The effect of agglomeration economies and geography on the survival of accommodation businesses in Sicily

ABSTRACT

The study explores the geographical pattern of the accommodation industry in the Italian insular region of Sicily, focusing on the determinants of the risk of market exit. We adopt a standard framework of business survival analysis where agglomeration economies play an important role. We then extend the analysis by considering the role of geography to explore whether the risk of market exit depends on nearness to desirable amenities. The geography is here measured by the distance from the coast and the altitude of the place where the firm is located. When we look at the entire population of accommodation firms that started between 2010 and 2014, we find evidence that the risk of failure increases for those which are over 2 km from the coast.

KEYWORDS:

• firm survival
• tourism industry
• geographical location
• insular regions

JEL:

• C41
• R12
• Z30

ACKNOWLEDGMENTS

The authors are especially grateful to Giuseppe Espa for discussing with them the research idea and encouraging them to develop it. They thank all those attending the 49th Scientific Meeting of the Italian Statistical Society, especially Giuseppe Arbia and Roberto Patuelli, for their suggestions on a previous version of the article. The authors also thank two anonymous referees and the Editor-in-chief, Paul Elhorst, for their useful and constructive comments.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

Notes

1 However, they consider an extensive definition of localization economies including inside also the related industries, that is, the diversification.

2 We referred to the following NACE codes: 5510 (hotels); 5520 (holiday homes, cottages, youth hostels, etc.); 5530 (campgrounds, recreational camps, space for recreational vehicles); and 5590 (student residences, worker hostels, rooming and boarding houses, etc.).

3 We excluded from the analysis firms located in the smaller islands off Sicily (the Eolian, Egadi and Pelagian islands). Moreover, we observe firms that started in any calendar year within the period 2010–14 and whether or not they survived until the end of the period. Therefore, the data set consists of right-censored longitudinal data. However, according to the typical approach of survival analysis, the dependent variable in this study is the survival time in each year (i.e., 1, 2, 3, etc. years), regardless of the calendar year of the company’s birth and death.

4 See Appendix A in the supplemental data online for more details.

5 Specifically, we consider the following industries: restaurants and mobile food service activities (5610); beverage service activities (5630); renting and leasing of motor vehicles (7711, 7721); travel agency and tour operator activities (7911, 7912); other reservation service and related activities (7990); creative, arts and entertainment activities (9004); libraries, archives, museums and other cultural activities (9101–9104); sport activities (9319); and amusement and recreation activities (9321, 9329).

6 Since we exploit data at establishment level, there is no ambiguity about the attribution of annual turnover. However, using data at the registered office level could generate some bias in exploring the distribution of economic activities across space. Unfortunately, we do not know if firms belong to a group (e.g., a hotel chain), thus we cannot capture those potential economies of scale.

7 In place of distance from the coast, we could use distance from the beach since some parts of the coast could simply be rocks and therefore less attractive to tourists. However, it would be difficult to find an objective measure of distance from the beach, and this could significantly affect our results. The Sicilian coast is quite accessible almost everywhere and accommodation businesses are mostly located in all these accessible areas, but at different distances from the coast. In other words, different distances correspond in our analysis to different investments by firms in terms of location choice.

8 Other categories (e.g., from 2 to 3 km and so on) were considered, but no significant results emerged. Thus, we decided to stop at 2 km since it seems to be the significant threshold of the risk of failure.

9 All the firms observed remain in the market for at least one year after their entry.

10 See Tables C1–C7 in Appendix C in the supplemental data online for details on survival probabilities estimates.

11 All statistical analyses were performed using the ‘survival’ package (Therneau, 2015) in the R statistical environment (R Core Team, 2018).

12 From a macroeconomic perspective on the US economy, Saito and Wu (2015) find that employment growth may be negatively affected by employment density, suggesting an offset effect of congestion on localization.