Assessing the Effect of Sentiment on Fiscally Induced Migration: A Tiebout Hypothesis Approach within EU 27

Article

Christos Kallandranis and Socrates Karidis*

Assessing the Effect of the Consumer-VoterSentiment on Tiebout-Like Migration: TheEU 27 Case

Abstract: The paper revisits the determinants of intra-European migration using aconditional logit model for the period 2000–2010. With all moving restrictionslifted, a substantial Union enlargement in 2004, and a severe economic crisis inprogress, new evidence is being revealed regarding the decision making process ofindividuals who choose to relocate. We examine Tiebout’s proposition of votingwith one’s feet in the context of the Union, and we show that fiscal policies in theform of fiscal packages offered to a country’s residents as well as consumerconfidence indicators play a decisive role in affecting that process. With 26alternative destinations over a period of 11 years, we find that people will chooseto move to a country that offers them a better fiscal surplus, higher confidence aseconomic agents and higher potential earnings. In light of our results, and afterestablishing the migration patterns, we believe that a new discussion should openregarding efficiency concerns and government policy implications.

Keywords: fiscal surplus, migration, consumer confidence indicator, Tiebouthypothesis, conditional logitJEL Classification: E62, E690, F22, F60

*Corresponding author: Socrates Karidis, Department of Accounting and Finance,University of Bedfordshire, Park Square, Luton, Bedfordshire LU1 3JU, UK,E-mail: [email protected] Kallandranis, Department of Accounting Finance and Economics, Regent’s UniversityLondon, Inner Circle, Regent’s Park, London NW1 4NS, UK, E-mail: [email protected]

1 Introduction

A number of studies have empirically investigated the determinants of fiscallyinduced migration, with most of the interest directed to the intra-state migration

doi 10.1515/gej-2013-0009 Global Economy Journal 2014; 14(1): 31–55

Authenticated | [email protected] author's copyDownload Date | 7/31/14 9:41 AM

cases of US and Canada (Clark, Hatton, and Williamson 2002; Hunt and Mueller2004). However, little work has been done regarding migration within theEuropean Union despite the significant changes that have taken place, particu-larly over the past decade, and the overall absence of legal distortions. Moststudies in the area concentrate on a specific country and primarily Germany andthe UK (Hatton 1995; Fertig 2001; Brucker and Schroder 2007).

While looking at both the enlargement that took place in two phases (2004,2007) and the economic crises, affecting mainly countries in the periphery, onewould expect a sizable movement, initially from the residents of “New Europe”toward the traditional European economies, followed by clearly defined migra-tion patterns from the countries that are mainly affected by the crises to the onesthat took the hit more moderately. Nevertheless, one would rightfully argue thatmigration patterns developed due to the economic crisis might be difficult todetect, as the new economic map in the European continent has not yet maturedin its formation.

With a significant and still growing empirical literature on migration, twomain trends can be identified: one that looks at the direct effect of migration inboth the origin and the destination countries in terms of income redistributionand efficiency and one that tries to identify the driving factors that would inducepeople to migrate. Those can be either pushing or pulling, for outward or inwardmigration respectively.

This paper deals with the latter and tries to empirically investigate the fiscaland other economic determinants of intra-European migration for the period2000–2010. For instance, public goods provision and locational choice havebeen of interest for a number of researchers led by Samuelson (1954) andparticularly by Tiebout (1956) who, in an effort to overcome the problem ofidentifying a consumer’s preferences as posed by Samuelson (and Musgrave in1939), proposed that those preferences are revealed by the locational choice.Individuals will act as consumer-voters and choose to locate in areas or com-munities that best satisfy their preferences for public goods. According toTiebout, “the consumer-voter may be viewed as picking that community thatbest satisfies his preference pattern for public goods”. Tullock (1971) extendsTiebout’s thesis by including the notion that individuals would also assess thetax burdens associated with a specific locational choice: “the individual decidingwhere to live will take into account the private effects upon himself of the bundle ofgovernment services and taxes”. It is plausible to argue that fiscal factors shouldbe considered along with other variables such as wage, unemployment, educa-tion, cost of living, and so forth.

A number of studies examine the impact of fiscal factors on internal migra-tion (Buchanan and Goetz 1972; Flatters, Henderson, and Mieszkowski 1974;

32 C. Kallandranis and S. Karidis


Starrett 1980; Cebula and Karoglis 1986; Shaw 1986; Koven and Shelley 1989;Day 1992; Cushing 1993; Ott and Shadbegian 1993; Day and Winer 2001; Liebigand Sousa-Poza 2006). However, little attention has been paid to the EU case.Recent papers by Peridy (2006) and Karidis and Quinn (2006) examine theimpact of fiscal factors on flows of immigrants within or into the EuropeanUnion. Peridy (2006) focuses on health and education spending by the govern-ment while Karidis and Quinn (2006) utilize a broader measure of social spend-ing by governments but also include the impact of tax rates. Jackson, Ortmeyer,and Quinn (2013) examine the impact of fiscal policies on both the size and theeducational levels of immigrants in destination countries for a sample of OECDcountries.

We believe that the Tiebout–Tullock hypothesis is the appropriate approachfor this study given our thesis that one of the driving forces of migration is thedifference in fiscal surpluses offered to the consumer-voter. The absence of legalrestrictions within the European Union allows for the assessment of the hypoth-esis in this context. We overcome what Oates (2006) refers to as a “puzzlingfailure” of the original Tiebout model where there is no real treatment of taxesby employing the notion of fiscal surplus, a variable that has been used in theliterature in an effort to incorporate Tullock’s extension. In addition, we attemptto take the Tiebout–Tullock hypothesis a step further and extend the usualquestion of whether fiscal factors are useful determinants of one’s choice tomigrate. We focus on the indirect impact of sentiment indicators as an amplifi-cation mechanism over and above the usual macroeconomic variables. In parti-cular, we empirically test the existence of a positive and significant impact ofconsumer confidence, as a leading indicator, on people’s motivation to relocate.Consumer spending depends not only on current income and wealth but also onconsumer confidence regarding households’ personal financial situation as wellas the general economic situation. Thus, in line with the Tiebout–Tullock inter-pretation of consumer-voter, we try to explore the possibility of an asymmetry inone’s decision to migrate, not only ex post, i.e. after facing the tight conditionsin his/her country of origin, but also ex ante via people’s expectations regardingfuture economic conditions. Undoubtedly, expectations are bound to be impor-tant in this context, given that people’s decisions are by default forward-looking.In other words, people will discount the future, while employing efficiently anyrelevant information regarding the probability to migrate. Consumer confidencereflects specific attitudes related to particular events and/or uncertainty aboutthe future financial and economic situation. A number of previous studies showthe relation between sentiment indices and economic stance. However, there arehardly any studies modeling the migration choice as a function of individuals’expectations for a European panel of countries. To the best of our knowledge,

Effect of Consumer-Voter Sentiment on Tiebout-Like Migration 33


there is no study focusing on migratory responses associated with agents’expectations of future conditions.

The paper contributes to the literature in the following ways. Firstly, it addsto the limited literature that incorporates the Tiebout–Tullock hypothesis in aninternational setting and in particular to that of the European Union. Accordingto Cebula (2009), this is one of the “fertile and effectively unexplored opportu-nities to which the Tiebout–Tullock hypothesis can be applied”. Secondly, we gobeyond the internationally oriented Tiebout–Tullock hypothesis literature byincorporating a forward-looking index as an indicator of the probable relocatingdecision. This broadened interpretation of the Tiebout notion works in parallelwith the traditional approach and enriches the range of the hypothesis itself in away that makes it more comprehensive. Finally, we work with an up to dateextensive sample of 27 countries and 11 years utilizing a concrete econometricmethodology in order to capture the determinants that drive the choice betweendifferent destinations.

This paper is organized as follows: Section 2 revisits, in short, the existingempirical Tiebout literature and presents a theoretical model that incorporatesfiscal factors, where the consumer-voter is viewed as a utility maximizing agentbetween the country of origin and that of destination. The theory providestestable hypotheses with regard to fiscal policy, consumer sentiment, earningsand migration. In Section 3, we discuss our data and the empirical method usedfor testing these hypotheses. Section 4 presents the results of our primaryempirical analysis and further discusses some alternative sensitivity tests. Weconclude in Section 5.

2 Theory and evidence

2.1 Theoretical framework

Individuals are seeking to maximize a utility function consisting of a bundle ofprivate and public goods. Their utility is constrained by the prices they have topay through direct purchases and taxation. The demand for private goodsdepends on the migrant’s income and prices. Similar theoretical developmentsare provided by Houtenville and Conway (2001) in the context of elderly migra-tion across states and by Farnham and Sevak (2006) who tested a lifetime modelof Tiebout sorting. Cebula (2005, 2009), on the other hand, uses the modeldeveloped by Sjaastad (1962) to consider the migration decision as an invest-ment one in which the consumer-voter expects his discounted present value of



the migration chosen to be the maximum possible from moving from area A toarea B. Nakosteen and Zimmer (1980) assume that potential migrants behave asthough they seek maximization of their present value of net gains resulting fromlocational choice. Following Karidis and Quinn (2006), we incorporate a randomutility model, and we consider the individual utility function to be given by:

Uij ¼ Ui Xj;Gj

� � ½1�

where Xj and Gj stand for private and public goods in country j, respectively. Thevariable j is defined as j ¼ A, B with A being the country of origin and B thedestination.

Over the process of maximizing their utility, individuals believe that theirdecision to migrate will not affect the distribution of tax burdens and consump-tion of public good among the citizens of the two countries in question. We alsoassume that the cost of moving is negligible.

The demand for private goods is given by

Xj ¼ X E Ij� �

;Pj� � ½2�

where E(Ij) is the expected income of individual i in country j and Pj is theaggregate price level in country j. Prior studies assume that although theindividual is uncertain about his/her income in the destination country, thereis no uncertainty involved when he/she assesses the income level associatedwith the home country. In this paper we argue that changing economic condi-tions, especially in times of economic uncertainty, create doubts regarding one’sconfidence even in the home country. An individual’s decision process takesinto account the fact that increasing unemployment rates as well as continuousausterity measures in countries of trouble create severe doubts about incomesecurity and therefore this uncertainty has to be taken into account.

Therefore, indirect utility functions at home and abroad are

ViA ¼ Vi EðIAÞ;PA;GA½ � j ¼ A ðoriginÞ ½3a�

ViB ¼ Vi E IBð Þ;PB;GB½ � j ¼ B ðdestinationÞ ½3b�

Dividing by Pj yields

ViA ¼ Vi EðIAÞ

PA;GA

PA

� �½4a�

ViB ¼ Vi E IBð Þ

PB;GB

PB

� �½4b�



Expected disposable income is given by

E Ij� � ¼ 1� ρj

� wjLj � tjBj ½5�

where wj is the average wage in country j, Lj is the number of working hours, tj isthe tax rate and Bj the tax base with Bj < wjLj.

ρj represents an effort to capture income uncertainty at home and abroad. Itexpresses the probability of being unemployed at home or abroad, and it isbasically shown in the unemployment rate. However, it could be proxied by aconsumer sentiment indicator as well. The individual worries about being ableto maintain a certain standard of living, and although the information set athome is richer and more accessible, as it reflects the person’s current state, theviews expressed in a consumer-voter’s indicator of confidence can be taken intoconsideration along with the unemployment rate in order to give a sense of howuncertain the person is. Therefore ρj represents the probability that individual iwill not be able to secure his/her income at home or abroad to a level that willmaximize his/her utility function unless if it is zero.

Considering Lj to be given and by substituting eq. [5] in eq. [4] we canrewrite the indirect utility function as

ViA ¼ Vi ð1� ρAÞwA

PA� tABi

A

PA

� �;GA

PA

�½6a�

ViB ¼ Vi 1� ρBð ÞwB

PB� tBBi

B

PB

� �;GB

PB

�½6b�

Assuming that the income tax base equals the wage base and rearranging theterms, the indirect utility functions of individual i take the following form for theorigin and destination countries:

ViA ¼ Vi ρA;

wA

PA; tA

GA

PA

� ½7a�

ViB ¼ Vi ρB;

wB

PB; tB

GB

PB

� ½7b�

We can determine that Vρ<0, Vw>0, Vt<0, VG >0, implying that a person’sutility increases with the wage rate, and the level of public goods provided whileit decreases with higher uncertainty, and higher prices of private and publicgoods.



Individual i will choose to migrate from country A to country B only if theutility derived is higher by doing so than staying at home:

ViA ρA;

wA

PA; tA

GA

PA

� <Vi

B ρB;wB

PB; tB

GB

PB

� ½8�

When in a position to choose from a set of different options, the individualwill choose the destination that maximizes his/her utility. The inclusion of fiscalfactors in the theoretical model implies that any choice involves a Tiebout–Tullock like process of adjustment. This is captured by the two fiscal policyvariables. For this process to be optimum, efficiency concerns should be con-sidered according to Buchanan and Goetz (1972). However, these concerns arebeyond the scope of the present study. We extent the idea that the choice of adestination depends on policy decisions by implementing two arguments thatreflect market conditions directly but contain also elements of indirect policydecisions: Labor earnings and uncertainty factors regarding the state of theeconomy. Therefore, the theoretical model yields the following testablehypotheses:1. Labor earnings in a specific country should, ceteris paribus, have a positive

effect on the likelihood of an individual choosing this country as potentialdestination (V 0

w >0).2. Uncertainty in securing a level of labor income in a particular country is,

ceteris paribus, negatively associated with the decision of a potentialmigrant to choose this country as destination (V 0

ρ <0).3. The marginal income tax rate in a country should, ceteris paribus, have a

negative relationship with the individual’s decision to choose this countryas a migration destination (V 0

t <0).4. The level of public goods provision should, ceteris paribus, have a positive

effect on the probability of an individual migrating to that country (V 0G >0).

2.2 Empirical evidence

There is a number of studies incorporating Tiebout’s notion and attesting to itsimplications.1 However, the Tiebout model remained essentially inactive for overa decade after its first publication. When the model reappeared again in theliterature, it took the form of a testable proposition concerning local fiscalbehavior. Existing tests of the Tiebout model can be grouped into two broad

1 For an extensive review, see Dowding, John, and Biggs (1994) and Oates (2006).



categories. Indirect tests, the most common, have focused on deductive implica-tions of the model. For example, Oates (1969) and Edel and Sclar (1974) searchedfor evidence of capitalization of local tax and service levels in house values.According to Farnham and Sevak (2006), “such studies tend to find that prop-erty taxes (public services) are capitalized negatively (positively) in local housevalues, although the authors disagree on whether such capitalization is evidencefor or against Tiebout’s theory”. On the other hand, Epple, Zelenitz, and Visscher(1978), Yinger (1982), Rubinfeld (1987) and Yinger et al. (1988) criticize capitali-zation studies on the basis that they do not account for housing supply or theadjustment of community boundaries and therefore cannot be used to drawconclusions on the efficiency of the observed distribution of local public goodsprovision.

The number of studies examining the direct effect of fiscal variables onmigration, the main Tiebout–Tullock argument of people voting with their feet,is much smaller. As mentioned above, economic theory states that individualswill choose to relocate to communities where they can maximize their utilitywhich consists of both economic and non-economic factors. However, whenstudying migration decisions, economists typically focus on the impact of eco-nomic determinants while trying to control for non-economic factors (Gabriel,Shack-Marquez, and Wascher 1992; Greenwood and Hunt 1984; Sasser 2010;Kennan and Walker 2011).

A number of papers have been published regarding domestic migrationwithin the United States and Canada. Works by Buchanan and Goetz (1972),Flatters, Henderson, and Mieszkowski (1974), Starrett (1980), Shaw (1986), Islam(1989), Day (1992), Ott and Shadbegian (1993) and Day and Winer (2012) examinethe impact of aggregate government spending variables on the number ofimmigrants a county attracts.

Surprisingly, there is little evidence regarding the fiscal aspects of migrationwithin the European Union. Cuthbertson, Foreman-Peck, and Gripaios (1982)and Twomey (1987) test the Tiebout hypothesis with data from different bor-oughs in the United Kingdom and find that government spending has a sig-nificant impact on location decisions. Evidence that fiscal factors affectinternational migration is included in the Liebig and Sousa-Poza (2006) study,where the Tiebout hypothesis is examined using data from migrants in differentparts of Switzerland. Recent papers by Peridy (2006) and Karidis and Quinn(2006) examine the impact of fiscal factors on flows of immigrants into theEuropean Union. Peridy (2006) finds among other things that public spendingin the EU is a significant variable in accordance with the welfare magnet theory.Karidis and Quinn (2006) propose that a country’s fiscal policies regarding taxesand spending affect international migration flows. In a recent paper, Jackson,



Ortmeyer, and Quinn (2013) examine both government spending priorities andspecific social programs on international migration flows. They find that publicspending has a significant impact on total migration flows.

3 Data and empirical framework

According to our theoretical framework, the individual seeks to maximizehis/her utility by choosing to migrate to a specific country instead of another.The ith migrant derives utility by moving from country A to country Baccording to:

Vij ¼ β0jzþ ej where j ¼ A;B and only if Vi

B >ViA ½9�

Although we cannot observe the individual utilities, individual choices (Y¼0,1)can be observed and reveal that if Vi

B >ViA, then Y ¼ 1 (migrate), otherwise Y¼0

(non, migrate, in which case ViA >V

iB). If country B is chosen as migration

destination, then

Prob Y ¼ 1jz½ � ¼ Prob ViB >V

iAjz

� � ½10�z is a vector of characteristics specific to the region. The conditional choicemodel will provide information on which characteristics in vector z played asignificant role in the individual’s decision. Y will take the value of 1 for thechosen country and 0 for the ones he/she rejects, i.e. the rest of the choices. Ifwe assume that Yi is a random variable indicating the choice made, McFadden(1974) has proven that under certain assumptions

ProbðYij ¼ 1Þ ¼ eβ

0zij

PJj¼1

eβ0zij

½11�

The McFadden model considers the effects of choice characteristics (countrydifferences in our case) on top of individual ones, allowing for the unobserveddifferences to play a role when determining the probabilities. The conditionallogit model performs a maximum likelihood estimation of models withdichotomous dependent variables. Using the observed revealed choice (in thiscase the destination country), it captures the unobserved attributes of thechoice. It is in a sense a fixed effects model and in our case it helps identifyingthe unobserved region specific factors that induced the migrant to choose adestination country.



If we denote by Y ¼ 1, the individual’s choice to move from country A tocountry B, eq. [10] is proven to be equal to

Prob Y ¼ 1jz½ � ¼ Prob β0zþ ε>0jz½ � ½12�where β equals βB – βA.

The conditional choice model deals with a number of problems we mighthave had encountered had we chosen to employ either a linear regression oranother logistic approach. It overcomes the problem of serial correlationbetween locations by assuming heterogeneity across them. It also allows formoving beyond the potential problems of a linear probability estimation such aslack of efficiency of the estimating method and the potentiality of the condi-tional expectations lying outside the [0,1] interval. Country specific character-istics can be used instead of individual ones as the multinomial choice modelwould imply. Finally, as pointed out by Davies, Greenwood, and Li (2001), usingthe nested logit model as developed by Maddala (1983) and McFadden (1984), inwhich a person’s migration is treated as two separate decision procedures (moveor stay followed by the choice of destination), would imply that the migrationdecision is made sequentially. We agree with the authors that a decision aboutmigrating is intimately related to the possible destination choices.

The four hypotheses of the theoretical model are tested for the period 2000–2010. In our approach, Tiebout’s notion of local becomes national and thenational becomes international. Countries are included in the sample upontheir entrance in the European Union. Therefore, growing overtime, the samplebecomes an unbalanced panel. The number of countries goes from 15 in 2000 to25 in 2004, reaching a maximum of 27 in 2007. The total number of country tocountry available pairs over that period is included and it is 3,081. The structureof the conditional model suggests that it is irrelevant whether nonexistentcountry pairs are included or not. The model is tested based on the availablemacroeconomic level data for the countries in our sample.

The ability of the conditional logit to weight observations is importantbecause of the way in which the dependent variable is constructed. The depen-dent variable is a dichotomous zero/one variable representing the migrationchoice. With the data in this paper being macroeconomic flows and not amicroeconomic survey, the level of observation is country to country and notindividual. Each observation is a country pair in a given year.

For instance, let us consider the case of Austria in 2005. Let us assume thatin that year, a total of 3,000 Austrian residents decided to migrate. Let us furtherassume that 1,500 Austrians went to Spain, 1,000 to the UK and 500 to Greece.Austria will have nine country pairs for 2005 [(Austria–Spain)*3, (Austria–UK)*3and (Austria–Greece)*3]. The choice variable will take the value of 0 or 1 for



each of the pairs. It will be weighted by the number of individuals making thatchoice. For instance, there will be an Austria–Spain observation equal to 1 witha weight of 1,500 on that observation. This represents the fact that 1,500individuals chose this Austria–Spain migration. At the same time though, theyrejected UK and Greece as their destination so for the pairs Austria–UK andAustria–Greece the choice variable will have the value of zero with a weight of1,500. The process continues until the dependent choice variable is constructedfor 3,081 country pairs.

Data on migration flows are obtained from OECD’s International MigrationDatabase. All explanatory variables are introduced in the model in differencesbetween the destination and the origin country. Data are obtained from Eurostat.The hypotheses are tested with government expenditure and tax variables asratios. We follow Buchanan and Goetz (1972), Ott and Shadbegian (1993) andKaridis and Quinn (2006) in using the fiscal surplus variable (Fiscal) which isdefined as the ratio of spending to tax revenue, both as percentage of GDP.Fiscal surplus variables are constructed for origin and destination countries andtheir difference (destination – origin). Results produced in this paper are con-sistent with the use of tax and spending separately as suggested by Fox, Herzog,and Schlottman (1989). We expect an increase in the difference of fiscal sur-pluses between the destination and home to increase the probability of migrat-ing to that particular country and therefore to have a positive sign.

The effect of labor earnings (Earnings) on the probability under question iscaptured by a net earnings variable which is also entering the model in the formof the difference between the destination and the origin country. Annual netearnings are valued in Euros. We choose to employ this variable instead of thetraditionally used GDP per capita for the sake of consistency with our theoreticalmodel and because we believe that the part of a person’s decision to migratewhich is induced by the income factor is closely related to labor income earnedat home compared to what can potentially be earned abroad.

We believe that a main contribution of this study to the existing literature isthe evaluation of consumer confidence (Confidence) as an additional impactfactor on the relocation choice. Although there has been an extensive andenlightening empirical literature on the issue of economic fundamentals andtheir effects on output fluctuations, more needs to be done in order to captureadequately some crucial aspects of agents’ decision making. Confidence has noeffect on the real economy, under the standard neo-classical economics assump-tion on production and utility function, that is, a unique equilibrium is deter-mined. On the other hand, when sunspot equilibrium exists, a random variablethat does not directly affect economic fundamentals (i.e. preferences, technol-ogy, etc.) can cause economic fluctuations (Utaka 2003). There is a growing



literature on confidence indicators and their use in monitoring or forecastingshort-term economic developments.2 In other words, output can fluctuate justbecause everyone expects it to Matsusaka and Sbordone (1995). In this respect, a“pessimistic” agent would be keener to migrate, if he/she expects a lowerquality of life either in terms of income/employment or in terms of publicprovisions in his/her home country as a consequence of a worsening economicenvironment.

We circumvent the obstacle of measuring expectations, by utilizing thewidely used consumer confidence indicator (CCI hereafter), which relieves usfrom having to rely on ad hoc assumptions regarding the expectations’ forma-tion mechanism. Although business surveys3 have a relatively longer historythan consumer ones, much of the empirical research on sentiment surveysfocuses on the expectations of consumers. Such surveys receive considerableattention because of their potential to be utilized as indicators of future eco-nomic performance and the overall state of the economy. Hence, in order todetect turning points in the economic cycle and the overall agents’ expectationswe include the CCI which is taken from Eurostat and is the arithmetic average ofthe seasonally adjusted balances (%) referring to the questions on the financialsituation of households, general economic situation, unemployment expecta-tions (with inverted sign) and savings, all over the next 12 months. The variableenters the model in standardized form with a mean of zero and a standarddeviation of one.

The present study explores the information contained in the consumersentiment survey conducted for the European economies. In particular, we tryto determine how people’s expectations, regarding the difference in economicperformances between the origin and the destination countries, are related tothe formation of a migration trend. We call this the indirect effect on migration,and we believe that it goes beyond the traditional direct tests of how economicfactors affect migration patterns.

The extent to which sentiment indicators can, in a sense, forecast economicactivity has been a recurrent and controversial topic in economic research. Anumber of researchers advocate that the predictive capacity of consumer senti-ment indices provides little gain in forecasting real economic activity (see e.g.Stock and Watson 1993; Emerson and Hendry 1994; Weale 1996; Fan and Wong1998; Bram and Ludvigson 1998; Howrey 2001). Against this line, a vast number of

2 For an extensive review, see Kallandranis (2007).3 For a review on business surveys regarding Economic Sentiment Indicator as a forwardlooking indicator see Mourougane and Roma (2003), Hüfner and Schröder (2002), Drakos(2006) and Drakos and Kallandranis (2005, 2007).



empirical studies have found a consistent connection between CCIs and economicactivity. They find evidence that overall sentiment indicators summarize prior orcontemporary information and that consumer confidence has significant explana-tory power for the growth rate of consumption (see e.g. Acemoglu and Scott 1994;Carroll, Fuhrer, and Wilcox 1994; Matsusaka and Sbordone 1995; Berg andBergstrom 1996; Huth, Eppright, and Taube 1994; Lovell and Tien 2000; Dunnand Mirzaie 2006; Easaw and Heravi 2004; Easaw, Garratt, and Heravi 2005; Celik,Aslanoglu, and Uzun 2010). Thus, they argue, what happens in the real world hasrepercussions on the behavior of consumers.

Two other variables are employed in our set of factors that might inducepeople to choose a specific migration destination. Price level differentialsbetween home and abroad are considered in the form of a Price Level Index(PLI) derived from Purchasing Power Parities against the European average, andthey are expected to negatively influence the decision to migrate. In an effort tocapture any potential competition that the migrant might face in the destinationcountry we utilize an education variable as well (Education). The total numberof tertiary education graduates (per 1,000) of population aged between 20 and29 years of age gives us a good proxy of how tough the competition can be forpeople who will try to establish themselves in a new market. We argue that thehigher the skills in the destination country’s labor market, the lower the prob-ability of being competitive for people who exhibit a moderately risk neutral orrisk averse behavior. The expected sign for this variable is negative, indicatingthat a more skilled labor force in the destination country could be a discoura-ging factor for a potential migrant.

Finally, there is evidence in the literature that migration flows depend,ceteris paribus, on whether a common language is spoken between the originand the destination countries (Greenwood and McDowell 1991). To capture thepossible relevant effect we are using the corresponding variable (Language)which takes the value of 1 when a common language is spoken between theorigin and the destination countries and 0 elsewhere. We believe that a commonlanguage adds a level of comfort for the potential migrant and shortens theadjustment period. Therefore, we expect to find that it will have a positive effecton the decision of an individual to migrate to a specific destination.

Given the framework provided, the migration decision is shown as:

ChoiceAB ¼ b0 þ b1Fiscalþ b2Unemploymentþ b3Earningsþ b4Confidence

þ b5PLIþ b6Educationþ b7Languageþ ε

½13�Descriptives are shown in Table 1.



4 Empirical results

Estimation results are largely consistent with our expectations. This holds for allalternative specifications of the basic model as well, which are presented in thesecond part of this section. The variables of main interest, such as fiscal surplusdifferences and consumer confidence, justify their use and seem to play animportant role in the decision making process of potential migrants. All vari-ables are statistically significant mainly due to the way the conditional logitmodel incorporates the weight variable into each country pair observation.

Results of the basic model are presented in Table 2. All variables are statis-tically significant at the 1% level and of the expected sign, with the exception ofthe unemployment variable which surprisingly turns out to be positive. Our fiscalvariable being the largest in size and of the expected sign confirms the existingliterature on people’s decision to migrate, indicating that an increase of a onepercentage unit in the fiscal surplus differential will increase the probability ofmigrating to the specific destination by 11.1 percentage points (pp). This result

Table 1: Descriptive statistics

Variable Mean Std. Dev. Min Max

Fiscal surplus origin 1.07 0.11 0.86 1.88Fiscal surplus destination 1.05 0.09 0.86 1.88Fiscal surplus difference −0.02 0.11 −0.88 0.82Unemployment origin 7.83 3.34 2.3 20.07Unemployment destination 7.62 3.33 2.3 20.07Unemployment difference −0.21 4.53 −15.67 15.67Net earnings origin 29.70 16.15 3.22 54.99Net earnings destination 35.64 14.94 6.45 54.99Net earnings difference 5.92 22.69 −44.76 50.44PLI origin 92.13 24.59 40.1 140.5PLI destination 99.75 22.15 48.8 140.5PLI difference 7.62 33.17 −85.3 96.00Consumer confidence origin −13.09 17.01 −64.38 23.51Consumer confidence destination −11.29 17.09 −64.38 23.51Consumer confidence difference 1.84 24.46 −83.2 87.8Education origin 58.65 18.73 12.1 100.2Education destination 56.92 17.50 12.1 100.2Education difference −1.52 24.18 −77.90 77.90Language 0.12 0.33 0 1

Notes: Mean, Std. Dev., Min, Max, stand for sample average, standard deviation, minimum andmaximum, respectively, for all pairs.



confirms prior findings in the intra-country context and that of Karidis and Quinn(2006) and Peridy (2006) in an international setting. However, in our case,extending the sample to include all 27 countries members of the EuropeanUnion and utilizing the fiscal surplus variable in the context of spending–revenuetrade off gives a new boost to the Tiebout–Tullock original idea of people votingwith their feet.

Our second variable of interest, consumer confidence, has also a positiveand significant effect on the decision to migrate. People’s perception about theirability and willingness to consume in a certain environment, as measured by ourconsumer confidence variable, is proven to affect the decision making processwhen choosing a destination. An increase of a one standard deviation in thedifference of our consumer confidence variable between home and abroadincreases the probability of migrating by 2.1 pp.

As anticipated, the net earnings difference between the two countriesinduces people to choose the country where they believe they will obtain higherlabor income while the opposite is true for the cost of living as this is measuredby the PLI variable which gives a negative and significant result.

As mentioned in the empirical formulation of our model, we expect thatpeople consider their potential competition in the labor market when they areabout to make a decision about migrating. A higher level of tertiary educationgraduates would be a discouraging factor in the process of choosing one’sdestination. Results show that an increase in the differential between highereducation graduates between home and abroad by 1 pp will decrease the

Table 2: The decision to migrate: dependent variable is choice

Variable Estimated coefficient SE Marginal effect

Fiscal surplus 1.815 0.0070 0.111Unemployment 0.138 0.0001 0.008Net earnings 0.114 0.0001 0.007PLI −0.050 0.0007 −0.003Consumer confidence 0.349 0.0006 0.021Education −0.027 0.0002 −0.002Language 1.383 0.0012 0.084

Pseudo R-squared 0.18LR chi2(7) 6.18Observations (country pairs) 3081p(1–p) 0.061

Notes: All variables are significant at the 1% level. All LRs are expressed in millions.



probability of migrating to that particular destination by 2 pp. Finally, ourfindings regarding the effect of a common language between the origin andthe destination country are consistent with our expectations and the existingliterature: the presence of a common language will increase the probability ofthe relevant destination country to be chosen.

4.1 Robustness tests

4.1.1 Income groups

The effects of our determinants on the decision to migrate are tested for a total of27 countries over a period of 11 years. Taking into account the substantialdifferences in country characteristics in terms of economic structure, economicpolicy schemes as well as individual potential behavior, as those are revealed byexamining the descriptive statistics of our variables, we believe that the robust-ness of our model should be examined after considering those disparities, atleast to some extent. In addition, the whole character of the European Union hasbeen altered after the substantial enlargement that took place in 2004 followedby a smaller increase in the number of members in 2007. We find it plausible toconsider the year 2004 as a structural break in the Union’s composition andtherefore the two resulting sub-periods (pre- and post-enlargement) could revealattributes of the choice that are not visible when the whole sample is considered.

For the above reasons we conduct two kinds of robustness tests. Firstly, weplace all country members in three income groups, according to their GDPper capita and try to identify different patterns of behavior within those groups.The criterion used for grouping the countries is pretty standard in the literature.Countries within a standard deviation of the Union’s average GDP per capita areconsidered to be middle income while the rest are classified as low and highincome if their income is below and above the average by more than a standarddeviation, respectively. This classification is presented in Table 3 and refers tothe period 2000–2010.

Results are fairly consistent throughout all three samples. They are reportedin Table 4. When low income countries are considered, the effect of our fiscalsurplus variable strengthens. A one percentage unit change in the fiscal surplusdifference between origin and destination increases the probability of migratingto that particular country by 19.1 pp (compared to 12 pp when the whole sampleis considered). Potential migrants seem eager to exploit fiscal differences, andthey choose to move to a country that offers a more generous and cheaper fiscalscheme. Similarly, they are interested in securing their income by moving to a



country where consumer confidence is revealed to be higher. An increase inconsumer confidence differential by one standard deviation increases the prob-ability of choosing the specific destination by 2.5 pp. Net earnings differentialseems to be a decisive factor as well for low income countries. High percentageof tertiary education graduates discourages potential migrants, as expected,while a common language remains a strongly encouraging factor. On the otherhand, the unemployment rate differential continues to surprise us by coming outpositive and significant.

Similar conclusions can be drawn when we look at middle and high incomecountries. It is worth noticing how our fiscal surplus variable, although itremains positive and significant, loses a good part of its strength in terms ofsize when origins in the top income group are considered. A plausible explana-tion would be that people from higher income countries, being more confidentregarding their potential earnings in the destination, are less interested in whatthe government can provide. Little attention is paid to educational differencesbetween home and abroad when it comes to moving from high income countriesleading us to a speculative proposition that high skilled people in the agesbetween 20 and 29 are particularly mobile and not afraid to face the competition.

There is an apparent consistency among all three income groups when itcomes to consumer confidence and net earnings. Residents of all countries are

Table 3: Country classification according to GDP per capita average GDP per capita for the union(2000–2010): $26910

Groups Income Country

Class 1: “Low income” Yj � �Y<� σj RomaniaBulgariaLatviaPoland

LithuaniaHungaryEstonia

Class 2: “Middle income” Yj � �Y< σj�� Slovak Republic

PortugalMaltaCzech RepublicGreeceSloveniaCyprus

SpainItalyFinlandFranceGermanyUnited KingdomBelgium

Class 3: “High income” Yj � �Y>σj DenmarkSwedenAustria

NetherlandsIrelandLuxembourg

Source: Eurostat.



Table4:

Thede

cision

tomigrate:incomeclasses¼

1–3,

depe

nden

tvariab

leis

choice

Variab

leClas

s1

Clas

s2

Clas

s3

Estimated

coefficien

tSE

Margina

leffect

Estimated

coefficien

tSE

Margina

leffect

Estimated

coefficien

tSE

Margina

leffect

Fiscal

surplus

3.18

60.0111

0.191

0.946

0.0108

0.057

0.580

0.026

50.037

Une

mploy

men

t0.138

0.0002

0.008

0.146

0.0002

0.009

0.172

0.0006

0.011

Net

earnings

0.164

0.0002

0.010

0.089

0.0001

0.005

0.108

0.0003

0.007

PLI

−0.085

0.0001

−0.005

−0.031

0.0001

−0.002

−0.033

0.0002

−0.002

Con

sumer

confiden

ce0.417

0.0009

0.025

0.302

0.0010

0.018

0.266

0.0020

0.017

Education

−0.025

0.0004

−0.002

−0.025

0.0004

−0.001

−0.023

0.0008

−0.001

Lang

uage

2.49

40.0021

0.150

0.717

0.0018

0.043

1.144

0.0035

0.072

Pseu

doR-sq

uared

0.24

0.13

0.24

LRch

i2(7)

4.38

1.62

0.76

Obs

ervation

s(cou

ntry

pairs)

719

1594

768

p(1–

p)0.060

0.060

0.063

Notes:Allvariab

lesaresign

ifican

tat

the1%

level.AllLR

sareexpressedin

millions

.

Q16



seeking to secure their incomes in their destination in a better way than they doat home and that is shown by the relatively large size of the coefficient of ourconsumer confidence variable in all three cases. Finally, higher price levelindices at destination are, as expected, a discouraging factor when it comes tochoosing a destination.

4.1.2 Structural break 2004

Observing migration flows before and after the enlargement of 2004 leads to theexpected substantiation of the fact that along with the higher number of possibledestination there was a substantial increase in the number of migrants from 1.3million between 2000 and 2003 to 6.8 million in the period 2004–2010.Regression results are presented in Table 5.

Although the number of alternative destinations has increased, the determi-nants of choosing a particular country to migrate are consistent with the initialmodel specification throughout the period under consideration. In both periods,our fiscal surplus variable is proven to be the most significant factor in choosingthe destination country followed by the CCI. Prior to 2004 and given the longerhistory of population mobility among European countries, people seem to place

Table 5: The decision to migrate: pre and post 2004, dependent variable is choice

Variable Pre 2004 Post 2004

Estimatedcoefficient

SE Marginaleffect

Estimatedcoefficient

SE Marginaleffect

Fiscal surplus 5.336 0.0245 0.342 1.549 0.0076 0.093Unemployment 0.184 0.0005 0.012 0.134 0.0002 0.008Net earnings 0.094 0.0002 0.006 0.124 0.0001 0.007PLI −0.019 0.0002 −0.001 −0.059 0.0001 −0.004Consumer confidence 0.392 0.0016 0.025 0.366 0.0007 0.022Education −0.016 0.0008 −0.001 −0.028 0.0001 −0.002Language 0.533 0.0029 0.034 1.629 0.0014 0.098

Pseudo R-squared 0.12 0.20LR chi2(7) 0.58 5.87Observations (country

pairs)904 2177

p(1–p) 0.064 0.060



a very high importance on the fiscal package offered, providing a very strongverification to the Tiebout–Tullock argument.

All variables are significant and of the expected sign with the exception ofthe unemployment rate once again. In fact, the consistency with which theunemployment rate contradicts our expectations should lead us to alternativeconsiderations. It seems that other characteristics of the home and destinationcountries outweigh the negative impact that unemployment might have on one’sdecision. People seem to be more interested in securing their income throughfiscal provisions and are willing to compete for a place in the labor market,probably being confident enough to do so due to their skills. In addition, acommon language seems to be providing a substantial level of comfort to thepotential migrant. Although it goes beyond the scope of this paper, it would beparticularly interesting to estimate the model using instead of the unemploy-ment rate, the unemployment spell. A possible consideration from the part of thepotential migrant of the duration he/she would have to wait until finding a placein the labor market, while being entitled to fiscal provisions as it is the case forEU citizens across the Union, would provide a useful insight on the reasons theunemployment rate variable seems to work in a way that is unanticipated in ourmodel.

In conclusion, and in both sub-periods, potential migrants, when in theprocess of choosing a destination country, seem to prefer a cheaper destinationwhich would provide them with a higher fiscal surplus, and satisfactory netearnings which they would be able to secure when they find a place in the jobmarket. They seem to be more confident when there are no language barriers.They also believe that they have better chances in doing so by moving to acountry where the number of higher education graduates is comparable or worseto theirs, indicating that their skills can be appreciated in their destination ofchoice.

5 Concluding remarks

This paper uses a conditional logit model to investigate the determinants ofintra-European Union migration. The results show that, in all sample specifica-tions, fiscal policies, as measured by our fiscal surplus variable, as well as theconsumer–voter sentiment, derived from the confidence indicator used, play asignificant role in a potential migrant’s decision making process when choosinga destination. Other factors as net earnings, prices, education and a commonlanguage seem to also have an effect on that process.



Our results are consistent with prior studies indicating the importance offiscal policy variables when testing Tiebout’s notion of moving with one’s feet.However, with the inclusion of the consumer confidence as an additional impactfactor on the relocation choice, we attempt to move beyond the traditionalapproach, by trying to capture the indirect effects of sentiment indicators onthe migration choice. We have established that although fiscal policies areprimarily assessed by the potential migrant looking for a destination that offershigher but cheaper benefits, people’s expectations about the economic perfor-mance at home and abroad have a significant share in the decision makingprocess.

Clearly, there is more to be done in identifying how the trends in Europeanmigration are formed. Literature on the subject, although limited, is growing.Extending the vector of characteristics on which the choice is based, beyond thecountry level, to include personal attributes of the migrant is something that theconditional model can do, when a database of appropriate size is provided.Additionally, when trends in European migration are established along withtheir determinants, efficiency concerns and government policy implicationsshould be considered. After all, being able to reveal one’s preferences whenhe/she assesses government policies regarding public goods provision andtaxation, and the effect of transforming these effects into actions, was Tieboutand Tullock’s main scope of study.

Acknowledgments: We benefited from the helpful comments of an anonymousreferee. We would also like to thank George Koutsoudis of Tesco Bank for hisassistance during the data manipulation process. Responsibility for any remain-ing errors is our own.

References

Acemoglu, D., and A. Scott. 1994. “Consumer Confidence and Rational Expectations: AreAgents’ Beliefs Consistent with the Theory?” The Economic Journal 104:1–19.

Berg, L., and R. Bergstrom. 1996. “Consumer Confidence and Consumption in Sweden.”Uppsala University, Department of Economics Working Papers Series 1996:7.

Bram, J., and S. Ludvigson. 1998. “Does Consumer Confidence Forecast HouseholdExpenditure? A Sentiment Index Horse Race.” Economic Policy Review 4:59–79.

Brucker, H., and P. J. H. Schroder. 2007. “International Migration with Heterogeneous Agents:Theory and Evidence.” Institute for Employment Research Discussion Paper 2007:27.

Buchanan, J. M., and C. J. Goetz. 1972. “Efficiency Limits of Fiscal Mobility: An Assessment ofthe Tiebout Model.” Journal of Public Economics 1:25–43.

Carroll, C. D., J. C. Fuhrer, and D. W. Wilcox. 1994. “Does Consumer Sentiment ForecastHousehold Spending? If So, Why?” The American Economic Review 84:1397–408.



Cebula, R. J. 2005. “Internal Migration Determinants: Recent Evidence.” International Advancesin Economic Research 11:267–74.

Cebula, R. J. 2009. “Migration and the Tiebout–Tullock Hypothesis Revisited.” American Journalof Economics and Sociology 68:541–51.

Cebula, R. J., and M. Z. Kafoglis. 1986. “A Note on the Tiebout–Tullock Hypothesis: The Period1975–1980.” Public Choice 48:65–9.

Celik, S., E. Aslanoglu, and S. Uzun. 2010. “Determinants of Consumer Confidence in EmergingEconomies: A Panel Cointegration Analysis.” Paper presented at the 30th annual meetingof the Middle East Economic Association, Allied Social Sciences Association, Atlanta, GA,January 3–6.

Clark, X., T. J. Hatton, and J. G. Williamson. 2002. “Where Do US Immigrants Come from, andWhy?” NBER Working Paper 8998.

Cushing, B. J. 1993. “The Effect of the Social Welfare System on Metropolitan Migration in theUS, by Income Group, Gender and Family Structure.” Urban Studies 30:325–37.

Cuthbertson, K., J. Foreman-Peck, and P. Gripaios. 1982. “The Effects of LocalAuthority Fiscal Decisions on Population Levels in Urban Areas.” Regional Studies16:165–70.

Davies, P. S., M. J. Greenwood, and H. Li. 2001. “A Conditional Logit Approach to US State toState Migration.” Journal of Regional Science 41:337–60.

Day, K. M. 1992. “Interprovincial Migration and Local Public Goods.” Canadian Journal ofEconomics 25:123–44.

Day, K. M., and S. Winer. 2001. “Policy-Induced Migration in Canada: An Empirical Study”,Human Resources Development Canada, Applied Research Branch.

Day, K. M., and S. Winer. 2012. Interregional Migration and Public Policy in Canada: AnEmpirical Study, Vol. 223. Montreal, Canada: McGill-Queen’s Press-MQUP.

Dowding, K., P. John, and S. Biggs. 1994. “Tiebout: A Survey of the Empirical Literature.” UrbanStudies 31:767–97.

Drakos, K. 2006. “A Note on Uncertainty and Investment across the Spectrum of Irreversibility.”Applied Economics Letters 13:873–76.

Drakos, K., and C. Kallandranis. 2005. “Investment and Cash Flow: Evidence from Greek ListedCompanies.” Applied Economics Quarterly 51:387–96.

Drakos, K., and C. Kallandranis. 2007. “Investment and Cash Flow: Evidence for Asymmetries inEuropean Manufacturing.” Applied Financial Economics 17:1191–200.

Dunn, L. F., and I. A. Mirzaie. 2006. “Turns in Consumer Confidence: An Information AdvantageLinked to Manufacturing.” Economic Inquiry 44:343–51.

Easaw, J. Z., D. Garratt, and S. M. Heravi. 2005. “Does Consumer Sentiment Accurately ForecastUK Household Consumption? Are There Any Comparisons To Be Made with the US?”Journal of Macroeconomics 27:517–32.

Easaw, J. Z., and S. M. Heravi. 2004. “Evaluating Consumer Sentiments as Predictors of UKHousehold Consumption Behavior: Are They Accurate and Useful?” International Journal ofForecasting 20:671–81.

Edel, M., and E. Sclar. 1974. “Taxes, Spending, and Property Values: Supply Adjustment in aTiebout–Oates Model.” The Journal of Political Economy 82:941–54.

Emerson, R. A., and D. Hendry. 1994. “An Evaluation of Forecasting Using Leading Indicators.”Journal of Forecasting 15:271–91.

Epple, D., A. Zelenitz, and M. Visscher. 1978. “A Search for Testable Implications of the TieboutHypothesis.” The Journal of Political Economy 86:405–25.



Fan, C. S., and P. Wong. 1998. “Does Consumer Sentiment Forecast Household Spending?: TheHong Kong Case.” Economics Letters 58:77–84.

Farnham, M., and P. Sevak. 2006. “State Fiscal Institutions and Empty-Nest Migration: AreTiebout Voters Hobbled?” Journal of Public Economics 90:407–27.

Fertig, M. 2001. “The Economic Impact of EU-Enlargement: Assessing the Migration Potential.”Empirical Economics 26:707–20.

Flatters, F., V. Henderson, and P. Mieszkowski. 1974. “Public Goods, Efficiency, and RegionalFiscal Equalization.” Journal of Public Economics 3:99–112.

Fox, W. F., H. W. Herzog, and A. M. Schlottman. 1989. “Metropolitan Fiscal Structure andMigration.” Journal of Regional Science 29:523–36.

Gabriel, S. A., J. Shack-Marquez, and W. L. Wascher. 1992. “Regional House-Price Dispersionand Interregional Migration.” Journal of Housing Economics 2:235–56.

Greenwood, M. J., and G. L. Hunt. 1984. “Migration and Interregional EmploymentRedistribution in the United States.” The American Economic Review 74:957–69.

Greenwood, M. J., and J. M. McDowell. 1991. “Differential Economic Opportunity, Transferabilityof Skills, and Immigration to the United States and Canada.” The Review of Economics andStatistics 73:612–23.

Hatton, T. J. 1995. “A Model of UK Emigration, 1870–1913.” The Review of Economics andStatistics 77:407–15.

Houtenville, A., and K. S. Conway. 2001. “Elderly Migration and State Fiscal Policy: Evidencefrom the 1990 Census Migration Flows.” National Tax Journal 54:103–23.

Howrey, P. E. 2001. “The Predictive Power of the Index of Consumer Sentiment.” BrookingsPapers on Economic Activity 1:175–216.

Hüfner, F. P., and M. Schröder. 2002. “Forecasting Economic Activity in Germany-How UsefulAre Sentiment Indicators?” ZEW Discussion Paper 2:56.

Hunt, G. L., and R. E. Mueller. 2004. “North American Migration: Returns to Skill, Border Effects,and Mobility Costs.” Review of Economics and Statistics 86:988–1007.

Huth, W. L., D. R. Eppright, and P. M. Taube. 1994. “The Indexes of Consumer Sentiment andConfidence: Leading or Misleading Guides to Future Buyer Behavior.” Journal of BusinessResearch 29:199–206.

Islam, M. N. 1989. “Tiebout Hypothesis and Migration-Impact of Local Fiscal Policies.” PublicFinance 44:406–18.

Jackson, A. L., D. L. Ortmeyer, and M. A. Quinn. 2013. “Are Immigrants Really Attracted to theWelfare State? Evidence from OECD Countries.” International Economics and EconomicPolicy 10:1–29.

Kallandranis, C. 2007. “Topics on the Dynamics of Capital and Labour Demand: An EmpiricalAnalysis.” PhD diss., Department of Economics, University of Patras.

Karidis, S., and M. A. Quinn. 2006. “Fiscal Harmonization and Migration in the EuropeanUnion.” Brussels Economic Review 49:333–52.

Kennan, J., and J. R. Walker. 2011. “The Effect of Expected Income on Individual MigrationDecisions.” Econometrica 79:211–51.

Koven, S. G., and M. C. Shelley. 1989. “Public Policy Effects on Net Urban Migration.” PolicyStudies Journal 17:705–18.

Liebig, T., and A. Sousa-Poza. 2006. “The Influence of Taxes on Migration: Evidence fromSwitzerland.” Cambridge Journal of Economics 30:235–52.

Lovell, M. C., and P.-L. Tien. 2000. “Economic Discomfort and Consumer Sentiment.” EasternEconomic Journal 26:1–8.



Maddala, G. S. 1983. Limited-Dependent and Qualitative Variables in Econometrics, Vol. 3.Cambridge, UK: Cambridge University Press.

Matsusaka, J. G., and A. M. Sbordone. 1995. “Consumer Confidence and EconomicFluctuations.” Economic Inquiry 33:296–318.

McFadden, D. 1974. “Conditional Logit Analysis of Qualitative Choice behavior.” In Frontier inEconometrics, edited by Paul Zarembka, 105–42. New York and London: Academic Press.

McFadden, D. L. 1984. “Econometric Analysis of Qualitative Response Models.” Handbook ofEconometrics 2:1395–457.

Mourougane, A., and M. Roma. 2003. “Can Confidence Indicators Be Useful to Predict ShortTerm Real GDP Growth?” Applied Economics Letters 10:519–22.

Musgrave, R. 1939. “The Voluntary Exchange Theory of Public Economy.” The Quarterly Journalof Economics 53:213–37.

Nakosteen, R. A., and M. Zimmer. 1980. “Migration and Income: The Question of Self-Selection.” Southern Economic Journal 46:840–51.

Oates, W. E. 1969. “The Effects of Property Taxes and Local Public Spending on Property Values:An Empirical Study of Tax Capitalization and the Tiebout Hypothesis.” The Journal ofPolitical Economy 77:957–71.

Oates, W. E. 2006. “The Many Faces of the Tiebout Model.” In The Tiebout Model at Fifty: Essaysin Public Economics in Honor of Wallace Oates, edited by William A. Fischel, 21–45.Cambridge MA: Lincoln Institute of Land Policy.

Organization for Economic Co-operation and Development (OECD). 2013. “InternationalMigration Database.” Accessed July 2. http://stats.oecd.org/.

Ott, A. F., and R. Shadbegian. 1993. “Fiscally Induced Migration and Public Goods Provision: AnEmpirical Test of the Buchanan-Goetz Model.” Cahiers D’economie 7:36–57.

Peridy, N. J. 2006. “Welfare Magnets, Border Effects or Policy Regulations: What DeterminantsDrive Migration Flows into the EU?” Global Economy Journal 6:1–35.

Rubinfeld, D. L. 1987. “The Economics of the Local Public Sector.” Handbook of PublicEconomics 2:571–645.

Samuelson, P. A. 1954. “The Pure Theory of Public Expenditure.” The Review of Economics andStatistics 36:387–89.

Sasser, A. C. 2010. “Voting with Their Feet: Relative Economic Conditions and State MigrationPatterns.” Regional Science and Urban Economics 40:122–35.

Shaw, R. P. 1986. “Fiscal versus Traditional Market Variables in Canadian Migration.” TheJournal of Political Economy 94:648–66.

Sjaastad, L. A. 1962. “The Costs and Returns of Human Migration.” The Journal of PoliticalEconomy 70:80–93.

Starrett, D. A. 1980. “On the Method of Taxation and the Provision of Local Public Goods.” TheAmerican Economic Review 70:380–92.

Stock, J. H., and M. W. Watson. 1993. “A Procedure for Predicting Recessions with LeadingIndicators: Econometric Issues and Recent Experience.” In Business Cycles, Indicators andForecasting, edited by James H. Stock and Mark W. Watson, 95–156. Chicago: University ofChicago Press.

Tiebout, C. M. 1956. “A Pure Theory of Local Expenditures.” The Journal of Political Economy64:416–24.

Tullock, G. 1971. “Public Decisions as Public Goods.” The Journal of Political Economy79:913–18.



Twomey, J. 1987. “Local Authority Fiscal Stance and the Pattern of Residential Migration in theNorth West of England.” Applied Economics 19:1391–401.

Utaka, A. 2003. “Confidence and the Real Economy-the Japanese Case.” Applied Economics35:337–42.

Weale, M. 1996. “An Assessment of OECD and UK Leading Indicators.” National InstituteEconomic Review 156:63–71.

Yinger, J. 1982. “Capitalization and the Theory of Local Public Finance.” The Journal of PoliticalEconomy 90:917–43.

Yinger, J., H. S. Bloom, A. Boersch-Supan, and H. F. Ladd. 1988. Property Taxes and HouseValues: The Theory and Estimation of Intrajurisdictional Property Tax Capitalization.Boston, MA: Academic Press.



Assessing the Effect of Sentiment on Fiscally Induced Migration: A Tiebout Hypothesis Approach within EU 27

Documents