![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
Abstract
The use of a second home may result in different outcomes for households, ranging from financial profit and holiday use to uses that are clearly unprofitable. We contribute to the literature on second homes by exploring the case of second homes that are not let out, representing the least profitable outcome from an economic viewpoint. The empirical investigation relies on the 2002–2012 Bank of Italy Survey on Household Income and Wealth (SHIW), which also provides extensive information on property, including the actual use. Our results highlight: a gender gap, whereby the unprofitable use of second homes tends to be more clearly associated with male decision-makers; a lack of association with the economic characteristics of the household; and a strong association with the specific characteristics of the property, with inherited property more likely to be used unprofitably. In addition to casting doubt on the effectiveness of second homes as an investment vehicle, our results may have important policy or regulatory implications for housing and rental markets.
Acknowledgements
The authors thank for helpful comments and suggestions the Editor, the referees, and participants at MAF International Conference (Vietri, 2014), Modena-Netspar Conference on Advances in Household Finance (Modena, 2015) and Benjamin Guin for his extremely useful discussion. Usual disclaimers apply.
Funding
Costanza Torricelli acknowledges financial support from the University of Modena and Reggio Emilia [grant number E96D14000720005].
Notes
1 As recently pointed out by Muller (Citation2011, p. 137) in a review of second home research “Certainly, second homes are difficult to define, and the terminology is contested”.
2 Sierminska & Doorley (Citation2013) show that on average Italian households have a propensity to own investment property (i.e. other than the main residence) second only to Spanish households (higher than the US, Canada and Germany, and similar to Luxembourg).
3 The affluence explanation is investigated in detail in a multi-country perspective in Paris (Citation2011).
4 They thus find evidence of the compensation hypothesis, which Dijst et al. (Citation2005) previously found also for the Netherlands but not for Germany.
5 For more details on the SHIW see http://www.bancaditalia.it/statistiche/indcamp/bilfait.
6 In the final sample more than 80 per cent of the additional properties are (at least in part) legally owned by the head of the household. Hence, also considering the demographic characteristics of the legal owner of the second home rather than of the head of the household the empirical evidence obtained remains largely unchanged (results available on request).
7 We also disregard observations in which the household declares a negative consumption (6 observations) or in which the head of the household is under 18 years old or older than 90 (307 observations). Additionally, we disregard observations in which the household owns the second home but not the primary home (1987 observations), corresponding to 4.18 per cent of the original sample. In their study on China, Huang & Yi (Citation2010) report that 5.1 per cent of their sample consists of households renting their primary dwelling and owning additional homes. As a robustness check, we also run the same analyses including these observations, obtaining similar results: see Section 5.
8 Usufruct, which is typical of Roman-based legal systems, is the temporary right (normally until the death of the usufructuary beneficiary) to the use and enjoyment of the property of another, without changing the character of the property.
9 E.g. for the US, an estimate of the average cost of a second home is around 700$ per month, not including mortgage, repairs or improvements (http://realestatescorecard.com/library/story/cost-living/cost-owning-second-or-vacation-home, http://www.wsj.com/articles/SB10001424127887323463704578495193247679474).
10 Similar evidence is found when the empirical strategy relies on a multinomial logit model where the final use of the second home is modelled as a categorical variable taking four unordered values, namely: Holiday, Profitable (when rented or used for work), Unprofitable (consistent with our main definition, when unrented) and Other. However, both the Hausman and the Small-Hsiao tests provide evidence in favour of the Independence of Irrelevant Alternatives (IIA) assumption, required by the multinomial model, only for three cases out of four. Although Kropko (Citation2011) concludes that the IIA should not be a major concern when using multinomial logit, since it “provides […] accurate point estimates […] even when the IIA assumption is severely violated”, we prefer the probit specification here and use the multinomial model as a robustness check.
11 Due to data availability, we are not able to distinguish between mortgages on primary homes and second homes.
12 All monetary amounts are expressed in real terms using the 2012 Consumer Price Index provided by ISTAT.
13 Risky financial assets include stocks, corporate bonds and foreign assets in line with other studies for Italy (e.g. Brunetti & Torricelli, Citation2010, p. 485).
14 In the remaining cases, the share of property of the households is generally 50–50.
15 This is in line with Choi et al. (Citation2014) who inter alia report that second homes are the ultimate local-biased investment.
16 The use of sample weights is suggested in order to take into account the possible non- and under-reporting in the survey data (see e.g. Cannari & D’Alessio, Citation1993; Cannari & Faiella, Citation2008; Faiella & Gambacorta, Citation2007).
17 Our results are robust to clustering the standard errors at fewer clusters, namely at the regional level. In deciding the level of clusters, we follow Cameron & Miller (Citation2015), who suggest determining the level of clusters based on:( i) the degree of correlation between the regressors within the cluster; and (ii) the number of clusters, that should not be “too few”, whereby “depending on the situation, “few” may range from less than 20 to less than 50 clusters”. Since our unit of observation is the second home, for all households owning more than one second home, the household-level regressors (e.g. all the socio-economic controls and the portfolio controls) are perfectly correlated. Furthermore, Italy has 20 regions, so that clustering at the regional level leads to the minimum number of clusters allowed according to Cameron & Miller’s (Citation2015) guidelines. For the above reasons, we opt for clustering at the household level in our main specification.
18 Alternative specifications for age (age-class dummies rather than age in linear and quadratic terms) and for income and wealth (in quartile dummies rather than in logs) do not change the results (available upon request).
19 We also try a model specification including the interaction term between the inherited dummy and gender, the only significant control among the demographic and economic controls. However, no statistical significance was found, ruling out the presence of a further gender difference in the management of inherited second homes.
20 In order to test the robustness of our estimates to the inclusion of year fixed effects, we also run all specifications without time fixed effects, with basically unchanged results (available on request).