Abstract
In this paper we investigate whether better information about the macroeconomic environment of an economy has a positive impact on its capital inflows, namely portfolio and foreign direct investment (FDI). The purpose of our study is to explicitly quantify information asymmetries by compliance with the IMF's Special Data Dissemination Standard (SDDS). For FDI, we find statistically significant and robust support for this hypothesis: SDDS subscription increased inflows by an economically relevant magnitude of about 60 percent. We also find evidence of aversion against political and macroeconomic risk as determinants of portfolio and FDI flows anduse a non-parametric test for spatial correlation in the residual of capital flows.
I. Introduction
Economic theory attributes positive welfare effects to capital flowing from capital-abundant countries to those which have potentially productive assets, but where the capital necessary to employ them is scarce. This implicitly assumes that (foreign) investors are aware of these assets, i.e. they have the information necessary to make an optimal decision. Some investors, especially foreign direct investors, however, are not specialized in acquiring information and those who are, especially larger portfolio funds and credit rating agencies, have to deal with non-excludable information they acquired, due to herd behavior in financial markets (cf. inter alia Banerjee, 1992; Bikhchandani et al., 1992, and Avery and Zemsky, 1998). On the demand side, countries with productive assets, but lack of capital may find it difficult to signal their productivity while less productive countries may whitewash their signaled information, and the international asset market may turn out to be a market for lemons (cf. Akerlof, 1970).1 The price mechanism may also fail in this context because countries with the highest interest rate and hence the most productive investment opportunities may be perceived as especially unstable so that risk averse investors may rather prefer to invest into safe havens (cf. Stiglitz and Weiss, 1981).
Previous research has already investigated the role of information for capital flows, both empirically and theoretically. However, most of the early empirical studies could not convincingly identify a parameter for the quantitative impact of information on capital flows. Similar to the ‘Solow residual’ (cf. Vaizey, 1964, p. 5), they attributed patterns in capital flows that models could not explain to informational frictions. Are these unexplained patterns really a measure of the impact of information asymmetries or are they simply a ‘measure of our ignorance’ about the determinants of international capital flows? Probably the most convincing identification strategies have been provided by studies such as Gelos and Wei (2005), Daude and Fratzscher (2008), and Harding and Javorcik (2011). All these studies have a somewhat different focus and methodology from each other and from our investigation, which is probably the most related to the study of Daude and Fratzscher (2008).
More precisely, we look at the impact that compliance with the IMF’s Special Data Dissemination Standard (SDDS) had on international capital flows, specifically portfolio and foreign direct investments (FDI). The SDDS, established in 1996 with the aim of enhancing member countries’ access to the international capital market, is about macroeconomic data provision to the public. Institutional investors’ decision on investments are based on macroeconomic and financial data, but not all the investors have time and money to collect information they need. A first look at the data supports this view, at least for FDI flows: As depicted in the left panel of figure 1, average levels of FDI inflows (relative to GDP) were higher for almost all subscriber countries after SDDS subscription than before.2 The picture is similar, though less definite for portfolio flows (right panel). We substantiate these descriptive findings in a sophisticated econometric framework where we find statistically significant and robust evidence of an economically relevant impact of providing more (accurate) information about the mac-roeconomic and financial environment under the umbrella of the SDDS on FDI inflows, but fail to find the same evidence for portfolio flows. Furthermore, we find evidence for macro-economic risk-aversion for portfolio and for FDI flows and more robust evidence of political-risk aversion for portfolio flows.
Capital Flows (% of GDP) prior to and after SDDS Subscription
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
Capital Flows (% of GDP) prior to and after SDDS Subscription
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
Capital Flows (% of GDP) prior to and after SDDS Subscription
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
Our contribution further adds to the literature by looking at systematic differences between FDI and portfolio flows and by proposing new measures for productivity that may be especially relevant for the FDI literature. Finally, we also consider spatial interdependencies in our investigation. Contrary to previous studies on spatial interdependences in FDI flows, such as Coughlin and Segev (2000), Blonigen et al. (2007), or Baltagi et al. (2007), our approach relies on less stringent assumptions about the potentially underlying spatial process. In line with the results implied by Baltagi et al. (2008), we do not find evidence for significant spatial patterns in our empirical models.
We review the previous literature on information and international investment in section II. The empirical model and variables used in this paper are introduced together with the data in section III. A variable list with summary statistics of the data can be found in appendix B. We present our results in section IV, showing that better data dissemination through SDDS increases FDI inflows by about 60 percent, but has no significant aggregate effect on portfolio flows. In section V, we show robustness checks that provide strong supplementary support for the impact of information on FDI. In section VI, we discuss the implications of our findings for macroeconomic stability and growth as well as potential lines of future research.
II. Information and Investment: A Literature Review
Previous (macro-)economic studies have already highlighted the role of information in international capital markets, but have mostly failed to provide a convincing empirical identification methodology for the impact of information. French and Poterba (1991), for example, note that even when being risk-averse, few investors diversify their portfolio internationally despite potential nontrivial risk-reduction by cross-border holdings. Their results suggest that investors expect domestic returns to be systematically higher than those of a diversified portfolio by imputing an “extra risk to foreign investments because they know less about foreign markets, institutions, and firms” (p. 225). Tesar and Werner (1995) found that foreign equity portfolios were turned over much faster than domestic equity portfolios. They argue that transactions costs associated with trading foreign securities hence cannot be the reason for the observed reluctance of investors to diversify their portfolios internationally.3 They conclude that informational constraints may play a role, but also argue that the observed lack of international diversification may have less to do with ‘international’ investment choices or transaction costs, but simply reflect the tendency of individuals to hold ill-diversified portfolios.4 Mody and Taylor (2003) find high probabilities of capital crunches for certain episodes in emerging economies and argue that this is not only influenced by default risk but also by asymmetric information; however, the paper fails to convincingly identify this channel. Bren nan and Cao (1997) have argued that positive correlations of international equity flows with the returns on the markets of the destination countries can be due to information asymmetry between foreign and domestic investors and provide micro-level evidence for this hypothesis in follow-up work (Brennan et al., 2005). Ahearne et al. (2004) find that countries with higher stock market listing in the US play a larger role in the US portfolios whereas Pagano et al. (2002) find that foreign-listed European companies perform better in the US, but without significant leveraging effects. The role of information is also emphasized by the empirical evidence of Hau (2001a,b), showing that foreign traders perform worse on German stock markets. Finally, Byard et al. (2011) provide some evidence that the adoption and strong enforcement of the European Union’s International Financial Reporting Standards (IFRS) reduced financial analysts’ absolute forecast errors when domestic accounting standards differ significantly from the IFRS.
On the macro level, Portes et al. (2001) find that distance matters in gravity models using two different data sets of gross bilateral equity transactions. Contrary to what one would expect from portfolio diversification, the impact is negative.5 They attribute this finding to the hypothesis that “distance is seen as a proxy for informational frictions” (p. 784). While distance has often been used for this purpose thereafter, it is questionable to what extent this proxy is appropriate. Savastano (2000, p. 157), for example, already noted “that distance (and hence gravity-type equations) is probably not among the factors that will help us understand the geography of capital flows”.
Daude and Fratzscher (2008) use a pseudo-fixed effects model of the Anderson and van Win-coop (2003) class for bilateral capital stocks to address the “pecking order” of different types of capital with emphasis on information and the quality of host country institutions. As information friction measurements, they use distance, the volume of bilateral telephone calls, bilateral newspapers’ and periodicals’ trade, and the stock of immigrants from the source country in the host country.6 They find that all investigated forms of capital respond significantly to information, but that the elasticity is higher for FDI than for other forms of capital which is evidence against the models of Razin et al. (1998) and Goldstein and Razin (2006) that suggests that portfolio should be more elastic to informational frictions.7 However, the information proxies of Daude and Fratzscher (2008) cover a whole range of potential transaction costs that may include but are not limited to information. This may cause an omitted variable bias. For example, newspaper circulation and telephone traffic will be correlated with immigrant stocks and if immigrants have a ‘home bias’ in consumption, using these measures is likely to provide biased and inconsistent estimates in the presence of horizontal FDI. The same applies to the study of Milesi-Ferretti and Lane (2004) who use a similar model and try to capture information by a number of cultural and physical proximity variables.
In their study on the effect of information frictions on portfolio holdings of emerging market funds (relative to the host country’s share in the world market portfolio, proxied by Morgan Stanley’s Emerging Markets Free Index), Gelos and Wei (2005) construct a measure for macroeconomic data opacity that is based on Agça and Allum (2001) and comes closest to our SDDS variable. Their ovall results indicate that portfolio funds prefer to hold more assets in more transparent emerging markets. Furthermore, the authors (p. 3003ff) conduct a quasi-event study, where a dummy variable takes on the value 1 once a country either voluntarily publishes its IMF Article IV reports, publishes the IMF’s “Reports on Standards and Codes”, or adopts SDDS, and find a statistically significant, albeit moderate increase of the respective country’s portfolio weighting.
Considering FDI, Wheeler and Mody (1992) find support for agglomeration economies as a driving factor for US manufacturing multinational corporations (MNCs). Head et al. (1995, p. 226) attribute the agglomeration behavior of 751 Japanese multinationals in the USA to lowering the cost of acquiring information about the local market. Blonigen et al. (2005) find empirical results that information exchange in “Presidential Council” meetings of Japanese MNCs may lower information costs and thus implies positive impacts on FDI and find empirical support for agglomeration. In a similar vein, Kinoshita and Mody (2001) find Japanese investment in Asian emerging markets to be positively correlated with Japan’s own previous investment and the current investment by competitors and argue that this cannot be explained by industrial agglomeration, but by the value of private information. Bobonis and Shatz (2007) find FDI agglomeration within the US and conclude that it would be desirable in future research to disentangle different economic motives for this behavior such as technological spillovers, information sharing or other externalities. The FDI-agglomeration literature hence shows that the role of information is also important for the assessment of foreign market potential in the multinationals’ FDI decision,8 although Davies and Kristjánsdóttir (2010), for example, do not find similar agglomerative type effects for FDI flows to power-intensive industries in Iceland for the period 1989 - 2001.9,10 Other potential evidence for the importance of information for FDI can be derived from the results of Davies et al. (2009). Their finding that tax treaties only increase the extensive margin of FDI may in part be driven by the fact that information asymmetries decrease with tax treaties so that entry into the potential host market becomes easier. In fact, they conclude that their “results suggest that the impact of treaties might be greatest due to their impact on issues of uncertainty, not by adjusting the effective tax rates firms face” (p. 108). Finally, in a recent contribution Harding and Ja-vorcik (2011) find that investment promotion agencies (IPAs) have a positive impact on FDI flows11 from the US to developing (but not to industrialized) countries. While they do not generally take a stand on whether IPAs play an informative or persuasive role (cf. footnote 29 on p. 1469), they also provide some evidence that IPAs may alleviate information asymmetries. However, they use the common measures for information such as language, cultural and power distance and newspaper circulation which are likely to also capture other impacts, as discussed above.12
In summary, previous macro-studies on capital flow determinants have highlighted the potential role of information. But we find our contribution going way beyond these preliminary efforts in that our conclusions are drawn from the direct estimation of capital flows on information measurement using the SDDS, one of the IMF’s data standard pillars. Our contribution is also related to other literature investigating impacts of the SDDS subscription (Cady, 2005; Cady and Gonzalez-Garcia, 2007). That is, our finding that countries’ subscription to the SDDS results in an increase in FDI inflows would support their underlying hypotheses that the SDDS subscription would decrease the borrowing cost for emerging markets.
III. Investment: Model and Data
In our empirical strategy, we focus on the investors’ motives (i.e. the supply side) toward host country effects. This is not to say that home country effects do not matter,13 but our aim is to focus on a given country’s policy options to attract investment.14 This means that source country fundamentals have to be taken as externally given15 and we can focus on overall inflows instead of bilateral flows. In line with the study of Harding and Javorcik (2011), we use flow data instead of stock data. This has the advantage of being plagued less intensively with autocorrelation and potential spurious regression problems (cf. Wacker, 2012). Furthermore, the use of flow data allows us to rule out reversed causality to a large degree and we expect to see more responsive changes in the flow data than in the stock data.16
A. Portfolio and foreign direct investment
In our study, we use both FDI and portfolio investment flows. In the balance of payments, which is the source for international capital flow data, the distinction between FDI and portfolio investment is a rather pragmatic one: a host-country enterprise in which a foreign investor owns 10 percent or more of the voting power is classified (or: should be classified) as a direct investment enterprise. FDI thus usually implies a long-term relationship between investor and the direct investment enterprise in the host country.17 The residual category of cross-border transactions involving debt or equity securities is then classified as portfolio investment which covers (but is not limited to) securities traded on financial markets.18
Economically, portfolio and FDI hence refer to different concepts of investment and they could therefore respond differently to changes in the explanatory variables. In our context, for example, it might be the case that it is part of the job of portfolio investors to acquire information about potential markets and this leads to considerable economies of scale because of large externalities. MNCs, on the other hand, can barely benefit from such externalities of information and they would hence respond stronger to public provisioning of macroeconomic data.
Our data on capital flows come from the International Financial Statistics (IFS) balance of payments data. Since IFS data for most countries do not start before 1993, we use data from the IMF’s World Economic Outlook (WEO) where the IFS data are not available but WEO data are.19 In order to correct for potential errors from this procedure, we use a dummy variable that equals 1 if WEO data are used and equals 0 if IFS data are used. We use data in constant USD and take the natural logarithm thereof.
B. Econometric model
We estimate a log-linear20 static model with real capital flows on the left hand side and account for potential autocorrelation in inference using a heteroscedasticity and autocorrelation consistent (HAC) variance estimator based on Huber (1967) and White (1980), commonly referred to as ‘cluster-robust’ standard errors. We thus model the investment of type j in country i at year t as given by
where
C. Determinants of international investment flows
A basic textbook (domestic) investment equation (e.g. Blanchard, 2010) describes investment through demand Y and interest rate i:
The rationale of equation (2) is to capture the (expected) returns of investment and its (expected) cost as the determinants positively or negatively influencing the investment decision. In what follows, we discuss those aspects of international investment costs and returns that have shown to be very robust in the literature and we also present the data we use to control for them.
Like for domestic investment, current and future market potential are a main driving force for international investment. It is well documented in the literature that capital flows thus positively react to the size of the market, usually measured by GDP, market capitalization and/or its growth rate (cf., inter alia, Blonigen et al., 2003; Portes and Rey, 2005), to the investment rate and savings rate in the economy (cf. Hernández et al., 2001), and to the overall competitiveness of the economy (Stehrer and Woerz, 2009).
In our regressions, we use GDP data from WEO and take the natural logarithm of its real value in USD to account for current market size. We proxy future market potential by short and long run factors influencing GDP growth. Short-run growth is measured as the percentage change of real GDP per capita in national currency, taken from WEO. The investment rate, measured as gross capital formation at current national prices to GDP at current national prices (both taken from WEO) proxies for long-run growth.
It is standard in FDI models to account for education as a measure for the overall competitiveness of the economy. Supplementary to using the classical dataset of Barro and Lee (2010) on educational achievements, we also look at high-tech exports (from the World Bank Development Indicators), total number of patents in a country (from OECD) and export unit values (from IMF WEO), because we think they are more appropriate than an overall education measure in capturing the competitiveness of sectors of interest to foreign investors. We compare the performance of these proposed measures to the average years of schooling from the Barro and Lee (2010) data set.
We also account for the trade share in our model, which is measured as the sum of imports (including c.i.f.) and export from and to the rest of the world in current USD (from IFS) relative to GDP in current USD (from WEO). One reason is that as economies become more open, they might have larger markets. Furthermore, there might be important interdependen-cies between FDI and trade. In fact, one main argument in the FDI literature is that foreign affiliates of multinational firms can overcome trade costs and trade restrictions such as tariffs and non-tariff barriers to trade (cf. Blonigen, 2002, although the empirical evidence is somewhat mixed). For our purpose as a control variable, our measure of trade share should largely account for trade openness.22
The role of international trade in the context of investment decisions also gives rise to look at impacts of the exchange rate, as done by Froot and Stein (1991) who developed a model of informational imperfections where a depreciation leads to FDI inflows and provide evidence for this phenomenon. Blonigen (1997) shows that a real depreciation of the host country’s real exchange rate may increase profits of multinational firms that would (also) sell affiliate’s products in the home market (or process them there). The depreciation also allows foreign firms to make higher bids for host country’s assets than domestic firms because the multinational can realize profits of the acquisition in its home currency. Since most FDI comes in the form of M&A, real exchange rate depreciations will thus have a positive effect on FDI. The relationship has also been addressed by other studies such as Chakrabarti (2001) and Pain and van Welsum (2003). We also control for these effects by taking the implied PPP exchange rate, measured in national currency per USD from WEO.23
Although the above equation (2) highlights the role of the interest rate, it is remarkable that the interest rate has not made it into the standard set of control variables for FDI models.24 As Lehmann and Kang (2004) argue, local leveraging in the host economy is of high relevance for MNCs’ affiliates and hence a low interest rate will be preferable because it provides them easier access to credit or capital. The situation is completely different when the only focus is the capital flow component: in the Mundell-Fleming model, capital responds positively to the spread of the domestic to the foreign interest rate. This highlights an important potential difference between portfolio and foreign direct investment (cf. Wacker, 2012).25 We hence include the spread of the money market rate (MMR)26 over LIBOR in percent p.a. (both taken from IFS) to proxy for the interest rate.
Finally, it is important for our exercise to account for financial openness of the host country. Governments that attribute more positive growth effects to open financial markets may be more likely to open their capital account on one hand. But they may also be more likely to join SDDS on the other hand. Since both, capital account openness and SDDS will potentially increase capital flows, omission of controling for capital openness could cause an omitted variable bias. We hence control for financial openness using the index of Chinn and Ito (2006, 2008, 2011), which measures a country’s degree of capital account openness and is available up to 2009. It is based on binary dummy variables that codify the tabulation of restrictions on cross-border financial transactions reported in the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions. It is hence a de jure measure. A higher index value27,28
D. Risk
In our analysis, we look at two broad categories of risk: political and macroeconomic risks in international capital markets. Political risk (or instability) is an obvious cost to investors already outlined by Lucas Jr. (1990) and has both direct and indirect effects. Political risk may increase the direct cost of doing business. According to the theoretical model and empirical results of Kesternich and Schnitzer (2010), for example, ownership shares in multinational firms decrease as political risk increases. Thus, while the number of MNCs may not necessarily decline, the amount of FDI will.29 Indirect negative effects of political instability may arise in situations where governments set spending priorities to short-term projects that politically pay off immediately, instead of undertaking necessary long-run infrastructure and education spendings.
Empirical studies addressing the role of political instability include Wei (2000a,b); Papaioannou (2009) and Daude and Fratzscher (2008).30 All of these studies find negative impacts of political risk on FDI inflows. Daude and Fratzscher (2008) furthermore show that portfolio investment is much more sensitive to institutional indicators and market openness than FDI and that investor protection has a large effect on portfolio investment but not on FDI. This is in line with the predictions of the model of Albuquerque (2003) that FDI is harder to expropriate because of inalienability of its proprietary asset.31,32
Our data set includes the political risk ratings provided by the International Country Risk Guide (ICRG). It takes into account factors of government stability, socioeconomic conditions, investment profile, internal conflict, external conflict, corruption, military in politics, religious tensions, law and order, ethnic tensions, democratic accountability, and bureaucracy quality. Data are provided on monthly basis and averages over one year were taken. Risk ratings range from a high of 100 (least risk) to a low of 0 (highest risk), though the lowest de facto rating in the sample is 56.
As our measure for macroeconomic risk, we look at exchange rate volatility. Exchange rate volatility can have a negative indirect effect on productivity, at least when financial markets are poorly developed, as recently pointed out by Aghion et al. (2009). Furthermore, exchange rate volatility usually does not come on its own and might thus be a good indicator that something else is going on in the economy. Finally, risk-averse MNCs will directly be affected by changes in the exchange rate when affiliates are not operating independently of each other, but are part of complex vertical production networks and export platforms (cf., for example, Cushman, 1985; Schmidt and Broll, 2009; Campa, 1993; Kiyota and Urata, 2004 on the issue).
Our calculation of exchange rate volatility is based on monthly data from the IFS and uses data on the national currency per Special Drawing Right (SDR) instead of per USD, in order to avoid variation that arises from volatility in one single reference currency. We take the squared deviations from the expected exchange rate for each month, divide it by last month’s exchange rate and sum these deviations over the first 6 months of year t and of the last 6 months of year t – 1. As Engel and West (2005) show, even if exchange rates respond to economic fundamentals, their fluctuations should be nearly unpredictable, especially in the short run, so that today’s exchange rate is a reasonable predictor for tomorrow’s exchange rate. Hence, our measure for volatility of the exchange rate e is
where t(1/12) denotes the first month of year t.33 The intuition of our measure is that investors will make their investment decision based on previous volatility in exchange rates that serve as an estimate of future exchange rate volatility.
E. Information
As explained in section I, the International Monetary Fund has launched the Special Data Dissemination Standard (SDDS), as one of two data transparency standards, in March, 1996. Compliance with this data standard is voluntary for member states that are interested in getting or expanding access to international capital markets by signaling data of a certain quality in 18 macroeconomic and financial categories outlined in appendix A. On February 19, 1999 Canada and the United States were the first SDDS subscribers that met the requested data standard specifications. To date (2012), SDDS has 71 member countries listed in appendix A with their exact timing of subscription, metadata posting and SDDS compliance.
Although improvements in data provisioning may have taken place prior to official compliance with SDDS, we assign a dummy variable equal 1 to an observation if the country i has met the SDDS specification34 in year t and a 0 otherwise:
Accordingly, 1999 is the first year where 1-values are observed for at least some of the countries in the sample.
F. Spatial interdependencies
Our discussion so far has referred to distance as an exclusive measure for informational frictions for capital flows. However, other studies have introduced spatial structures in international capital flow models with the objective to account for interdependent shocks (Coughlin and Segev, 2000; Hernández et al., 2001; Blonigen et al., 2007; Baltagi et al., 2007). This raises two issues. First, it further adds to the argument that distance can measure many other aspects than informational frictions, hence making the interpretation of a related parameter estimate economically doubtful. Second, it points out that one should take into account potential spatial interdependencies in the econometric framework since untreated spatial correlation may bias the estimated covariance matrix (cf. Conley, 1999) similar to the time-series case of autocorrelation.
To deal with this issue, we apply an approach based on Conley (1999) and Conley and Ligon (2002) that has never been applied in studies on cross-boarder capital flows to the best of our knowledge.35 Correcting for such spatial heteroscedasticity, however, is only necessary if spatial correlation is present in the first place. Our approach therefore is to non-parametrically estimate the correlation of the Pearson-transformed error terms with respect to their distance in space. Since it is virtually impossible to analytically derive a null hypothesis for such a test against a meaningful alternative, we bootstrap a (pointwise) 90 percent confidence region for the null hypothesis of no spatial correlation. Contrary to the approach of using a moving-average estimator for the spatial correlation (Conley, 1999; Conley and Ligon, 2002 etc.), we add to the existing literature by using a smoothing spline. More details on the exact procedure can be found in section 5.6 of Hashimoto and Wacker (2012). Our results indicate that there is no need to correct inference for spatial interdependencies.
IV. Main Empirical Results
The estimation results of equation (1) for FDI and portfolio investment are depicted in tables 1 and 2, respectively. The first columns show baseline results not including the SDDS variable which enters the model in the second column. The difference between the third column and the first two columns is that the latter use the standard Barro and Lee (2010) years of schooling while our proposed measures for skill intensity in the sector(s) of comparative advantage can be found in the third column which is our preferred specification. The last column provides the results using random effects instead of fixed effects regression specification. Note that the Hausman test does by no means suggest that random effects would provide consistent estimates, however, we think it is interesting to see what happens to the model if cross-section variation is taken into account.36
FDI determinants Dependent Variable: ln
(FDI flow)
FDI determinants Dependent Variable: ln
(FDI flow)
| model | (1a) | (2a) | (3a) | (4a) |
|---|---|---|---|---|
| SDDS | 0.4462*** (0.1522) | 0.4760*** (0.1355) | 0.4590*** (0.1310) | |
| ln(GDP) (-1) | 0.7724*** (0.0995) | 0.7732*** (0.0952) | 0.7116*** (0.0795) | 0.8287*** (0.0550) |
| GDP growth | 4.1104*** (0.9197) | 4.4241*** (0.8887) | 2.9280*** (0.9967) | 3.8167*** (1.2328) |
| investment rate (-1) | 3.1872*** (0.9126) | 3.4468*** (0.8909) | 2.1197* (1.1974) | 2.0197** (1.0252) |
| capital account open | 0.1445** (0.0646) | 0.1299** (0.0625) | 0.1399** (0.0677) | 0.1377** (0.0554) |
| political risk | 0.1143** (0.0458) | 0.1140** (0.0473) | 0.0207 (0.0380) | 0.0021 (0.0240) |
| exchange rate volatility | −2.9418*** (0.7488) | −2.8234*** (0.7428) | −3.3692*** (0.8635) | −3.4416*** (1.0731) |
| exchange rate volatility (-1) | −0.7471 (0.7864) | −0.7105 (0.7611) | −1.4820** (0.5685) | −1.1573 (0.7046) |
| interest rate (-1) | 0.0000 (0.0000) | 0.0000 (0.0000) | −0.0002*** (0.0001) | −0.0003*** (0.0001) |
| real exchange rate (-1) | −0.0001 (0.0001) | −0.0001 (0.0001) | −0.0003* (0.0001) | −0.0003*** (0.0001) |
| trade share | 0.1873 (0.2047) | 0.0990 (0.1899) | −0.1835 (0.2890) | 0.4382*** (0.1227) |
| yrs. of schooling | −0.1155 (0.1383) | −0.0926 (0.1322) | ||
| high-tech exports (-1) | −0.0000 (0.0000) | −0.0000 (0.0000) | ||
| export unit value (-1) | −0.0098** (0.0042) | −0.0043 (0.0036) | ||
| # of patents (-1) | 0.0000 (0.0000) | −0.0000 (0.0000) | ||
| WEO data dummy | 0.8398* (0.4916) | 0.8663* (0.5097) | −0.2562 (0.4287) | −0.6221 (0.4193) |
| constant | −6.1518 (3.8131) | −6.3316 (3.8362) | 2.4603 (3.2122) | 0.0000 (.) |
| estimation | FE | FE | FE | RE |
| time dummies | yes | yes | yes | yes |
| observations | 1,084 | 1,084 | 634 | 634 |
| (N x avg. T) | (70 x 15.5) | (70 x 15.5) | (55 x 11.5) | (55 x 11.5) |
| within R-squared | 0.38 | 0.40 | 0.44 | 0.43 |
| Hausman (p-value) | 0.000 | |||
FDI determinants Dependent Variable: ln
(FDI flow)
| model | (1a) | (2a) | (3a) | (4a) |
|---|---|---|---|---|
| SDDS | 0.4462*** (0.1522) | 0.4760*** (0.1355) | 0.4590*** (0.1310) | |
| ln(GDP) (-1) | 0.7724*** (0.0995) | 0.7732*** (0.0952) | 0.7116*** (0.0795) | 0.8287*** (0.0550) |
| GDP growth | 4.1104*** (0.9197) | 4.4241*** (0.8887) | 2.9280*** (0.9967) | 3.8167*** (1.2328) |
| investment rate (-1) | 3.1872*** (0.9126) | 3.4468*** (0.8909) | 2.1197* (1.1974) | 2.0197** (1.0252) |
| capital account open | 0.1445** (0.0646) | 0.1299** (0.0625) | 0.1399** (0.0677) | 0.1377** (0.0554) |
| political risk | 0.1143** (0.0458) | 0.1140** (0.0473) | 0.0207 (0.0380) | 0.0021 (0.0240) |
| exchange rate volatility | −2.9418*** (0.7488) | −2.8234*** (0.7428) | −3.3692*** (0.8635) | −3.4416*** (1.0731) |
| exchange rate volatility (-1) | −0.7471 (0.7864) | −0.7105 (0.7611) | −1.4820** (0.5685) | −1.1573 (0.7046) |
| interest rate (-1) | 0.0000 (0.0000) | 0.0000 (0.0000) | −0.0002*** (0.0001) | −0.0003*** (0.0001) |
| real exchange rate (-1) | −0.0001 (0.0001) | −0.0001 (0.0001) | −0.0003* (0.0001) | −0.0003*** (0.0001) |
| trade share | 0.1873 (0.2047) | 0.0990 (0.1899) | −0.1835 (0.2890) | 0.4382*** (0.1227) |
| yrs. of schooling | −0.1155 (0.1383) | −0.0926 (0.1322) | ||
| high-tech exports (-1) | −0.0000 (0.0000) | −0.0000 (0.0000) | ||
| export unit value (-1) | −0.0098** (0.0042) | −0.0043 (0.0036) | ||
| # of patents (-1) | 0.0000 (0.0000) | −0.0000 (0.0000) | ||
| WEO data dummy | 0.8398* (0.4916) | 0.8663* (0.5097) | −0.2562 (0.4287) | −0.6221 (0.4193) |
| constant | −6.1518 (3.8131) | −6.3316 (3.8362) | 2.4603 (3.2122) | 0.0000 (.) |
| estimation | FE | FE | FE | RE |
| time dummies | yes | yes | yes | yes |
| observations | 1,084 | 1,084 | 634 | 634 |
| (N x avg. T) | (70 x 15.5) | (70 x 15.5) | (55 x 11.5) | (55 x 11.5) |
| within R-squared | 0.38 | 0.40 | 0.44 | 0.43 |
| Hausman (p-value) | 0.000 | |||
Portfolio determinants Dependent Variable: ln
(portfolio flow)
Portfolio determinants Dependent Variable: ln
(portfolio flow)
| model | (1b) | (2b) | (3b) | (4b) |
|---|---|---|---|---|
| SDDS | −0.2079 (0.2185) | −0.0545 (0.2425) | 0.0713 (0.2022) | |
| ln(GDP)(-1) | 1.0519*** (0.1024) | 1.0516*** (0.1064) | 1.2999*** (0.2473) | 1.1935*** (0.0924) |
| GDP growth | 3.2752 (2.3892) | 3.2010 (2.4053) | 6.1644*** (2.1640) | 4.4890* (2.3619) |
| investment rate(-1) | 2.9249** (1.3292) | 2.7624** (1.3441) | 2.3474 (2.2172) | −0.0970 (1.4085) |
| capital account open | 0.0198(0.0799) | 0.0298 (0.0809) | −0.0552 (0.0853) | 0.1462*** (0.0559) |
| political risk | 0.1107*** (0.0413) | 0.1221*** (0.0398) | 0.1121** (0.0512) | −0.1451*** (0.0381) |
| exchange rate volatility | −0.6470 (1.6459) | −0.6689 (1.6538) | −1.3807 (1.4529) | −2.5401** (1.0597) |
| exchange rate volatility (-1) | −5.4158(5.5546) | −5.2558(5.6120) | −22.1873** (10.7241) | −22.5561** (8.9123) |
| interest rate (-1) | 0.0000 (0.0000) | 0.0000 (0.0000) | 0.0007** (0.0003) | 0.0007** (0.0003) |
| real exchange rate(-1) | −0.0000 (0.0001) | −0.0000 (0.0001) | 0.0002 (0.0002) | −0.0001 (0.0001) |
| trade share | 0.0586 (0.4282) | 0.0851(0.4244) | −0.7122(0.7054) | 0.0382(0.1856) |
| yrs. of schooling | 0.0893(0.1803) | 0.0869 (0.1794) | ||
| high-tech exports (-1) | 0.0000 (0.0000) | 0.0000 (0.0000) | ||
| export unit value (-1) | −0.0055 (0.0071) | 0.0010 (0.0045) | ||
| # of patents (-1) | −0.0000 (0.0000) | −0.0001** (0.0000) | ||
| WEO data dummy | 0.4811* (0.2723) | 0.4806* (0.2759) | 0.4611 (0.5525) | 1.0629 (0.7039) |
| constant | −14.7047*** (3.0642) | −15.3078*** (3.1258) | −19.1741** (7.3630) | 0.0000 |
| estimation | FE | FE | FE | RE |
| time dummies | yes | yes | yes | yes |
| observations | 805 | 805 | 520 | 520 |
| (N x avg. T) | (66 x 12.2) | 66 x 12.2) | (53 x 9.8) | (53 x 9.8) |
| within R-squared | 0.27 | 0.27 | 0.31 | 0.28 |
| Hausman (p-value) | 0.067 |
Portfolio determinants Dependent Variable: ln
(portfolio flow)
| model | (1b) | (2b) | (3b) | (4b) |
|---|---|---|---|---|
| SDDS | −0.2079 (0.2185) | −0.0545 (0.2425) | 0.0713 (0.2022) | |
| ln(GDP)(-1) | 1.0519*** (0.1024) | 1.0516*** (0.1064) | 1.2999*** (0.2473) | 1.1935*** (0.0924) |
| GDP growth | 3.2752 (2.3892) | 3.2010 (2.4053) | 6.1644*** (2.1640) | 4.4890* (2.3619) |
| investment rate(-1) | 2.9249** (1.3292) | 2.7624** (1.3441) | 2.3474 (2.2172) | −0.0970 (1.4085) |
| capital account open | 0.0198(0.0799) | 0.0298 (0.0809) | −0.0552 (0.0853) | 0.1462*** (0.0559) |
| political risk | 0.1107*** (0.0413) | 0.1221*** (0.0398) | 0.1121** (0.0512) | −0.1451*** (0.0381) |
| exchange rate volatility | −0.6470 (1.6459) | −0.6689 (1.6538) | −1.3807 (1.4529) | −2.5401** (1.0597) |
| exchange rate volatility (-1) | −5.4158(5.5546) | −5.2558(5.6120) | −22.1873** (10.7241) | −22.5561** (8.9123) |
| interest rate (-1) | 0.0000 (0.0000) | 0.0000 (0.0000) | 0.0007** (0.0003) | 0.0007** (0.0003) |
| real exchange rate(-1) | −0.0000 (0.0001) | −0.0000 (0.0001) | 0.0002 (0.0002) | −0.0001 (0.0001) |
| trade share | 0.0586 (0.4282) | 0.0851(0.4244) | −0.7122(0.7054) | 0.0382(0.1856) |
| yrs. of schooling | 0.0893(0.1803) | 0.0869 (0.1794) | ||
| high-tech exports (-1) | 0.0000 (0.0000) | 0.0000 (0.0000) | ||
| export unit value (-1) | −0.0055 (0.0071) | 0.0010 (0.0045) | ||
| # of patents (-1) | −0.0000 (0.0000) | −0.0001** (0.0000) | ||
| WEO data dummy | 0.4811* (0.2723) | 0.4806* (0.2759) | 0.4611 (0.5525) | 1.0629 (0.7039) |
| constant | −14.7047*** (3.0642) | −15.3078*** (3.1258) | −19.1741** (7.3630) | 0.0000 |
| estimation | FE | FE | FE | RE |
| time dummies | yes | yes | yes | yes |
| observations | 805 | 805 | 520 | 520 |
| (N x avg. T) | (66 x 12.2) | 66 x 12.2) | (53 x 9.8) | (53 x 9.8) |
| within R-squared | 0.27 | 0.27 | 0.31 | 0.28 |
| Hausman (p-value) | 0.067 |
For the FDI models (2a) - (4a) the SDDS dummy enters the model highly significant.37 The increase in the explanatory power when moving from model (1a) without SDDS to (2a) with SDDS is relatively small but it is important to stress that the impact of information is nevertheless economically highly relevant: Conditional on other factors, providing high-quality information about the macroeconomic and financial environment under the umbrella of SDDS increases FDI inflows by 56.2 to 61 percent.38 Furthermore, one should note that the change in the estimated parameters for the control variables is of minor importance when SDDS enters the model. This suggests that omitted variable biases in previous investigations that failed to account for informational asymmetries were negligible.
Considering the control variables, we first look at the performance of our proposed measures for human capital and technology in the export-relevant sector in model (3a) relative to the standard education variable of Barro and Lee (2010) in model (2a). As one can see, years of schooling do not turn out to be significant and the estimated sign is contrary to the expected effect, supporting our claim that overall education averages on the macro level may not be as relevant for a multinational firm’s investment decision. On the other hand, the number of patents is at the borderline of weak statistical significance (t-statistic 1.46) and shows both the expected sign and an economically relevant impact.39 The impact of the world market share of high-tech exports is far from being statistically significant, maybe reflecting offsetting positive effects from high knowledge in the sector with comparative advantage and negative impacts from too competitive markets.
The considerable negative (and statistically significant) effect of (lagged) export unit values may be surprising on a first view if one thinks of commodity prices that are reflected in the export unit values. However, our sample mainly consists of advanced economies for which the large country assumption is reliable at least in their sectors of comparative advantage. Lower export unit values could then simply reflect high total factor productivity and hence high international competitiveness in this sector which would attract FDI.40
As expected, FDI responds positive to current and potential future market size measured by GDP, the growth of GDP p.c. and the investment rate, where the latter is only weakly significant in model (3a). Also, the positive impact of de jure capital account openness is statistically significant and very robust, capturing the incentives to transfer capital.41 On the other hand, the effect of local leveraging as measured by the spread of the interest rate is only statistically significant in models (3a) and (4a). There, however, it is highly significant but of minor economic relevance: An increase in the spread of one percentage point increases FDI inflows by about 0.02 percent. Nevertheless, we find it important to control for the host country interest rate in FDI models, especially since data availability should not pose a problem in most applications. The insignificant effect of the interest spread in model (2a) may be driven by sample effects: The larger sample includes more less developed economies where multinationals have advantages over local competitors by having larger access to capital markets in their source countries. In the sample of more advanced (and more homogeneous) countries, access to local leverage might influence the multinationals’ location decision more clearly.
A similar effect may influence the results for political risk: It is unlikely that it has a relevant impact in the subsample of more advanced economies but when a larger and more heterogeneous sample is investigated, FDI shies away from political risk.42
Considering exchange rate volatility, FDI clearly resiles from macroeconomic risk: The appropriate test statistic is an F-test for joint insignificance of both lags of exchange rate volatility and we can reject this null hypothesis both in model (2a) and (3a) at the 1 percent level of statistical significance.43
Contrary to Blonigen (1997), we find a negative impact of a real exchange rate devaluation on FDI inflows. This might be driven by the fact that he disaggregates the effect down to the industry level while we look at the aggregate effect in the whole economy. Furthermore, his rationale is only one of many potential channels between the real exchange rate and FDI. For example, under Dixit-Stiglitz preferences, a real exchange rate appreciation will ceteris paribus increase the relative demand for imported varieties (because they become relatively cheaper). This increased demand can then either be served by imports or by horizontal FDI, so both of them will increase.44
The results for portfolio investment are generally not as appealing as the ones obtained for FDI flows. We do not find a significant impact of SDDS compliance on inflows, neither is the size of the estimated parameter very relevant in economic terms. This contrasts with previous macro studies that held asymmetric information responsible for “too low” international portfolio capital flows (but could not empirically justify this assumption). We will discuss this issue in the concluding section VI. However, portfolio flows too respond positive to current and potential future market size, although the growth rate of GDP p.c. or the investment rate are not statistically different from 0 in models (1+2b) or (3b), respectively.
We can reject the null hypothesis that both lags of exchange rate volatility have no impact on portfolio flows at the 10 percent level in model (3b) but not in model (2a). As expected, portfolio investment responds positive to spreads in the interest rate in model (3b). Somewhat surprising, we do not find a statistically significant impact of de jure financial liberalization on portfolio inflows. We find very robust evidence of portfolio investment shying away from political risk in the fixed effects models. The positive significant correlation with political risk in the random effects model (4b) may be due to an omitted variable bias.
Finally, we find it interesting to notice that portfolio investment reacts more elastic to changes in the (current) market size than FDI. This corresponds to the theoretical assumption that for FDI more firm-internal considerations play a role in the investment decision while portfolio investment tends to view the potential host markets more isolated. The evidence is more mixed when looking at potential future market size development but this is not too surprising: Since portfolio investment should be more flexible than FDI, its location decision is not as binding as for the latter.
V. Robustness and Further Results
In section IV we have estimated a highly significant effect of SDDS subscription on FDI inflows, obtaining a parameter estimate around 0.45. The aim of this section is to discuss the robustness of the parameter estimate and its statistical inference.
Therefore, we first used QQ and PP plots and a kernel density estimate to investigate how much the estimated residuals (of our preferred model (3a)) deviate from a normal distribution. The results are presented in figures 2 and 3 and show no specific pattern, thereby lending support to the overall model specification and the assumption that our main test statistic in fact follows a t-distribution.
Plots for normality of residuals from model (3a)
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
Plots for normality of residuals from model (3a)
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
Plots for normality of residuals from model (3a)
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
Kernel density estimate for residuals from model (3a) with corresponding normal distribution
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
Kernel density estimate for residuals from model (3a) with corresponding normal distribution
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
Kernel density estimate for residuals from model (3a) with corresponding normal distribution
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
To deal with the potential problem of omitted variables influencing both, capital inflows and SDDS subscription discussed in subsection III.B, it should be highlighted that such omitted variables have to be country-specific and to vary over time (see subsection 5.1 of Hashimoto and Wacker, 2012 for further details). Accordingly, a trend in the unexplained part of the model that differs between subscriber and non-subscriber countries would be evidence of an omitted variable bias. To investigate this possibility, we estimate the model
for FDI flows up until certain points in time45 and perform a Wald test to check equality of parameters, H0: δSDDS = γnon–SDDS. The estimates for the different parameters and the p-values of the F-statistic are displayed in table 3. The results do not provide any evidence whatsoever that there would be an underlying time-dependent process that was omitted from equation (1) and influenced the probability of joining SDDS, i.e. there is no evidence for an omitted variable bias.46
Different time trends between SDDS subscribers and non-subscribers?
Different time trends between SDDS subscribers and non-subscribers?
| before year… | 1996 | 1997 | 1998 | 1999 |
|---|---|---|---|---|
| SDDS trend (δ) | −0.0764 | −0.0898 | −0.1173 | 0.0534 |
| non-SDDS trend (γ) | −0.0468 | −0.0890 | −0.1295 | −0.0046 |
| difference significant (p-val)? | 0.563 | 0.989 | 0.822 | 0.377 |
| observations | 150 | 190 | 234 | 278 |
Different time trends between SDDS subscribers and non-subscribers?
| before year… | 1996 | 1997 | 1998 | 1999 |
|---|---|---|---|---|
| SDDS trend (δ) | −0.0764 | −0.0898 | −0.1173 | 0.0534 |
| non-SDDS trend (γ) | −0.0468 | −0.0890 | −0.1295 | −0.0046 |
| difference significant (p-val)? | 0.563 | 0.989 | 0.822 | 0.377 |
| observations | 150 | 190 | 234 | 278 |
To account for the problem of selection bias, we check what would happen if we excluded the early subscribers from our sample. Countries that experience developments in potentially omitted variables influencing both SDDS subscription and FDI inflows should be those that are more likely to join early. The parameter estimate obtained when excluding the bulk of countries that subscribed to SDDS in 1996 is presented in the first column of table 4 and equals 0.50. Still being statistically significant, this is very strong evidence for our findings that SDDS has a strong positive impact on FDI inflows since the number of observations decreases considerably.47
Parameter robustness
Parameter robustness
| SDDS (general) | 0.5016** (0.1907) | 0.4935** (0.2362) | 0.2056(0.1550) |
| SDDS (high inc) | 0.6905*** (0.2340) | ||
| Note | w/o 1996 subscribers | subscr. instead of compl. | parameter heterogeneity |
| observations | 160 | 634 | 587 |
Parameter robustness
| SDDS (general) | 0.5016** (0.1907) | 0.4935** (0.2362) | 0.2056(0.1550) |
| SDDS (high inc) | 0.6905*** (0.2340) | ||
| Note | w/o 1996 subscribers | subscr. instead of compl. | parameter heterogeneity |
| observations | 160 | 634 | 587 |
Since subscribers will generally improve their data quality already after (or even slightly before) subscription to SDDS and it may take a while before official specifications are met, we also look at the impact when the dummy variable starts equaling 1 after countries subscribe to SDDS. As we can see from column 2 of table 4, most of the action seems to take place after subscription already.
This gives rise to exploring the dynamics of the process in more detail. Therefore, we first introduce dummy variables that that are specific to SDDS subscribers and measure the impact on FDI flows at each year before and after subscription. More precisely the model
is estimated with the same control variables as in model (3a), and where ζt is a SDDS subscriber-specific time dummy. The interpretation of this variable is the effect of subscription, conditional on other factors, at time period t. The results are depicted in the left panel of figure 4, where the subscription year is taken as reference year 0. As can be seen, in the first four years after subscription, capital inflows considerably increase but the impact does not remain as robust thereafter with negative estimates for the years 5 and 7 and an overall picture that suggests somehow increased inflows.
Dynamic effects of SDDS subscription
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
Dynamic effects of SDDS subscription
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
Dynamic effects of SDDS subscription
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
Since year-specific effects are probably too volatile because they might be influenced by various other noise steaming from “global” effects in the specific years or different countries having somewhat differing dynamics, we also construct period-specific dummies, that is a dummy that is equal to 1 at the year of subscription and two years thereafter (period 1), 3 to 5 years (period 2), 6 to 8 years (period 3), and 8 to 10 years (period 4) after subscription, respectively. Furthermore, we control for effects in the three years prior to subscription (period 0). The overall picture in the right panel of figure 4 is less volatile but shows the same general pattern as before: Most of the increased inflows occur in the first years after subscription, the effect decays afterwards but suggests slightly higher FDI inflows also at later periods in time.
In fact, this is also the dynamic we would expect from a theoretical perspective: Increased information should result in a permanently higher capital stock and to reach this higher stock level, an adjustment process has to take place once, which operates through increased inflows. After the new equilibrium level is reached, inflows should only be slightly higher in subscriber countries since replacement of the (now increased) capital stock will be higher than otherwise (cf. Wacker, 2012).
To investigate parameter heterogeneity, we allow the parameter estimate to vary between different country income categories based on the World Bank classification 1987. We find that the impact of SDDS was stronger for high income countries (0.69) than for upper medium (0.20) and lower medium (0.22) countries and that the fit of this extended model is “better” in terms of standard model selection criteria (AIC/BIC) and a likelihood ratio (LR) test statistic (13.24 with 2 degrees of freedom). Since the parameter estimate for upper medium and lower medium countries is fairly similar, we perform the same exercise comparing this extended model to one that has one parameter estimate for high income countries and another one for all other countries. We find the latter to outperform the extended model along all three lines (LR statistic 0.01 with 1 degree of freedom), so we report the corresponding coefficients of the latter in the last column of table 4. We find that the impact of SDDS on FDI inflows is in fact driven by high-income countries. This, however, should not be too surprising: Economically, the better the overall performance of the economy, the better capital markets will respond to the provision of data. Statistically, most SDDS subscribers are high-income countries so that parameter identification is easier for these countries. This result does not imply that countries with a lower income level could not benefit from SDDS or from the signaling of macroeconomic data. In fact, the estimated parameter for other countries is still positive and economically relevant (+23 percent) and the estimated standard error is of reasonable size (t-statistic 1.33). The result, however, highlights that data-provision on its own will not be sufficient to acquire FDI inflows but should be based on sound macroeconomic fundamentals.48
We also performed a recursive regression to investigate the stability of our estimated parameter. Detailed results are reported in section 5.3 of Hashimoto and Wacker (2012) and indicate that for samples truncated in the years 2000 - 2005 the impact is somewhat smaller than our parameter estimate of approximately 0.45, however, it never falls short of 0.16.
Finally, we estimate the spatial correlation pattern as outlined at the end of section III.F. Figure 5 depicts the results. The estimated spline function is almost identically with a horizontal line of zero-correlation. The estimated spatial correlation does by no means reach a level that is beyond randomness so that one could not reject the null hypothesis of no spatial correlation in the error terms. Accordingly, the inference in our models is not plagued by spatial correlation in the residuals.
Smoothing spline estimate of spatial correlation in FDI flow residuals with 90 percent pointwise confidence band s based on 1,000 bootstrap replications
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
Smoothing spline estimate of spatial correlation in FDI flow residuals with 90 percent pointwise confidence band s based on 1,000 bootstrap replications
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
Smoothing spline estimate of spatial correlation in FDI flow residuals with 90 percent pointwise confidence band s based on 1,000 bootstrap replications
Citation: IMF Working Papers 2012, 242; 10.5089/9781475568660.001.A001
VI. Discussion and Conclusion
A. Main findings
Our analysis has shown that countries which committed themselves to provide macroeco-nomic data with a certain accuracy and timeliness, as requested by the IMF’s Special Data Dissemination Standard, received more foreign direct investment inflows than other countries. The impact that accounts for a set of standard control variables, time- and country-specific effects, is both statistically significant and economically relevant: Compliance with the SDDS increases FDI inflows by about 60 percent.49
The most important impact occurs in the first years after subscription to (or compliance with) the SDDS, especially for those countries that are able to submit data which is solid not only with respect to technical accuracy, but also considering the underlying economic fundamentals. Most industrialized countries hence experience an even larger impact of 100 percent as reported in table 4, while the comparable impact for other countries that do not possess such strong fundamentals may be somewhat above 20 percent. However, our results also indicate that in the longer run these countries can also catch up to the overall parameter of 60 percent, probably driven by the fact that once financial markets are monitoring macroeconomic and financial data, there is a stronger incentive for countries to get these fundamentals straight.
Portfolio and FDI flows both shy away from political and macroeconomic risk, though political risk does not seem to matter much for FDI in the most advanced countries and the aversion against macroeconomic risk in the form of exchange rate volatility is less robust for portfolio investment.
B. Relation to other findings in the literature
Our main finding is generally in line with the result of Daude and Fratzscher (2008) that FDI is more responsive to information than other forms of capital flows. We should also highlight that information about the macro environment is probably more important for FDI, whereas portfolio investment is likely to be more connected with firm-level information in the host economy.50 Portfolio investors are supposed to acquire information and this is essentially what portfolio funds are paid for. As the model of van Nieuwerburgh and Veldkamp (2009) points out, investors profit more from information others do not know. Especially, large institutional investors might have an incentive to acquire information about potential host countries on their own. They may therefore lose short-run arbitrage gains from informational asymmetries once information becomes public. On the other hand, smaller investors with more long-run perspective, that cannot acquire information on their own, may intend to rely on public information.51 In our analysis, these two opposing public information effects on portfolio decisions - large institutional investors do not need public information and small investors rely on public information - might simply offset each other resulting in an overall impact that is small and statistically not significantly different from zero. Hence, our findings do not necessarily contradict the results of Gelos and Wei (2005), which underscore that emerging market portfolio funds respond positively to information, because their sample only captures one third of the total portfolio flows to the relevant countries (p. 2989) and their specific measure for macrodata opacity, which is comparable to our SDDS measure, loses statistical significance when including fixed region effects (p. 3002).
C. Further results
Our study empirically supports the distinction of portfolio flows from FDI flows in the balance of payments. We find empirical evidence that the elasticity of portfolio investment toward current market size is significantly higher than for foreign direct investment. We also find evidence of other differences between portfolio and FDI flows, most notably their contrary response toward the interest rate, highlighting that FDI is not simply a capital flow but also that MNCs’ finance strategy for investments should play a more important role in the (micro-)economic attempt to understand the behavior of the multinational firm. Our study also suggests that the empirical modeling of (firm demand for) education should probably focus more specific on know-how in sectors with comparative advantages and shows that spatial correlations do not play a significant role when FDI flows are estimated at an aggregate level using fixed effects.52
D. Perspectives on further research and policy issues
Our results highlight the need for further, more disaggregated research about the role of information for portfolio (and potentially other forms of) investment. As mentioned above, the finding that there is no significant overall effect of information does not imply that the structure of portfolio investment stays unaffected. Informational asymmetries may attract market makers and hence more short-term oriented portfolio flows, which would generally result in more volatility in capital markets (cf. e.g. Diamond and Verrecchia, 1991; Du and Wei, 2004) and can cause adverse externalities (cf. Bianchi, 2011). Public dissemination of more accurate and timely economic and financial data is hence a potentially important macroeconomic tool to manage capital inflows since it may attract more long-term oriented portfolio flows. As our evidence strongly suggests that more FDI will be attracted by stronger macro fundamentals and by the macroeconomic information, public dissemination of accurate economic data will initiate more stable and one of the most advantageous forms of capital inflows to countries. This should result in a more sustainable and healthy international investment position, may help to lessen the degree of international balance of payments imbalances, and would develop the productive resources of capital-scarce countries. Accordingly, we also think that informational frictions may account to a certain extent for the fact that growth effects of financial openness have turned out to be somewhat disappointing and are generally far from being robust (cf. Jeanne et al., 2012, ch. 3 and 4 for a recent survey and investigation), although this channel still has to be investigated in more detail.
Appendix A. Information on SDDS
List of SDDS subscribers
List of SDDS subscribers
| Country | subsc | meta | spec | Country | subsc | meta | spec |
|---|---|---|---|---|---|---|---|
| Argentina | 1996 | 1996 | 1999 | Korea | 1996 | 1998 | 1999 |
| Armenia | 2003 | 2003 | 2003 | Kyrgyz Republic | 2004 | 2004 | 2004 |
| Australia | 1996 | 1998 | 2001 | Latvia | 1996 | 1997 | 1999 |
| Austria | 1996 | 1997 | 2001 | Lithuania | 1996 | 1997 | 1999 |
| Belarus | 2004 | 2004 | 2004 | Luxembourg | 2006 | 2006 | 2006 |
| Belgium | 1996 | 1997 | 2001 | Macedonia | 2011 | 2011 | 2011 |
| Brazil | 2001 | 2001 | 2001 | Malaysia | 1996 | 1996 | 2000 |
| Bulgaria | 2003 | 2003 | 2003 | Malta | 2009 | 2009 | 2009 |
| Canada | 1996 | 1996 | 1999 | Mauritius | 2012 | 2012 | 2012 |
| Chile | 1996 | 1997 | 2000 | Mexico | 1996 | 1996 | 2000 |
| Colombia | 1996 | 1997 | 2000 | Moldova | 2006 | 2006 | 2006 |
| Costa Rica | 2001 | 2001 | 2001 | Morocco | 2005 | 2005 | 2005 |
| Croatia | 1996 | 1996 | 2001 | Netherlands | 1996 | 1996 | 2000 |
| Cyprus | 2009 | 2009 | 2009 | Norway | 1996 | 1996 | 2000 |
| Czech Republic | 1998 | 1998 | 1999 | Peru | 1996 | 1996 | 1999 |
| Denmark | 1996 | 1996 | 2000 | Philippines | 1996 | 1996 | 2001 |
| Ecuador | 1998 | 1998 | 2000 | Poland | 1996 | 1996 | 2000 |
| Egypt | 2005 | 2005 | 2005 | Portugal | 1997 | 1998 | 2000 |
| El Salvador | 1998 | 1998 | 1999 | Romania | 2005 | 2005 | 2005 |
| Estonia | 1998 | 1999 | 2000 | Russia | 2005 | 2005 | 2005 |
| Finland | 1996 | 1996 | 2000 | Singapore | 1996 | 1996 | 2001 |
| France | 1996 | 1996 | 2000 | Slovak Republic | 1996 | 1998 | 1999 |
| Georgia | 2010 | 2010 | 2010 | Slovenia | 1996 | 1996 | 2000 |
| Germany | 1996 | 1997 | 2000 | South Africa | 1996 | 1996 | 2000 |
| Greece | 2002 | 2002 | 2002 | Spain | 1996 | 1998 | 2000 |
| Hong Kong SAR | 1996 | 1997 | 2000 | Sweden | 1996 | 1996 | 2000 |
| Hungary | 1996 | 1997 | 2000 | Switzerland | 1996 | 1996 | 2001 |
| Iceland | 1996 | 1998 | 2004 | Thailand | 1996 | 1996 | 2000 |
| India | 1996 | 1997 | 2001 | Tunisia | 2001 | 2001 | 2001 |
| Indonesia | 1996 | 1997 | 2000 | Turkey | 1996 | 1996 | 2001 |
| Ireland | 1996 | 1996 | 2001 | Ukraine | 2003 | 2003 | 2003 |
| Israel | 1996 | 1996 | 2000 | United Kingdom | 1996 | 1996 | 1999 |
| Italy | 1996 | 1996 | 2000 | United States | 1996 | 1996 | 1999 |
| Japan | 1996 | 1996 | 2000 | Uruguay | 2004 | 2004 | 2004 |
| Jordan | 2010 | 2010 | 2010 | West Bank and Gaza | 2012 | 2012 | 2012 |
| Kazakhstan | 2003 | 2003 | 2003 |
List of SDDS subscribers
| Country | subsc | meta | spec | Country | subsc | meta | spec |
|---|---|---|---|---|---|---|---|
| Argentina | 1996 | 1996 | 1999 | Korea | 1996 | 1998 | 1999 |
| Armenia | 2003 | 2003 | 2003 | Kyrgyz Republic | 2004 | 2004 | 2004 |
| Australia | 1996 | 1998 | 2001 | Latvia | 1996 | 1997 | 1999 |
| Austria | 1996 | 1997 | 2001 | Lithuania | 1996 | 1997 | 1999 |
| Belarus | 2004 | 2004 | 2004 | Luxembourg | 2006 | 2006 | 2006 |
| Belgium | 1996 | 1997 | 2001 | Macedonia | 2011 | 2011 | 2011 |
| Brazil | 2001 | 2001 | 2001 | Malaysia | 1996 | 1996 | 2000 |
| Bulgaria | 2003 | 2003 | 2003 | Malta | 2009 | 2009 | 2009 |
| Canada | 1996 | 1996 | 1999 | Mauritius | 2012 | 2012 | 2012 |
| Chile | 1996 | 1997 | 2000 | Mexico | 1996 | 1996 | 2000 |
| Colombia | 1996 | 1997 | 2000 | Moldova | 2006 | 2006 | 2006 |
| Costa Rica | 2001 | 2001 | 2001 | Morocco | 2005 | 2005 | 2005 |
| Croatia | 1996 | 1996 | 2001 | Netherlands | 1996 | 1996 | 2000 |
| Cyprus | 2009 | 2009 | 2009 | Norway | 1996 | 1996 | 2000 |
| Czech Republic | 1998 | 1998 | 1999 | Peru | 1996 | 1996 | 1999 |
| Denmark | 1996 | 1996 | 2000 | Philippines | 1996 | 1996 | 2001 |
| Ecuador | 1998 | 1998 | 2000 | Poland | 1996 | 1996 | 2000 |
| Egypt | 2005 | 2005 | 2005 | Portugal | 1997 | 1998 | 2000 |
| El Salvador | 1998 | 1998 | 1999 | Romania | 2005 | 2005 | 2005 |
| Estonia | 1998 | 1999 | 2000 | Russia | 2005 | 2005 | 2005 |
| Finland | 1996 | 1996 | 2000 | Singapore | 1996 | 1996 | 2001 |
| France | 1996 | 1996 | 2000 | Slovak Republic | 1996 | 1998 | 1999 |
| Georgia | 2010 | 2010 | 2010 | Slovenia | 1996 | 1996 | 2000 |
| Germany | 1996 | 1997 | 2000 | South Africa | 1996 | 1996 | 2000 |
| Greece | 2002 | 2002 | 2002 | Spain | 1996 | 1998 | 2000 |
| Hong Kong SAR | 1996 | 1997 | 2000 | Sweden | 1996 | 1996 | 2000 |
| Hungary | 1996 | 1997 | 2000 | Switzerland | 1996 | 1996 | 2001 |
| Iceland | 1996 | 1998 | 2004 | Thailand | 1996 | 1996 | 2000 |
| India | 1996 | 1997 | 2001 | Tunisia | 2001 | 2001 | 2001 |
| Indonesia | 1996 | 1997 | 2000 | Turkey | 1996 | 1996 | 2001 |
| Ireland | 1996 | 1996 | 2001 | Ukraine | 2003 | 2003 | 2003 |
| Israel | 1996 | 1996 | 2000 | United Kingdom | 1996 | 1996 | 1999 |
| Italy | 1996 | 1996 | 2000 | United States | 1996 | 1996 | 1999 |
| Japan | 1996 | 1996 | 2000 | Uruguay | 2004 | 2004 | 2004 |
| Jordan | 2010 | 2010 | 2010 | West Bank and Gaza | 2012 | 2012 | 2012 |
| Kazakhstan | 2003 | 2003 | 2003 |
SDDS Data Coverage
SDDS Data Coverage
| Category | Component (example) |
|---|---|
| Real Sector | |
| National Accounts | GDP by categories |
| Production Indices | idustrial / commodity production |
| Labor market | (un)employment, wages |
| Price Indices | CPI, producer price index |
| Fiscal Sector | |
| General Government Operations | revenue, expenditure, financing |
| Central Government Operations | revenue, expenditure, financing |
| Central Government Debt | domestic / foreign (by currency) |
| Financial Sector | |
| Analytical Banking Accounts | money, credit |
| Analytical Central Bank Accounts | reserve money, domestic claims, external position |
| Interest Rates | government security rates |
| Stock Market | share price index |
| External Sector | |
| Balance of Payments | goods and services |
| Reserves | reserves |
| Merchandise Trade | exports and imports |
| International Investment Position | |
| Exchange Rates | spot rates, 3- and 6-month forward markets |
| External Debt | debt of different sectors (government, banking etc.) |
| Population |
SDDS Data Coverage
| Category | Component (example) |
|---|---|
| Real Sector | |
| National Accounts | GDP by categories |
| Production Indices | idustrial / commodity production |
| Labor market | (un)employment, wages |
| Price Indices | CPI, producer price index |
| Fiscal Sector | |
| General Government Operations | revenue, expenditure, financing |
| Central Government Operations | revenue, expenditure, financing |
| Central Government Debt | domestic / foreign (by currency) |
| Financial Sector | |
| Analytical Banking Accounts | money, credit |
| Analytical Central Bank Accounts | reserve money, domestic claims, external position |
| Interest Rates | government security rates |
| Stock Market | share price index |
| External Sector | |
| Balance of Payments | goods and services |
| Reserves | reserves |
| Merchandise Trade | exports and imports |
| International Investment Position | |
| Exchange Rates | spot rates, 3- and 6-month forward markets |
| External Debt | debt of different sectors (government, banking etc.) |
| Population |
Appendix B. Sample, Variables and Descriptive Statistics
List of Variables
List of Variables
| Variable | Explanation | Mean(Std. Dev.) | Source |
|---|---|---|---|
| ln(FDI) | Logarithm of real foreign direct investment inflows in USD | 22.04 (1.70) | IFS (WEO) |
| ln(portfolio) | Logarithm of real foreign portfolio investment inflows in USD | 22.10 (2.40) | IFS (WEO) |
| SDDS | dummy variable equal 1 if country i met ‘Special Data Dissemination Standard’ in yeart t, see table 5 | 0.40 (0.49) | IMF |
| ln(GDP) | Logarithm of real GDP in USD | 25.95 (1.79) | WEO |
| GDP growth | percentag change ((yt – yt–1)/yt–1) of real GDP per capita in national currency | 0.031 (0.036) | calc from WEO |
| investment rate | gross capital formation (at current prices) / GDP (also at current prices) | 0.233 (0.058) | calc from WEO |
| capital account open | Chinn and Ito (2006) index for capital account openness (higher values mean higher openness | 1.12 (1.40) | Chinn/Ito (2011) |
| political risk | political risk rating of International Country Risk guide (ICRG, yearly average), 0 (high risk) -100 | 65.89 (3.49) | ICRG |
| exchange rate volatility | see equation (3) on page 16 | 0.0026 (0.024) | calc from IFS |
| interest rate | spread of money market rate over LIBOR (both in percent p.a.) | 19.08 (231.37) | calc from IFS |
| real exchange rate | implied purchasing power parity exchange rate measured in national currency per USD | 85.90 (371.63) | WEO |
| trade share | sum of imports (including c.i.f.) and export from and to the world in current USD / GDP in current USD | 0.696 (0.601) | calc from IFS, WEO |
| yrs. of schooling | Years of Schooling from Barro and Lee (2010) using a smoother with weights 25, 16, 9, 4, and 1 | 9.04 1.99 | Barro/Lee (2010) |
| high-tech exports | country i’s share of high-tech exports in global high-tech exports at year t | 1.84e+10 (3.47e+10) | calc from WDI |
| export unit value | export unit value index | 87.31 (20.59) | WEO |
| # of patents | total number of patents | 841.73 (2,592.6) | OECD |
| WEO data dummy | dummy variable equal 1 if WEO data was used for dependent variable (instead of IFS data) | 0.121 (0.327) | own calc. |
List of Variables
| Variable | Explanation | Mean(Std. Dev.) | Source |
|---|---|---|---|
| ln(FDI) | Logarithm of real foreign direct investment inflows in USD | 22.04 (1.70) | IFS (WEO) |
| ln(portfolio) | Logarithm of real foreign portfolio investment inflows in USD | 22.10 (2.40) | IFS (WEO) |
| SDDS | dummy variable equal 1 if country i met ‘Special Data Dissemination Standard’ in yeart t, see table 5 | 0.40 (0.49) | IMF |
| ln(GDP) | Logarithm of real GDP in USD | 25.95 (1.79) | WEO |
| GDP growth | percentag change ((yt – yt–1)/yt–1) of real GDP per capita in national currency | 0.031 (0.036) | calc from WEO |
| investment rate | gross capital formation (at current prices) / GDP (also at current prices) | 0.233 (0.058) | calc from WEO |
| capital account open | Chinn and Ito (2006) index for capital account openness (higher values mean higher openness | 1.12 (1.40) | Chinn/Ito (2011) |
| political risk | political risk rating of International Country Risk guide (ICRG, yearly average), 0 (high risk) -100 | 65.89 (3.49) | ICRG |
| exchange rate volatility | see equation (3) on page 16 | 0.0026 (0.024) | calc from IFS |
| interest rate | spread of money market rate over LIBOR (both in percent p.a.) | 19.08 (231.37) | calc from IFS |
| real exchange rate | implied purchasing power parity exchange rate measured in national currency per USD | 85.90 (371.63) | WEO |
| trade share | sum of imports (including c.i.f.) and export from and to the world in current USD / GDP in current USD | 0.696 (0.601) | calc from IFS, WEO |
| yrs. of schooling | Years of Schooling from Barro and Lee (2010) using a smoother with weights 25, 16, 9, 4, and 1 | 9.04 1.99 | Barro/Lee (2010) |
| high-tech exports | country i’s share of high-tech exports in global high-tech exports at year t | 1.84e+10 (3.47e+10) | calc from WDI |
| export unit value | export unit value index | 87.31 (20.59) | WEO |
| # of patents | total number of patents | 841.73 (2,592.6) | OECD |
| WEO data dummy | dummy variable equal 1 if WEO data was used for dependent variable (instead of IFS data) | 0.121 (0.327) | own calc. |
List of countries in the sample(model 3a):
Algeria, Argentina, Australia, Austria, Brazil, Bulgaria, Canada, Chile, Colombia, Croatia, Czech Republic, Estonia, Finland, France, Germany, Greece, Hong Kong (China), Iceland, Indonesia, Ireland, Italy, Jamaica, Japan, Jordan, Korea, Rep., Kuwait, Latvia, Lithuania, Malaysia, Mexico, Moldova, Morocco, Netherlands, New Zealand, Norway, Pakistan, Peru, Philippines, Poland, Romania, Russian Federation, Singapore, Slovak Republic, Slovenia, South Africa, Spain, Sri Lanka, Switzerland, Thailand, Turkey, Ukraine, United Kingdom, United States, Uruguay, Venezuela
References
Agça, Mehmet and Peter Allum (2001), “Economic data dissemination: What influences country performance on frequency and timeliness?” IMF Working Papers 01/173, International Monetary Fund.
Aghion, Philippe, Philippe Bacchetta, Romain Ranciére, and Kenneth Rogoff (2009), “Exchange rate volatility and productivity growth: The role of financial development.” Journal of Monetary Economics, 56, 494 – 513.
Ahearne, Alan G., William L. Griever, and Francis E. Warnock (2004), “Information costs and home bias: an analysis of US holdings of foreign equities.” Journal of International Economics, 62, 313–336.
Akerlof, George A. (1970), “The market for ‘lemons’: Quality uncertainty and the market mechanism.” Quarterly Journal of Economics, 84, 488–500.
Albuquerque, Rui (2003), “The composition of international capital flows: risk sharing through foreign direct investment.” Journal of International Economics, 61, 353–383.
Albuquerque, Rui, Norman Loayza, and Luis Serven (2005), “World market integration through the lens of foreign direct investors.” Journal of International Economics, 66, 267–295.
Anderson, James E. and Eric van Wincoop (2003), “Gravity with gravitas: A solution to the border puzzle.” American Economic Review, 93, 170–192.
Ausubel, Lawrence M (1990), “Insider trading in a rational expectations economy.” American Economic Review, 80, 1022–41.
Avery, Christopher and Peter Zemsky (1998), “Multidimensional uncertainty and herd behavior in financial markets.” American Economic Review, 88, 724–48.
Badinger, Harald and Peter Egger (2010), “Horizontal vs. vertical interdependence in multinational activity.” Oxford Bulletin of Economics and Statistics, 72, 744–768.
Baltagi, Badi H., Peter Egger, and Michael Pfaffermayr (2007), “Estimating models of complex FDI: Are there third-country effects?” Journal of Econometrics, 140, 260–281.
Baltagi, Badi H., Peter Egger, and Michael Pfaffermayr (2008), “Estimating regional trade agreement effects on FDI in an interdependent world.” Journal of Econometrics, 145, 194–208.
Banerjee, Abhijit V. (1992), “A simple model of herd behavior.” The Quarterly Journal of Economics, 107, 797–817.
Barro, Robert and Jong-Wha Lee (2010), “A new data set of educational attainment in the world, 1950-2010.” Working Paper No. 15902, National Bureau of Economics Research, Cambridge, MA.
Berg, Andrew, Rafael Portillo, and D. Filiz Unsal (2010), “On the optimal adherence to money targets in a New-Keynesian framework: An application to low-income countries.” Working Paper 10/134, International Monetary Fund, URL http://www.imf.org/.
Bianchi, Javier (2011), “Overborrowing and systemic externalities in the business cycle.” American Economic Review, 101, 3400–3426.
Bikhchandani, Sushil, David Hirshleifer, and Ivo Welch (1992), “A theory of fads, fashion, custom, and cultural change as informational cascades.” Journal of Political Economy, 100, 992–1026.
Blanchard, Olivier (2010), Macroeconomics, 5. edition. Prentice Hall, Upper Saddle River, New Jersey.
Blonigen, Bruce A. (1997), “Firm-specific assets and the link between exchange rates and foreign direct investment.” American Economic Review, 87, 447–65.
Blonigen, Bruce A. (2002), “Tariff-jumping antidumping duties.” Journal of International Economics, 57, 31 – 49.
Blonigen, Bruce A. (2005), “A review of the empirical literature on FDI determinants.” Atlantic Economic Journal, 33, 383–403.
Blonigen, Bruce A., Ronald B. Davies, and Keith Head (2003), “Estimating the knowledge-capital model of the multinational enterprise: Comment.” American Economic Review.
Blonigen, Bruce A., Ronald B. Davies, Glen R. Waddell, and Helen T. Naughton (2007), “FDI in space: Spatial autoregressive relationships in foreign direct investment.” European Economic Review, 51, 1303 – 1325.
Blonigen, Bruce A., Christopher J. Ellis, and Dietrich Fausten (2005), “Industrial groupings and foreign direct investment.” Journal of International Economics, 65, 75–91.
Blonigen, Bruce A. and Jeremy Piger (2011), “Determinants of foreign direct investment.” NBER Working Papers 16704, National Bureau of Economic Research, Inc.
Bobonis, Gustavo J. and Howard J. Shatz (2007), “Agglomeration, adjustment, and state policies in the location of foreign direct investment in the United States.” The Review of Economics and Statistics, 89, 30–43.
Bond, Eric W. and Larry Samuelson (1986), “Tax holidays as signals.” American Economic Review, 76, 820–26.
Brennan, Michael J. and H. Henry Cao (1997), “International portfolio investment flows.” Journal of Finance, 52, 1851–1880.
Brennan, Michael J., H. Henry Cao, Norman Strong, and Xinzhong Xu (2005), “The dynamics of international equity market expectations.” Journal of Financial Economics, 77, 257 – 288.
Byard, Donal, Ying Li, and Yong Yu (2011), “The effect of mandatory IFRS adoption on financial analysts’ information environment.” Journal of Accounting Research, 49, 69–96.
Cady, John (2005), “Does SDDS subscription reduce borrowing costs for emerging market economies?” IMF Staff Papers, 52, 503–517.
Cady, John and Jesus Gonzalez-Garcia (2007), “Exchange rate volatility and reserves transparency.” IMF Staff Papers, 54, 741–754.
Cady, John and Anthony J. Pellechio (2006), “Sovereign borrowing cost and the IMF’s data standards initiatives.” IMF Working Papers 06/78, International Monetary Fund.
Calvo, G., L. Leiderman, and C. Reinhart (1993), “Capital inflows and the real exchange rate appreciation in Latin America: The role of external factors.” IMF Staff Papers, 40, 108–151.
Campa, Jose Manuel (1993), “Entry by foreign firms in the United States under exchange rate uncertainty.” The Review of Economics and Statistics, 75, 614–22.
Chakrabarti, Avik (2001), “The determinants of foreign direct investment: Sensitivity analyses of cross-country regressions.” Kyklos, 54, 89–113.
Chinn, Menzie D. and Hiro Ito (2006), “What matters for financial development? Capital controls, institutions, and interactions.” Journal of Development Economics, 81, 163–192.
Chinn, Menzie D. and Hiro Ito (2008), “A new measure of financial openness.” Journal of Comparative Policy Analysis, 10, 309–322.
Chinn, Menzie D. and Hiro Ito (2011), “Notes on the Chinn-Ito financial openness index 2009 update.” mimeo.
Chuhan, Punam, Stijn Claessens, and Nlandu Mamingi (1998), “Equity and bond flows to Latin America and Asia: the role of global and country factors.” Journal of Development Economics, 55, 439–463.
Conley, Timothy G. (1999), “GMM estimation with cross sectional dependence.” Journal of Econometrics, 92, 1–45.
Conley, Timothy G and Ethan Ligon (2002), “Economic distance and cross-country spillovers.” Journal of Economic Growth, 7, 157–87.
Coughlin, Cletus C. and Eran Segev (2000), “Foreign direct investment in China: A spatial econometric study.” The World Economy, 23, 1–23.
Coughlin, Cletus C., Joseph V. Terza, and Vachira Arromdee (1991), “State characteristics and the location of foreign direct investment within the United States.” The Review of Economics and Statistics, 73, 675–683.
Cushman, David O. (1985), “Real exchange rate risk, expectations, and the level of direct investment.” The Review of Economics and Statistics, 67, 297–308.
Dabla-Norris, Era, Jiro Honda, Amina Lahreche, and Geneviéve Verdier (2010), “FDI flows to low-income countries: Global drivers and growth implications.” Working Paper 10/132, International Monetary Fund, Washington, DC, URL http://www.imf.org/.
Daude, Christian and Marcel Fratzscher (2008), “The pecking order of cross-border investment.” Journal of International Economics, 74, 94 – 119.
Davies, Ronald B. and Helga Kristjánsdóttir (2010), “Fixed costs, foreign direct investment, and gravity with zeros.” Review of International Economics, 18, 47–62.
Davies, Ronald B., Pehr-Johan Norbáck, and Ayca Tekin-Koru (2009), “The effect of tax treaties on multinational firms: New evidence from microdata.” World Economy, 32, 77–110.
di Giovanni, Julian (2005), “What drives capital flows? The case of cross-border M and A activity and financial deepening.” Journal of International Economics, 65, 127 – 149.
Diamond, Douglas W. and Robert E. Verrecchia (1991), “Disclosure, liquidity, and the cost of capital.” Journal of Finance, 46, 1325–1359.
Du, Julan and Shang-Jin Wei (2004), “Does insider trading raise market volatility?” Economic Journal, 114, 916–942.
Engel, Charles M. and Kenneth D. West (2005), “Exchange rates and fundamentals.” Journal of Political Economy, 113, 485–517.
Fernandez-Arias, Eduardo (1996), “The new wave of private capital inflows: Push or pull?” Journal of Development Economics, 48, 389 – 418.
French, Kenneth R. and James M. Poterba (1991), “Investor diversification and international equity markets.” American Economic Review, 81, 222–26.
Froot, Kenneth A. and Jeremy C. Stein (1991), “Exchange rates and foreign direct investment: An imperfect capital markets approach.” The Quarterly Journal of Economics, 106, 1191–1217.
Gelos, R. Gaston and Shang-Jin Wei (2005), “Transparency and international portfolio holdings.” Journal of Finance, 60, 2987–3020.
Giles, David E. A. (1982), “The interpretation of dummy variables in semilogarithmic equations: Unbiased estimation.” Economics Letters, 10, 77–79.
Glennerster, Rachel and Yongseok Shin (2008), “Does transparency pay?” IMF Staff Papers, 55, 183–209.
Goldstein, Itay and Assaf Razin (2006), “An information-based trade off between foreign direct investment and foreign portfolio investment.” Journal of International Economics, 70, 271 – 295.
Gordon, Roger H and A Lans Bovenberg (1996), “Why is capital so immobile internationally? Possible explanations and implications for capital income taxation.” American Economic Review, 86, 1057–75.
Harding, Torfinn and Beata S. Javorcik (2011), “Roll out the red carpet and they will come: Investment promotion and FDI inflows.” Economic Journal, 121, 1445–1476.
Hashimoto, Yuko and Konstantin M. Wacker (2012), “The role of risk and information for international capital flows: New evidence from the SDDS.” Courant Research Centre: Poverty, Equity and Growth – Discussion Paper 124, University of Göttingen.
Hau, Harald (2001a), “Geographic patterns of trading profitability in Xetra.” European Economic Review, 45, 757–769.
Hau, Harald (2001b), “Location matters: An examination of trading profits.” Journal of Finance, 56, 1959–1983.
Haufler, Andreas and Ferdinand Mittermaier (2011), “Unionisation triggers tax incentives to attract foreign direct investment.” Economic Journal, 121, 793–818.
Head, Keith, John Ries, and Deborah Swenson (1995), “Agglomeration benefits and location choice: Evidence from Japanese manufacturing investments in the United States.” Journal of International Economics, 38, 223–247.
Hernández, Leonardo, Pamela Mellado, and Rodrigo Valdés (2001), “Determinants of private capital flows in the 1970s and 1990s: Is there evidence of contagion?” Working Paper 01/64, International Monetary Fund, Washington, DC, URL http://www.imf.org/.
Huber, Peter J. (1967), “The behavior of maximum likelihood estimates under nonstandard conditions.” in: Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability. Berkeley, CA: University of California Press, Vol. 1: 221âĂŞ233.
IMF (1991), “Determinants and systemic consequences of international capital flows: a study.” Occasional Paper no. 77, International Monetary Fund, Research Department, Washington, DC, URL http://www.imf.org/.
Javorcik, Beata S. and hang-Jin Wei (2009), “Corruption and cross-border investment in emerging markets: Firm-level evidence.” Journal of International Money and Finance, 28, 605–624.
Jeanne, Olivier (2011), “Capital account policies and the real exchange rate.” John Hopkins University, mimeo.
Jeanne, Olivier, Arvind Subramanian, and John Williamson (2012), Who Needs to Open the Capital Account? Peterson Institute for International Econometrics, Washingon, DC.
Kesternich, Iris and Monika Schnitzer (2010), “Who is afraid of political risk? Multinational firms and their choice of capital structure.” Journal of International Economics, 82, 208 – 218.
Kinoshita, Yuko and Ashoka Mody (2001), “Private information for foreign investment in emerging economies.” Canadian Journal of Economics/Revue canadienne d’économique, 34, 448–464.
Kiyota, Kozo and Shujiro Urata (2004), “Exchange rate, exchange rate volatility and foreign direct investment.” The World Economy, 27, 1501–1536.
Lehmann, Alexander and Hyoung G. Kang (2004), “Multinational affiliates and local financial markets.” Working Paper 04/107, International Monetary Fund, Washington, DC, URL http://www.imf.org/.
Lipsey, Robert E. (2002), “Home and host country effects of FDI.” Working Paper No. 9293, National Bureau of Economics Research, Cambridge, MA.
Lucas Jr., Robert E. (1990), “Why doesn’t capital flow from rich to poor countries?” American Economic Review, 80, 92–96.
Maddala, Gangadharrao Soundalyarao (1971), “The use of variance components models in pooling cross section and time series data.” Econometrica, 39, 341–358.
Milesi-Ferretti, Gian Maria and Philip R. Lane (2004), “International investment patterns.” IMF Working Papers 04/134, International Monetary Fund, URL http://ideas.repec.org/p/imf/imfwpa/04-134.html.
Mody, Ashoka, Efraim Sadka, and Assaf Razin (2003), “The role of information in driving FDI flows: Host-country transparency and source-country specialization.” IMF Working Papers 03/148, International Monetary Fund.
Mody, Ashoka and Mark P. Taylor (2003), “International capital crunches: The time-varying role of informational asymmetries.” Discussion Papers 3757, C.E.P.R. Discussion Papers.
Pagano, Marco, Ailsa A. Röell, and Josef Zechner (2002), “The geography of equity listing: Why do companies list abroad?” The Journal of Finance, 57, 2651–2694.
Pain, Nigel and Desirée van Welsum (2003), “Untying the Gordian Knot: The multiple links between exchange rates and foreign direct investment.” Journal of Common Market Studies, 41, 823–846.
Papaioannou, Elias (2009), “What drives international financial flows? Politics, institutions and other determinants.” Journal of Development Economics, 88, 269 – 281.
Pellechio, Anthony and John Cady (2006), “Differences in IMF data: Incidence and implications.” IMF Staff Papers, 53, 326–349.
Portes, Richard and Héléne Rey (2005), “The determinants of cross-border equity flows.” Journal of International Economics, 65, 269 – 296.
Portes, Richard, Helene Rey, and Yonghyup Oh (2001), “Information and capital flows: The determinants of transactions in financial assets.” European Economic Review, 45, 783–796.
Razin, Assaf, Efraim Sadka, and Chi-Wa Yuen (1998), “A pecking order of capital inflows and international tax principles.” Journal of International Economics, 44, 45–68.
Savastano, Miguel (2000), “Comment on Gosh and Wolf.” in: Sebastian Edwards (ed.): Capital Flows and the Emerging Economies: Theory and Evidence, and Controversies, University of Chicago Press, 156–158.
Schmidt, Christian and Udo Broll (2009), “Real exchange-rate uncertainty and us foreign direct investment: an empirical analysis.” Review of World Economics, 145, 513–530.
Scholes, Myron and Mark Wolfson (1990), “The effects of changes in tax law on corporate reorganization activity.” Journal of Business, 63, S141–S164.
Stehrer, Robert and Julia Woerz (2009), “’Attract FDI!’ – A universal golden rule? Empirical evidence for OECD and selected non-OECD countries.” European Journal of Development Research, 21, 95–111.
Stiglitz, Joseph E and Andrew Weiss (1981), “Credit rationing in markets with imperfect information.” American Economic Review, 71, 393–410.
Tesar, Linda L. and Ingrid M. Werner (1995), “Home bias and high turnover.” Journal of International Money and Finance, 14, 467 – 492.
Vaizey, John (1964), “The residual factor and economic growth.” Study Group in the Economics of Education, Paris: OECD.
van Nieuwerburgh, Stijn and Laura Veldkamp (2009), “Information immobility and the home bias puzzle.” Journal of Finance, 64(3), 1187–1215.
Wacker, Konstantin M. (2012), “Measurement and Determinants of Activities of Multinational Firms and Foreign Direct Investments.” mimeo.
Warnock, Francis E. (2002), “Home bias and high turnover reconsidered.” Journal of International Money and Finance, 21, 795–805.
Wei, Shang-Jin (2000a), “How taxing is corruption on international investors?” The Review of Economics and Statistics, 82, 1–11.
Wei, Shang-Jin (2000b), “Local corruption and global capital flows.” Brookings Papers on Economic Activity, 31, 303–354.
Wheeler, David and Ashoka Mody (1992), “International investment location decisions: The case of U.S. firms.” Journal of International Economics, 33, 57–76.
White, Halbert (1980), “A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity.” Econometrica, 48, 817âĂŞ830.
This paper was prepared when Konstantin Wacker was an intern in the Statistics Department during the summer of 2011. We would like to thank Emma Angulo, John Cady, Natasha Che, Ron Davies, Philipp Grosskurth, Stephan Klasen, Silvia Matei, Anna Orthofer, reviewers from the IMF’s Research Department and seminar participants at the IMF’s Statistics Departmental and at the University of Göttingen for helpful comments, inputs and ideas. All remaining errors are ours.
Bond and Samuelson (1986) and Gordon and Bovenberg (1996) provide models where productive countries can use tax-holidays to identify themselves to foreign investors.
Unweighted average over the 5 years prior to and after subscription. Outliers were deleted to provide the graph on a meaningful scale.
Because variable transaction costs would not explain the high turnover and because it seems generally improbable that the cumulated return on a well-diversified portfolio does not exceed the fixed barriers to entry in most markets.
Warnock (2002) argues that an underestimation of foreign-equity holdings drove some results of Tesar and Werner (1995), but also concludes that variable investment costs cannot explain the home-bias puzzle.
Economies being geographically close tend to have higher correlations, portfolio diversification would thus suggest investing in distant economies.
It should be noted that Portes et al. (2001) also look at the impact of of bilateral telephone call traffic to account ‘explicitly’ for information so that their contribution takes more effort to identify the impact of information than other studies of that time. However, we find the empirical strategy of Daude and Fratzscher (2008) more convincing so that we focus on their results.
Mody et al. (2003) also develop a model where FDI has an advantage over other forms of foreign investment in case of information asymmetries.
Furthermore, more accurate information may even increase GDP and thus market potential in an economy because it allows efficient flexible inflation targeting by the central bank, whereas the readily observed interest rate may only bear loose connection to the true interest rate in an economy with information gaps (cf. Berg et al., 2010 on the issue).
However, they find that fixed market entry costs play an important role and can lead to a bias in simple OLS estimation (instead of a Heckit procedure) when bilateral flows are used, especially for the parameter estimate of distance, which makes a strong statement for investigating the effect that information may play in this context.
Note that there is also a potential channel how information could negatively affect FDI: Models of vertical FDI motivate such investments, inter alia, by the problem of contract enforcement in vertical market relationships. IMF (1991, p. 24) thus argues that asymmetrical information provides a clue why FDI has been such an important component of capital flows. Hence, more accurate information might also lower incentives for FDI. This is also the rationale of the models of Goldstein and Razin (2006) and Mody et al. (2003). However, we find this argument – while potentially adequate in some special circumstances – not very important on the macro level, especially considering the fact that most FDI is driven horizontally.
More precisely, they use the first difference of BEA stock data.
In an older study using data from the early 1980s, Coughlin et al. (1991) also find that US state government spending to attract FDI had a positive, statistically significant effect on FDI attraction.
In fact, studies such as Calvo et al. (1993), Fernandez-Arias (1996), di Giovanni (2005) or Dabla-Norris et al. (2010) show that external push factors are highly relevant, although the literature also suggests that the importance of push vs. pull factors depends on the time period and countries analyzed. See, e.g., Chuhan et al. (1998); Hernández et al. (2001); Albuquerque et al. (2005); Dabla-Norris et al. (2010).
Hence, it is implicitly assumed that attraction of foreign capital is a policy motive. Potential gains from higher capital inflows generally include positive growth effects, higher resources for temporary fiscal stimulus in case of a domestic recession under constrained tax revenues and low saving of private households, or for inter-temporary utility maximization when future consumption is less valued than actual one.
We econometrically control for these ‘global’ effects by using year dummies.
Note that in equilibrium flows will equal the depreciation of existing stock. Flows will hence (also) depend on levels of economic activity in the host country. Furthermore, the adjustment from one equilibrium to another will occur via flows which is exactly our research focus in the context of a policy change in informational asymmetries. See Wacker (2012) for details.
Graham and Krugman (1989: 10) showed that raising the classification criterion to 20 or even 50 percent would only have a minor impact on the measurement of US firms classified as being under foreign control. However, this does not necessarily tell us much how the picture would look like for, say, applying a 5 percent rule instead, especially for non-US firms in more recent years. We find this to be another reason worth addressing the question whether FDI and portfolio flows differ in their determinants.
Precise definitions of FDI and portfolio investment can be found in the IMF’s “Balance of Payments and International Investment Position Manual”, 6th edition (2009).
WEO data are compiled by IMF staff based on the information gathered by the IMF country desk officers in the context of their missions to IMF member countries and through their ongoing analysis of the evolving situation in each country. Historical data are updated on a continual basis, as more information becomes available, and structural breaks in data are often adjusted to produce smooth series with the use of splicing and other techniques. See Pellechio and Cady (2006) on the general differences between IFS and other data sets.
Note that logs are taken for GDP so it can be interpreted as an elasticity.
We test for an omitted (time-dependent) variable problem in the robustness checks. Exclusion of the latter channel is trivial: On one hand, SDDS is a multilateral initiative and most countries joined at a single point in time (1996), so exogeneity can be assumed. The concern that international investors grew very strong over time and pushed both FDI flows and the implication of SDDS in 1996 is controlled for by the time fixed effect. Furthermore, we look at the date when SDDS specifications are met by subscribers, which usually takes place three to four years after countries’ subscription to SDDS so that our main explanatory variable is predetermined.
Also note that we are using fixed effect estimation, so the fact that larger countries will generally have a lower trade share than smaller economies will not pose a problem.
As for many other variables, we lag the PPP exchange rate by one year to avoid the problem of reversed causality since a capital inflow will automatically lead to an increase in the exchange rate if the latter is allowed to float freely, although it is ultimately the net inflow of all forms of capital that is relevant.
For example, it is not among the determinants discussed by Blonigen (2005) or Blonigen and Piger (2011).
Furthermore, previous studies on the impact of political instability that failed to control for the interest rate are likely to suffer from an omitted variable bias: instable countries are more likely to have higher interest rates. The relationship between stability and FDI hence also captures the cost of financing, not only an “instability tax”.
The MMR is the rate at which banks lend to each other for short term.
Since the index also takes into account restrictions on the current account, one may argue that it is too broad for our purpose. However, according to Jeanne (2011), import restrictions can have exactly the same effect as controls on capital inflows and reserve accumulation.
We do not include measures for other investment costs such as wages or taxes because both theoretical models and empirical evidence are ample for these variables. Despite cost-minimization playing an important role (cf. Badinger and Egger, 2010), MNCs do not necessarily shy away from paying high wages (Lipsey, 2002) and Haufler and Mittermaier (2011) even argue that governments in countries with high unionization rates (and thus probably higher wages) will have more incentives to attract FDI, e.g. by tax incentives. Scholes and Wolf-son (1990) provide a framework where FDI flows grow as a result of a tax increase. The results of Davies et al. (2009) highlight that MNCs’ response to taxation is very complex. Since tax systems are usually highly persistent, our fixed effect should absorb most of their impact. An omitted variable bias for SDDS stemming from omitting wage data is economically less likely. Furthermore, wages should be highly correlated with GDP which is among our control variables so that we control for this potential problem at least partially.
Similarly,