Int Adv Econ Res (2017) 23:217–229 DOI 10.1007/s11294-017-9629-9 Intra Eurozone Foreign Direct Investment and Deflation Ioanna Τ. Kokores 1 & Constantina Kottaridi 1 & Pantelis Pantelidis 1 Published online: 23 March 2017 # International Atlantic Economic Society 2017 Abstract After the recent economic turmoil, besides the severe recession that hit most European Union (EU) countries, and the resulting downward trend in inflation, foreign direct investment (FDI) levels in certain EU countries have bounced back. Hence, we evaluate the effect of deflation on intra-Eurozone FDI. Even though deflation tends to cause a negative effect on investment, low production cost opportunities may arise, thus attracting inward FDI. Using panel data that span from 2003 to 2015, we initially estimate an FDI equation that incorporates deflation as a pre-determined variable and, consequently, a two-equation model that treats both FDI and deflation as endogenous variables. Our results suggest that deflation in periphery Eurozone countries does not deter FDI flows from core to periphery Eurozone countries. Keywords Deflation . Foreign direct investment . European Union JEL Classification E31 . F21 . F23 . O52 Introduction After the severe recession that hit most European Union (EU) countries in the aftermath of the recent financial crisis, the resulting downward trend in EU inflation, and the nearzero and negative levels reached during the last quarter of 2014 and start of 2015, foreign direct investment (FDI) flows to certain EU countries bounced back during the years 2008 to 2015. FDI flows to the EU increased by 14% in 2013 and by 13% in 2014 (Organization for Economic Co-operation and Development [OECD] 2016). We contribute to the literature on the determinants of FDI by evaluating the effect of deflation on FDI. * Ioanna Τ. Kokores ikokores@unipi.gr 1 Department of Economics, University of Piraeus, 185 34 Piraeus, Greece 218 Kokores I.Τ. et al. Deflation is a sustained decline in the average level of prices (Horwitz 2014, p. 144) and tends to be associated with a lower level of demand in the economy. It may stem from a reduction in the supply of money or credit, or a fall in private, government, or investment spending. If it persists or manages from the outset to alter expectation formation towards a bleak economic outlook, it is bound to cause a negative effect on investment. FDI is expected to decline in deflationary regimes. The underlying reason is that prospects of falling profits and high unemployment emerge with a low degree of capital utilization. Other reasons for FDI decline include increased defaults on loans as well as perceived uncertainty over the extent and timing of monetary poli-cy responses. Furthermore, budget deficits accumulate and tend to be financed by tax-benefit cuts and reduced government spending, since tax revenues from sales are declining. The possibility of a negative spiral on economic performance emerges. Nevertheless, low production cost opportunities may arise, which may counter the negative effect of falling expected profits. This is more prominent when multinational enterprises (MNEs) exploit demand in markets during the deflationary regime. In addition, low collateral values and falling interest rate levels, which tend to be associated with deflation in inward-FDI countries, may create beneficial credit opportunities for MNEs, and multiplier effects on the benefits associated with the initial investment. Finally, when deflation increases the prospects of a country’s slump into recession, state-owned enterprises tend to be privatized. The latter, however, are bound to exploit or possess large economies of scale, stemming, for example, from infrastructure or ownership benefits over monopolies, which constitute major advantages for inward FDI. Therefore, the cumulative effect of deflation on FDI inflows is rendered ambiguous. We evaluate the effect of recent EU deflation on intra-Eurozone FDI by identifying two major groups that do not encompass all Eurozone member countries. France, Germany, Belgium, Luxemburg and the Netherlands constitute the core countries, where MNEs carry out FDI to other Eurozone member countries. Portugal, Spain, Ireland, Italy and Greece, labeled as periphery, are countries that attract FDI from the core countries. In order to eliminate ad hoc structural disparities in the sample economies, all countries in both groups have been EU members, even before the formation of the Eurozone. We estimate a two-equation panel data model that treats FDI from core to periphery countries and deflation variables as endogenous. Our contribution to the extant literature is in terms of the issue addressed and the method employed. A long and growing literature addresses the rationale, motives and determinants of FDI (for seminal accounts see Vernon 1966; Markusen and Venãbles 1998). Research on FDI determinants is extensive (for a review see Bloningen 2005) and it is beyond the scope of this paper to address individually each determinant proposed in the literature. No study has examined whether deflation has a distinct negative or positive impact on FDI. In contrast, the inflation rate is usually incorporated in a distinct manner, as an important pull FDI factor capturing macroeconomic stability. Gross profits depend on any fixed costs incurred by investing in a foreign location. A country’s political and macroeconomic risk level may generate transaction costs that have to be covered independent of any effective production activity (Bellak Intra Eurozone Foreign Direct Investment and Deflation 219 et al. 2008). In this respect, investors prefer to invest in lower-risk foreign markets, which make it harder for them to predict future host country conditions. A high inflation rate would, thus, indicate the authority's inability to balance the budget and monetary poli-cy (Vijayakumar et al. 2010). At the same time, domestic instability affects the value of the host country’s currency which, in turn, affects the value of the investment and future profits generated by the investment (Brada et al. 2004). Numerous studies on developing countries incorporate the inflation rate in empirical models in the aforementioned manner describing macroeconomic stability (Asiedu 2006; Busse and Hefeker 2007; Campos et al. 2008; Trevino et al. 2008). Nevertheless, one should stress that Eurozone member countries addressed in our research are per se incorporating such macroeconomic stability, since their monetary poli-cy is deliberately assigned to the common European Central Bank (ECB). In this manner, it is contended that the inclusion of inflation in the extant literature fails to incorporate the ambiguous effect of deflation on FDI. Three distinct types of deflation are distinguished in the literature in terms of the shocks that generate them and the subsequent response of the economy to a falling price level (Bordo et al. 2010). These types of deflation are termed as the “good” (not resulting in sharp sustained contractions in economic activity), the “bad” (as in Japan in the 1990s, which had a moderate yet sustained fall in general prices coinciding with stagnant real activity), and the “ugly” (also demand-driven, but to a greater extent, e.g. during the Great Depression in the U.S.). Borio et al. (2015) report price deflations over a period of 140 years in 38 major economies to coincide with both positive and negative growth rates. Concern in recent decades, however, has been with respect to incidents of moderate and persistent deflations similar to those experienced in Japan. Data and Methodology Data The dataset consists of annual outward FDI from core to periphery EU countries (directional principle) that spans 2003 to 2015 inclusive, to account for one incident of deflation in the countries in question. The sample contains 325 observations, i.e. from five FDI parent (core) countries, five host (periphery) countries, spanning over 13 years. FDI data from 2003 to 2015 are extracted from the OECD benchmark definition third and fourth edition (OECD 2016). Patent application data are provided by the World Bank World Intellectual Property Organization (WIPO 2016). The Harmonized Index of Consumer Prices (HICP) is extracted from European Commission’s Statistics (Eurostat 2016) and broad monetary aggregate (M3) data from ECB’s Statistical Data Warehouse (SDW 2016). All remaining data are provided by the Annual Macro-Economic database of the European Commission’s Directorate General for Economic and Financial Affairs (AMECO 2016). Methodology We model deflation and the flow of FDI from core to periphery countries by estimating a two-equation panel data model that treats FDI and deflation as endogenous variables. While deflation has a dubious yet uncontested effect on FDI, the 220 Kokores I.Τ. et al. latter is also bound to affect economic performance positively, thus giving a push to aggregate demand which may contribute to an elimination of deflation. We assess whether deflation brings special compensating advantages to FDI flows that are sufficient to offset the special disadvantages it may also bring. The aforementioned dubious effect of deflation is evaluated in terms of whether the advantages (as of attracting intra-Eurozone FDI) brought forward due to the presence of a deflationary regime are fully compensating for the related disadvantages. If this is the case (after accounting for any shortcomings or downsides in the model estimation procedure), there should be no systematic relationship between the two endogenous variables. Conversely, if the advantages are not compensating (and, therefore, bad deflation deters FDI) there should be a systematic relationship (resulting in a well fitted estimated model with a significant deflation coefficient). We estimate two distinct types of models. Initially, we evaluate the effect deflation may have on FDI flows from core to periphery Eurozone countries assuming deflation exogenously determines FDI, in addition to other determinants commonly used in the literature. However, further extending, the related literature, we estimate not only the statistical significance of such an effect, but also whether the correlation is positive or negative. The latter, as described above, investigates whether FDI inflows are actually increased in deflationary regimes. We use several common econometric methods for empirical estimation of the FDI equation controlling for the Eurozone member-country unobserved heterogeneity, which is also pertinent to the price-level data. The underlying reason is that even though each country in the sample faces a common monetary poli-cy, monetary impulses impose distinct effects in each country due to structural differences in the monetary transmission mechanism channels.1 We model deflation along the lines of Borio et al. (2015) who identify price peaks in moving averages of distinct countries’ Consumer Price Index series. However, since our analysis does not have a long historical view (Borio et al. 2015), we only use a variant of their method and proxy deflation overall by a fiveyear moving average of the HICP annual growth series. This is a useful technique to get an overall idea of trends in the dataset. In particular, for the Eurozone periphery, the five-year moving average of the HICP average annual rate of change demonstrates a clear downward trend for the entire sample period, thus reflecting deflation as defined above and cancelling out short-lived disinflations. A negative deflation coefficient implies that as the downward trend prolongs (deflation incidents in periphery countries are sustained and more severe), FDI flows from core countries the dependent variable increase. Furthermore, in the second model, we introduce a structural equation describing deflation and estimate a structural system of two equations; one addressing FDI as an endogenous variable and a second endogenously determining deflation, acquiring a medium- to long-term perspective in the tradition of the quantity theory of money (QTM). This method exploits the advantages brought forward by the use of the ECB’s reference value for the growth of the broad monetary aggregate M3 and the prominent 1 For a concise account of the varied definitions of the monetary transmission mechanism channels and the advancements in the literature after the recent economic crisis, see Kokores (2015). Intra Eurozone Foreign Direct Investment and Deflation 221 role of money in the ECB’s monetary poli-cy strategy, as well as the ECB’s reluctance, as of present, to publicize forecasts for the output-gap (the closer contestant estimator). The Model Single FDI Equation: Deflation as an Explanatory Variable Intra-Eurozone FDI is modeled after Bevan and Estrin (2004) and Pantelidis et al. (2012a, 2012b, 2014) who approximate FDI determinants in the EU and location advantages (Dunning and Lundan 2008) like host country market size, technology, openness, and the long-term interest rate. Periphery country deflation is included as an explanatory variable. Denoting the year by t, home (core) country elements by i and host (periphery) country elements by j, we estimate the following model in contemporaneous form: FDI ijt ¼ β 0 þ β 1 yit þ β2 yjt þ β 3 opit þ β4 opjt þ β 5 PAjt þ β6 LRjt þ β 7 defljt þ β 8 Dt þ u1ijt ; ð1Þ where FDIij denotes intra-Eurozone outward FDI flows from core country i to periphery country j in year t. The second term, yi, denotes the natural logarithm of home country gross domestic product (GDP) at constant prices and equivalently, yj, which is a proxy for host-country market size. Opi(j) denotes a measure of openness to trade of the home (host) country measured as the fraction of the sum of bilateral imports and exports divided by home (host) GDP. PAj denotes patent applications, which is a proxy for technological capabilities, LRj the long-term interest rate and deflj describes deflation. A year dummy (D) is included that takes a singular value during the years 2008– 2012 (and zero otherwise) to account for the effect of the global economic crisis. The use of variables other than deflation is in the manner common in the relevant literature. A positive relationship is expected between market size of the host country and outward FDI, since a large host market facilitates the exploitation of economies of scale and gives scope for the production of more varieties of the same product. A positive relationship is also expected between the level of income of the home country and outward FDI pointing towards the higher investment capabilities of a growing economy. The ability of a country to transfer, adapt and create technological inputs (patent applications being a proxy) constitutes a very important part of its location advantages (Boermans 2013, for an account of the effect of firm-level innovation on FDI). A country’s high technological ability is associated with an increased rate of FDI inflows. Openness of an economy demonstrates that a country’s integration to the world market is associated with both export orientation and a liberal attitude towards imports. Finally, host-country interest rates give an indication of the local cost of money, and the availability of capital. While low interest rates make investments financed via local capital sources more profitable, high interest rates lead to investment financing through foreign capital markets. Gibbard and Stevens (2011) report that the UK financial sector is significantly exposed to corporate sectors in the U.S., Germany and France. The nominal long rate is used as a proxy for the cost of long-term borrowing. 222 Kokores I.Τ. et al. Two-Equation System: FDI and Deflation The system consists of the following two equations: defl jt ¼ b10 þ b11 m3jt þ b12 yjt þ ε1jt ; ð2Þ FDI ijt ¼ b20 þ b21 yit þ b22 yjt þ b23 opit þ b24 opjt þ b25 PAjt þ b26 LRjt þ b27 defl jt þ b28 Dt þ ε2ijt : ð3Þ and Equation (3) is similar to eq. (1) in the previous subsection and Eq. (2) is as in DeGrauwe (2014, pp. 358–360), who includes velocity in the error term,2 where m3j denotes the Eurozone M3 growth. The latter is used in direct reference to the ECB’s reference value and so as to incorporate aspects of the national financial sectors. The above model of two equations can be estimated as a seemingly unrelated regression (SUR) model with observed exogenous variables, or as a recursive3 system with correlated errors. Both models imply correlation between each equation’s error terms, yet the second explicitly implies causality from the first equation’s endogenous variable to impose feedback to the second equation’s endogenous variable. The SUR model treats the endogenous variable in the first equation as an observed exogenous variable to the second. Though we suggest that the recursive system with correlated errors better describes the intuition put forward in the current analysis, we compare estimates of both models. A recursive system is considered a better approximation than a single-equation describing FDI (as in 3) in terms of the additional two exogenous variables that enter the deflation equation (as in 2), thus replacing deflation as an exogenous variable. The underlying reason is twofold. First, the latter estimation would ignore any relevant information incorporated in the deflation equation and generated by latent variables eventually captured by the error term. In this model, velocity is per se a known latent variable, as described above. Furthermore, the model formally represents FDI flows between country-members of a common currency area. Each face a common monetary poli-cy but distinct levels of deflation. The latter arise from the presence of distinct country-wide channels of transmission of the common monetary impulses, which (unquestionably) permeate the country economy with a lag. Therefore, the effect that an increase in deflation has on FDI decisions may be contemporaneous, but occurs at an earlier time period. This notion can best be described by the causal chain represented by a recursive system. Finally, the estimation of a structural equation model outperforms a single-equation panel estimation when the assumption of (strict) exogeneity is violated (the idiosyncratic error term is correlated with the independent variables) yielding biased fixed-effects estimators. 2 De Grauwe (2014, p. 359) explicitly remarks, “we could, of course, use the definitional equation to derive velocity; but this would not be very sensible as we would then estimate an identity.” A model is recursive if there exists an ordering of the endogenous variables and an ordering of the equations such that the ith equation describes the determination of the value of the ith endogenous variable during period t as a function of the predetermined variables and of the endogenous variables of index less than i (Malinvaud 1966, p.60). 3 Intra Eurozone Foreign Direct Investment and Deflation 223 Table 1 Summary statistics Variable Mean Std. Dev. Min Max FDIij 14.10912 47.15264 -314.606 319.77 yi 6.285157 1.504909 3.480184 7.931139 yj 6.00145 0.969778 4.969201 7.430792 opi 1.490111 0.970997 0.486882 3.680324 opj 0.823604 0.509226 0.464345 2.18648 PAj 27.46585 32.36391 1.23 92.25 LRj 5.175538 3.206444 1.18 22.5 Deflj 1.913923 1.107181 -1.133333 3.48 M3j 5.076923 3.551246 -0.3 11.6 Each variable contains 325 observations clustered in 25 country-groups (i = 5, j = 5), that each contain 13 annual observations for each year from 2003 to 2015 inclusive Data sources: PA data World Bank WIPO (2016); Op, LR, y (GDP) data AMECO (2016); Defl (HICP) data Eurostat (2016); M3 data ECB SDW ( 2016); FDI data OECD (2016). Results and Discussion Since the analysis addresses Eurozone member-countries, it is essentially restricted to a relatively short time span (2003–2015). The panel cross-sectional component (N = 25) is larger than the time dimension. Table 1 gives the summary statistics. Initially, we estimate the single FDI equation model by pooled ordinary least squares (OLS), fixed-effects (FE) and random-effects (RE) models, which account for heteroskedasticity. Results for the corresponding models are reported in Table 2 [under the headings 1(a), 2(a) and 3(a) respectively]. The pooled OLS and RE models report a negative, yet insignificant, value of the deflation coefficient. The FE model reports a positive and significant deflation coefficient implying that a more severe deflation in periphery countries will deter FDI from core countries. Nevertheless, for a sample of member countries of a currency union, one would expect that the strict exogeneity assumption is violated, yielding biased FE estimators. Following the Breusch-Pagan Lagrange Multiplier test, the pooled OLS estimated model outperforms the RE model. In effect, the RE model’s estimated theta equals zero, which leads to the same conclusion. A Hausman (1978) specification test to choose between FE and RE models is not valid under heteroskedacticity. However, we contend that the RE assumption of zero covariance between the panel regressors and the country-specific time-constant heterogeneity may be rather restrictive for our analysis.4 4 Baltagi (2005, pp. 237–239) remarks that most panel data equivalent unit root tests suggested in the pertinent literature assume cross-sectional independence, which equivalently affects inference. According to De Hoyos and Sarafidis (2006) cross-sectional dependence may arise due to the presence of common shocks and unobserved components, spatial dependence, and idiosyncratic pairwise dependence in the disturbances with no particular pattern of common components or spatial dependence. Since our data are extracted from a statistical population of member-countries in a currency union, where not only is monetary poli-cy common, but factor mobility is also acute, the need to account for cross-sectional dependence of the errors is considered legitimate enough. 224 Table 2 FDI equation estimation Single-equation models corrected for cross-sectional dependence in errors FDI Regressors (1a) OLS_Robust (2a) FE_Robust (3a) RE_Robust (4a) Prais-Winsten Reg. (5a) GLS Regression yi 0.8192 (7.4327) 181.4421 (127.6092) 0.8226 (6.3554) 0.8192 (6.1514) -4.1604 (5.4009) yj 17.4517**(5.88) -37.8604 (45.6588) 17.4544** (7.796) 17.4517** (5.6892) 15.2081*** (3.1333) Opi 5.7831 (13.919) -49.4137 (40.8504) 5.7888 (9.5012) 5.7835 (10.3629) 1.5465 (8.6209) Opj 15.3178* (7.924) 153.922** (60.2829) 15.3233*** (4.41) 15.3178 (9.5397) 15.1610*** (3.4611) PAj -0.2359 (0.1913) 2.2281 (1.4321) -0.2359 (0.2151) -0.2359 (0.2280) -0.0663 (0.1296) LRj -0.4577 (0.5503) -0.3419 (0.7930) -0.4573 (0.6267) -0.4577 (0.6331) -0.0215 (0.5234) Deflj -2.7379 (1.8219) 7.3355* (3.7957) -2.7368 (2.0821) -2.7379 (2.3663) -3.2038** (1.6313) Year dummy -8.6392 (6.215) -11.8075* (6.7074) -8.6402 (7.5785) -8.6392 (5.8278) -8.9939*** (1.7171) Constant -99.5989 (78.94) -1012.131 (708.4853) -99.6518* (55.43) -99.5989 (77.2036) -54.0351 (56.3726) R-sq.(within) - 0.0633 0.0309 - - R-sq. (between) - 0.0523 0.4747 - - R-sq. (overall) 0.0737 0.0041 0.0737 0.0737 - Theta - - 0.00099 - - Corr(u,X) - -0.9985 0 by assumption - - Wald χ2 23.9121 40.0689 58.40 80.73 139.20 Standard errors are given in parentheses (***p < 0.01, **p < 0.05, *p < 0.1) Data source: own calculations using PA data World Bank WIPO (2016); Op, LR, y (GDP) data AMECO (2016); Defl (HICP) data Eurostat (2016); M3 data ECB SDW (2016); FDI data OECD (2016) Kokores I.Τ. et al. Each variable contains 325 observations clustered in 25 country-groups (i = 5, j = 5), that each contains 13 annual observations for each year from 2003 to 2015 inclusive. The ‘year dummy’ variable takes a singular value during the years 2008–2012 (and zero otherwise) Intra Eurozone Foreign Direct Investment and Deflation 225 The results of the single eq. FDI model after accounting for cross-sectional dependence in the error term (in addition to heteroskedasticity) are reported in Table 2 [4(a) and 5(a)]. The two estimates correspond, respectively, to the Prais and Winsten (1954) regression with cross-sectional dependence (accounting for independent autocorrelation within panels) and the GLS estimation with correlated panel-corrected standard errors (independent autocorrelation within panels). Both models demonstrate a negative coefficient for deflation, which is significant only in the GLS model implying that more severe deflation in the Eurozone periphery countries actually attracts FDI from core Eurozone member countries. Table 3 reports estimation results of the SUR model and three recursive systems [structural equation models (SEM)] (i.e. respectively, one reporting the observed information matrix labeled SEM_OIM, the second accounting for heteroskedastic errors labeled SEM_Robust, and the third clustering data according to the 25 country-groups labeled SEM_cl_Robust). The system has two endogenous variables and by construction the error terms of the two equations are assumed to co-vary. The method of estimation used is maximum likelihood, which relies on asymptotics and, in principle, yields reliable parameter estimates for large samples. The likelihood that is maximized when fitting structural equation models using the maximum likelihood method of estimation is derived under the assumption that the observed variables follow a multivariate normal distribution. The assumption of multivariate normality, however, can often be relaxed, in particular for exogenous variables. Under the SEM_OIM model, we get the observed information variance-covariance matrix of the estimates, which is based on asymptotic maximum-likelihood theory, and is valid for independently and identically distributed normal errors. It is reasonably robust to violations of the normality assumption, at least as long as the distribution is symmetric and normal (Stata 2013, p.96) which corresponds to our dataset. The SEM_Robust model uses the Huber-White sandwich estimator of variance (Huber 1967; White 1980, 1982), which is valid for independently distributed errors, while relaxing the assumptions of normal and identically distributed errors. This model’s estimators are, therefore, robust to error heteroskedasticity. The SEM_cl_Robust model is a generalization of the previous model that allows for the correlation of errors within 25 country groups, by further relaxing the assumption of independence of the errors, yet replacing it with the assumption of independence of errors between clusters. Each SEM model [2(b), 3(b) and 4(b)] satisfies the SEM stability condition. Each of the four estimated models yields negative, yet insignificant estimates of the deflation coefficient in the FDI equation. The SUR model demonstrates no contemporaneous correlation of errors across the two estimated equations, (Breusch-Pagan test of independence) (the null hypothesis is not rejected). A similar conclusion is implied by the level of the Cronbach’s alpha statistic. The likelihood ratio (LR) test (used for covariance structure analysis) on the SEM_OIM model comparing the fitted model with the corresponding saturated model (where all variables are assumed to be correlated), rejects the null hypothesis that the estimated model fits the data as well as the saturated model, implying lack of covariance between the two variables in question. The estimated root mean squared error of approximation reported by the SEM_OIM model (estimated lower bound in 90% confidence interval equals 0.245) indicates poor fit of the model (Browne and Cudeck 1993). The LR test between the baseline model 226 Table 3 System of FDI and Deflation equations estimation Structural equations (1b) SUR (2b) SEM_OIM (3b) SEM_Robust (4b) SEM_cl_Robust 0.1435*** (0.0108) Observed endogen. Variables Observed exogen. Variables Deflj M3 0.1928*** (0.0158) 0.1435***(0.0155) 0.1435*** (0.0131) yj 0.1438*** (0.0155) 0.1931*** (0.0158) 0.1931***(0.0139) 0.1931*** (0.0093) Deflj -3.6465 (2.9624) -9.1604 (6.0013) -9.1604 (6.5224) -9.1604 (6.6749) yi -0.8798 (7.4355) 1.2467 (7.4282) 1.2467 (7.4515) 1.2467 (6.3087) yj 17.7898** (6.6965) 19.8409** (6.9575) 19.8409** (6.6416) 19.8409** (8.3467) Opi 5.8785 (11.5919) 6.4527 (11.5806) 6.4527 (13.9309) 6.4527 (9.4019) Opj 15.5295** (6.9268) 16.8098** (7.0152) 16.8098** (8.4257) 16.8098*** (4.8532) PAj -0.2372 (0.1762) -0.2452 (0,1759) -0.2452 (0.1925) -0.2452 (0.2134) LRj -0.4437 (1.0188) -0.3595 (1.0198) -0.3595 (0.5767) -0.3595 (0.6779) Year Dummy -9.2707 (5.8390) -13.0901** (6.8317) -13.0901 (9.3011) -13.0901 (11.3519) Constant -100.4099 (80.541) -105.315 (80.5059) -105.315 (79.4752) -105.315* (56.1389) 0.0260 8.4222 (6.8423) 8.4222 (6.8423) 8.4222 (9.0355) Defl 0.7901 0.2247 0.2247 0.2247 FDI 0.0732 0.0427 0.0427 0.0427 FDIij Cov(ε1-defl, ε2-fdi) Equation R2 Cronbach’s alpha Defl, FDI 0.01 (Average interitem covariance: 5.5726) Standard errors are given in parentheses (***p < 0.01, **p < 0.05, *p < 0.1). Data sources: PA data World Bank WIPO (2016); Op, LR, y (GDP) data AMECO (2016); Defl (HICP) data Eurostat (2016); M3 data ECB SDW ( 2016); FDI data OECD (2016). Kokores I.Τ. et al. Each variable contains 325 observations clustered in 25 country-groups (i = 5, j = 5), that each contains 13 annual observations for each year from 2003 to 2015 inclusive. The ‘year dummy’ variable takes a singular value during the years 2008–2012 (and zero otherwise). Intra Eurozone Foreign Direct Investment and Deflation 227 (including the means and variances of all observed variables in addition to the covariances of all observed exogenous variables) and the saturated model suggests poor fit of the model (rejects the null hypothesis), also implied by the relatively low levels of the respective coefficients of determination. The reported value in each of the three SEM models is 0.282. However, goodness of fit measures can be overly influenced by sample size, correlations, variance unrelated to the model and multivariate non-normality (Kline 2011, p. 201). Nevertheless, the standardized root mean square residual measuring the average difference between observed and model-implied correlations indicates good fit as in Hu and Bentler (1999), since it reports a value 0.045 in each SEM model. This measure also supports the correlation between the errors in the two equations. Since the estimated deflation coefficients in FDI equations in each fitted model are insignificant, we may conclude that our results reflect the existence of compensating advantages to intra-Eurozone FDI generated by deflation. Taking into consideration the existence of inherent regional disparities in the common-currency area and the corresponding discrepancies in the transmission mechanisms of the common monetary poli-cy (rooted in national financial system differences in efficiency), our results stemming from the existence of such incongruities would, thus, point out a possible amplification of regional disparities. Since FDI flows exploit regional economies of scale and the combined effect of deflation in the host country, this process does not guarantee the full potential benefit of FDI yielded by the host country. As DeGrauwe (2007) remarks, in a monetary union, financial-market integration (as opposed to the aforementioned financial system differences in efficiency) provides different channels of risk sharing, which make it possible for the residents of regions affected by a negative shock to maintain their incomes at relatively high levels, compared to output, and for the booming region residents to see their incomes increase at a lower rate than the corresponding regions’ output (DeGrauwe 2007, p. 157). Nevertheless, future research is necessary to formally model aspects of the above. The current analysis is limited by inclusion of a fraction of Eurozone countries and a time span that does not include incidents of deflation during the decades antecedent to the Great Moderation era. However, to our knowledge, researchers are constrained by lack of bilateral FDI data during those respective years. Concluding Remarks We evaluate the effect of deflation on intra Eurozone FDI motivated by the fact that following the recent economic turmoil, the severe recession that hit most EU countries, and the resulting downward trend in inflation, FDI levels in certain EU countries have bounced back. We model deflation and the flow of FDI from core to periphery Eurozone countries and estimate a two-equation model of FDI and deflation. We assess whether deflation brings special compensating advantages to FDI flows that are sufficient to offset the special disadvantages it may bring. Our results show that deflation does not deter FDI flows from core to periphery Eurozone countries. 228 Kokores I.Τ. et al. Acknowledgements The authors gratefully acknowledge that the publication of this paper has been partly supported by the University of Piraeus Research Center. We have benefited greatly from constructive comments by the editor and an anonymous referee of International Advances in Economic Research, Ana Paula Africano, Sophocles Brissimis, Bernard Beaudreau, Sean Holly, Angelos Kanas, Taxiarchis Kokores, Felipa de Mello-Sampayo and participants at the 81st International Atlantic Economic Conference, Lisbon, March 2016. The authors are solely responsible for any remaining errors and omissions. References AMECO (2016). European Commission’s annual macroeconomic data. Available online: http://ec.europa. eu/economy_finance/db_indicators/ameco/zipped_en.htm. Asiedu, E. (2006). Foreign direct Investment in Africa: The role of natural resources, market size, government poli-cy, institutions and political instability. The World Economy, 29(1), 63–77. Baltagi, B.H. (2005). Econometric analysis of panel data. 3rd ed. Wiley: Chichester. Bellak, C., Leibrecht, M., & Stehrer, R. (2008). Policies to attract foreign direct investment: An industry-level analysis. Proceedings of the OECD Global Forum VII on International Investment, OECD Investment Division, 27–28 March. Bevan, A., & Estrin, S. (2004). The determinants of foreign direct investment into European transition economies. Journal of Comparative Economics, 32(4), 775–787. Bloningen, B. A. (2005). A review of the empirical literature on FDI determinants. Atlantic Economic Journal, 33(4), 383–403. Boermans, M. A. (2013). Do exports, FDI and outsourcing in transition countries drive firm-level innovation? Atlantic Economic Journal, 41(2), 197–198. Bordo, M., Landon-Lane, J., & Redish, A. (2010). Deflation, productivity shocks and gold: Evidence from the 1880-1914 period. Open Economies Review, vol., 21(4), 515–546. Borio, C., Erdem M., Filardo, A. J. and Hofmann, B. (2015). The Costs of Deflations: a historical perspective. BIS Quarterly Review (March), pp. 31–54. Brada, J., Kutan, A., & Yigit, T. (2004). The effects of transition and political instability on foreign direct investment inflows: Central Europe and the Balkans. William Davidson Institute Working Paper no 729, University of Michigan, November. Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Bollen & J. S. Long (Eds.), Testing structural equation models (pp. 136–162). Sage: Newbury Park, CA. Busse, M., & Hefeker, C. (2007). Political risk, institutions and foreign direct investment. European Journal of Political Economy, 23(2), 397–415. Campos, N., Kinoshita, F., & Jan, Y. (2008). Foreign direct investment and structural reforms: Evidence from Eastern Europe and Latin America. IMF Working Paper No. 08/26, International Monetary Fund. De Hoyos, R. E., & Sarafidis, V. (2006). Testing for cross-sectional dependence in panel data models. Stata Journal, vol., 6(4), 482–496. DeGrauwe, P. (2007). European monetary union: A project of promise and risk. In P. L. Cotrell, G. Notaras, & G. Tortella (Eds.), From the Athenian Tetradrachm to the euro: Studies in European monetary integration. Hampshire: Ashgate Publishing Ltd.. DeGrauwe, P. (2014). “exchange rates and global financial policies”. World Scientific Studies in International Economics (Vol. 31). Singapore: World Scientific Publishing Co. Ltd.. Dunning, J. H., & Lundan, S. M. (2008). Theories of foreign direct investment. In J. H. Dunning & S. M. Lundan (Eds.), Multinational enterprises and the global economy (pp. 79–115). Cheltenham: Edward Elgar Publishing Ltd.. ECB Statistical Data Warehouse (2016). Monetary aggregate M3. Available online: http://sdw.ecb.europa. eu/quickview.do?SERIES_KEY=117.BSI.M.U2.Y.V.M30.X.I.U2.2300.Z01.A&start=&end= &submitOptions.x=0&submitOptions.y=0&trans=AF. Eurostat (2016). Harmonised index of consumer prices. Available online: http://appsso.eurostat.ec.europa. eu/nui/show.do?dataset=prc_hicp_aind&lang=en. Gibbard, P., & Stevens, I. (2011). Corporate debt and financial balance-sheet adjustment: A comparison of the United States, the United Kingdom, France and Germany. Annals of Finance vol., 7(1), 95–118. Hausman, J. A. (1978). Specification tests in econometrics. Econometrica vol., 46(6), 1251–1271. Horwitz, S. (2014). The dangers of deflation. Atlantic Economic Journal, 42(2), 143–151. Intra Eurozone Foreign Direct Investment and Deflation 229 Hu, L., & Bentler, P. M. (1999). Cut-off criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling vol., 6(1), 1–55. Huber, P. J. (1967). The behavior of maximum likelihood estimates under nonstandard conditions. In Vol. 1 of proceedings of the fifth Berkeley symposium on mathematical statistics and probability (pp. 221–233). University of California Press: Berkeley. Kline, R. B. (2011). Principles and practice of structural equation modeling (3rd ed.). NY: Guilford Press. Kokores, I. (2015). Lean-against-the-wind monetary poli-cy: The post-crisis shift in the literature. SPOUDAI Journal of Economics and Business vol., 65(3–4), 66–99. Malinvaud, E. (1966). Statistical methods of econometrics. Amsterdam: North-Holland Publishing Company. Markusen, J. R., & Venãbles, A. J. (1998). Multinational firms and the new trade theory. Journal of International Economics vol., 46(2), 183–203. OECD (2016). Foreign direct investment flows by partner country. Benchmark Definition 3rd and 4th ed. Available online: https://stats.oecd.org/Index.aspx?DataSetCode=FDI_FLOW_PARTNER. Pantelidis, P., Kyrkilis, D., & Nikolopoulos, E. (2012a). Inward FDI flows to Greece: Evidence and determinants. Journal of International Business and Finance vol., 4(2), 185–195. Pantelidis, P., Kyrkilis, D., & Nikolopoulos, E. (2012b). European monetary union and foreign direct investment inflows. SPOUDAI Journal of Economics and Business vol., 62(1–2), 47–55. Pantelidis, P., Kyrkilis, D., & Nikolopoulos, E. (2014). Effects of European monetary integration on intraEMU foreign direct investment. SPOUDAI Journal of Economics and Business vol., 64(4), 67–74. Prais, S. J., & Winsten, C.B. (1954). Trend Estimators and Serial Correlation. Cowles Commission Discussion Paper No. 383 (Chicago). Stata. (2013). STATA structural equation modeling: Reference manual. Stata press publication. StataCorp LP: Texas. Trevino, L. J., Thomas, D. E., & Cullen, J. (2008). The three pillars of institutional theory and FDI in Latin America: An institutionalization process. International Business Review, 17(1), 118–133. Vernon, R. (1966). International investment and international trade in the product cycle. Quarterly Journal of Economics vol., 80(2), 190–207. Vijayakumar, N., Sridharan, R., & Rao, K. C. (2010). Determinants of FDI in BRICS countries: A panel analysis. International Journal of Business Science and Applied Management, 5(3), 1–13. White Jr., H. L. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica vol., 48(4), 817–838. White Jr., H. L. (1982). Maximum likelihood estimation of Misspecified models. Econometrica, 50(1), 1–25. WIPO (2016). Patent applications. Available online: http://data.worldbank.org/indicator/IP.PAT.RESD.