DEPARTMENT OF ECONOMICS DISCUSSION PAPER SERIES

ISSN 1750-4171 DEPARTMENT OF ECONOMICS DISCUSSION PAPER SERIES Loation of Foreign Diret Investment in the Central and Eastern European Countries: A Mixed Logit and Multilevel Data Approah Simona Rasiute and Eri J. Penteost WP 2008-04 Dept Eonomis Loughborough University Loughborough LE11 3TU United Kingdom Tel: + 44 (0) 1509 222701 Fax: + 44 (0) 1509 223910 http://www.lboro.a.uk/departments/e 1

The Loation of Foreign Diret Investment in the Central and Eastern European Countries: A Mixed Logit and Multilevel Data Approah Simona Rasiute and Eri J. Penteost Department of Eonomis Loughborough University Loughborough Leiestershire LE11 3TU UK First draft May 2008 Abstrat This paper uses the Mixed logit (ML) model and a novel three-level dataset to examine the fators explaining 1,108 foreign diret investment (FDI) loation deisions into 13 Central and Eastern European ountries (CEECs) over an eleven-year period between 1997 and 2007. The ML model approah is superior to other disrete hoie methods in that it allows for random taste variation, unrestrited substitution patterns and orrelation in unobserved fators over time. The highly signifiant empirial results, based on a general underlying eonomi model of imperfet ompetition, show that the responsiveness of the probabilities of hoies to invest in a partiular ountry in CEE to ountry-level variables differs both aross setors and aross firms of different sizes and profitability. The results generalise previous studies that used only ountry-level data or only industry- and firm-level data to give a more aurate explanation of the firm-speifi investment loation deisions. JEL lassifiation Nos: F23, P33 Keywords: Mixed logit model, random parameters, foreign diret investment, multi-level data, Halton draws 2

1. Introdution The disrete hoie eonometri methodology has beome an inreasingly popular tehnique for investigating the loation deisions of foreign diret investment (FDI), espeially in the regional siene literature (see for example, Head et al, 1999, Guimaraes et al, 2000 and Kim et al, 2003). This literature, however, is subjet to two prinipal limitations. First, both Multinomial logit (MNL) and Nested logit (NL) models are not suffiiently flexible to satisfatorily model the investment loation hoies of multinational enterprises (MNEs). The MNL model is subjet to restritive assumptions regarding the substitution patterns aross investment loation alternatives and the absene of random taste heterogeneity aross deision-makers, while the NL model only partially relaxes the independene of irrelevant alternatives (IIA) assumption in order to aommodate the substitution aross alternatives to a limited degree. A seond limitation is that the existing literature on the determinants of FDI usually uses only ountry- and industry-speifi data and does not inorporate investing firm harateristi, whereas greater estimation effiieny an be ahieved by using multi-level data, on the firm, industry and ountry. This paper therefore generalises the existing literature by making two prinipal ontributions. First, it uses the Mixed logit (ML) model to investigate investment loation hoies by MNEs for the first time, merely allowing for random taste variation and unrestrited substitution patterns. Seond, it makes use of a multi-level data set allowing firm, industry (or setor) and ountry effets to simultaneously determine the firm-level FDI loation deisions. The data overs 1,108 investment loation hoies of firms in the EU(15), Norway, Switzerland, Russia, Japan and the USA into 13 Central and Eastern European Counties (CEECs) the 12 reent EU member states exluding Cyprus and Malta, but inluding Croatia, Russia and Ukraine - over an eleven year period from 1997 to 2007. The estimation result shows that firms investing in different setors and firms of different size and profitability benefit from loation fators to a different degree. For example, larger firms and firms investing in sale-intensive industries will prefer to invest in larger ountries in order to exploit their eonomies of sale, as ompared to smaller firms and firms investing in other setors. 3

The rest of the paper is set out as follows. Setion 2 outlines a new-trade theory model of the firm s profit funtion, used to identify the eonomi variables that determine the FDI deision and Setion 3 explains the ML model. Setion 4 disusses the dataset and the onstrution of these variables and Setion 5 presents the eonometri results. Finally, Setion 6 onludes. 2. The Theoretial Model A lot of theoretial literature exist on the extensiveness of firms prodution, for example, the firms deision to serve only domesti market, to serve domesti market and to export or to serve domesti market and to establish a subsidiary abroad (Baldwin, 2005; Helpman, 2006; Helpman et al., 2004; Jean, 2002; Melitz, 2003; Montagna, 2001). However, one the deision to invest abroad has been made, investment loation hoies have not been theoretially modelled. As a result, a model of imperfet ompetition has been applied to model where MNEs hoose to loate foreign apital. The deision where to invest does not only depend on opportunities offered by foreign markets and industries but also on firms individual harateristis. As a result, the model makes three important assumptions: investing firms are heterogeneous aross industries in respet to produtivity, industries vary in fator intensities and ountries differ in relative fator abundane (but idential in terms of preferenes and tehnologies). Eah firm produes a different produt and the demand funtion for firm s produt in the foreign market an be expressed as in (Helpman, 2006): ε q i = D pis (2) where q i is the quantity of output by firm i in ountry and p is is the prie of one unit of output, D is a measure of the demand level in ountry, and ε is the elastiity of demand defined as ε 1/(1-λ). The demand elastiity is assumed to be onstant (ε >1, as 0<λ<1), where λ is a mark-up over marginal osts. Produers are assumed to be small relative to the industry, and therefore, D is treated as exogenous by produers. Firms that loate their apital abroad have produtivity θ i and variable osts per unit of output of a/θ i. The firm that enters foreign ountry also inurs fixed osts af, where a is the ost of resoures and f is a measure of fixed prodution osts in terms of 4

resoures. Then the profit-maximising prie for firm i is p i =a/λθ i. Variable osts vary with firm s produtivity, while fixed foreign market entry osts are the same for all firms, as many fixed osts suh as building or equipping a fatory with mahinery are unlikely to vary substantially with firm produtivity ((Bernard et al., 2007)). The presene of fixed prodution osts implies that, in equilibrium, eah firm hooses to produe a unique variety (Bernard et al., 2007). Firms use three fators of prodution: skilled labour, unskilled labour and apital. It is assumed that skilled labour and apital are alloated to fixed osts (refleting osts of R&D and osts of fatories and equipment), while unskilled labour are alloated to variable osts (apturing standard prodution). The relative intensity of fator use varies aross industries (therefore, α and β will also vary aross industries). The minimum total osts an be speified as: 1 s α TC = ( w ) ( r ) is s β f + u ( w ) s θ 1 α β where w s s is an hourly wage rate of skilled labour in ountry and industry s; w s u is an hourly wage rate of unskilled labour in ountry and industry s, r is a return on apital in ountry and the α and β parameters are the shares of skilled labour and apital in total ost respetively. The first term on the right hand side of (3) represents the fixed osts of aquiring information about foreign markets, developing appropriate marketing strategies, building distribution networks (Bernard et al., 2007), also aquiring fatories and equipment. MNEs enterprises use foreign apital in the form of foreign diret investment and host ountry skilled and unskilled labour, L s and L u respetively. The labour market equilibrium requires equating labour supply, whih is exogenous ( L ), with the total demand for skilled (L s ) and unskilled (L u ) labouring ountry : s u i q i (3) L = L + L (4) The prie an be written as a onstant mark-up over marginal ost: 1 This is a standard formulation following Krugman (1991), where the total minimum ost is derived from a standard ost minimization problem assuming a Cobb-Douglas prodution funtion. 5

p i ( θ) τ = d u ( w ) s λθ 1 α β Ieberg type transport osts τ d, are assumed between the soure ountry d and host ountry (Samuelson, 1954) 2. The after-tax profit in eah loation is defined as total revenue (TR) less total osts (TC) net of tax and less the osts arising from the institutional, legal, politial and maroeonomi environment prevalent in the host ountry. It an be written as: where T is the tax rate in loation and is i ( TRis TCis) G (5) π = (1 T ) (6) G is a term that aptures the osts that firms inur due to the maroeonomi investment environment prevalent in the host ountries. TRis is the revenue reeived by firm i in ountry and industry s from selling output q, and π TC is are the osts of produing q. Substituting (4) and (5) into (6) we obtain: is = u 1 α β 1 ε τ w 1 λθ i (7) d ( ) ( ) ( s) s α T D β 1 λ ( ws) ( r) f G This speifiation of the profit funtion of investment abroad, in ontrast to the majority of the urrent theoretial literature on FDI, allows for the heterogeneity of firms in different setors and ountries. Country speifi equation (7) in respet to T, D, τ, w, r and G will also vary aross firms in respet to θ and aross industries in respet to α and β. Therefore, it is assumed that partiular loation advantages do not have the same value for all multinational enterprises, as firms operating in different setors and firms of different size and profitability benefit from loal resoures to different degrees. Loation advantages vary for MNEs with different harateristis, and the interation between loation and firm together with industry attributes, rather than eah of the firm and industry fators independently affets loation hoies. For example, smaller firms are 2 For example, when goods are shipped from ountry d to ountry, only a fration 1/τ d of the original unit is assumed to arrive ( > 1). Hene, other things being equal, the more remote loations are at a disadvantage. τ d 6

expeted to invest in ountries with strong historial ties and similar ulture and language, while larger firms and firms investing in sale-intensive industries are expeted to invest in more remote and larger ountries in order to exploit their eonomies of sale. 3. Eonometri Estimation The ML model has been applied in environmental, health and transport studies (Greene and Hensher, 2003; Hole, 2007; Meijer and Rouwendal, 2006), however, its appliation in eonomis is still limited. Furthermore, the ML model has never been used to investigate the investment loation hoies of MNEs. The ML model, whih is probably the most flexible disrete hoie model, has beome appliable only reently with the development of simulation and inreased omputer speed. The ML 3 model is a very flexible model that approximates any random utility model (MFadden and Train, 2000). In ontrast to the MNL model, the ML model allows for random taste variation, unrestrited substitution patterns and orrelation in unobserved fators over time. Furthermore, the ML probability an still be derived from utility-maximising behaviour. In ontrast to the Probit model, it is not restrited to normal distribution. The derivation of ML is straightforward, and simulation of its hoie probabilities is omputationally simple (Train, 2003). In the most general form, the ML probabilities are the integrals of standard logit probabilities over a density of unobserved random parameters (Train, 2003): P ( ) f( β ) i = Li i i dβi β (1) where f(β i ) is the random parameter density funtion, whih is speified to be ontinuous. The mixed logit probability is a weighted average of the logit formula evaluated at different values of β i, with the weights given by the density f(β i ) (Train, 2003). The logit probability evaluated at parameters β i is expressed as L i ( β ) i = e β x d ' i e i ' i β x ig (2) 3 In the literature also referred to random parameter logit, mixed multinomial logit, kernel logit and hybrid logit. 7

Where x i is a vetor of observed hoie-speifi variables and β i are individual firmspeifi parameters. Sine the integral annot be alulated analytially, it has to be approximated through simulation by maximising the simulated log-likelihood funtion. A value of β is drawn form the distribution f(β ψ) for a given value of ψ and labelled β r with the subsript r=1 referring to the first draw. The logit formula L i (β r ) is alulated with this draw. Finally, the two steps are repeated many times, and the results are averaged, whih gives the simulated probability: ( P = 1 R i L i R r= 1 r ( β ) where R is a number of draws. P ( i is an unbiased estimator of P i. Its variane dereases as R inreases. It is stritly positive (ln P ( i is defined) and smooth (twie differentiable) in the parameters ψ and the variables x, whih failitates the numerial searh for the maximum likelihood funtion and the alulation of elastiities. Furthermore, P ( i sums to one over alternatives whih is helpful when interpreting the results. The simulated probabilities are substituted into the log-likelihood funtion to give the simulated log likelihood: SLL = C I = 1 i= 1 (3) d i ln P ( i (4) where d i =1 if i hose and zero otherwise. The maximum simulated likelihood estimator (MSLE) is the value of ψ that maximizes SLL. The ML struture an be derived in two equivalent ways: by allowing flexible substitution patterns aross alternatives (error-omponents struture) (Brownstone, 2000) and by aommodating unobserved heterogeneity aross individuals (random-oeffiients struture) (Bhat, 1998, 2002, Revelt and Train,1998, Train, 1998). 4. The Data Set and the Variable Speifiation Table 1 gives a summary of variable definitions and soures. There are 1,108 firm-level data observations on FDI flows from firms of 20 market eonomies (EU15 ountries, 8

USA, Japan, Russia, Norway and Switzerland) to the firms into 13 transition eonomies (12 new EU member states (exept for Malta and Cyprus) plus Croatia, Russia and Ukraine) from 1997 to 2007. Of all 13 host CEECs in the sample, Poland has been hosen the majority of times by MNEs to loate their investment (about 21 perent) and it was followed by Russia with about 17 perent of foreign investment loation hoies. Slovenia and Latvia, on the other hand, have reeived the smallest share foreign apital alloations (2 and 3 perent respetively). The two major soure ountries for investment in CEE in the sample are Finland and the UK with the shares of approximately 12 and 11 perent respetively. MNEs from Japan and Ireland were at the other end of the sale regarding investment loation hoies in CEE of about 1 perent eah. Almost a third of the firms in the sample are small (with turnover up to 100,000), 30 perent of the firms are medium (with turnover from 100,000 to 1mln), 26 perent of the firms are large (turnover from 1mln to 10mln) and 12 perent of the firms are very large firms (turnover above 10mln). The majority- almost half- of very large firms in the sample have invested in sale-intensive industries. The rest of the firms have seleted traditional setors to loate most of their investment (33 perent of small firms, 41 perent of medium firms and 43 perent of large firms). 14 perent of the firms in the sample have inurred losses, 44 perent of the firms have earned profits up to 50,000 and 42 perent of the firms have earner profits above 50,000. Regardless of the investing firms profitability traditional setors reeived most investment alloations. The investment loation harateristis may have a different effet on firms investing in different setors. Sale-intensive setors (Sale) inlude typial oligopolisti, large firm industries, with high apital intensity, extensive eonomies of sale and learning, high tehnial and managerial omplexity, for example, automobiles, airrafts, hemials, petrol and oal produts, shipbuilding, industrial hemials, drugs and mediines, petrol refineries, non-ferrous metals and railroad equipment (Midelfart- Knarvik et al., 2000). Siene-based setors (Siene), on the other hand, are haraterised by innovative ativities diretly linked to high R&D expenditures, for example, fine hemials, eletroni omponents, teleommuniations, and aerospae (Midelfart-Knarvik et al., 2000). Traditional (supplier-dominated) setors (Tradit) inlude suh industries as textiles, lothing, furniture, leather and shoes, eramis, and the 9

simplest metal produts. Finally, banking insurane and retail are examples of servie setors (Servie). Firms loating investment in traditional setors may be more onerned about the availability of unskilled labour and may pay lower wages as ompared to the firms in other setors, for example, siene-based industries, where more skilled labour is employed and higher wages are paid for higher skills. Four groups of industries have been hosen to loate foreign investment more or less evenly in the sample. The largest number of foreign apital alloations in CEE took plae in traditional setors (36 perent), followed by sale-intensive industries (24 perent) and servie setors (23 perent). Siene-based industries have reeived the smallest share of FDI (18 perent) in the sample. However, when looking at the distribution of investment loation hoies among the four groups of industries in separate ountries, traditional setors have not neessarily attrated most foreign apital alloations. For example, in the Czeh Republi, Hungary and Slovakia sale-intensive industries have reeived the largest share of FDI (30 perent, 26 perent and 47 perent respetively), while in Estonia and Lithuania the servie setor has attrated most foreign apital alloations (35 perent and 33 perent respetively). The ountry-speifi determinants of FDI into the CEECs an be loosely divided into the traditional determinants and the transition-speifi determinants. The transitionspeifi determinants are proxied by the risk assoiated with eah host ountry, G, in equations (5) and (7). The institutional, legal and politial environment, i.e. transpareny and effetiveness of legal system, are important for the deision of foreign investors to loate their apital abroad. The Transpareny International Corruption Pereption Index (TICP) is used as a measure of the extent of orrupt praties in the host ountry. This index pools information from ten different surveys of business exeutives, risk analysis and the general publi. The TICP index ranks ountries in terms of the degree to whih orruption is pereived to exist among publi offiials and politiians and it varies from 1 (high orruption) to 10 (no orruption). In order to make the interpretation of the parameter more intuitive the TICP index is multiplied through by minus one, so that the smaller the number the higher risk. Following the theoretial model desribed in setion 2, the traditional determinants are the market size of the host ountry, D, the ost of apital in the host 10

ountry, r, distane τ d, and tax rates T in the host ountry. As Table 1 shows, market size is simply the real GDP of the ountry and the rate of return is measured as the real disount (interest) rate. Distane an be onsidered as a measure of the transation osts of undertaking foreign ativities, suh as the osts of transport and ommuniations, the osts of dealing with ultural and language differenes, the osts of sending personnel abroad, and the informational osts of institutional and legal fators, e.g., loal property rights, regulations and tax systems. These kinds of osts are assumed to inrease with distane. The orporate inome tax rate affets the profitability of foreign diret investment and hene influenes the investment loation hoies of MNEs. Few studies analyse the effet of taxation on the loation hoies of foreign firms in the CEECs (Bellak and Leibreht, 2005; Carstensen and Toubal, 2004; Clausing and Dorobantu, 2005; Wei, 2000). The studies that do inlude tax rates as loation hoie fators in CEECs usually use statutory orporate inome tax rates. Statutory orporate inome tax rates are not an appropriate indiator of the tax burden espeially in the ase of FDI, beause they are only one of the determinants of total tax burden, while the tax base is also influened by depreiation shemes, treatment of losses and valuation of inventories among others. In this paper, in ontrast to the majority of other studies, the tax burden in the 13 CEECs is measured as the effetive orporate inome tax rate whih is alulated by dividing revenue taxes paid by orporations and other enterprises by a host ountry s GDP. This approah allows omparisons of different tax systems, taking into aount suh important aspet as untaxed reserves, tax enforement and the treatment of losses. In addition to the above mentioned fators, three other ountry-speifi fators are inluded in the empirial model: the national rate of unemployment and two dummy variables, one for European Union (EUD) and another for a ommon border (CBD) between the investing and the investment reeiving ountry. A dummy variable for ommon border between the soure and the host ountry is inluded, as it is expeted that the host ountry is more likely to be hosen to loate investment if it shares the border with the soure ountry. Usually ountries sharing the same border have similar ulture and language and stronger historial ties. 11

Countries that joined the EU by January 2007 had to satisfy the eonomi (market eonomy), politial (demoray and human rights) and administrative (well-funtioning institutions) riteria set at the Copenhagen European Counil in 1993. The aession of a CEE ountry into EU meant free trade with EU member states and the adoption of Western business and legal environment, whih provided foreign investors with onfidene in suess of eah ountry s reforms. As a result, the parameter of EU dummy variable is expeted to have a positive sign. Although, unemployment is not important for the individual firm s profit funtion, it may still be of signifiane at the ountry level as an indiator of labour market flexibility and availability of labour fore. Countries with high loal demand for goods and servies and high labour market flexibility are likely to fae relatively low rates of unemployment, whih may enourage firms to invest in a partiular host ountry. On the other hand, a high unemployment rate may mean that although it is easy to reruit labour, there is low demand loally and labour market rigidities. The impat of unemployment on the investment loation deision is therefore stritly ambiguous and it may have a different effet on firms investing in different industries. For example, firms investing in traditional setors employ less skilled labour and may be more onerned about the availability of workers, while firms investing in siene-based industries, whih employ more skilled labour, may be disouraged by higher unemployment, as unemployed people loose their skills through time. Industry-level real wage rates, w s, are inluded as a proxy for the average variable osts of firms and they impliitly assume that workers are not fully mobile aross setors, at least in the short run. The profitability of the firm investing abroad is expeted to be higher if the labour osts are lower in the hosen ountry than in the rest of the destination ountries. On the other hand, higher wages may reflet higher skill and, therefore, may have a positive effet for firms investing in siene-based industries, whih employ more skilled labour as ompared to other industries. As a result, the sign of the parameter is amgiuous. The harateristis of individual investing firms an also have an influene on the responsiveness of ountry-level variables. The firm-level variables inlude the turnover of the investing firm as a proxy for its size ( s i ) and earnings before interest and tax as a 12

proxy for its profitability ( e i ). Firms of different sizes and profitability possess different resoures and apabilities (Dean et al., 1998). Small firms are assumed to be haraterised by speed, flexibility and nihe-filling apabilities due to their strutural simpliity and faster deision making, entrepreneurial-orientation and less risk aversion (Woo, 1987). As a result, smaller firms respond quiker to the dynamis of the industry environment. Larger firms, whih are usually more profitable, are able to aquire larger market share by exploiting sale eonomies, bargaining power, patents, reputation and finanial resoures to deal with shoks and business downturns (Dean et al., 1998). Larger firms investing in sale-intensive industries are expeted to invest in ountries with larger markets in order to exploit their eonomies of sale, while more profitable firms are expeted to be less disouraged to invest in remote ountries, as more funds are available to over transation osts, suh as osts of transport and ommuniation, the osts of dealing with ultural and linguisti differenes and information osts of institutional and legal fators, et. 5. Estimation and Results The analysis starts by treating eah ountry level variable and the variable that varies among ountries and industries (w s ) separately as random by imposing various distributions (Table 2). The random parameters with most appropriate distributions are ombined in the final speifiation, whih has the best model fit (the largest loglikelihood value) and whih avoids distributions that ause flat log-likelihood at the estimates. Triangular distribution is imposed on the variables for market size in the host ountry, D, the wage variable, w s, the distane variable, τ d, and the unemployment variable, u. Restrited uniform distribution is imposed on the two dummy variables: the dummy variable for ommon border and the dummy variable for EU membership (Table 3). The means of the uniform distribution for the dummy variables are restrited to be equal to their varianes; as a result, the sign of the estimate random parameters of the two dummy variables will be the same for all the investing firms. Initially 100 Halton intelligent draws are used to estimate the model, while inreasing the number to 1000 for the final model speifiation. 13

The ML is not only able to determine the existene of heterogeneity around the mean parameter, through the estimation of the standard deviation parameter, but it an also indiate the soure of the heterogeneity through the interation between a random parameters and other attributes (moderator variables). For example, an observed heterogeneity in some ountry level determinants of investment loation hoie an be due to differenes in industry or/and investing firms harateristis. The following interation terms are statistially signifiant: the interation terms between the dummy variable for traditional setors and unemployment rate in the host ountry, Tradit u ; the interation term between the dummy variable for traditional setors and the wage rate in the host ountry, Tradit w s ; the interation term between investing firms profitability and distane between investing and investment reeiving ountries, e i τ d, and finally, the interation term between investing firm s size and host ountry s market size, s i GDP. Despite the widespread use of interation terms in Disrete Choie Methodology, the majority of applied researhers misinterpret oeffiients of interation terms (Ai and Norton, 2003). Unlike in linear models, the interation effet in nonlinear models is a funtion of not only the oeffiient for the interation, but also the oeffiients for eah interated variable and the values of all the variables in the model (Greene, 2008). Therefore, the sign of the interation oeffiient may not indiate the diretion of the interation effet, as the interation effet may have different signs for different values of ovariate. Furthermore, the interpretation of a separately inluded in the model variable if it is also a part of an interation term hanges (Jaard, 2001). It does not represent a main effet but a onditional effet instead: the effet of the variable when the values of the moderator variable (the other interated variable) are zero. For example, the variable, w s is not only inluded in the model separately but also interated with the dummy variable, Tradit. As a result, while the interation, w s Tradit, represents the effet of wage rate in traditional setors in a partiular host ountry on the probability of seleting the ountry to loate foreign apital, the variable, w s, represents the effet of wage osts in other setors (Siene-based, Servie and Sale-intensive setors). 14

As neither the sign nor the magnitude of the interations and separately inluded in the model variables if they are also inluded in interation terms are informative, elastiities and marginal effets have to be estimated for ontinuous and dummy variables respetively (Table 4) 4. Negative estimated elastiities for variable, Wage s, indiate that the higher wage osts are in non-traditional setors in a host ountry, the less likely the ountry will be hosen to loate foreign apital. The effet of the wage rate on the probability of loating investment in a partiular ountry if a firm hooses to invest in traditional setors is a sum of the estimated elastiities for Wage s and w s Tradit. The sums of the estimated elastiities are negative and larger in absolute values showing that firms that hoose to invest in traditional setors are more sensitive to higher wages rates in the host ountry than ountry, whih hoose to invest in non-traditional setors.. The unemployment variable, u, is not only inluded in the model separately but also interated with the dummy variables for traditional setors, Tradit u. Negative elastiities for, u, indiate that the higher the unemployment rate is in the host ountry, the less likely the ountry to be hosen by foreign firms to loate their apital, if they hoose to invest in non-traditional setors. Negative but smaller in absolute values or even positive sums of the elastiities for Tradit u indiate that higher unemployment in a host ountry has a less negative or even positive effet on the probability of seleting the ountry to loate foreign apital for firms that invest in traditional setors, as ompared to the firms that invest in non-traditional setors. Typially, traditional setors employ more unskilled labour, as ompared to other setors, for example, siene-based industries, whih employ more skilled labour and pay higher wages that reflet a skill premium. When two ontinuous variables are interated, for example, e i τ d and s i GDP, the interpretation of the interation terms is muh more ompliated. As a result, the hanges in market shares in different ountries and the hanges in a number of firms investing in a partiular ountry are investigated due to the gradual hange in the 4 The elastiities for separately inluded in the model market size and distane variables do not have muh explanatory value, as they indiate the effet when moderator variables (investing firm s size and profitability respetively) are equal to zero. The tax variable and the risk variable do not appear statsitially signifiant. The standar deviation of the retrun on apital variable is not statistially signifiant, indiating that all information is aptured by the mean. The higher is the return on apital in the ost ountry, the more likely the ountry to be hosen by foreign investors. 15

moderator variable with the help of simulation. The results presented in Table 5 show the effet of 1, 10, 50 and 100 perent inrease in variables e i and s i on the hange in the market shares and in the number of firms investing in a partiular ountry. The estimated hanges in market shares and in the number of investing firms for both interation terms show onsistently positive and inreasing effets. The results for e i τ d indiate that more profitable investing firm are less likely to be deprived to invest in more remote ountries as ompared to less profitable firms. More profitable firms usually have more resoures to pay for transation osts assoiated with investment in more remote ountries, for example, ost of transport of ommuniation, the osts of dealing with ultural and linguisti differenes, the ost of sending personnel abroad, and information osts of institutional and legal fators. The results for s i GDP indiate that the larger the host ountry is, the more likely it is to be hosen by an investing firm to loate its apital, and the effets is stronger for larger investing firm, as ompared to smaller firms. Larger firms are usually haraterized by high eonomies of sale; therefore, they searh for larger foreign markets to exploit these eonomies. Statistially signifiant interation terms between ountry-level variables and industry-level dummies together with the firmlevel variables onfirm the existene of heterogeneity revealed by statistially signifiant standard deviations of the parameters of ertain ountry-level variables. Positive and statistially signifiant estimated elastiities for EU dummy variable indiate that ountries whih beame members of the EU by January 2007 are more likely to be hosen by foreign investors to loate their apital. An investment reeiving ountry that has a ommon border with an investing ountry is more likely to be hosen as an investment loation, and this is refleted in the positive elastiities for CBD d (Table 4). Neighbouring ountries usually have strong historial ties and less linguisti and ultural barriers. 6. Conlusions This paper applies the Mixed logit (ML) model, whih is probably the most flexible disrete hoie model, to investigate investment loation hoies by MNEs for the first time. It also makes use of a novel multi-level data set allowing firm, industry (or 16

setor) and ountry effets to simultaneously determine the firm-level FDI loation deisions. The highly signifiant empirial results support the presene of heterogeneity in the investment loation deisions, whih is not only revealed by statistially signifiant interation terms, but also by statistially signifiant standard deviations of the random parameters. The results show that firms investing in traditional setors are less likely to be disouraged to invest in ountries with higher unemployment rate but more likely to be disouraged by higher wage rates as ompared to MNEs that invest in non-traditional setors. The larger the host ountry is the more likely it is to be hosen by foreign investors to loate their apital and the effet is larger for larger investing firms. On the other hand, more profitable firms are less likely to be disouraged to invest in more remote ountries, as ompared to less profitable firms. This more general approah to the FDI deision shows that to allow for firm heterogeneity is important if robust estimates are to be found for their omplex effets. These results ast doubt on the robustness of earlier empirial studies that foused on maroeonomi features of the FDI loation deision not inorporating investing firms arateristis, and applied Multinomial logit and/or Nested logit models. 17

Referenes Ai, C., and Norton, E. C. (2003). Interation terms in logit and probit models. Eonomi Letters 80, 123-129. Baldwin, R. E. (2005). Heterogeneous Firms and Trade: Testable and Untestable Properties of the Melitz model. nber Working Paper Series, Working Paper 10471. Bellak, C., and Leibreht, M. (2005). Do low orporate inome tax rates attrat FDI? Evidene from Eight Central- and East European Countries. In "Researh Paper Series, Globalisation, Produtivity and Tehnology. Researh Paper 2005/43", The University of Nottingham. Leverhulme Centre for Researh on Globalisation and Eonomi Poliy Bernard, A. B., Redding, S. J., and Shott, P. K. (2007). Comparative advantage and heterogeneous firms. Review of Eonomi Studies 74, 31-66. Brownstone, D. (2000). Disrete Choie Modeling for Transportation. In "9th IATBR Travel Behavoir Conferene", Australia. Carstensen, K., and Toubal, F. (2004). Foreign diret investment in Central and Eastern European ountries: a dynami panel analysis. Journal of Comparative Eonomis 32, 3-22. Clausing, K. A., and Dorobantu, C. L. (2005). Re-entering Europe: Does European Union andiday boost foreign diret investment? Eonomis of Transition 13, 77-103. Dean, T. J., Brown, R. L., and Bamford, C. E. (1998). Differenes in large and small firm responses to environmental ontext: Strategi impliations from a omparative analysis of business formations. Strategi Management Journal 19, 709-728. Greene, W. H. (2008). "Eonometri Analysis," 6th/Ed. Pearson-Prentie Hall. Greene, W. H., and Hensher, D. A. (2003). A latent lass model for disrete hoie analysis: ontrasts with mixed logit. Transportation Researh Part B- Methodologial 37, 681-698. Helpman, E. (2006). Trade, FDI, and the organization of firms. Journal of Eonomi Literature 44, 589-630. Helpman, E., Melitz, M. J., and Yeaple, S. R. (2004). Export versus FDI with heterogeneous firms. Amerian Eonomi Review 94, 300-316. Hole, A. R. (2007). Modelling heterogeneity in patients' preferenes for the attributes of a general pratitioner appointment. National Primary Care Researh and Development Centre, Centre for Health Eonomis, University of York. Jaard, J. (2001). "Interation effets in logisti regression," Sage Publiations, London, New Delhi. Jean, S. (2002). International trade and firms' heterogeneity under monopolisti ompetition. Open Eonomies Review 13, 291-311. MFadden, D., and Train, K. (2000). Mixed MNL models for disrete response. Journal of Applied Eonometris 15, 447-470. Meijer, E., and Rouwendal, J. (2006). Measuring welfare effets in models with random oeffiients. Journal of Applied Eonometris 21, 227-244. Melitz, M. J. (2003). The impat of trade on intra-industry realloations and aggregate industry produtivity. Eonometria 71, 1695-1725. 18

Midelfart-Knarvik, K. H., Overman, H. G., Redding, S. J., and Venables, A. J. (2000). The Loation of European Industry. European Eonomy - Eonomi Papers 142, Commission of the EC, Diretorate-General for Eonomi and Finanial Affairs (DG ECFIN). Montagna, C. (2001). Effiieny gaps, love of variety and international trade. Eonomia 68, 27-44. Samuelson, P. (1954). The Transfer Problem and Transport Costs, Eonomi Journal. Eonomi Journal 64, 264-89. Train, K. (2003). "Disrete Choie Methods with Simulation," Cambridge University Press, Cambridge. Wei, S. J. (2000). How taxing is orruption on international investors? Review of Eonomis and Statistis 82, 1-11. Woo, C. Y. (1987). Path-Analysis of the Relationship between Market Share, Business- Level Condut and Risk. Strategi Management Journal 8, 149-168. 19

Table 1: List of variables, definitions and soures Variable Definition Soure Choie a CEEC, in whih firm n hooses to loate its Bureau van Dijk Zephyr investment over the period of time from 1997 to 2007 (it gets the value of 1 if the ountry reeived investment and 0 otherwise) database τ d distane between the apital ities of the soure http://www.indo.om/distane/ ountry d and the host ountry in kilometres D Real GDP of the host ountry of the year IFS investment took plae G Corruption pereption index of the host ountry of the year investment took plae Transpareny International unemployment rate of ountry (perentage per IFS u annum) of the year investment took plae T effetive orporate inome tax rate in ountry of the year investment took plae r the real disount (interest) rate IFS CBD d a dummy variable that takes a value 1 if both onstruted soure ountry d and host ountry share a border, and 0 otherwise EU Sale s Siene s Tradit s Servie s Wage Size n Earnings n dummy variable that takes value 1 if ountry joined EU before January 2007, and 0 otherwise dummy variable that takes a value 1 if industry s is a sale-sale industry, and 0 otherwise dummy variable that takes a value 1 if industry s is a siene-based industry, and 0 otherwise dummy variable that takes a value 1 if industry s is a traditional industry, and 0 otherwise dummy variable that takes a value 1 if industry s is a servie setor, and 0 otherwise hourly real wage rates in the industry s in the ountry of the year investment took plae turnover of the investing firm i in Euros of the year investment took plae earnings before interest and taxes of the investing firm i in Euros of the year investment took plae Calulated using data from IFS onstruted onstruted onstruted onstruted onstruted International Labour Organisation Bureau van Dijk Zephyr database Bureau van Dijk Zephyr database 20

Table 2 The Imposition of Various Distributions for Country-level Random Variables Distane Risk Coef z-stats Coef z-stats Log-likelihood Coef z-stats Coef z-stats Log-likelihood Normal Log-likelihood is flat 0.3906 {5.354} 0.5161 {6.012} -2507.072 Log-normal Log-likelihood is flat Log-likelihood is flat Triangular -1.1821 {-11.007} 3.0309 {7.569} -2501.440 0.3964 {5.394} 1.2748 {6.218} -2506.846 Restrited Triangular -1.0522 {-11.635} 1.0522 {11.635} -2510.590 0.2894 {4.309} 0.2894 {4.309} -2513.508 Dome -1.1291 {-10.192} 3.1230 {7.490} -2502.545 0.4579 {5.571} 1.6673 {6.261} -2504.697 Erlang Log-likelihood is flat 0.4193 {5.097} 0.5565 {4.884} 2507.592 Weibull Log-likelihood is flat 0.9415 {6.076} 0.3256 {5.190} -2509.200 Exponential Log-likelihood is flat 0.3280 {4.888} 0.3845 {4.454} -2510.319 Unemployment Interest Coef z-stats Coef z-stats Log-likelihood Coef z-stats Coef z-stats Log-likelihood Normal -10.3786 {-5.634} 18.3814 {8.110} -2498.291 0.1900 {1.122} 2.7461 {0.727} -2514.210 Log-normal 1.4548 {1.2147} 1.2147 {4.719} -2508.554-0.3434 {-0.077} 0.9028 {0.178} -2514.339 Triangular -10.5425 {-5.674} 44.5571 {8.379} -2498.102 1.2036 {1.151} 6.7393 {0.955} -2514.191 Restrited Triangular -5.7529 {-4.164} 5.7529 {4.164} -2513.707 1.0176 {1.029} 1.0176 {1.029} -2514.376 Dome -11.1859 {-5.827} 47.8832 {7.987} -2499.240 1.4166 {1.363} 10.1212 {1.114} -2514.019 Erlang -9.9504 {-5.427} 15.9837 {6.419} -2502.942 1.1047 {1.124} 1.6004 {0.455} -2514.222 Weibull 14.3140 {5.474} 13.5585 {7.337} -2501.263 Log-likelihood is flat Exponential -8.7071 {-4.840} 21.9603 {7.439} -2496.474 1.0517 {1.040} 1.3075 {0.405} -2514.332 Tax Wage Coef z-stats Coef z-stats Log-likelihood Coef z-stats Coef z-stats Log-likelihood Normal -0.0766 {-1.503} 35.1249 {3.971} -2511.830-0.105 {-1.990} 0.0392 {1.451} -2514.294 Log-normal 0.0975 {0.051} 1.7071 {1.552} -2513.612-2.9966 {-3.077} 1.9047 {2.696} -2510.452 Triangular -0.8978 {-0.347} 87.1765 {4.013} -2511.847-0.1071 {-1.990} 0.1191 {1.550} -2514.260 Restrited Triangular 2.3872 {1.181} 2.3872 {1.181} -2514.386 Log-likelihood is flat Dome -0.7759 {-0.282} 99.8594 {4.096} -2511.453-0.1466 {-2.521} 0.3523 {1.187} -2513.993 Erlang 0.2126 {0.077} 28.6690 {2.416} -2512.262-0.2053 {-3.152} 0.3396 {3.478} -2511.963 Weibull Log-likelihood is flat 0.0275 {0.228} 0.0894 {0.825} -2513.937 21

Exponential 0.3441 {0.140} 31.1936 {2.704} -2512.805 0.1725 {-2.123} 0.1725 {1.689} -2513.408 GDP Coef z-stats Coef z-stats Log-likelihood Normal 0.4970 {7.635} 0.9543 {6.857} -2485.322 Log-normal Log-likelihood is flat Triangular 0.4956 {7.497} 2.2860 {6.988} -2484.991 Restrited Triangular 0.5579 {11.120} 0.5579 {11.120} -2504.503 Dome 0.4803 {5.752} 2.1914 {5.160} -2490.409 Erlang 0.3918 {4.174} 0.9384 {4.661} -2494.652 Weibull 1.7012 {8.995} 0.6940 {0.1214} -2494.405 Exponential 0.4301 {5.791} 0.8612 {4.652} -2500.163 22

Table 3 The Results of the Mixed Logit Model Estimation Variables Distribution Mean Stand. Dev. Wage Triangular -0.285 {-4.493} 0.7252 {3.021} GDP Triangular 0.575 {10.760} 1.4868 {5.511} Distane Triangular -1.252 {-10.625} 1.9154 {3.223} Unempl Triangular -3.5644 {-2.684} 20.7788 {2.653} Border Restrited Uniform 0.5651 {4.205} 0.5651 {4.205} EU Restrited Uniform 0.8028 {5.204} 0.8028 {5.204} Interest - 4.5863 {3.945} - - Prof_Dist - 0.1471 {2.922} - - Size_GDP - 0.1334 {4.090} - - Tr_Unemp - 8.4893 {4.308} - - Tr_Wage - -1.4718 {-5.065} - - Table 4 Elastiities and Marginal Effets Country Wage Tr_Wa Sum Unem Tr_Un Sum GDP Dist Border EU BG -0.2202-0.1107-0.3309-0.2445 0.3796 0.1351 0.0519-1.6297 2.7083 7.1613 CZ -0.5055-0.2631-0.7686-0.3289 0.1867-0.1422 0.3748-1.7266 1.5842 7.8312 EE -0.3121-0.1521-0.4642-0.1797 0.0607-0.119 0.0222-1.7427 0.3026 7.1535 CR -0.4737-0.3035-0.7772-0.193 0.4491 0.2561 0.0798-2.0011 0 0 HU -0.4693-0.2257-0.695-0.329 0.1733-0.1557 0.2714-1.7798 1.2098 6.2476 LT -0.4095-0.2069-0.6164-0.3091 0.198-0.1111 0.0528-2.0335 0 4.5774 LV -0.3861-0.2-0.5861-0.3177 0.211-0.1067 0.0262-2.0144 0.2689 5.1733 PL -0.3865-0.2599-0.6464-0.076 0.3714 0.2954 1.0206-1.3697 1.1459 10.8886 RO -0.1836-0.0732-0.2568-0.3222 0.2252-0.097 0.0956-1.8359 0 5.0919 RU -0.0886-0.039-0.1276-0.1628 0.1214-0.0414 1.3678-1.9985 1.9985 0 SI -0.3268-0.2823-0.6091-0.321 0.1284-0.1926 0.0686-1.7367 0.8931 3.96 SK -0.4688-0.293-0.7618-0.1991 0.3965 0.1974 0.093-1.7468 1.0103 5.8174 UA -0.1212-0.0617-0.1829-0.3373 0.3159-0.0214 0.1277-1.9746 0.4912 0 23

Table 5 Simulation Results of Changes in Market shares and the Number of Firms Size_GDP 1% 10% 50% 100% BG 0.001 0 0.005 0 0.022 0 0.045 1 CZ 0.004 0 0.038 0 0.202 2 0.435 4 EE 0 0 0.002 0 0.011 0 0.023 1 CR 0 0 0.003 0 0.017 0 0.035 1 HU 0.002 0 0.021 0 0.111 1 0.236 3 LT 0.001 0 0.004 0 0.021 1 0.042 0 LV 0.001 0 0.003 0 0.011 0 0.023 1 PL 0.018 0 0.185 2 1.007 11 2.148 24 RO 0.001 0 0.007 0 0.039 1 0.081 1 RU 0.025 0 0.254 2 1.23 3 2.373 26 SI 0 0 0.003 0 0.018 0 0.037 0 SK 0.001 0 0.008 0 0.039 0 0.079 1 UA 0.001 0 0.007 0 0.034 1 0.07 1 Prof_dist 1% 10% 50% 100% BG 0.008 0 0.089 1 0.973 11 1.788 20 CZ 0.007 0 0.09 1 0.899 10 1.64 18 EE 0.007 0 0.071 1 0.732 8 1.723 19 CR 0.003 0 0.034 1 0.538 6 1.21 14 HU 0.007 0 0.085 1 0.914 10 1.667 18 LT 0.006 0 0.063 0 0.747 8 1.639 18 LV 0.006 0 0.062 1 0.738 8 1.633 18 PL 0.022 0 0.261 3 1.33 15 2.164 24 RO 0.008 0 0.102 1 1.051 12 1.905 21 RU 0.032 0 0.3 3 1.267 14 2.103 23 SI 0.003 0 0.037 0 0.508 5 1.134 12 SK 0.005 0 0.059 0 0.727 8 1.476 16 UA 0.005 0 0.06 1 0.833 10 1.626 18 24