ISCTE-IUL

This study investigates the recent trend of the Feldstein-Horioka puzzle and the underlying reasons for moderation in capital flows. This issue is analysed quite inadequately for the period after the Global Financial Crisis, which represents a crucial turning point for economic climate and policies. The Feldstein-Horioka Puzzle is estimated using the World’s 13 largest economies, with panel GMM regression, between 1996 and 2016. We uncover that the Global Financial Crisis had a persistent detrimental effect on capital liberalization, after which the Feldstein-Horioka puzzle has revived and capital mobility has decreased. We suggest two possible explanations for such moderation in capital flows: the increasing risk perception and risk aversion behaviour of fund supplying countries, which increases the home bias, and capital controls against free flow of capital that have been applied after the Global Financial Crisis of 2008/2009.


Introduction
In a perfectly open international macroeconomic setting (without frictions and interest rates disparities) domestic savings are expected to flow freely to any country in order to seek for the highest return (Debarsy and Ertur, 2010;Raza et al., 2018). 4 In such a case, capital should become perfectly mobile across countries. Indeed, related exchange rate models (i.e., Mundell-Fleming), presumably accept full capital mobility (Fleming, 1962;Mundell, 1963;Dornbusch, 1991;Taylor, 1995;Coakley et al., 1998). However, a controversial finding by Feldstein and Horioka (1980) has introduced a puzzle: in their study on 16 OECD economies over the period betwen 1960 and 1974, investments have been found closely related to domestic savings, opposing to the proposition of perfect capital mobility. From a policy standpoint, capital mobility might bring both benefits and problems to countries. The benefits rely on a neoclassical perspective, such that it may rise economic welfare as the capital is allocated more efficiently and, therefore, provides consumption smoothing (Koepke, 2015;Hannan, 2018). Moreover, FDI inflows may stimulate productivity by financing investment projects, by triggering knowledge transfers, transmission of new products, processes and adoption of novel management methods (Unctad,1999). However, there are also disadvantages -sudden capital surges and rapid outflows may cause serious boom-boost cycles and macro-financial instability, particularly in emerging economies (i.e. currency crisis, devaluation, hyperinflation, like in the case of Turkey in 1994 and 2001 and Argentina in 2001). Moreover, FDI inflows might lead to collapse of many domestic companies due to increased competition (Moosa, 2002;Hunya and Geishecker, 2005) Hence, given its political importance, it is important to investigate/understand the nature and evolution of capital mobility and its relationship with the FH puzzle. It is known that during crisis times, corporations are often willing to possess liquidity in order to keep it as a buffer and to avoid cash imbalances and operative problems (Faulkender and Wang, 2006;Opler et al., 1999;Allen and Gale, 1994;Brown, 2000;Pulvino, 1998;Shleifler and Wishny, 1992). Due to the liquidity possession, during the Global Financial Crisis, many firms have tended to cut the corporate dividend pay outs in Western and developing economies (Huang et al., 2021;Bliss et al., 2015). Additionally, a pressure on asset selling in stock markets was observed (Anton and Polk, 2014). Consequently, due to the high costs and uncertainty (volatility), external funding opportunities of firms in Western economies (i.e., the US and the EU) have been constrained during the Global Financial Crisis (Faulkender and Wang, 2006;Opler et al., 1999;Campello et al., 2010). This has pushed many firms to rely more on domestic/internal credits and savings. (Opler et al.,1999;Almeida et al., 2011) All these financial circumstances might have changed the linkage between investments and domestic savings, i.e., the Feldstein Horioka puzzle and capital liberalization tendencies during 3 and after the Global Financial Crisis. Hence, the Global Financial Crisis experienced in 2008/09 might have significant impact on the evolution of the Feldstein Horioka puzzle. Moreover, the more recent developments in the World markets, which are considered to slow down/hamper the economic globalization, (such as Covid-19, Brexit, etc.), might have accelerated the attention to the topic.
From an economic policy perspective, the relationship between domestic savings and investments might have been stronger since GFC. This might be due to the fact that many countries have revised their trade and finance policies. The capital account stance, cross banking transactions and other international financial relationships have been changed substantially. They possibly became more precautionary. Moreover, investors risk perception has increased over this period. All these may have changed the evolution of FH puzzle.
The theoretical reasons of why such an FH puzzle occurs have heatedly been debated by scholars. Firstly, home bias is claimed as a major reason. It arises when fund managers are reluctant to invest in foreign securities due to fear from exchange rate volatility, political risks, and asymmetric information. (Niehans, 1992;Tesar and Werner, 1995;Obstfeld and Rogoff, 2000). Second, capital controls, regulatory and macro-prudential policies imposed by countries are put forward as another reason behind low capital mobility (Feldstein (1983;. Controls on capital and macro-prudential policies include restrictions against the uncontrolled free flow of financial capital that may cause artificial asset bubbles and over/under valuation of financial assets. Third, the existence of tight monetary policy, significant disparities in interest rate across countries are likely to cause less mobility (Frankel, 1991;Raza et al. 2018).
Regarding previous results, the initial findings of Feldstein and Horioka (1980) indicate a high savings retention coefficient of 0.88. Saving retention coefficient is the parameter that shows the strength of the relationship between savings and investments. In studies with more recent data, it has generally been found lower levels for this coefficient, pointing to a greater capital mobility. Some examples of these studies are Feldstein and Bachetta (1991) who analyzed OECD countries for the period between 1960-and 1989, Caprio and Howard (1984) who analyzed 23 economies for the 1961-1981 period, Tesar (1991) who analyzed 23 OECD countries, Blanchard and Giavazzi (2002) who analyzed the Eurozone, Amirkhalkali et al. (2003) who analyzed 19 OECD countries for the 1971-1999 period. Overall, the common result obtained is that capital mobility shows a tendency to increase since the 1970s that is an observation supporting the view that the FH puzzle is not strongly valid as in the past.
Our contribution to the literature is four-fold. First, there is a limited number of studies that investigate the evolution of the FH hypothesis in the era post 2008/2009 (Ma and Li, 2016). However, since the GFC has had a destroying effect on markets, increased risk perception by investors and regulatory policies imposed by countries, might have altered significantly the direction and magnitude of the global capital flows. Moreover, researchers claimed that the FH puzzle was solved during the period 2000-2007, as the savings retention coefficients were found low in several studies. Is it still the case? Or the paradox has revived after the Global Financial 4 Crisis? One exceptional study that analyze the post crisis era is implemented by But and Morley (2017) who analyzed 27 countries over the period 1980-2012. They found that the savings retention coefficient has declined before the crisis and risen afterwards, validating the revival of the FH puzzle most recently. Our study departs from the referred one, by analyzing the more recent period  and by cross-validating the evolution of the FH puzzle with corresponding rises/stops in FDI and portfolio investments data. There are some other studies that analyze the same issue for a single or a restricted group of countries, for example Johnson and Lamdin (2014), for European countries, Phiri (2019), for South Africa, Kaur and Sarin (2019), for Asian economies, but they do not provide evidence on global trends, which we intend to do, since the countries in our dataset cover about 80 % of the World GDP.
Our second contribution is to analyze the following research questions: have these flows re-surged in the post era of the crisis? Or the decline in capital mobility is persistent even after the Global McQuade and Schmitz, 2016). As a second hypothesis, rising risk perception on investments is referred to as key explanation (Milesi-Ferreti and Tille 2011; Lane 2013). Low economic growth in emerging economies, high public debt, increased exchange rate volatility, number of bank interrupts have created a perception of danger, which has created reluctant foreign investors. As a third argument, it is claimed that inopportune monetary policy actions may have created such a pattern. However, interest rates have considerably decreased after the Global Financial Crisis, creating high money supply and liquidity, which contrasts with this hypothesis (Giambacorta et al., 2014;Ahmed and Zlate, 2014;Bremus and Fratzscher, 2015). We try to shed light also into this issue by providing descriptive evidences on the validity of these hypotheses.
The fourth contribution regards a methodological improvement. In the literature, it has been discussed that the connection between Investment (I) and Savings (S) is endogenous, due to simultaneity or omitted variables such as exogenous global shocks, heterogeneity across countries in market size and level of development. (Harberger, 1980;Obsfeld, 1986;Finn, 1990;Levy, 1990; Apergis and Tsoumas, 2009). To overcome such a problem, we use a GMM technique, which is superior, compared to other methods. We also instrument the variables by using a proper set of exogenous variables.
Hence, the aim of this work is to investigate the evolution of capital mobility and to test the FH puzzle for a panel of the World's largest (13) economies over a very recent period 1996-2016.The reason why it starts from 1996 but not earlier is due to data unavailability.
This work is divided in 5 sections. In section 2, we show empirical evidence on capital flows movements. In section 3 we present the empirical approach of our work, namely data, statistical tests and econometric method. In section 4 we present and analyze the results and section 5 concludes.

The Evolution of Capital Flows
From an historical perspective, financial globalization has followed a U-shaped pattern since1870 to 2008. During this period, integration of capital markets had four different episodes (Bordo et al., 2004;Bordo, 2017). From 1870 to the beginning of the World War I in 1914, the classical era of the gold standard ensured stable exchange rates and highly mobile capital (Obsfeld and Bordo and Rockoff, 1996). The Interwar period, between 1914 and 1945, is a period that globalization had come to a halt due to capital restrictions and the Great Depression in 1929 (Reinhart et al., 2016). From 1945 to 1973, the Bretton Woods system enabled exchange rate stability and trade volume has resumed. However, Keynesian policies, based on strong domestic demand and capital restrictions, hampered the free flow of capital (Reinhart et al., 2016). We depict in Figure 1 the evolution of FDI flows and Portfolio Investment flows (in percentage of GDP) and their long-term trends. (Zhou et al., 2011). 5 The data refers to the world data. Regarding FDI flows depicted in Figure 1, it may be inferred that strong growth prospects in Emerging markets has attracted capital inflows during the 1970s. This in line with the Portfolio Theory that points to the importance of expected returns and risk premium in determination of the capital flows and their direction. (Ahmed and Zlate, 2014). Sudden stop and surges have been observed before and after the oil shock, which interbank lending had increased. Fast 6 adoption/spillover of technology and increased volume of trade has contributed to the capital markets' integration process. The debt crisis in Latin American countries occurred in 1975-1981, which has slowed down partially capital flows (Hannan, 2017;2018). In the mid-1980s, it has accelerated again until 1990. In early 1990s, a restrain in financial flows has been observed, due possibly to the Gulf War and political instability.
Afterwards, total FDI flows/GDP has risen from 0.6 % in 1992 to 4.3 % in 2000 ( Figure 1). The reasons of such an increase may be related to strong growth and high returns in emerging markets, loose monetary policies applied by advanced countries, end of the Cold War, and increasing trade deficits. The foundation of the TWO and removal of tariffs and quotas, the emergence of the NAFTA, the MAI (Multilateral Agreement on Investments), and productivity increase driven by the invention of internet systems, are also among the factors that contributed to this process. However, a tendency to decline has been observed from 4. More importantly, capital mobility has not recovered well in the period of post Global Financial Crisis. The long-term trend of FDI flows has been declining since 2008 until 2016. Similarly, portfolio investments have been decreasing as well, even at a faster pace. To be able to document the severity of the moderation, we calculate for each country the ratio of FDI and portfolio flows (scaled to GDP) between selected years, presented in Table 1. The first (upside) panel shows the corresponding results for FDI flows, while the second panel (located at the bottom) displays the results for portfolio flows. Regarding the results for FDI, it is first calculated the ratio of the period of 2016 to 2007 for each country, and some aggregate groups such as total, developed, and emerging economies. 2007 is widely recognized as the economic peak year before the crisis, which is why we select it as a reference year. The calculated ratio is 0.52 for the total of countries, 0.53 for developed and 0.74 for emerging economies. Hence, it means that by 2016, FDI flows have declined significantly, almost to its half level compared to peak year. The decrease is more pronounced in the developed world than in emerging economies. When we look at the result for each country, the ratio ranges between 0.46 (Eurozone) and 2.06 (Japan). In only 3 (out of 13) countries (Japan, Brazil, and South Korea), this ratio is above 1, indicating an increase in FDI flows. Furthermore, we calculate the same ratios by dividing 2016 value to 2006, the pre-peak year. At the aggregate level, the decline is confirmed by the ratio of 0.67.
An even more seriously decreasing trend is observed for portfolio flows, as can be seen in the right-hand side of Figure 1 and in Table 1. We calculate the ratio of 2016 to 2007, which is 0.32 at the aggregate level, 0.30 for advanced countries, and 0.32 for emerging economies. Hence, portfolio flows have declined greatly to almost one-third of their level after the GFC Then, we calculate the same ratios by dividing the 2016 value by the 2006 level. At the aggregate level, it is 0.29.
( Table 1 About Here) In order to be able to cross verify the trends, it is illustrated in Figure 2, the evolution of capital flows (FDI and Portfolio) and its subparts (out-and inflows) in developing (emerging) and in developed economies. In the first two graphs (Figures 2a and 2b), the sum of absolute value of FDI or Portfolio inflows and outflows are scaled with respect to GDP.

Empirical Methodology
In this section we present the data, the statistical tests done to the data, namely structural break tests, panel unit root tests, co-integration tests, and the econometric approach of our work.

Data
The dataset covers a panel that includes the World's 13 largest countries (measured by GDP) over the period 1996-2016. These countries are Australia, Brazil, Canada, China, Eurozone, India, Indonesia, Japan, Mexico, the Russian Federation, Turkey, and the United States of America. US, Canada, Australia, Eurozone and Japan as assumed as developed economies, whereas China, India, Brazil, Russia, South Korea, Turkey, Indonesia and Mexico referred to as developing/emerging economies. These countries represent about 80 % of the World GDP, hence, the results obtained from the dataset are likely to give a general idea about the global patterns.
Variables and data sources are defined in Table 2. In our work we use the Share of Savings relative to Gross National Income (S/GNI), the Share of Investment relative to GDP (I/GDP), Total Real GDP (size), GDP per capita (dev) and growth of Real GDP are the variables adopted.

Structural Break Tests
The idea that the GFC had a persistent and structural impact on the size of capital liberalization needs to be validated. To do so, structural break tests on FDI flows are applied. We were able to apply this test to FDI data only, as the time span is extremely short for portfolio data. The variable tested in this case is the sum of FDI flows (in absolute value) scaled to GDP for each country (fdi), such that: = + + fdi=|FDI Inflows + FDI Outflows|/GDP (1) Bai and Perron (1998)'s famous algorithm is widely used to detect multiple and unknown structural breaks in regression models. We regress each countries' fdi variable on a linear time trend and detect the breaks in a and b.
The detected break years are summarized in Table 3. The first column documents all break years detected, the second column focuses only the ones around 2008/09. For the aggregate series, it is detected "2009" as a structural break. The trend of fdi, following the global and country level break is also summarized in columns 3 and 4. It has been detected in 7 of the 13 countries a structural break that includes the Global Financial Crisis (between 2004 and 2010). In column 3, it is seen that in 7 countries fdi trend is downward following 2009 and only in 3 of them has an upward trend. In column 4, we can see that the fdi trend after the country-level break, indicates that only 1 country has an upward trend while in 6 countries the trend is downward after the Global Financial Crisis.

Panel Unit Root Tests
In order to select the proper estimation method, there is the need to investigate the unit root features of the variables. To do so, we implement various tests to examine panel unit root processes (4 different types to ensure robustness). These tests are developed by Levin, Lin, and Chu (2002) Fisher, 1932). In detail, unit root tests take the following general form: i represents the countries, t represents years. Z includes exogenous variables. ∃ , denotes the identically, independent, and normally distributed random error terms. Hence, if ∅ = 1, a nonstationary evolution is present in the form of a random walk. Thus, it means that exists a nonstationary process in the evolution of the variable of interest. If, however, ∅ < 1 it indicates a stationary evolution around a constant.
The main distinction between the four tests arises in the assumption of unit processes, either common or individually varying ones. To start with, Levin, Lin, and Chu (2002) presume a unit root parameter ∅ =∅, constant across countries. In contrast, the remaining techniques, Im, Pesaran, and Shin (2003), Fisher-ADF and Fisher-PP, adopt a cross-varying unit root parameters (∅ ) between countries. 7 Results are presented in Table 4. We observe that all our variables are found stationary in levels. Thus, we use the variables in levels in our estimations.

Panel Cointegration Tests
Next, we investigate whether the two variables (S/GNI and I/GDP) have a cointegrating long-term relationship. We apply a Johansen (1988) cointegration test, which investigates stationarity of the random error terms. The combinations of different test specifications and corresponding results are in Table 5. An intercept term is always added to the test, except in the first specification. We also consider the no trend, linear trend, and quadratic trend options alternatively. The Eigen and Trace statistics help define how many cointegrating relationships are found. Critical values are determined on the basis of MacKinnon, Haug, and Michelis (1999)'s work at 5 %. According to the results, there is no cointegrating relationship, regardless of the specification. Hence, we can safely continue to the analysis with conventional panel data regression analysis. 10 ( Table 5 About Here)

Econometric Model
A way of examining the degree of capital mobility and the Feldstein-Horioka puzzle is through the following equation: I/GDP = α + β S/GNI + e I: Investment, S: Savings where β is referred to as the savings retention coefficient. If β=0, it is implied that investments are not related to domestic savings, which indicates the existence of perfect capital mobility. The adopted panel data regression is the following: ( / ) , = + ( / ) , + ( / ) , + First, we apply a Hausman test, which is useful in choosing the relevant model. It helps testing the following null hypothesis on the basis of a Chi-Square test statistic (Hausman, 1978):

Ho: Both Fixed Effects and Random Effects models have identical coefficient estimations; consistency in both models Ha: Fixed and Random Effects models have different coefficient estimations; one model is inconsistent
If the null hypothesis is accepted, one may use either model since fixed or random effects estimators provide consistency. In contrast, if the alternative hypothesis is accepted, the fixed effects model should be preferred.
Results for the Hausman test are in Table 6. We observe that there is a large difference in coefficients between cross-sectional fixed effects and random effects estimations. However, no sizable difference between coefficients is observed when period fixed and random effects are compared. Overall, total effects suggest the rejection of the null and acceptance of the alternative hypothesis. Therefore, we use the cross-sectional fixed effects model, since it provides consistency.
( Table 6 About Here) In terms of the estimation technique, we prefer the GMM rather than the OLS (Hansen, 1982;Hansen, Heaton, and Yaron, 1996). GMM is known to be superior in dealing with endogenous relationships. Indeed, in the literature about the FH puzzle, several factors causing endogeneity are discussed. First, common exogenous shocks and business cycles are likely to affect simultaneously both investment and savings decisions. For instance, in the case of an unanticipated negative change in expectations, a jump in commodity and energy prices or occurrence of political instability, both savings and investments will fall spontaneously, which will give rise to a strong association between investment and savings (Frankel and Razin, 1986;Baxter and Crucini, 1993). Second, the level of economic development might simultaneously affect both investment and savings (Apergis and Tsoumas, 2009;Obsfeld, 1986;Finn, 1990;Harberger, 1980;Levy, 1990). A developed country's saving and investment rate is expected to be different from an underdeveloped country. Third, the economic size of countries might create endogeneity, hence large countries can have an influence on investments and on the formation of prices (Apergis and Tsoumas, 2009;Obsfeld, 1986;Finn, 1990;Harberger, 1980;Levy, 1990). To account for these effects, we employ dev, size and growth variables as exogenous instruments. They represent the level of development, market size, and short-run economic growth respectively (as explained in Table 2).

Panel GMM Fixed Effects Estimations
The results of GMM estimation are in Table 7. The savings retention coefficient is 0.56 for the aggregate dataset, which is significantly different from zero. Moreover, we apply a Wald Test and reject the hypothesis that saving retention equals to 1. The Jarque Bera test indicates the normally distributed errors. Thus, the main result is that capital mobility is not as restricted as found by Feldstein and Horioka (1980) but it is not as high as recent claims by economists. So, one may argue that the FH puzzle is still not solved.
Once we split our dataset in two parts -developed countries and emerging economies -and re-run the model, we obtain different results. The savings retention coefficient is observed to be smaller for developed countries (0.52) than emerging economies (0.57). It is observed that capital mobility is higher in advanced economies compared to emerging economies.
( Table 7 About Here) Additionally, we investigate the timely evolution of the FH puzzle. To be able to investigate it, we estimate the GMM for rolling windows of 8 years, which is a sufficiently long period to capture at least one business cycle. We apply exactly the same estimation procedure as in Table 7. We present the evolution of the estimated savings retention coefficients in Figure 3.
For the aggregate dataset, initially capital mobility was low, as β is high, around 0.8 in the 1996-2003 period. However, β starts declining since the 1990s and hits the lowest level (0.39) in the period 2003-2010, representing a period before the economic peak. This had led to the claims that the FH puzzle was over, as the savings retention coefficient has decreased substantially. However, after 2010, β rises again, above 0.6, which indicates falling capital mobility in the most recent period. The pattern is quite similar for emerging economies -while initially the savings retention coefficient is found around 0. the major reasons behind the moderation of equity flows across 50 economies, the domestic pull factors such as institutional quality, country risk profile, and stability were found important as well.
More hypotheses have been put forward in the literature, on the causes of the decrease of capital liberalization after the GFC.
The first hypothesis is the risk hypothesis. According to this hypothesis, capital mobility has declined since risk in emerging economies has risen. Global risk aversion behavior created a home bias; therefore, funds flow from the developed economies to emerging countries have slowed down. Milesi-Ferretti and Tille (2011)  It is clear that economic growth rebounded after the Global Financial Crisis, but tends to get lower afterwards. The annual growth rate in the developed world approaches 2 %, while in emerging economies to 3 %, far lower than the pre-crisis rates. The VIX and MSCI indexes increase sharply during the Global Financial Crisis, but gets lower afterwards, although it is still above the precrisis levels. Hence, one may argue that low economic growth rates and relatively higher risk indexes (compared to pre-crisis levels) validates the reality of this hypothesis, which might have created a home bias and the revival of the FH puzzle.
The second hypothesis is related to role of restrictive and macro-prudential policies -capital controls hypothesis. According to this hypothesis, regulations and controls on capital, such as on interbank lending and cross border exchange of securities, have lowered capital integration, giving rise to the FH puzzle, as investments are financed more by domestic savings (Lane,  It is observed that it has been rising for emerging market economies, but stationary for advanced countries. Developing countries have introduced many restrictive policies after the Global Financial Crisis. Trade of mortgage backed securities, uncontrolled sub-prime credit lending is hampered. Hence, as a result, it is supportive of this hypothesis that moderation of capital integration might be associated with capital controls. 14 ( Figure 5 About Here) The final argument is the monetary policy hypothesis, which states that unconventional monetary policy might contribute to the capital account closeness. We depict the evolution of lending interest rates (cross-country average) (in Figure 6, on the left-hand side) and money supply (broad money/GDP) (in Figure 6, on the right-hand side) in the emerging and advanced countries.
( Figure 6 About Here) It is observed that both developed and emerging market countries have adopted expansionary policies, since interest rates have declined substantially after the GFC. Accordingly, money supply has remarkably increased. Despite increasing liquidity, the downturn of capital flows contradicts the monetary policy hypothesis (Giambacorta, Hofmann, and Peersman, 2014).
Overall, increasing capital controls and risk perception seem to find support as two major hypotheses behind the moderation of capital flows.

Policy Discussion and Conclusions
This study has three major results. First, the FH puzzle is found increasingly evident, particularly after the Global Financial Crisis. Various arguments in the literature had claimed that the FH puzzle was over by the 2007 economic peak, when capital integration was very high and the savings retention coefficients was very low. Our findings contrast with this view and supports the revival of the puzzle, as the savings retention coefficients rise significantly after the GFC.
Second, consistent with the increase in the savings retention coefficients, FDI and portfolio flows have declined considerably during the GFC and were not able to recover afterwards. This decline is widespread for the majority of analyzed countries and particularly pronounced in portfolio flows. The impact of the GFC is persistent, since the structural breaks in FDI flows are observed for most of the countries around the years 2008/2009. Third, the two main reasons behind such moderation seem plausible. The first one is the fear and risk aversion behavior of fund supplying countries that increases the home bias. Hence, developed countries tend to hesitate investing in emerging countries, as the increased risk in these countries during the GFC has not yet dropped to desired levels. Low economic growth in the emerging world after the GFC has also contributed to the moderation of flows. The second reason is the controls and restrictions on capital mobility that have increased following the GFC, especially in emerging markets. Such macro-prudential policies and stringent applications against interbank lending and uncontrolled trade of mortgage-backed securities played a major role.
These results raise a policy discussion about the status and future of financial globalization. From a perspective of capital receiving countries, increasing capital controls might limit the fund flows. It, thus, may create moderate economic growth, as observed recently, but in a more sustainable fashion as sudden capital in/outflows and related boom-boost cycles are avoided. Consistently, a more stable growth path is possible, as observed recently (2010-2016).
From a fund supplying countries' standpoint, increased financial risk during the Global Financial Crisis has elevated the VIX and the volatility of MSCI indexes. The risk level has lowered substantially, but it is still above the pre-crisis level. Hence, this has created a home bias and selective behavior.
The post crisis may be termed as "controlled globalization", which might be seen as a beneficial set of policies, as it helps smoothing output in both developed and emerging economies. These policies help avoiding artificially created asset bubbles, sudden drains/surges of capital, and boombust cycles. Moderate output growth in this system might be perceived as a major drawback, but it should be perceived as a more economically sustainable process. As trade has an increasing trend, there will always be a need for external financing the trade deficits. Hence, capital flows will be need. However, this process might not be as significant as in the pre-crisis era and may be much controlled. Accordingly, the FH puzzle is expected to continue, as investment will still be largely financed by domestic savings, under limited cross border capital lending.           Figure 6. Evolution of interest rates and money supply Note: In interest rate graph Turkey and Eurozone is not included due to data unavailability in World Bank dataset, in broad money/GDP graph, Eurozone and Canada is excluded due to data unavailability in World Bank dataset. Data Source: World Bank Database