a view. The Journal of Real Estate Finance and Economics 16(1): 55–73.

Biagi

Detotto

(2014) Crime as tourism externality. Regional Studies 48(4): 693–709.

Biagi

Brandano

Caudill

(2016) Tourism and house prices in Italy: a latent class approach. Tourism Economics 22(5): 964–978.

Biagi

Brandano

Lambiri

(2015) Does tourism affect house prices? Evidence from Italy. Growth and Change 46(3): 501–528.

10.

Biagi

Lambiri

Faggian

(2012) The effect of tourism on the housing market. In: Uysal

Perdue

Sirgy

(eds) Handbook of tourism and quality-of-life research: enhancing the lives of tourists and residents of host communities. Dordrecht: Springer Netherlands, pp. 635–652.

11.

Bierens

Martins

(2010) Time-varying cointegration. Econometric Theory 26: 1453–1490.

12.

Bilotkach

(2015) Are airports engines of economic development? A dynamic panel data approach. Urban Studies 52(9): 1577–1593.

13.

Brida

Cortes-Jimenez

Pulina

(2016) Has the tourism-led growth hypothesis been validated? A literature review. Current Issues in Tourism 19(5): 394–430.

14.

Broock

Scheinkman

Dechert

, et al. (1996) A test for independence based on the correlation dimension. Econometric Reviews 15(3): 197–235.

15.

Cai

(2007) Trending time-varying coefficient time series models with serially correlated errors. Journal of Econometrics 136(1): 163–188.

16.

Cai

Park

(2009) Functional-coefficient models for nonstationary time series data. Journal of Econometrics 148(2): 101–113.

17.

Cebula

(2009) The hedonic pricing model applied to the housing market of the city of Savannah and its Savannah historic landmark district. Review of Regional Studies 39(1): 9–22.

18.

Ceron

J-P

Dubois

(2005) The potential impacts of climate change on French tourism. Current Issues in Tourism 8(2-3): 125–139.

19.

Chalermpong

(2010) Impact of Airport noise on property values: case of Suvarnabhumi international airport, Bangkok, Thailand. Transportation Research Record 2177(1): 8–16.

20.

Chen

Chi

(2018a) Do airports boost economic development by attracting talent? An empirical investigation at the subcounty level. Social Science Quarterly 99(1): 313–329.

21.

Chen

Gao

, et al. (2018b) Nonparametric estimation and forecasting for time-varying coefficient realized volatility models. Journal of Business & Economic Statistics 36(1): 88–100.

22.

Cheshire

Sheppard

(1998) Estimating the demand for housing, land, and neighbourhood characteristics. Oxford Bulletin of Economics and Statistics 60(3): 357–382.

23.

Cidell

(2015) The role of major infrastructure in subregional economic development: an empirical study of airports and cities. Journal of Economic Geography 15(6): 1125–1144.

24.

Clemente

Montañés

Reyes

(1998) Testing for a unit root in variables with a double change in the mean. Economics Letters 59(2): 175–182.

25.

Cohen

Coughlin

(2008) Spatial hedonic models of airport noise, proximity, and housing prices. Journal of Regional Science 48(5): 859–878.

26.

Conroy

Milosch

(2011) An estimation of the coastal premium for residential housing prices in San Diego County. The Journal of Real Estate Finance and Economics 42(2): 211–228.

27.

Crouch

(1994) The study of international tourism demand: a review of findings. Journal of Travel Research 33(1): 12–23.

28.

Crouch

(1995) A meta-analysis of tourism demand. Annals of Tourism Research 22(1): 103–118.

29.

Daily Mail (2018) Activists hijack tourist buses in Barcelona and Valencia in protest against mass tourism. Daily Mail. https://www.dailymail.co.uk/news/article-5933575/Activists-hijack-bus-Barcelona-protest-against-mass-tourism.html

30.

Debrezion

Pels

Rietveld

(2007) The impact of railway stations on residential and commercial property value: a meta-analysis. The Journal of Real Estate Finance and Economics 35(2): 161–180.

31.

Grudnitski

(1995) Golf courses and residential house prices: an empirical examination. The Journal of Real Estate Finance and Economics 10(3): 261–270.

32.

Dustmann

Fitzenberger

Schönberg

, et al. (2014) From sick man of Europe to economic superstar: Germany’s resurgent economy. Journal of Economic Perspectives 28(1): 167–188.

33.

Espinet

Saez

Coenders

, et al. (2003) Effect on prices of the attributes of holiday hotels: a hedonic prices approach. Tourism Economics 9(2): 165–177.

34.

Fleischer

Tchetchik

(2005) Does rural tourism benefit from agriculture? Tourism Management 26(4): 493–501.

35.

Fourie

Santana-Gallego

(2011) The impact of mega-sport events on tourist arrivals. Tourism Management 32(6): 1364–1370.

36.

Gallent

Tewdwr-Jones

(2001) Second homes and the UK planning system. Planning Practice and Research 16(1): 59–69.

37.

Gill

(2000) From growth machine to growth management: the dynamics of resort development in Whistler, British Columbia. Environment and Planning A 32(6): 1083–1103.

38.

Ryan

(2008) Place attachment, identity and community impacts of tourism—the case of a Beijing hutong. Tourism Management 29(4): 637–647.

39.

Hailemariam

Smyth

Zhang

(2019) Oil prices and economic policy uncertainty: evidence from a nonparametric panel data model. Energy Economics 83: 40–51.

40.

Hamilton

(2007) Coastal landscape and the hedonic price of accommodation. Ecological Economics 62(3–4): 594–602.

41.

Härdle

Vieu

(1992) Kernel regression smoothing of time series. Journal of Time Series Analysis 13(3): 209–232.

42.

Henley

(2020) Overtourism in Europe’s Historic Cities Sparks Backlash. Surry Hills: The Guardian.

43.

Ivanovski

Hailemariam

Smyth

(2020) The effect of renewable and non-renewable energy consumption on economic growth: non-parametric evidence. Journal of Cleaner Production 286: 124956.

44.

Johansen

(1991) Estimation and hypothesis testing of cointegration vectors in Gaussian vector autoregressive models. Econometrica: Journal of the Econometric Society 59: 1551–1580.

45.

Jordà

Knoll

Kuvshinov

, et al. (2019) The rate of return on everything, 1870–2015. The Quarterly Journal of Economics 134(3): 1225–1298.

46.

Jordà

Schularick

Taylor

(2017) Macrofinancial history and the new business cycle facts. NBER Macroeconomics Annual 31(1): 213–263.

47.

Kapetanios

Shin

(2008) GLS detrending-based unit root tests in nonlinear STAR and SETAR models. Economics Letters 100(3): 377–380.

48.

Kapetanios

Shin

Snell

(2003) Testing for a unit root in the nonlinear STAR framework. Journal of Econometrics 112(2): 359–379.

49.

Kopper

(2009) The breakthrough of the package tour in Germany after 1945. Journal of Tourism History 1(1): 67–92.

50.

Le Goffe

(2000) Hedonic pricing of agriculture and forestry externalities. Environmental and Resource Economics 15(4): 397–401.

51.

Chen

Gao

(2011) Non-parametric time-varying coefficient panel data models with fixed effects. The Econometrics Journal 14(3): 387–408.

52.

Lin

De Sousa

(2013) Examining the economic impact of park facilities on neighboring residential property values. Applied Geography 45: 322–331.

53.

López Díaz

(2017) Why Barcelona Locals Really Hate Tourists. London: The Independent.

54.

Marjavaara

Müller

(2007) The development of second homes’ assessed property values in Sweden 1991–2001. Scandinavian Journal of Hospitality and Tourism 7(3): 202–222.

55.

Matos

JCB

Flindell

Le Masurier

, et al. (2013) A comparison of hedonic price and stated preference methods to derive monetary values for aircraft noise disturbance and annoyance. Transportation Research Part D: Transport and Environment 20: 40–47.

56.

Milano

Cheer

Novelli

(2018) Overtourism: A Growing Global Problem. Parkville: The conversation.

57.

Muller

Yin

Kim

, et al. (2008) The dynamics of land development in resort communities: a multiagent simulation of growth regimes and housing choice. Environment and Planning A 40(7): 1728–1743.

58.

Naudé

Saayman

(2005) Determinants of tourist arrivals in Africa: a panel data regression analysis. Tourism Economics 11(3): 365–391.

59.

Nelson

(2004) Meta-analysis of airport noise and hedonic property values. Journal of Transport Economics and Policy 38(1): 1–27.

60.

Nelson

(2010) Valuing rural recreation amenities: hedonic prices for vacation rental houses at Deep Creek Lake, Maryland. Agricultural and Resource Economics Review 39(1203-2016-95442): 485–504.

61.

OECD (2016) OECD Tourism Trends and Policies 2016. Paris: OECD.

62.

OECD (2017) Focus on House Prices. Paris: OECD.

63.

OECD (2018) Housing: Germany Policy Brief. Paris: OECD.

64.

Orbe

Ferreira

Rodríguez-póo

(2000) A nonparametric method to estimate time varying coefficients under seasonal constraints. Journal of Nonparametric Statistics 12(6): 779–806.

65.

Orbe

Ferreira

Rodriguez-Poo

(2005) Nonparametric estimation of time varying parameters under shape restrictions. Journal of Econometrics 126(1): 53–77.

66.

Orbe

Ferreira

Rodriguez-Poo

(2006) On the estimation and testing of time varying constraints in econometric models. Statistica Sinica 16: 1313–1333.

67.

Paramati

Roca

(2019) Does tourism drive house prices in the OECD economies? Evidence from augmented mean group estimator. Tourism Management 74: 392–395.

68.

Park

Phillips

(2001) Nonlinear regressions with integrated time series. Econometrica 69(1): 117–161.

69.

Phillips

(2001) Trending time series and macroeconomic activity: some present and future challenges. Journal of Econometrics 100(1): 21–27.

70.

Pope

(2008) Buyer information and the hedonic: the impact of a seller disclosure on the implicit price for airport noise. Journal of Urban Economics 63(2): 498–516.

71.

Qiu

Zhang

(1995) Determinants of tourist arrivals and expenditures in Canada. Journal of Travel Research 34(2): 43–49.

72.

Rahlf

(2016) The German time series dataset, 1834–2012. Journal of Economics and Statistics 236(1): 129–143.

73.

Raza

Shah

(2017) Tourism growth and income inequality: does Kuznets Curve hypothesis exist in top tourist arrival countries. Asia Pacific Journal of Tourism Research 22(8): 874–884.

74.

Recher

Rubil

(2020) More tourism, more crime: evidence from Croatia. Social Indicators Research 147(2): 651–675.

75.

Robinson

(1989) Nonparametric estimation of time-varying parameters. Statistical Analysis and Forecasting of Economic Structural Change. Berlin: Springer, pp. 253–264.

76.

Rush

Bruggink

(2000) The value of ocean proximity on barrier island houses. The Appraisal Journal 68(2): 142–150.

77.

Sautter

Leisen

(1999) Managing stakeholders a tourism planning model. Annals of Tourism Research 26(2): 312–328.

78.

Schäfer

Hirsch

(2017) Do urban tourism hotspots affect Berlin housing rents? International Journal of Housing Markets and Analysis 10(2): 231–255.

79.

Semmens

(2005) Seeing Hitler’s Germany: Tourism in the Third Reich. Basingstoke: Palgrave Macmillan.

80.

Shin

Greenwood-Nimmo

(2014) Modelling asymmetric cointegration and dynamic multipliers in a nonlinear ARDL framework. Festschrift in honor of peter Schmidt. New York, NY: Springer, pp. 281–314.

81.

Silverman

(1970) A pledge unredeemed: the housing crisis in Weimar Germany. Central European History 3(1/2): 112–139.

82.

Taylor

Smith

(2000) Environmental amenities as a source of market power. Land Economics 76(4): 550–568.

83.

Tsui

KWH

Tan

Chow

CKW

(2016a) Will air transport or tourism affect urban property prices? The case of regional cities in New Zealand. In: The 57th Annual Conference of New Zealand Association of Economists. Auckland, New Zealand, 29 Jun 2016, pp. 1–21.

84.

Tsui

WHK

Tan

DTW

Shi

(2016b) Impacts of airport traffic volumes on house prices of New Zealand’s major regions: a panel data approach. Urban Studies 54(12): 2800–2817.

85.

Vanslembrouck

Van Huylenbroeck

Van Meensel

(2005) Impact of agriculture on rural tourism: a hedonic pricing approach. Journal of Agricultural Economics 56(1): 17–30.

86.

Vismara

Paleari

Ritter

(2012) Europe’s second markets for small companies. European Financial Management 18(3): 352–388.

87.

Voicu

Been

(2008) The effect of community gardens on neighboring property values. Real Estate Economics 36(2): 241–283.

88.

T-P

(2019) Does tourism affect house prices? Evidence from china. Journal of China Tourism Research 15(4): 490–502.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.07 MB