The effect of tourism on crime in Italy: A dynamic panel approach

Biagi, Bianca; Brandano, Maria Giovanna; Detotto, Claudio Working Paper The effect of tourism on crime in Italy: A dynamic panel approach Economics Discussion Papers, No. 2012-4 Provided in Cooperation with: Kiel Institute for the World Economy – Leibniz Center for Research on Global Economic Challenges Suggested Citation: Biagi, Bianca; Brandano, Maria Giovanna; Detotto, Claudio (2012) : The effect of tourism on crime in Italy: A dynamic panel approach, Economics Discussion Papers, No. 2012-4, Kiel Institute for the World Economy (IfW), Kiel This Version is available at: https://hdl.handle.net/10419/54945 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by-nc/2.0/de/deed.en Discussion Paper No. 2012-4 | January 13, 2012 | http://www.economics-ejournal.org/economics/discussionpapers/2012-4 The Effect of Tourism on Crime in Italy: A Dynamic Panel Approach Bianca Biagi, Maria Giovanna Brandano, and Claudio Detotto University of Sassari and CRENoS Abstract The purpose of this paper is to demonstrate that, all else being equal, for the case of Italy, tourist areas tend to have a greater amount of crime than non-tourist areas in the long run. Following the literature of the economics of crime à la Becker (1968) and Ehrlich (1973) and using a system GMM approach for the time span 1985–2003, the authors empirically test whether total crime in Italy is affected by tourist arrivals. Findings confirm the initial intuition of a positive relationship between tourism and crime in destinations. When controlling for the difference between tourists and residents in the propensity to be victimized, no relevant differences are found: the likelihood to be victimized is quite similar for the two groups. As a consequence, agglomeration and urbanisation effects seem to be the main explanation for the impact of tourism on crime. One can image that overcrowded cities provide more opportunities to criminals to commit illegal activities regardless of the number of visitors and residents in destinations. Paper submitted to the special issue Tourism Externalities JEL D62, K00, L83 Keywords Tourism; crime; externalities Correspondence Bianca Biagi, e-mail: [email protected]. Claudio Detotto, e-mail: [email protected]. Maria Giovanna Brandano acknowledges the financial support provided by Regione Autonoma della Sardegna (‘Master and Back’ research grant). © Author(s) 2012. Licensed under a Creative Commons License - Attribution-NonCommercial 2.0 Germany EconomicsDiscussionPapers Introduction Tourism demand has grown overtime. In 2010 international tourist arrivals reached 940 million, and tourism receipts generated US$ 919 billion; in the time span 1975-2000 international arrivals have increased at an average pace of 4.6 per cent per year (UNWTO, 2011). With approximately 43 million international tourists, Italy is the fifth most visited country worldwide, within Europe, Italy ranks third (UNWTO, 2011). Academic literature confirms the impact of the tourism sector on the economy; a wide strand of research finds a positive linkage between tourism and growth for developed and developing countries, in the short and in the long run1. Why does the tourist sector have a strong impact on national or regional environments? It has to do with the characteristics of the product demanded: a bundle of goods and services (Sinclair and Stabler, 1997); most of which are non–traded and include manmade and natural amenities. On one side, the increasing demand for the tourism good and the intrinsic characteristic of the tourism product boosts local economy and makes residents better off. On the other side, the same features might generate negative environmental or social externalities that make residents worse off. When these negative impacts are not properly taken into account, development-led tourism becomes unsustainable. This paper investigates a possible source of unsustainability, which can occur when criminal activity is stimulated by the presence of tourists. In this case tourism imposes a social cost on residents, but it is also detrimental or the tourism market as a whole, negatively affecting potential tourism emand. f d  Why should crime increase with the presence of visitors? Following Fuji can be found: tourists carry valuable objects lidaymakers is more imprudent; tourists are and Mak (1980) many reasons and money; the attitude of ho  1 For an extensive literature review on tourism led-growth hypothesis see Brida et al. (2011). www.economics‐ejournal.org  2 EconomicsDiscussionPapers “safer” targets for criminals because they rarely report crime to the police; the presence of tourists alters the local environment, for instance, by generating a reduction of social responsibility for surveillance. Ryan (1990) and Kelly (1993) add that in some cases crime is driven by demand for illegal good or services in destinations. According to the Routine Activity Theory of Cohen and Felson (1979), crime depends on the opportunities; as a consequence, the presence of visitors increases the set of available occurrences. Overall, there are not many studies that explore this topic with the aid of econometric models. The assumption is usually that criminals are rational à la Becker (1968) and Ehrlich (1973) and respond to incentives; as such, the presence of tourism is seen as a further incentive for illegal activities. The seminal work is due to Jud (1975) who, controlling for urbanization, investigates the impact of foreign tourist business on total crime per capita in a cross-section of 32 Mexican States for the year 1970. The study confirms that total crime and property-related crime (fraud, larceny, and robbery) are strongly and positively linked to tourism, while crime against persons (assault, murder, rape, abduction, and kidnapping) is only marginally linked to it. Along the same line of Jud (1975), McPheters and Stronge (1975) use time series analysis to investigate whether seasonal crime reacts to seasonal tourism in Miami. They find that the tourism-crime relationship is significant and that economics –related crime such as robbery, larceny and burglary have a similar seasonality to tourism. Fuji and Mak (1980) reach the same conclusions for the case of Miami. For a cross section of 50 US States, Pizam (1982) finds a weak relationship between tourism and crime, suggesting that perhaps at national level the relationship is not supported. Van Tran and Bridges (2009), controlling for the degree of urbanization, the rate of unemployment, and the spatial position of the each state within Europe, analyse the relationship between tourist arrivals and crime against persons in 46 European nations. They find that, on average, the increase in the number of tourists reduces the rate of crime against persons. More recently, using panel data on crime and visitors of National Parks in every county in the US, Grinols et al. (2011) conclude that some tourist type has no impact on crime. Campaniello (2011) again using a panel approach explores the role of the 1990 Football World Cup in Italy; the results www.economics‐ejournal.org  3 EconomicsDiscussionPapers indicatethat hosting the Football World Cup has led to a significant increase in property crimes. On the same line, Biagi and Detotto (2012) find for a cross section of Italian provinces a positive relationship between tourism and pick-pocketing. The main aim of the present paper is to test whether the positive tourismcrime relationship found in Biagi and Detotto (2012) for a cross-section of Italian provinces2, is persistent over time. Applying a System GMM approach to a panel of Italian provinces for the time span 1985-2003, we empirically test whether total crime in Italy is affected by tourist arrivals. In other words, the purpose of the paper is to demonstrate that all else being equal, for the case of Italy in the long run, tourist areas tend to have a greater amount of crime that non-tourist ones. A connection between tourism and crime does not tell us whether the victims are visitors or residents, it merely indicates the presence of a link between tourism and crime as a potential source of negative externalities. Knowing which group of people is more affected may give essential information to better quantify the externality and to identify possible solutions. For instance, criminal activity that mainly target tourists would impact on the image of a tourist destination as a whole, decreasing its future tourist demand; if on the contrary, the crime is largely committed against residents, the externality affects the quality of life of locals. Unfortunately, due to shortage of crime data worldwide, this analysis is not often undertaken; the few papers available use descriptive statistics (for the case of Hawaii see Chesnay-Lind and Lind, 1986; for the case of Barbados see de Albuquerque and McElroy, 1999). In this paper we investigate whether the propensity to be victimized is higher for tourists than residents. Since data on the victimization rate of visitors and residents are not provided by the National Institute of Statistics (from now on ISTAT), we apply a method that indirectly estimates such propensities: we calculate the “equivalent population” of each province as the total number of individuals actually fore, re-estimating the model using the crime lent population, we obtain that the difference present in the territory. There index corrected for the equiva  2 The analysis focuses on the tourism-related crime in the Italian provinces for the time span 1985-2003. The number of provinces has changed overtime; from 1974 until 1992 national territory was divided into 95 provinces, which become 103 in 1992 and 107 in 2006. To have a balanced panel, the study considers the classification at 95. www.economics‐ejournal.org  4 EconomicsDiscussionPapers between the coefficients associated with residents and tourists is not significantly different from zero; hence, the probability to be victimized is quite similar for the two groups. The paper is structured as follows. Section 1 offers a descriptive analysis of the evolution of tourism and crime in Italy. Section 2 focuses on the data and empirical model; section 3 describes the results. Section 4 highlights the main concluding remarks. 1 Tourism and crime in Italy Tourism and crime are two relevant phenomena in Italy. As reported by the National Institute of Statistics (ISTAT) tourists in Italy have constantly increased: 57 million of arrivals (international and domestic visitors) were accounted in the official tourist accommodations in 19853, and they reached 83 million in 2003 (a growth rate of 47%); the number of nights in official accommodations was about 333 million in 1985, and reached 344 million in 2003 (+3%)4. During this time span tourist arrivals have increased on average by 2.2% and tourist nights by 0.5%. As a result, the average length of stay decreased from about 6 days to 4. This downward trend of tourist nights is in line with the EU trend where the number of nights has decreased more than the number of trips (-1.6 % and - 1.0 % respectively; Eurostat, 2011). In 2003, more than fifty per cent of tourist arrivals and nights are concentrated in the Northern part of the country. As far as crime is concerned, Italy experienced a rather exceptional increase over the last 25 years (+35.7%); this trend is in contrast with what occurs for the same time span in many other Western countries such as US (-20.4%), Canada (-15.8%), the UK (-10.9%), France (-7.5%) and Germany (-6.9%; Eurostat, 2009).  3 Official accommodations include: hotels, campsites, guesthouses, Bed & Breakfasts, and other types of accommodations. 4 Tourist arrivals are the number of visitors -domestic and foreignregistered in the official accommodations; tourist nights are the total number of nights spent by visitors in official accommodations (ISTAT). www.economics‐ejournal.org  5 EconomicsDiscussionPapers The comparison of tourism arrivals and total crime series for the time span 19852003 highlights a common upward trend of the two variables (Figure 1). Even though, as the Figure clearly shows, crime increases at a higher pace than tourism. Furthermore, a counter cyclical relationship can be observed between the two series indicating a possible negative correlation among them. [FIGURE 1 HERE] To better understand the underlying tourism-crime relationship, for each Italian province we calculate a tourism index and a crime index as follows: LQ Tourism = Total _Arrivals i Total _Arrivals ⎞ ⎠ ⎟ ⎛ ⎝ ⎜ Sur f ace i Total _Sur f ace ⎞ ⎠ ⎟ ⎛ ⎝ ⎜ (1) LQCrime = Total _crimei Total _Crime ⎞ ⎠ ⎟ ⎛ ⎝ ⎜ Populationi Total _Population ⎞ ⎠ ⎟ ⎛ ⎝ ⎜ (2) where j= 1, 2,…95 Quartiles are the values used to divide each index into four equal groups. Table 1 shows the cross tabulation of the quartile distribution of the two indexes. The first quadrant in the Table displays the number of provinces with low level of crime and tourism (14 in total). The principal diagonal contains 47% of the Italian provinces, indicating a positive correlation between tourism and crime. The chi-squared test ( =45.5) indicates that the k groups are dependent. χ 2 [TABLE 1 HERE] www.economics‐ejournal.org  6 EconomicsDiscussionPapers As a result, crime and tourism seem to move in the same direction: low levels of tourism correspond to low levels of crime and vice versa. This descriptive analysis gives a first hint on the relationship between the two phenomena; this relationship needs to be further explored by using appropriate econometric techniques. 2 Data and empirical model Following the empirical literature on crime, this study proposes the dynamic panel data approach illustrated below to explore the relationship between tourism and crime for Italian provinces in the time span 1985-2003: CRIMEi,t= β 0+ β 1CRIMEi,t−1+ β 2GROWTHi,t + β 3INCOMEi,t + β 4UNEMPLi,t + β 5DENSITY i,t+ β 6TOURISMi,t+ β 7DIPLOMAi,t+ β 8DETERRENCEi,t+ β 9SOU TH i+ β 10YEARt+ η i+ ε i,t (3) CRIMEit is the number of total crimes per 100,000 inhabitants in the i-th province at time t. GROWTH and INCOME indicate the growth rate and level, respectively; of Gross Domestic Product (GDP) per capita at 1995 constant prices UNEMPL is the unemployment rate. Cantor and Land (1985) theorize the macroeconomic relationship between the economic performance and criminal activity, indicating two opposite sources of incentive to criminal behaviour: opportunity and motivation effect. The former is linked to fluctuations in INCOME and GROWTH; the opportunities to commit crime increase with economic performance, which leads to widespread availability of goods and profitable illegal activities. The latter works in the opposite way: the incentive to commit crime is caused by bad economic conditions. In other words, during recessions, the unemployment rate raises inducing individuals to increase their disposable income via illegal activities. DENSITY refers to the population per square kilometre; it is used as an indicator of urbanisation. According to Masih and Masih (1996), crime rises with urbanisation. TOURISM measures tourist arrivals (nationals and foreigners) per square kilometre; tourist arrivals, weighted by province size, gauge the attractiveness of a given destination. According to the empirical www.economics‐ejournal.org  7 EconomicsDiscussionPapers literature on the crime-tourism relationship, a positive correlation is expected5. DIPLOMA indicates the average level of education in the i-th province at time t; a higher level of education might indicate a higher level of social cohesion, which could reduce crime offences. DETERRENCE is the ratio of recorded offences committed by known offenders over the total crime recorded; it is a proxy of the deterrence effect “stemming from the efficiency of criminal investigation of the local police and from their knowledge of the local environment” (Marselli and Vannini, 1997; p.96). The expected sign is negative, therefore, a rise in the share of known offenders, due to an increase in deterrence or a higher level of efficiency/efficacy of police activity, reduces the crime rate. All variables are expressed in log-level terms, so that the coefficients can be interpreted as elasticities. SOUTH is a control variable which equals 1 if the province is located in the South of Italy and zero otherwise. YEAR is a set of time dummy variables; the inclusion of time dummies makes the assumption of no correlation across individuals in the idiosyncratic disturbances more likely to hold (Roodman, 2006). Finally, η i and ε it are the province fixed effect and the error term, respectively; we assume that E ( η i ) = 0 , E ( ε i,t η i ) = 0 and E ( η i ) = 0 . Table 2 and Table 3 provide detailed information and some descriptive statistics of the variables in use, respectively. [TABLE 2 HERE] [TABLE 3 HERE] As analysed by Buonanno (2006), crime series show strong persistence over time, indicating that the level of crime activity at time t affects crime behaviour at time t+1. To confirm this, we start our analysis running a basic Ordinary Least Squares (OLS) model, both random and fixed effect, and we apply the Wooldridge test (Wooldridge, 2002) to check serial correlation in panel data; we find that the null hypothesis of no serial correlation is strongly rejected  6. These arguments strongly suggest the use  5 See Biagi and Detotto (2012) for an extensive literature review on this topic. 6 The statistic tests are available on request. www.economics‐ejournal.org  8 EconomicsDiscussionPapers Notice that the difference between the coefficients of DENSITY and TOURISM is not significantly different from zero. When we re-estimate the model using different measures of tourism density, we obtain analogous results. This outcome gives a first suggestion on the possible source of negative externality when total crime is analysed: the impact of a rise in residents and visitors on crime is quite similar, which may indicate that the main forces driving tourism-crime relationship are the agglomeration and urbanisation effects. Overcrowded cities give more opportunities to criminals to commit illegal activities, regardless of the share of visitors and residents in the tourism destinations. Probably, as the empirical studies suggest, the presence of visitors provides an incentive for certain illegal activities; therefore, it is a possible substitution among crime types that should be further explored. 4 Concluding remarks The tourism-led growth hypothesis has been widely analysed by scholars, who overall agree on the significant role of the tourism sector in enhancing economic growth. While many studies focus in the short run relationship, a small number of them analyse the relationship between tourism and growth in the long run. There is also a wide concern about the negative impact of tourism activity in the host community in terms of social and environmental degradation. A possible source of negative externality exists when criminal activity develops in response to the presence of visitors. As Grinols et al. (2011) highlight, there are many reasonable theories stemming from economic and sociological studies of crime determinants that may explain the relationship between tourism and crime. Economic literature barely explores this issue, finding controversial results. The main aim of the present paper is to test whether the positive tourismcrime relationship that Biagi and Detotto (2012) find for property-related crime in a cross-section of Italian provinces, is persistent over time and holds when total crime is analysed. In other words, this study analyses the dynamic relationship between tourism arrivals and total crime. To do so, the OLS and System GMM approach are applied. www.economics‐ejournal.org  15 EconomicsDiscussionPapers We find that tourism positively affects criminal activity; in the short run, a one-per-cent increase in arrivals leads to a 0.018% rise in total crime, while, in the long run, the impact is about 0.11%. We further investigate whether the propensity to be victimized is higher for tourists than residents by re-estimating the model using the crime index corrected for the equivalent population; we obtain that the difference between the coefficients associated with residents (DENSITY) and visitors (TOURISM) is not significantly different from zero. As a consequence, when total crime is analysed, the probability to be victimized is quite similar for the two groups. One can image that overcrowded cities give more opportunities to criminals to commit illegal activities regardless of the share of visitors and residents in destinations. Unfortunately, this analysis has two limitations. First, it is based on the assumption that tourists and residents have the same propensity to report crime offences across the provinces. Second, aggregate crime data such as total crime rate, could fail to signal the presence of differences among crime typologies. In fact, it is reasonable that the impact of tourists is higher then the one of residents, on some typologies of crime, such as pick pocketing, bag snatching and fraud, and less on other types of illegal activity, such as financial crimes, handling and extortion. Finally, it is possible that the coefficients might underestimate the underlying relationship due to measurement error in the dependent variable. The crime data used in this paper are the total offences recorded by the Police, this represents the tip of the iceberg of this phenomenon. As further development, a state space approach (Hamilton, 1994) can be applied in order to estimate the unobservable component of crime series. In addition, other improvements may go in the direction of exploring how tourism-crime elationship changes according to the types of visitors (national and foreign) nd type of crime (against property or persons). r a   www.economics‐ejournal.org  16 EconomicsDiscussionPapers References Arellano,M.,andS.Bond(1991).SomeTestofSpecificationforPanel Data: Monte Carlo Evidence and an Application to Employment Equation.  ReviewofEconomicStudies 58: 277‐297.  URL: http://ideas.repec.org/a/bla/restud/v58y1991i2p277‐97.html. Becker, G.  (1968).  Crime and punishment: an economic approach. JournalofpoliticalEconomy 76: 169‐217.  URL: http://econpapers.repec.org/article/ucpjpolec/v_3a76_3ay_3a1968_3a p_3a169.htm. Biagi,B.,andC.Detotto(2012).CrimeasTourismExternality. r RegionalStudies,fo thcoming. Bimonte, S., Brida G., Pulina M., and L.  Punzo (2011).  Tourism and growth: stories of the two continents.  StructuralAdjustmentsand regionalIntegrationinEuropeandLatinAmerica.Routledge,UK (forthcoming). Blundell,R.,andS.Bond(1998).InitialConditionsandMoment RestrictionsinDynamicPanelDataModels.JournalofEconometrics87: 115‐143.URL:http://ideas.repec.org/a/eee/econom/v87y1998i1p115‐ 143.html. Buonanno,P.(2006).CrimeandLabourMarketOpportunitiesinItaly (1993–2002).  Labour 20: 601‐624.  URL: http://ideas.repec.org/a/bla/labour/v20y2006i4p601‐624.html. Campaniello,N.(2011).MegaEventsinSportsandCrime:Evidence From the 1990 Football World Cup.  JournalofSportEconomicsDOI: 10.1177/1527002511415536.  URL: http://cnr‐ www.economics‐ejournal.org  17 EconomicsDiscussionPapers it.academia.edu/NadiaCampaniello/Papers/368425/Mega_Events_in_S ports_and_Crime_Evidence_from_the_1990_Football_World_Cup. Cantor,D.andK.C.Land(1985).UnemploymentandCrimeRatesin thePost‐WorldWarIIUnitedStates:ATheoreticalandEmpirical Analysis.  AmericanSociologicalReview 50: 317‐332.  URL: http://www.jstor.org/stable/2095542. Chesney‐Lind, M., and I.Y.  Lind (1986).  Visitors as Victims: Crimes against Tourists in Hawaii.  AnnalsofTourismResearch 13:167‐191. URL:http://www.sciencedirect.com/science/article/pii/01607383869 00368. Cohen,L.E.,andM.Felson(1979).SocialChangeandCrimeRate Trends:ARoutineActivityApproach.AmericanSociologicalReview44: 588‐608.URL:http://www.jstor.org/stable/2094589. Cullen, J.  B.  and S.D.  Levitt (1999).  Crime, Urban Flight, and the Consequences for Cities.  TheReviewofEconomicsandStatistics 81: 159‐169.URL:http://ideas.repec.org/p/nbr/nberwo/5737.html. DeAlbuquerque,K.,andJ.McElroy(1999).TourismandCrimeinthe Caribbean.  AnnalsofTourismResearch 26:968‐984.  URL: http://www.popcenter.org/problems/crimes_against_tourists/PDFs/d eAlbuquerque_McElroy_1999.pdf. Detotto, C.  and E.  Otranto (2010).  Does Crime Affect Economic Growth? Kyklos 63: 330‐345.  URL: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1640229. Detotto,C.,andM.Pulina(2010).Testingtheeffectsofcrimeonthe Italianeconomy.EconomicsBulletin,Vol.30no.3pp.2063‐2074.URL: http://ideas.repec.org/a/ebl/ecbull/eb‐10‐00465.html. Dills,A.K., Miron J.A. andG.  Summers (2008). What Do Economists KnowAboutCrime?NBERWorkingPaper13759.NationalBureauof www.economics‐ejournal.org  18 EconomicsDiscussionPapers Economic Research, Cambridge, Mass, URL: http://www.nber.org/papers/w13759.pdf. cal  Ehrlich,I.(1973).Participationinillegitimateactivities:atheoreti andempiricalinvestigation.JournalofPoliticalEconomy81,521–65. l3/papers.cfm?abstract_id=961495.URL:http://papers.ssrn.com/so EUROSTAT,TourismStatistics. URL:http://epp.eurostat.ec.europa.eu/statistics_explained/index.php/ Tourism_statistics. Fajnzylber, P., Lederman D.  and N.  Loayza (2002).  Inequalityand Violent Crime.  JournalofLawandEconomics 45: 1‐40.  URL: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=303838. Fujii,E.  T., and J. Mak (1980). Tourismand Crime:Implicationsfor Regional Development Policy.  RegionalStudies 14: 27‐36.  URL: http://rsa.informaworld.com/srsa/content~db=all~content=a758918 960~frm=abslink. Griliches, Z.  and J.A.  Hausman (1986).  Errors in Variables in Panel Data.  JournalofEconometrics 31: 93‐118.  URL: http://www.nber.org/papers/t0037. GrinolsE.L.,Mustard,D.B.,andM.Staha(2011).HowdoVisitorsaffect Crime? JournalofQuantitativeCriminology 27: 363‐378.  URL: http://www.pressekatalog.de/visitors‐affect‐crime_ARTID_10.1007‐ ab‐s10940‐010‐9128‐0.htm. Hamilton,J.D.(1994).Statespacemodels.InEngleR.F.andD.L. McFadden (eds.) HandbookofEconometrics2.  Amsterdam: North Holland,Vol.4:3039‐3080. Hansen,L.(1982).LargeSamplePropertiesofGeneralizedMethodof Moments Estimators.  Econometrica 50: 1029‐1054.  URL: http://faculty.chicagobooth.edu/john.cochrane/teaching/35904_Asset_ Pricing/Hansen%20econometrica%20GMM.pdf. www.economics‐ejournal.org  19 EconomicsDiscussionPapers ISTA ,(1985‐2001). Statistichesulurismo,IS AT:Rom s T  tT a. Jud, D.G.  (1975).  Tourism and Crime in Mexico.  SocialScience Quarterly56:324‐330. URL:https://www.ncjrs.gov/App/publications/Abstract.aspx?id=3136 3. Kelly, I.  (1993).  Tourist destination crime rates: an examination of CairnsandtheGoldCoast.JournalofTourismStudies4:2‐11. Levin,A.,LinC.F.andC.S.J.Chu(2002).UnitRootTestsinPanelData: AsymptoticandFinite‐SampleProperties.JournalofEconometrics108: 1‐24.  URL: http://ideas.repec.org/a/eee/econom/v108y2002i1p1‐ 24.html. Marselli,R.andM.Vannini(1997).EstimatingaCrimeEquationinthe PresenceofOrganizedCrime:EvidencefromItaly.InternationalReview ofLawandEconomics 17: 89‐113.  URL: http://www.egov.ufsc.br/portal/sites/default/files/anexos/33009‐ 141386‐ ‐PB.pdf. Masih, A.M.M.  and R.  Masih (1996).  Temporal Causality and the Dynamics of Different Categories of Crime and their Socioeconomic Determinants: Evidence from Australia.  AppliedEconomics 28: 1093‐ 1104.URL:http://www.tandfonline.com/doi/abs/10.1080/000368496 327949#preview. Mauro,L.andG.Carmeci(2007).APovertyTrapofCrimeand Unemployment.ReviewofDevelopmentEconomics11:450‐462.URL: http://onlinelibrary.wiley.com/doi/10.1111/j.1467‐ 9361.2006.00350.x/abstract. McPheters,L.R.,andW.B.Stronge(1974).CrimeasanEnvironmental Externality of Tourism: Miami, Florida.  LandEconomics 50:288‐292. URL:http://www.jstor.org/stable/3145040. www.economics‐ejournal.org  20 EconomicsDiscussionPapers Mills,E.S.andL.S.Lubuele(1997).InnerCities.JournalofEconomic Literature35:727‐756.URL:http://www.jstor.org/stable/2729791. Mocan,H.N.,BillupsS.C.andJ.Overland(2005).ADynamicModelof DifferentialHumanCapitalandCriminalActivity.Economica72:655‐ 56667.681.URL:http://papers.ssrn.com/sol3/papers.cfm?abstract_id=8 Pizam,A.(1982).TourismandCrime:IsThereaRelationship?Journal ofTravelResearch20:7‐10.URL: agepub.com/content/http://jtr.s 20/3/7.abstract. Roodman, D.  (2006).  How to Do xtabond2: An Introduction to ‘Difference’and‘System’GMMinStata.CenterforGlobalDevelopment WorkingPaperNo.103.URL: http://repec.org/nasug2006/howtodoxtabond2.cgdev.pdf. Ryan, C.  (1993).  Crime, Violence, Terrorism and Tourism.  Tourism Management 14: 173 ‐ 183.  URL: http://www.sciencedirect.com/science/article/pii/026151779390018 G. Sargan, J.  (1958).  The Estimation of Economic Relationships Using Instrumental Variables.  Econometrica 26: 393‐415.  URL: ftp://iwh‐ halle.de/pub/vwgr/zipp_lit/econometrics/Past%20work/Sargan(1958 ).pdf. SinclairM.T.,andM.Stabler(eds)(1997).TheEconomicsofTourism, e.London:Routledg UNWTO(2011).WorldTourismBarometerVol.9,N.1. VanTran,X.,andF.S.Bridges(2009).TourismandCrimeinEuropean of Research(eRTR)Vol.7,No.3.Nations.eReview Tourism Wooldridge, J.M.  (2002).  EconometricAnalysisofCrossSectionand PanelData.Cambridge,MA:MITPress. www.economics‐ejournal.org  21 EconomicsDiscussionPapers WorldEconomicForum(2011).TheTravel&TourismCompetitiveness Report2011, Ginevra.  URL: http://www.weforum.org/reports/travel‐ tourism‐competitiveness‐report‐2011. Figure 1. Tourism and crime series (base year=1985) Notes: we use index numbers with a fixed base value 1985=100 Source: our elaboration on data from ISTAT Table 1. Number of Provinces for percentiles LQCrime LQTou 1s t 2n d 3r d 4th Total 1st 14 6 3 1 24 2nd 7 7 8 2 24 3rd 3 7 9 5 24 4th 0 4 4 15 23 Total 24 24 24 23 95 www.economics‐ejournal.org  22 EconomicsDiscussionPapers Table 2. List of explanatory variables Name Definition Type of variable Source Crime Total crime offences per 100,000 inhabitants Crime ISTAT, Statistiche Giudiziarie Penali Growth Growth rate of real value added per capita Economic Istituto Tagliacarne Income Value added per capita at a base prices (Year = Economic Istituto Tagliacarne Unemployment People looking for a job/labour force * 100 Economic Istituto Tagliacarne Density Density of population per square Kilometre Demographic ISTAT, Atlante statistico dei comuni Tourism Tourists official arrivals per square kilometre (tourists choosing official accommodations) Tourism ISTAT, Statistiche del turismo Diploma People with Italian diploma per 10,000 inhabitants Human capital ISTAT, Atlante statistico dei comuni Deterrence Ratio of incidents with unknown offenders over the total recorded per total crime Deterrence ISTAT, Statistiche Giudiziarie Penali South Dummy variable that values one if a province is located in the South and zero otherwise Geographic Our elaboration Table 3. Descriptive statistics of variables Name Mean SD Min Max Crime 3,091.80 1,374.01 745.48 13,255.08 Growth 0.01 0.07 -0.81 0.77 Income 14,113.17 3,925.78 4,517.04 26,025.37 Unemployment 10.94 6.71 1.7 33.2 Density 248.58 345.04 34.47 2,647.02 Tourism 283.65 392.85 14.76 2,529.23 www.economics‐ejournal.org  23 EconomicsDiscussionPapers Diploma 0.06 0.03 0.002 0.17 Deterrence 0.32 0.12 0.09 0.83 South 0.36 0.48 0 1 Table 4. OLS results (1) (2) (3) (4) (5) (6) FE RE RE FE-IV RE-IV RE-IV VARIABLES Growth -0.11 -0.050 -0.087 -0.098 0.020 -0.059 (0.071) (0.072) (0.072) (0.075) (0.077) (0.074) Income 0.23*** 0.16*** 0.23*** 0.21*** 0.0076 0.17*** (0.056) (0.050) (0.054) (0.069) (0.064) (0.057) Unemployment - 0.11*** - 0.047** - 0.072*** - 0.11*** - 0.042** - 0.084*** (0.021) (0.019) (0.020) (0.022) (0.020) (0.021) Density - 0.57*** 0.16*** 0.15*** - 0.56*** -0.033 0.0049 (0.19) (0.031) (0.031) (0.19) (0.058) (0.050) Tourism -0.050* 0.041** 0.062*** 0.028 0.27*** 0.24*** (0.028) (0.019) (0.019) (0.20) (0.059) (0.050) Diploma -0.028 -0.028 -0.027 -0.030 - 0.041** -0.036* (0.018) (0.019) (0.019) (0.019) (0.020) (0.019) Deterrence - 0.26*** - 0.28*** -0.28*** - 0.26*** - 0.28*** -0.28*** (0.022) (0.021) (0.021) (0.022) (0.022) (0.022) South 0.21*** 0.35*** (0.054) (0.068) Constant 8.89*** 5.20*** 4.00*** 8.55*** 6.32*** 4.67*** (1.18) (0.51) (0.54) (1.42) (0.61) (0.57) Observations 1,710 1,710 1,710 1,710 1,710 1,710 www.economics‐ejournal.org  24