The London School of Economics and Political Science The European Institute

Transcription

1 The London School of Economics and Political Science The European Institute MSc European Political Economy 2007/2008 Candidate Number: TRANSPORTATION COSTS AND WAGE CONVERGENCE IN THE ENLARGED EUROPEAN UNION: AN EMPIRICAL AND POLICY-BASED ASSESSMENT OF THE EFFECT OF LOW COST AIRLINE ACCESSIBILITY ON POLISH REGIONAL WAGE LEVELS, Dissertation completed by Candidate Number in completion of the requirements for MSc European Political Economy. Submission Date: Thursday 28 th August Word count: 10,000 words. I agree to my dissertation being made available to future European Institute students: Yes No

2 2 Abstract This study assesses the effect of low cost airlines on wage convergence between Western and Central Eastern European labour markets. An application of the Sjaastad Human Capital model is developed in order to substantiate the hypothesis that migratory externalities, caused by decreased transportation costs, should have a positive effect on wage developments in source migration regions. This hypothesis is tested using an ordinary least squares decomposition, in order to isolate the effect of low cost airline accessibility on wage levels in 379 Polish districts between 2002 and Findings confirm a negative and statistically significant relationship between a region s distance to low cost airline servicing airports, and wage levels, throughout the period in question. The paper concludes by assessing the implications of these findings on present EU Competition and Cohesion Policies, arguing that the effect of migratory externalities on regional development demonstrates the potential value of private public partnerships aimed at promoting low cost airline connections in underdeveloped European regions.

3 3 Table of Contents 1 Introduction Low Cost Airlines and Central Eastern Europe: A Background Literature Review Recent Studies of Migratory Trends in Central Eastern Europe Theoretical Approaches to Transportation Costs, Migration, and Wage Convergence The Effects of Transportation Costs on Out Migration Rates The Effect of Out Migration on Wage Levels in the Source Country Summary of Theoretical Works A Framework Of Migration And Convergence The Migration Decision (MD) Model The Source Country Labour Market (LM) Model Modelling the Hypothesis using the MD LM Framework Empirical Analysis Of The Effects Of Transportation Costs On Wage Convergence Variables and Data Sources Results Of An Ordinary Least Squares Regression Shortcomings of Methodology and Findings Low Correlation between Out Migration and Wage Levels Limited Availability of Control Variables Limits of Focusing on the Polish Case Summary of Empirical Analysis Political Economy Implications for EU Policies Incentive Packages as State Aid The Current Context Incentive Packages as Development Policy An Alternative Approach Summary of developmental implications Conclusions Appendix Works Cited... 35

4 4 1 Introduction This paper considers the wage augmenting effects of low cost airlines on underdeveloped regions in the European Union, with a particular focus on Central Eastern Europe and an empirical assessment of the Polish case. A modified Sjaastad Human Capital migration model and a Neoclassical labour market framework are used in order to model the hypothesis that reduced transportation costs identified by the presence of low cost airlines should positively affect wage levels in regions located closer to regional airports where low cost airlines operate. Using the wage decomposition proposed by García and Molina (2002), this study tests this hypothesis empirically, presenting findings that confirm a statistically significant relationship between regional wage levels and low cost airline accessibility, as measured in kilometre distance between (a) 379 LAU I 1 level districts in Poland and (b) the nearest airport providing low cost airline connections to Western European labour markets. These results suggest that the reduction in transportation cost provided by low cost airlines plays a role in regional wage gains under conditions of liberalized labour mobility. The political economy aspect of these findings is considered in the final section of this paper, which surveys the policy implications for EU Cohesion Policy as well the potential for the development of private public low cost airline partnerships in the context of EU Competition Policy and European integration. The decision to concentrate the empirical analysis of this work on Poland was prompted by two methodological considerations. First, as the geographically and demographically largest of the ten post communist states that joined the EU between 2004 and 2007, Poland offers access to a large dataset with uniform national characteristics. In contrast to other geographically smaller Central Eastern European states, Poland s size means that by 2006, low cost airlines Easyjet and Ryanair serviced connections to regional airports in 9 of Poland s 16 provinces (Figure 1.1). This geographic dispersion reduces Figure 1.1 the likelihood that this study s estimates of wage effects will be upwardly biased as a result of unrelated effects stemming from proximity to national capitals and urban centres, which are the locations where most low cost 1 Formerly NUTS IV.

5 5 airline connections operate from in smaller Accession states. The second consideration in selecting Poland is that of the scale of Polish emigration. Polish out migration to Western Europe exceeds that of other Accession states by a broad margin in absolute terms, with nearly three out of every four Central Eastern Europeans who emigrated to Western Europe between 2002 and 2006 of Polish origin (Appendix A.1). Polish out migration is also highest in relative terms, when compared to the other high population Viségrad states best suited for the comparative regional study conducted here 2. It is expected that this high level of absolute and relative out migration will improve the precision of this analysis, strengthening the expected relationship between reduced transportation costs and migratory externalities. Section 5 of this work, which contains the results of the empirical study, likewise outlines the limitations that arise from the single country approach specified here. This work adopts a number of terms native to migratory studies, which will be defined here as follows. The term region is used to denote a distinct migratory space (Kaczmarczyk, 2005) with consistent institutional and economic characteristics, which in turn push (pull) migrants out (into) the region. One important advantage of this approach (over a strictly nation based approach) is the ability to differentiate and study both subnational and national territorial units when discussing migratory trends. Next, regions from which (into which) emigration occurs are defined as source or out migration regions (host or in migration regions). Finally, this work adopts a broad definition of migratory externalities, which are defined as any and all effects of an individual emigrant s decision to relocate on his or her source region s economy. As will be demonstrated, these externalities fall into two categories: (a) direct effects, which refer exclusively to emigration s consequences for the source region s labour market(s), and (b) indirect effects, which include post migratory factors such as remittances, the brain drain syndrome, and productivity convergence. Further consideration of direct and indirect effects will be made in the course of this study. Having considered the purpose, scope, and terminology of this paper, this section concludes by presenting an outline of its structure. Section 2, which follows, serves to contextualize this study by providing an overview of Central Eastern European and low cost airlines developments during the 2002 through 2006 period that is the focus of this work. Section 3 reviews the existing literature that has previously assessed the relationship between wage convergence, migration, and transportation 2 These include Hungary, Slovakia, Poland, and the Czech Republic. Among the post communist Accession states that joined the European Union between 2004 and 2007, Slovenia and Lithuania exhibit the highest per capita out migration rates approximately 63 and 176 out migrants per 100,000 population, compared with 58 in the case of Poland (Appendix A.1). The interesting cases of Malta and Cyprus, which joined the European Union during this same period, are omitted in this work due to their unique political, developmental, and geographic circumstances.

6 6 costs, both in contemporary Europe and in the theoretical field. Section 4 presents the hypothesis of this study, which is then modeled theoretically using a Sjaastad Human Capital model and a Neoclassical labour market schedule under migratory shock. The hypothesis is empirically tested in Section 5, which contains the findings of an Ordinary Least Squares regression isolating the correlation between regional wage levels and proximity to airports servicing low cost airlines. Section 6 considers the main implications of these findings, assessing current frictions between low cost airlines and EU Competition Policy, and arguing that the migratory externalities of low cost airlines warrant their reconsideration as a developmental instrument promoting the EU s convergence objectives. Concluding remarks, including areas of interest for future research, are contained in Section 7. 2 Low Cost Airlines and Central Eastern Europe: A Background Between 1990 and 2004, the countries of Central Eastern Europe witnessed a dramatic opening of product markets, even as their labour markets continued to face legal and economic barriers to mobility. Since 2004 however, two enabling factors have emerged in the former Eastern Bloc with unique potential to advance labour market integration across the European Union. The first of these factors was the May 1 st Accession of eight former Eastern Bloc nations into the European Union, which began the gradual process 3 of liberalizing legal barriers to entry into highwage labour markets across Western Europe. The second factor was the widespread entrance of socalled low cost airlines into Central Eastern Europe, which lowered transportation costs between Eastern and Western European labour markets by as much as 60 to 80 percent (Lawton, 2002; Barrett, 2004). Both factors are consistent with a significant reduction in labour market frictions, the effects of which economists have traditionally identified with an acceleration of economic convergence. Indeed, surveys of real GDP per capita growth rate differentials between the EU 15 and EU 8 confirm that that such a convergence trend exists: CEE growth rates outpaced Western European rates by an average of 2.94 percent between 1996 and 2006, with recent differentials alone exceeding 4 percent (Appendix A.2). Isolating the causes of this convergence, and the magnitude of those causes, is therefore important for better understanding prospects for future convergence, not only in the new Member States, but also among prospective EU candidate 3 In 2004, the United Kingdom, Ireland, and Sweden were the first EU 15 countries to remove all restrictions for labour mobility from the eight post communist Accession states. In 2006, Finland, Greece, Portugal and Spain lifted labour mobility restrictions, while Belgium, France, and Luxembourg relaxed them (Source: EURES).

7 7 countries. In light of these considerations, the timing of the low cost airlines introduction in Central Eastern Europe makes it an attractive subject of study for assessing the role of transportation costs in promoting Central Eastern European convergence to EU 15 levels of development. Of interest is the fact that the emergence of low cost airlines in the early 1990s was itself a result of the European integration project. The decentralized business model of these airlines was made possible by the 1993 implementation of the European Commission s third and final airlines package, which integrated air travel into the Union s European Transportation Policy (Lawton, 1999: 105). In 1998, the first low cost airline connection was established to Prague, precipitating the widespread introduction of low cost flights to Central Eastern Europe in the immediate aftermath of the 2004 Accession agreements. Warsaw and Krakow opened their first Ryanair connections in 2004, and by 2007, low cost airlines Ryanair and Easyjet alone serviced nineteenth connections between the EU 15 and the ten post communist Member States, including nine to Poland (Appendix A.3). The diversity of connections to Poland point towards an important characteristic of the low cost airline business model: the unique complementary relationship that has emerged between low cost airlines and dispersed regional airports. As a number of studies have pointed out, high barriers to entry at existing major airports, including the existence of so called grandfather rights privileging existing national carriers in the allotment of annual airport slots, uniquely disadvantages new carriers, even in the wake of deregulation (Barrett, 2004; Gröteke and Kerber, 2004). As a result, low cost airlines were prompted to consider former military and regional airports, which were often willing to provide additional financial and contractual incentives in exchange for contracts assuring certain amounts of passenger traffic, continuity of service, and other guarantees (ibid). In addition to using smaller regional airports that directly competed with existing international hubs (such as Stansted and Luton airports in London and Charleroi airport in Brussels), low cost airlines opened connections to remote regional airports located in under served and often under developed aviation markets. To the author s knowledge, no existing study has focused on the broader developmental externalities that these new transportation regimes provide. An attempt to model, assesses, and contextualize these externalities forms the purpose of this paper.

8 8 3 Literature Review Economic consideration of the relationship between labour market frictions and convergence trends dates to the classical economic works of Smith and Cournot, among others 4. This literature review considers more recent studies that have utilized and expanded upon the contemporary Neoclassical approach, in order to provide an overview of both the empirical and theoretical literature informing this work. In doing so, separate consideration is given to two distinct bodies of literature. First, in Section 3.1, attention is directed toward recent studies of Central Eastern Europe, including a range of works that assess the interrelated effects of transportation costs, migration, and wage developments in the new, Post Accession political economy of EU migration. Given the diversity of works that have recently emerged in this field, and given space considerations, focus is concentrated on works which have examined the most recent trends and particularly those considering the Polish case. Section 3.2 then reviews theoretical developments in the field of economic migration research, examining a number of models that have informed this paper s hypothesis of a positive relationship between low cost airlines and regional wage rates. 3.1 Recent Studies of Migratory Trends in Central Eastern Europe Kaczmarczyk (2005) tests for the presence of homogenizing migratory trends in the former Eastern Bloc, providing a comparative overview of migration related economic and demographic indices within the ten post communist Accession states of 2004, as well as Romania and Bulgaria. Kaczmarczyk uncovers substantial contrasts between the states, both in degree of migratory tendency as well as overall development trajectory. These findings point toward what Kaczmarczyk identifies as the absence of a homogenous migratory space within Central Eastern Europe, suggesting the usefulness of differentiating between migratory externalities among the countries. Focusing on Poland, a number of works have recently identified unusually low levels of internal Polish migration, when compared to other post communist EU Member States. Recent demographic statistical works in this field include a comprehensive study by Deichmann and Henderson (2000), which draws two broad conclusions informing the present study. First, since the 4 For insight into the continued application and consideration of classical economic approaches predating the Neoclassical framework, please see W.A. Lewis (1954) and Ghatak (2003: 37).

9 9 early 1990 s, typical migratory developmental trends of rural to urban migration have seen a decline in Poland, resulting in a low level of urbanization by international standards; Deichmann et al. attribute this trend to a severe undersupply of housing in Polish cities (2000: 15). Second, the work also detects a parallel decline in general inter provincial Polish migration, which appears to suggest a low migratory tendency in contemporary Poland. But as Deichmann et al. observe, Polish international migration rates are by and far the highest in Central Eastern Europe. The authors consequently note that the factors constraining internal migration such a low availability of housing and under urbanization may be an important national push factor encouraging international migration. Finally, an econometric study by Ghatak, Mulhern and Watson (2007) complements these findings by considering the determinants motivating Polish migration. The work s findings confirm the observations made by Deichmann et al., suggesting that housing availability in the destination region is the most important pull factor for internal Polish migration, followed by source region income levels (GDP per capita), host region income levels, host region unemployment levels, and distance between source and host regions. These findings seem to qualify the unique attraction of international migration in Poland. International migrants are able to circumvent Polish housing shortages, and benefit from wage levels and unemployment levels which are much lower than those in Poland. 3.2 Theoretical Approaches to Transportation Costs, Migration, and Wage Convergence In addition to the empirical studies above, a number of works have taken the approach of modelling the relationship between transportation costs, migration, and wage convergence. Several of the most essential of these works are considered here. For purposes of clarity, this review is divided into two separate subsections. First, Section considers works modelling the effect of transportation costs on migratory rates, often incorporating the assumption of a dualistic economy with asymmetric levels of development. Section then considers studies which model the relationship between out migration and wage levels in the source country. This structure previews the framework modelled explicitly in Section 4, where this paper hypothesizes and models a twopart causal relationship between (a) transportation costs and migration rates, and (b) migration rates and wage levels in the source region.

10 The Effects of Transportation Costs on Out-Migration Rates Neoclassical wage convergence models continues to inform a number of theoretical and empirical studies examining the relation between interregional wage differentials, migration, and convergence. According to the Neoclassical approach, under conditions of liberalized labour mobility, wage differentials between regions converges in the long run (Cournot s Law of One Price) until wage differences between regional real wages reflect three factors: (a) local labour demand, (b) transportation costs, and (c) information costs of relocating from low to the high wage regions (Bauer, 1999: 13). Rosenbloom (1991, 1994, 1998), whose research inspires this study, has demonstrated the use of the Neoclassical approach in the study of interregional wage differentials in latter 19 th century United States. Rosenbloom s findings, which focus on transportation costs, reveal that these barriers played a significant role in perpetuating low wage conditions in the American South. At the same time, Pacific, North Atlantic, and Western European labour markets, interconnected by increasingly dense railroad and maritime networks, exhibited increased convergence across unemployment and wage indices. These results, in keeping with the expectations of the Neoclassical approach, suggest the continued usefulness of considering transportation costs when modelling determinants of wage differentials between asymmetricallydeveloped regions. Although the Neoclassical approach has proven useful in identifying leading causes of wage inequalities between regions, the persistence of inter regional wage differentials (including, notably, between and within European states) has prompted many studies to give greater consideration to the individual utility deliberations of potential migrants. This new approach allows for a more satisfying decomposition of the determinants of migration, incorporating individual or regionspecific transportation costs and psychic costs of relocation. Sjaastad (1962) has conducted foundational work in this field, and his work is widely credited with developing the Human Capital model, which treats the decision to migrate as an investment decision with ascertainable returns. According to this approach, potential migrants calculate the net present value of future earnings in a potential host country (the return from their stock of human capital), less the monetary and psychological costs of relocation. If the returns from relocation exceed a migrant s discounted future income stream in the country of residence, emigration occurs: Net Discounted Return from Migration Monetary Costs of Migration Non Monetary Costs of Migration 1

11 11 Equation (1) forms the basis for the Migration Decision model that will be developed in this work. Like the Neoclassical model, the Human Capital framework prominently incorporates transportation costs as a leading Monetary Cost of Migration. Also important for this work, Sjaastad argues that both gross out migration and net migration (out migration from a region less in migration) are an imperfect estimate of a labour market s ability to remove earning differentials (81). Sjaastad cites the fact that both measures are distorted by flows which include returning disillusioned migrants from previous periods, as well non economically motivated actors, such as retirees. To illustrate this observation, Sjaastad demonstrates that while 1949 out migration figures in the U.S. state of Georgia (the most under developed U.S. state at the time) were substantial, at 62,500, they were complemented by a nearly comparable degree of in migration at, 51,900 (ibid). The heterogeneity of non economically motivated migration inspires the unique approach used in the present study, which uses, instead, a third factor transportation costs to capture the effects of marginal economic migration, without measuring such migration directly. The next section will consider works that have surveyed the effects of such economic migration, particularly on the labour market of the source region, and especially under out migratory shocks such as those caused by reduced transportation costs The Effect of Out- Migration on Wage Levels in the Source Country Deichmann and Henderson (2000) and Taylor (2006), among others, have argued that there are two important developmental consequences stemming from out migration. First, out migration has the potential to ease labour market pressures, reducing surplus labour and unemployment, with the effect of raising wages. Second, once abroad, emigrants contribute to the economic development of their source region, sending home remittances as well as bringing home skills and productivity measures acquired abroad. Remittances have the effect of increasing disposable income and, presumably short run consumption, with the potential to increase long run growth if they are reinvested in the source economy (Taylor, 2006: 2). Factor productivity, on the other hand, is intrinsically associated with long run growth potential. Both the direct labour market effects and the indirect developmental effects identified here have the potential to raise wages in a migrant sending region. By contrast, D Costa (2008) provides an overview of conditions under which out migration may retard development in the out migration region. Among the factors considered, D Costa

12 12 identifies the well known fiscal loss of out migration, as the investment made by national authorities in schooling out migrants is lost from the stock of a region s human capital. In addition, outmigration of young workers shifts the demographics of the source region economy, increasing the age dependency ratio among the population that remains. Finally, D Costa notes that while increased incomes resulting from direct and indirect migratory externalities will likely contribute to economic growth, a wage cost push could lead to crowding out of small and medium sized enterprises by the larger domestic and multinational firms (68). Grubel (1966) complements D Costa s work by considering particular statistical effects that out migration may generate. According to Grubel, whereas the migration of high skilled individuals reduces a nation s stock of human capital (the brain drain effect), the emigration of under skilled workers may positively affect a nation s long term growth prospects by improving the state s capital to labour ratio (270). These migratory externalities, which are beyond the scope of the present study to ascertain, present an important frontier in research which would better ascertain the developmental impact of highly skilled migration to and from European regions with distinct degrees of human capital and productivity Summary of Theoretical Works A survey of positive and negative migratory externalities suggests two broad trends in the academic literature. First, both Neoclassical and Human Capital considerations of out migration stress the effects of transportation costs as an important barrier to labour mobility, with the resulting prediction that a reduction in transportation costs would lead to increased out migration. By contrast, the net effect of out migration on source regions remains nuanced, with substantive economic, statistic, and political economic dynamics. This study draws on both these trends in illustrating a model, developed in the following section, under which wage convergence between a source country and an exogenously specified high wage region (such as Western European labour markets) should result in upward wage convergence within the low wage region. The model focuses on the relaxation of labour market pressures as the primary cause of wage gains; however, the empirical approach which follows in Section 5 uses a unique proxy variable approach which captures net wage effects.

13 13 4 A Framework Of Migration And Convergence The present study hypothesizes that lower transportation costs increase migration, which in turn increases wages in the region of out migration. This section illustrates a two part Rational Choice framework substantiating this hypothesis, structured as follows. First, Section 4.1 develops a pairwise Migration Decision (MD) model, based on the Sjaastad specification, which illustrates a potential migrant s binary decision when determining whether to relocate to a region with higher wages (and, post migration, whether to return to the source region if wage conditions change). Section 4.2 introduces the second part of the framework, demonstrating a standard Labour Market (LM) schedule in order to illustrate short and long run effects of an out migratory shock on regional wage levels. Section 4.3 concludes by demonstrating the hypothesis of this work, showing how a decrease in transportation costs, such as the one caused on by the introduction of low cost airlines, increases migration under the MD model and leads to increased wages under the Labour Market schedule. Although the model specified here is highly technical in nature, its purpose is simple: to demonstrate, using contemporary economic theory, the effect of lowered transportation costs (such as those brought on by low cost airlines) on wage levels in relatively underdeveloped regions such as Central Eastern Europe. Section 5 will then test whether the hypothesis demonstrated here applies in the case of Poland, under the EU s liberalized labour mobility regime. 4.1 The Migration Decision (MD) Model The concept of a Migration Decision is used by Sjaastad (1962), Harris and Todaro (1970), Gabriel et al. (1993), Ghatak (2007), and others to denote the binary choice faced by individual workers in a low wage economy of whether or not to relocate to a potential host region with higher wages. According to the Sjaastad Human Capital model, under conditions of liberalized labour mobility, a potential migrant s Migration Decision may be intuitively modelled by comparing the net present returns from migration, to the monetary and non monetary costs of migration. This intuition may be specified as follows: Migration occurs if:,, 2

14 14 where NDR, the Net Discounted Return from migration, is a function of three factors: the expected host country wage, E[w H (t)]; the potential migrant s present wage in the source country, ; and a discount rate, i. Migration occurs when the Net Discounted Return from migration is greater than the sum of (1) the transport cost of relocating at time t 0, TC(t 0 ), and (2) the non monetary costs of relocation, which are individually specified for each migrant. In fully specifying the Migration Decision model, this section incorporates two refinements proposed by Harris and Todaro (1970) and Djajic and Milbourne (1988). In particular, Harris and Todaro propose modelling a potential migrant s wage expectation in the host region as a function of the region s unemployment rate, u H, at the time of migration t 0. This approach is adopted in the current study. Likewise, Djajic and Milbourne suggest that the discount rate may be modelled logarithmically as the diminishing area of e i, a function of the potential migrant s discount rate i (338). Incorporating these factors, and introducing the assumption that the potential migrant is presently employed 5 at equilibrium wage W H (t) in the source region, one arrives at: Migration occurs if:, ~..., 3 (Net Discounted Return from Migration) (Transportation & Non Monetary Migration Costs) which may be rewritten as: Migration occurs if:, ~..., 4 (Net Discounted Monetary Gain from Migration) (Non Monetary Migration Costs) where, E[W H (t 0 )] is the expected wage in the host region at time t 0, which is a function of the prevailing host region wage and (1 u H (t 0 )), where u H (t 0 ) is the host region s unemployment rate at period t 0. W S (t) is the real wage the potential migrant knows he or she will receive in the source region, at time t, i is the potential migrant s discount rate, n is the number of time periods in the migrant s planning horizon (Ghatak, 2003: 292), a measure of the potential migrant s age and individual characteristics, TC(t 0 ) is the transportation cost of relocating from source to host region at initial time t 0, r is the residual, capturing net non monetary psychic costs from relocation. Without loss of generalization, it is specified with a normal distribution and mean μ > 0. 5 That is, the potential migrant is not surplus labour; their involvement in the economy contributes to Gross Domestic Product.

15 15 Migration occurs at period t 0 if either inequality is valid. Two additional observations also deserve mention here (a full application of the model, incorporating the labour market framework specified next in Section 4.2, will be demonstrated in Section 4.3). First, a decrease in transportation costs has an unambiguous, positive effect on out migration from the source region, ceteris paribus. Second, it is observed that if wages in the source country increase, the model can and does allows for return migration as the relative Net Discounted Return from migration falls relative to monetary and nonmonetary costs of migration. Both consequences have important implications for the second half of this framework, which follows. 4.2 The Source Country Labour Market (LM) Model The second part of this framework assumes that a quantity of labour has emigrated, and demonstrates the effect of this emigration on wage levels in the source region using a simple Neoclassical model. In order to focus on key results, the simplifying assumption of homogenous inputs of capital and labour is made, as per Bauer (1999, 48 9). As before, wage levels in the host region are assumed to exceed those of the source region, and be rigid (fixed) in the short and long term for example, as a result of minimum wage legislation (Harris et al., 1970). In addition, the assumption that emigrants are employed in their source country prior to emigration is maintained. Based on these assumptions, this model considers the short and long run effects of out migration in turn. The primary observations are that wages increase in the short run under out migration, and decrease in the long run as a result of equilibrating forces. The short run effects of an out migratory shock are illustrated in Figure 4.4. Out migration of employed individuals reduces the supply of labour in the source region, causing an inward shift of the labour supply schedule. This results in increased wages and lower quantity of labour employed. Lower input of labour constrains production, causing an inward shift of the transformation curve, as per Figure 4.5. This results in a decrease in production on the national level.

16 16 Figure 4.4 Neoclassical Labour Market under Out-Migration (Short Run) Figure 4.5 Source Region Transformation Curve under Reduced Labour (Short Run) In the long run, two effects coincide to decrease wages. First, reduced population in the source regions decreases aggregate demand, resulting in a further reduction of domestic production. Reduced production results in a lower demand for the inputs of production, including labour. This forces an inward shift of the labour demand curve, as in Figure 4.6. Concurrently, the short run increase in wages arrived at in the short run reduces returns from migration, slowing out migration in the long run and leading to return migration, which causes an outward shift in labour supply. As a result, wages are again reduced. Figure 4.6 Neoclassical Labour Market under Out-Migration (Long Run)

17 Modelling the Hypothesis using the MD-LM Framework The purpose of the framework above has been to provide a theoretical foundation for the hypothesis that a reduction in transportation costs leads to an increase in migration, which in turn results in higher wages in the source region. In order to demonstrate how the framework above substantiates this hypothesis, this section traces the effects of a decrease in transportation costs through the two part framework specified above. Figure 4.7 Out-migration as a Function of Individual Non-Monetary Costs of Migration Suppose the existence a dual sector economy consisting of two regions, specified (for purposes of concreteness) as Warsaw (low wage) and London (high wage). The regions economies are closed, with the exception of labour flows, which occur as per the Migration Decision model illustrated in Equations (3) and (4). Warsaw has population N, all of whom are employed at equilibrium regional wage w S (t 0 ), have the same time horizon and discount rate n and i respectively, and differ only in their non monetary costs of relocation, which are, without loss of generalization 6, normally distributed with mean μ > 0 across the population. London, by contrast, has an exogenously specified and unchanging prevailing wage w H (t), such that w H (t) > w S (t), t. Due to the higher wage in London, no migration occurs from London to Warsaw, and London s unemployment rate is set at zero percent. Finally, A state of equilibrium is assumed in the moments preceding period t 0. This implies that Warsaw s labour market is in equilibrium after all potential migrants whose non monetary costs of migration satisfied the Migration Decision in Equation (3) have migrated to London. This initial migration is equal to area A in Figure 4.7, where r 0 is equal to the Net Discounted Monetary Gain from Migration at time t 0, as in Equation (4). The intuition behind this 6 A stochastic distribution, for example, does not substantively alter the results of this model.

18 18 approach is that all potential migrants whose (heterogeneous) non monetary costs of relocation were less than the Net Discounted Monetary Gain from Migration are located to the left of r 0, and have thus migrated. Now consider the introduction of a low cost airline connection at moment t 0, which lowers transportation costs from Warsaw to London to : 5 Under these conditions, Net Discounted Monetary Gain from migration increases by, so that Net Monetary Gain from Migration is equivalent to: 6 As the right part of Equation (6) illustrates, Net Discounted Monetary Gain from Migration increased by. This shifts rightward to, resulting in increased migration, as illustrated by Area B in Figure 4.7. Area B is positive and nonzero so long as TC 0 and there exist potential migrants with r (, ). The out migration of B employed individuals is felt in the source region labour market at period t 0 as a short run contraction of the labour supply. This results in an increase in wages, as illustrated in Section 4.2. These results demonstrate the hypothesis proposed in this study. Long term effects should partially negate this increase; however, given that the assumptions of a long term production drop off are largely inappropriate for the fast growing Polish economy of , the long effects demonstrated in this section serve a primarily illustrative purpose. The diagram above illustrates the logic of this section graphically. In the following section, the hypothesis is tested in the Polish case for the years 2002 through 2006.

19 19 5 Empirical Analysis Of The Effects Of Transportation Costs On Wage Convergence This section examines the findings of a standard Ordinary Least Squares regression used to measure the magnitude and significance of the relationship between accessibility to low cost airlines and wage convergence in 379 LAU I level Polish districts between 2002 and The results of this regression reveal the presence of a meaningful and statistically significant relationship between the variables. In addition to presenting these findings (Section 5.2), this section provides an overview of the variables used in this study (Section 5.1) as well as a discussion of potential methodological shortcoming (Section 5.3). Section 5.4 concludes. 5.1 Variables and Data Sources 7 The dependent variable used in this study, WAGE, is based on a district level survey of average gross monthly wages across Poland. Figures were adjusted against Poland s annual inflation rate, acquired from Eurostat. Observations are recorded in złoty, rather than a convergence term indexing Polish wages against an EU 15 = 100 term, because the approach of this empirical study is compare low cost airlines effects on Polish districts (with convergence trends captured implicitly), rather than to assess aggregate Polish convergence to EU 15 levels. Given this objective, a convergence term is deemed unnecessary. The primary independent variable in this study, AIRPORT, contains the distance in kilometres between each of Poland s 379 districts and the nearest airport servicing connections by Ryanair or Easyjet. Distances, calculated for the purposes of this work using computer mapping software 8, were adjusted throughout the period in order to account for reduced travel distances made possible by the introduction of new connections. In many cases, these reductions were substantial, with average airport accessibility improving by approximately 100 kilometres between 2004 and 2006 (Table 5.2). Since low cost airlines were not available in Poland until 2004, observations for 2002 and 2003 replicate observations for This decision is both reasonable as well as necessary, 7 Except where noted, all statistical data used in Section 5 was acquired from the Central Statistical Office of Poland (GUS). 8 Google Maps (

20 20 as the first low cost flights to open in 2004 (and hence the basis for observations) were to Warsaw and Krakow, two of Poland s primary international airports. It may be assumed that accessibility to competitive airport services was most readily available at these airports before 2004 as well. Additional control variables, modelled on the wage decomposition provided by García et al. (2002) were included to account for intervening factors, particularly given the likelihood that low cost airlines established connections to comparatively more developed areas. These control variables include: CAPITAL, the gross value in złoty of fixed assets in enterprise per 100,000 residents, which is used to capture capital stock, a leading determinant of long term growth; GRADS, the quantity of upper secondary school graduates per 100,000 residents, which is an education term used to capture the effect of human capital on wage rates; DENSITY, population density, which is included in order to proxy for urbanization as well as price levels (which were otherwise unavailable at the LAU I district level); GENDER, district level gender distribution, which captures the wage effects of gender inequalities within each district; BORDER, a dummy variable identifying regions with geographic proximity to EU 15 (i.e. German) labour markets, by marking districts within Poland s three westernmost provinces 9, as in George (2005); EMI_REG, the quantity of internal (that is, subnational) out migration, which is included to account for the important possibility that regional airport accessibility could inadvertently capture the labour market effects of inter district migration; EMI_INT, international out migration, which is included for comparative purposes; Four annual dummy variables, which were included to capture intertemporal effects stemming from the balanced panel data approach used in this study. 9

21 21 A matrix of correlation coefficients among the variables (Table 5.1) and a comparison of the regional distribution of the variables in 2002 and 2006 (Table 5.2) follows. Several points of particular interest emerge from these two tables. First, the correlations between EMI_REG and EMI_INT, and AIRPORT, are surprisingly low. Reasons for this are considered in the overview of this paper s shortcomings in Section 5.3. Due to the low correlation, the variable is included in the primary regression without the danger that it will confound results for the AIRPORT beta coefficient. A second point of interest is the strong regional contrast between districts sending internal versus districts sending international migration, with border regions responsible for a far larger share of international migration. This suggests the continued importance of geographic proximity and (presumably, non aviation based) international migration. Both observations are of value when interpreting the results of the regression analysis, which follows in the next section.

22 Figure 5.2: Relative Province-level Distribution of Variables in 2002 and

23 Results Of An Ordinary Least Squares Regression This section contains the results of a balance panel data ordinary least squares regression of the following function, as well as a number of additional regressions serving illustrative purposes: _ _ YEAR_06 _05 _04 _03 The results of this regression are displayed in Table 5.3. The primary regression (Regression A, first column) reveals a meaningful and statistically significant relationship between AIRPORT and WAGE at the 5% significance level. Even when accounting for wage augmenting factors, an additional Table 5.3: Estimated Coefficients from OLS Regression Analysis

24 24 kilometre of distance to regional airports is seen to reduce average district wages by groszy 10. To contextualize these results, one may consider the implication that a region 250 kilometres from a regional airport would see monthly wages złoty lower, per month, than a region 50 kilometres from a regional airport, ceteris paribus. At the upper end of this range, such a differential constitutes a nearly 5 percent reduction in average national wages. This, the primary finding the current study, is consistent with the expectations hypothesized in this work, as well as the framework modelled in Section 4. Alternative regressions, including those which contained lagged variables to account for delayed effects, do not substantively alter these findings, with the exception of Regressions E and F. These regressions, which constrain observations to those with AIRPORT observations below and above 100 kilometres, illustrate that the relationship posited in this work is strongly sensitive to distance. Additional research in this direction would certainly prove illuminating. Coefficient estimates for the controlling variables are largely in keeping with expectations, with some exceptions. Most importantly, the measure of human capital that was used, GRADS, is statistically significant, but carries the wrong sign. This is likely the effect of this study s choice of variable (annual secondary school graduates), which does not properly reflect the human capital stock of the broader workforce. Lagging this variable by one year does not meaningfully improve these results (Regression G), and the variable is omitted for comparison purposes in Regression B without substantively altering the fit of the equation. Interestingly, the effect of internal migration persists in the wage decomposition, reflecting a statistically significant positive correlation between internal out migration and wage gains. Problematically, international out migration does not appear to have a statistically significant relationship on wages. This, as well as other shortcomings, are discussed next. 5.3 Shortcomings of Methodology and Findings This section provides an outline of three important shortcomings in the methodology and findings of this study. First, the relationship between international out migration and distance to regional airport is problematic, with a correlation coefficient of just This complicates the causal relationship between transportation costs and migration specified in the theoretical outline. A second weakness stems from a lack of data. This is a problem, because distinctly higher wages are Groszy = 1 złoty. For comparison, the average nominal exchange rates of the złoty against the euro and the U.S. dollar were 4.08 and 3.45, respectively.

25 25 present in the vicinity of regional airports, which are themselves often based in regions industrial or commercial centres (so called Regional Hubs ). This reinforces the importance of the accompanying set of dependent variables. Unfortunately, these dependent variables were limited by the incomplete statistical series available at the degree of regional specificity employed by this study. Lastly, the study s focus on the Polish labour market limits the broader applicability of the study s findings to regions with certain stylized characteristics, including high out migration rates, relatively low internal migration, and a dispersed availability of regional airports. The ramifications of each of these factors is examined below Low Correlation between Out-Migration and Wage Levels The low correlation between international out migration and airport proximity contradicts part of the theoretical framework developed in this study, challenging the hypothesis of a causal relationship between out migration and wage gains modelled in Section 4. This is the case despite the fact that the broader relationship between low cost airlines and wage growth is otherwise confirmed by the empirical findings of this study. At least two possible explanations exist to explain this apparent contradiction. First, as Sjaastad has argued, migratory statistics are often a poor measure of economically motivated migration. This is due to the confounding effects of return migration, non economically motivated migration, as well as temporary migration not captured in official statistics. A second possibility is that the indirect externalities of migration, such as remittances and productivity convergence, play a more significant role in wage convergence than the direct effect of labour market adjustment that was modelled in Section 4.3. Taylor (2006), for example, has argued that remittance driven consumption alone can in theory increase local incomes by a factor of 2 3, much in the manner that Keynesian fiscal multiplier can lead to short term income growth (8 9). Both possibilities are reasonable, and future research in this area would be instructive Limited Availability of Control Variables A second shortcoming of this work stems from the possibility that the correlation between high wage regions and airport placement (the effect of airline connections being established in socalled Regional Hubs ) was not fully controlled for by the proxy variables included in this study. As Figure 5.4 demonstrates, districts within the so called 100km catchment area (Soltész, 2006: 724) of an airport appear to exhibit higher wages, particularly within a 25 kilometre radius of the airport,

26 26 with regions within 25 kilometres of a regional airport exhibiting wages over 20 percent higher than those found in regions 75 to 100 kilometres from an airport 11. These results, though expected, underline the importance of controlling for all developmental factors when using distance to airport as an instrument to measure low cost airline effects on development. Certain control variables, however, were unavailable. The most important of these were regional price levels, which would be expected to reveal lower wages as well as prices in regions further from Regional Hubs, without affecting purchasing power. This effect is proxied, imperfectly, using population density to identify urban zones, which characteristically feature higher wages and costs of living. In addition, a second factor of particular relevance for this paper historic and current PHARE 12, Structural Fund and Common Agricultural Policy payments by the European Union was unavailable at the LAU I district level. Both factors can be expected to play a significant wage augmenting role in affected regions. Unlike the effect of Regional Hubs, however, there is little reason to suspect systematic correlation between EU expenditures and placement of regional airports. The omission of EU expenditures is therefore likely to have only a limited distorting effect on the findings contained in this work Figure 5.4 Real Monthly Gross Average Wages and Salaries (zł), by Distance to LCA Airport in 2006 (21 observations) (33) (66) (72) (69) (46) (42) (10) (8) (6) (2) (4) Real Monthly Gross Average Wages and Salaries (number of observations above) 0 11 Author s calculations. 12 Poland and Hungary: Assistance for Restructuring their Economies

27 Limits of Focusing on the Polish Case A third shortcoming of this study lies in its empirical focus on the Polish case, which limits the immediate extrapolation of findings to countries with certain stylized characteristics. For example, migration externalities stemming from low cost airline accessibility will necessarily be a function of national migratory tendencies, which may be expected to differ across countries. This suggests that developmental externalities from low cost airlines will be greater in high out migration states (for example, Poland, Lithuania, and Slovenia) than low out migration states. In addition, the trends identified in this paper may change rapidly as wage differentials between Eastern and Western Europe fall, or as the Central Eastern European Member States adopt the low intra European migration rates currently characterizing older EU Member States (Puga, 2001; Baldwin, 2006). A final limitation of focusing on Polish case stems from the unique geographic dispersion of Polish regional airports. The effects of regional low cost airline servicing airports are expected to be particularly pronounced in rural and under developed regions. Smaller states, with relatively higher levels of urbanization and proximity to Western labour markets, such as the Czech Republic, may exhibit less pronounced benefits from low cost airline networks than, for example, Romania or Bulgaria. 5.4 Summary of Empirical Analysis Despite the shortcomings identified above many of which are endemic to the case study format this study nonetheless observes a large, statistically significant, and believable relationship between regional airport accessibility and wage levels. In particular, these findings confirm the hypothesis of this work by demonstrating that proximity to regional airports servicing low cost airlines matters, with each kilometre of additional distance decreasing average district wage levels by approximately 33 to 41 groszy. Given that the introduction of new regional airports more than halved this distance in the 2002 to 2006 period reducing average distance from 211 km to 104 km it is likely that the effect of low cost airlines was an important factor in promoting economic development in the Polish regions in the years following the political integration that brought Poland and seven other Central Eastern European states into the EU. The next and final section explores this concept further, arguing that the findings of this paper, if accurate, provide a new approach to convergence and development in an enlarged European Union.

28 28 6 Political Economy Implications for EU Policies In the budgetary period, the European Union will allocate 29 percent of its expenditures towards the promotion of economic convergence between the European regions over 250 billion euros in total (Leonardi, 2005: 156). Given the scale of these expenditures, it is important to consider whether the developmental impact of low cost airline networks demonstrated in this work can contribute to achieving European Union objectives in the area of inter regional convergence. This final section draws on the empirical findings of this paper in order to address this question. It argues that the results of this work, if correct, imply significant developmental externalities from low cost airline connections. As a consequence, it will be demonstrated that the use of government assistance to financially incentive low cost airline connections in underdeveloped regions (defined here as the provision of Developmental Incentive Packages ) has potential to become an important instrument in the EU s developmental policy. In order to relate these implications to Europe s contemporary political economy, this section proceeds in two parts. First, Section 6.1 discusses the present European context, with particular focus on a 2004 European Commission decision which severely constrained the ability of low cost airlines to negotiate beneficial incentive packages with fully or partially state owned airports. The Commission has argued that such expenditures must be categorized as state aid, the use of which is strongly constrained by the EU treaties and EU Competition Policy. As a result of this decision, it is not possible to promote low cost airline connections for developmental purposes, because the use of EU expenditures for this purpose would likewise potentially be classified as state aid on a supranational level. Section 6.2 addresses this conflict, arguing that a redefinition of low cost airlines as a public good or service addresses the (1) political economic and (2) legal conditions necessary to integrate low cost airline incentive packages into the EU s regional development policy. 6.1 Incentive Packages as State Aid The Current Context Between the 1993 deregulation of European airspace and 2004, low cost airlines were successful in negotiating unique financial and contractual incentives ( Incentive Packages ) at airports, many of which were in underserved and underdeveloped regions. In turn, airports compensated by drawing on increased traffic and high margin airport services (Barrett, 2004: 38).

29 29 Airports with full or partial public ownership likewise had stakeholders who considered broader positive externalities, such as increased regional tourism, airport employment, and the prestige of hosting an international airport. Given these benefits, low cost airlines were effective in negotiating broad incentive packages, which reduced their airport charges by as much as 90 percent in airports such as Dublin and Charleroi, Belgium (Barrett, 2000: 22). In 2004, however, a Commission decision largely discouraged the use of public funds to support Incentive Packages, limiting financial and contractual incentives to a five year period following the establishment of a new connection, and necessitating that Incentive Packages be available equally and transparently to low cost airlines as well as national carriers (Gröteke, 2004: 6 7; European Commission, 2004). The Commission s decision was prompted by a 2002 complaint against Brussels Charleroi airport, in which a rival Brussels airport, Zavantem, argued that the Incentives Package offered to Ryanair by Brussels South Charleroi Airport (BSCA) qualified as state aid, since the airport was partially owned by the Walloon region (Gröteke 2004). Under Article 87 of the Treaty Establishing the European Community, state aid which distorts or threatens to distort competition by favouring certain undertakings or the production of certain goods [...] is incompatible with the Common Market. The Commission recognized the complaint by the rival airport, and introduced the aforementioned guidelines, effectively limiting the use of Incentive Packages. Today, this decision substantially reduces the scope of support that low cost airlines can negotiate with regional airlines, undermining an important aspect of the complementarity between regional airports and low cost airlines and threatening the future establishment of low cost airlines in underdeveloped regions. Following the Commission s decision, a number of studies have emerged providing a critical consideration of this new environment. Gröteke (2004) argues that incentive packages, whether in the form of private contractual arrangements or in the form of state aid, are warranted under the principle of price differentiation. Since low cost airlines provide a differentiated service from traditional flag carrier airlines, there exists heterogeneity in carrier services which warrants differentiation of treatment by the airports. An alternative approach, drawing on Gröteke s conception of price differentiation but incorporating the implications of this study s findings, recognizes low cost airlines as a public service with development externalities. This approach is demonstrated below in considering the policy implications of this work s findings for EU Cohesion and Competition Policies.

30 Incentive Packages as Development Policy An Alternative Approach The findings of this study suggest that the two principle political economy conditions for welfare enhancing provision of public services that is, that (a) the service produces positive externalities which improve collective welfare, and (b) that in the absence of government support, suboptimal provision of the service would occur are likely satisfied in the case of low cost airline incentive packages. In particular, if the positive externalities of low cost airlines extend beyond the immediate municipality hosting the connection (as the findings in the present study suggest) then economy theory suggests that resource allocative welfare gains would accrue from government promotion of these services. Likewise, the European Union, which has pursued an explicit convergence policy since 1992, would likewise better achieve these goals by encouraging Developmental Incentive Packages to increase airport density in under developed regions within new and future Member States. If the externality promoting rationale of this support is recognized, it would redefine low cost airlines as a public service. This reclassification would have important legal ramifications in the context of Common Markets constraints on state aid. In particular, Article 73 of the Treaty Establishing the European Community makes explicit allowance for state aid when it is used to represent reimbursement for the discharge of certain obligations inherent in the concept of a public service (italics mine). That is, Developmental Incentive Packages would provide financial support to airlines, in exchange for the public service externalities that the airlines generate. Likewise, Articles 87(3)a and 87(3)c allows state aid to be used in the context of development within under developed regions. Although a rigorous survey of the legal environment surrounding state aid policies is beyond the scope of this paper, this short review suggests that the EU s legal environment is significantly altered if the migratory externalities of low cost airlines are taken under consideration. Based on these political economic and legal circumstances, it is argued that the findings of this paper warrant the consideration of Developmental Incentive Packages as a valuable instrument for the EU s present Cohesion Policy. This is particularly true, given that the EU s Structural Fund policy already has a comprehensive transportation infrastructure initiative: the Ten T Programme (Puga, 2002). A number of authors, however, have argued that present transportation development may have suboptimal developmental impact, resulting from the fact that large scale projects, such as throughways, improve the accessibility and development of regional hubs, without improving or reducing transportation costs in underdeveloped spokes (Puga and Venables, 1997; Fujita and Mori,

31 ). By introducing a direct, point to point network, low cost airlines have the potential to directly impact European regions, strongly in keeping with the framework of Europe s emergent objectives in the fields of multi level governance and subsidiarity. Moreover, and in broader normative legal terms, it may be strongly argued that the four fundamental freedoms of the European Union, which include the free movement of persons, are disproportionately applicable to those citizens, resident in border regions, who are provided with both public roads and public transportation enabling them to exercise the right of free movement across the European Union. This was evidenced in this study s findings, which found disproportionate international migration in Polish border regions (Table 5.2). By contrast, European citizens in the new Member States who reside in non border regions, or in countries that do not border EU 15 labour markets, do not have the benefit of public services promoting the right to free movement. Under such conditions, publicprivate partnerships enabling interregional mobility and migratory externalities regardless of the traveller s source region are consistent with the principles and objectives of European integration, cohesion, and development. 6.3 Summary of developmental implications It is argued that the migratory externalities identified from low cost airline connections in this work if accurate may justify using select government aid to support Developmental Incentive Packages. These packages are defined as the use of Cohesion Fund expenditures to promote low cost airline connections in underdeveloped regions. In particular, they are in strict contrast to the encompassing government regulation that preceded airline deregulation. The purpose of these packages would be to promote increased wage convergence across the enlarged EU, assisting growth and European economic integration. A low cost transportation infrastructure would, furthermore, promote the EU s policy of free movement of persons by significantly lowering transportation costs across the European regions. 7 Conclusions The purpose of this work has been to determine whether ascertainable positive externalities, particularly migration induced wage convergence, have been enhanced by the emergence of low cost airlines in Central Eastern Europe. Using a theoretical application of the Sjaastad human capital model and a Neoclassical labour market framework, this work presented a

32 32 hypothesis that such externalities should exist, and then tested for their presence using a wage decomposition function that controlled for a variety of factors potentially affecting wage levels. The paper s findings confirm a negative and statistically significant relationship between distance to low cost airline servicing airports and wage convergence, with a beta coefficient of approximately 33 to 41 groszy. These figures constitute a differential representing approximately a 5 percent difference in wages between regions 50 and 250 kilometres from a regional airport. The political economic and legal implications of these findings were likewise discussed, suggesting that the EU s Cohesion Fund programme objectives would benefit from a strategic policy aimed at promoting airline connections in underdeveloped regions. Broader EU ambitions, which protect the right of free movement of persons, would likewise benefit from a closer examination of the disproportionate accessibility of this right to those European citizens living closer to desirable labour markets, a circumstance that strategic public support of low cost airlines could address. In addition to future studies which might address the shortcomings discussed in Section 5.3, the author is able to identify a number of promising areas in this paper that would benefit from further research. For example, a broader assessment of Central Eastern Europe states, while controlling for significant differentials in out migration rates between them, would be an important step in broadening the scope and implications of this paper s findings. On the other hand, a more concentrated study of Rzeszow airport in south eastern Poland which has provided low cost airline connections since 2005 to a comparatively isolated and historically underdeveloped region would likewise illuminate the microeconomic impact of low cost airlines with a level of precision that is unavailable in the broader survey format employed here. Finally, the very recent nature of low cost airline connections to Central Eastern Europe suggests that any future study, implementing additional data and observations, would only improve the quality of the findings identified in this work.

33 33 Appendix Figure A.1: Out-migration to EU-15 Countries from Central Eastern Europe 2002 a Total Out Migration Average 2002 a Per Capita Out Migration Average a Lithuanian figures were not available in Figures for Hungary not available. S E t t