A Disaggregated View of The Capital
Inflow Effects on Domestic Credit
Allocation
A regional perspective
Djuwensi Passial
1University of Groningen, Faculty of Economics and Business
Supervisor:
Prof. dr. D. J. Bezemer Co-assessor:
dr. M.V. Nikolova
Abstract
Using panel data of 43 counties from 2000-2017 this research expands on earlier work by Samarina and Bezemer (2016) to investigate the substitutional effect capital inflow has on credit allocation by domestic banks, with a focus on the regional patterns. Using fixed-effects estimates I find a diverging effect of inflows on domestic credit allocation, based on inflow type. However these results are not robust to System-GMM estimation techniques. Direct Investment and Portfolio investment are found to have a statistically significant effect on credit allocation towards the Non-Financial private sector, with the former having a positive effect and the latter a negative effect. The interaction terms indicate regional heterogeneous patterns in the allocation behavior of banks within the regions.
1. Introduction
The concept of credit is an intriguing one, as described in their paper; “Towards a theory of shadow money” Gabor & Vestergaard (2016), define credit as a form of liability which acts to delay the settlement. This allows actors to economize on monies issued higher in the hierarchy and is prompted as “the secret of capitalism”. The hierarchy of money denotes a rank of forms in which debt settlement is accepted. As explained by Bezemer (2019) at the top of this ranking stands coins, notes and bank reserves, followed by bank deposits convertible at par. From these follow credit, which ties itself to economic uncertainty as the hierarchies in which it finds itself expands and contracts cyclically. Such expansions occur when more credit is extended and contracts as it is destroyed, through repayment or write-offs (Bezemer, 2019). As credit moves through its cyclical episodes so does the allocation and forms of credit, bringing about innovation.
As stated by Samarina and Bezemer (2016), since the 1990s domestic bank credit has been reallocated away from lending to non-financial business (Figure 1). An expanding literature discusses negative effects on growth and stability of this change in credit allocation (Calderón and Kubota, 2012; Bezemer, 2019; Samarina and Bezemer, 2016). This reallocation to more household credit causes lower private savings, slower economic growth, larger external imbalances (negative current account) and increased probability of crisis with a deeper and longer recession. Arcand, Berkes and Panizza (2012) find in their research that “too much finance” can indeed act as an encumbrance to economic growth. Here they found that private
sector credit above 100% of GDP has a “vanishing effect” that incites institutional and economic volatility.
The literature suggests that foreign capital inflow can be a driver of domestic credit reallocation (Samarina and Bezemer, 2016; Sugimoto and Enya, 2017). The creation of new securities and integration of financial markets facilitated the movement of capital in- and outflows globally leading to an expansion of credit. As indicated by Calderón and Kubota (2012) capital inflow can also be growth and development promoting by being a cheaper source of finance for firms, creating technology/ knowledge spillovers and enhancing international risk-sharing. Capital inflows however can also have pernicious effects, e.g. increasing macroeconomic volatility and currency mismatches while distorting asset prices (Furceri, Guichard, and Rusticelli, 2012; Calderón and Kubota, 2012; Brunnermeier, 2009).
To my knowledge the first paper to investigate these two phenomena is the one by (Samarina and Bezemer, 2016). In their research they hypothesize that the reallocation of domestic credit occurs when foreign capital competes with domestic bank loans in the non-bank-sector credit market. Domestic bank balance sheets are then recalibrated towards consumer credit creating the vulnerabilities discussed hitherto. Counties susceptible to the aforementioned events might seek out appropriate macro- and micro prudential policies to dampen capital inflows and their subsequent domestic credit allocation effects (OECD, 2011). The OECD Capital Report cautions taking such measures in isolation (OECD, 2011). The risk associated pertains to a less integrated capital market and retaliation measures by other countries, like those observed between China and the US through financial decoupling (Bloomberg, 2020). OECD (2011) highlights that international cooperation is essential to avoid undesired collective outcomes as countries take unilateral action to manage international capital flows.
As shown in Figure 22 regional heterogeneity exists in domestic composition of credit markets,
which is not yet highlighted in the literature explored. Storper (2018) encourages such research and argues that broader structural and development macro framework is needed to understand the current regional economic divergence around the world. To my knowledge this will be the first attempt at highlighting regional heterogeneity with respect to domestic credit allocation driven by different types of credit inflow. As aforementioned an international approach is desirable when looking to dampen the negative effect of such inflows. My choice in a regional approach is driven by increased regional cooperation in the free trade areas; custom unions; common markets and economic unions as the need to better understand the potential for regional development under global integration grows. My research will thus aim to answer the question:
“What are the regional effects of capital inflow on domestic credit allocation?”
The results indicate the presence of diverging patterns in the way different types of capital inflows interact with bank credit allocation to the Non-Financial sector (private). This
2 Countries within each region (2000-2019) Europe - North America [Canada, United States]- Middle/
South-America [Argentina, Brazil, Chile, Colombia, Mexico] – Africa [South Africa] – Asia [China, Hong-Kong SAR, India, Indonesia, Japan, Korea, Malaysia, Singapore, Thailand] – Middle East [Israel, Saudi Arabia, Turkey]
Figure 2 –Credit composition from a regional 2 perspective. Source: Author’s calculation based on
(BIS, 2020) statistics. 0% 20% 40% 60% 80% 100%
Europe North America Middle/
South-America Africa Asia Middle-East
Credit Composition by Region
Heterogeneity seems to hold for regions as well in a Fixed Effects model. The results are however not robust when estimated with a System GMM model to deal with endogeneity. The rest of this paper is divided into 5 sections. In section 2 I will explore the literature, methodology and results of previous researchers. Section 3, 4 and 5 will contain the methodology, data and empirical analysis of my research, respectively. And in section 6 I will discuss some concluding remarks.
2. Literature review
2.1 The effects of capital inflow
As aforementioned capital inflow can have diverging effects on the macroeconomic environment of the receiving country. The benefit of such movements lies in the efficient allocation of savings and investment between surplus and deficit countries. Its support in long-term growth does not exclude risk arising from the absorptive capacity of the home markets (OECD, 2011). The literature on the effects of capital inflow attributes the boom of domestic credit growth to its interaction with the current account (Davis, Mack, Phoa, & Vandenabeele, 2016). As Lane and McQuade (2014) indicate the current account imbalances can affect macroeconomic variables, such as the rate of output growth, the level of domestic spending, exchange rates, inflation, and asset prices, which in turn have an effect on equilibrium credit growth. International capital movement can increase the risks of economic overheating and end in abrupt reversals making domestic economies more susceptible to international boom-and bust cycles (OECD, 2011).
Taking global factors into consideration, Cuestas and Staehr (2017) found in their paper that capital inflow were key drivers of increased domestic credit in southern European countries in the pre-crisis period. Taking on a firm perspective Kaat (2016) investigates credit allocation to non-financial businesses with respect to capital inflow. Using linear regression analysis, he finds that increased capital inflow will lead to inefficient allocation of capital. The economic destruction coming from this allocation is because credit is allocated to businesses which are less profitable. He investigated this for 20,000 firm-year observations spanning 1995-2004 and notes that a current account deficit exacerbates this mechanism.
financial factors. Enya and Shinkaia (2018) highlights this for the Asian market 2000-2015,
utilizing a panel VAR model they find that FDI has a positive effect on household credit share, while other types of capital flows have weak or non-significant effect. Enya and Shinkaib
(2014) research the dynamics of disaggregated capital inflow for Asian economies. Focusing on five Asian economies (e.g. Korea, Hong Kong, Thailand, Malaysia and Indonesia) the authors find mixed results from the effects of capital inflow on macro-economic market indicators. Their data covering 2000-2010 shows portfolio investments being positively correlated to asset prices. The results from other types of capital were mixed.
2.2 The allocation of credit
The allocation of bank credit primarily towards households does not promote productivity growth, insofar as non-financial firms do, by way of innovation (King and Levine, 1993). Bezemer, Samarina and Zhangb (2017) highlights literature from the late 1990’s where
researchers already investigated the income growth effects of business loans vs mortgages. Here, business investment was the channel through which financial development affects growth and not household credit. The reallocation of credit towards household also increases property prices benefiting those who own the asset and creating inequality in wealth and income (Piketty, 2014)
Research regarding the drivers of domestic credit reallocation as a result of capital inflow are quite recent. The paper by Bezemer, Samarina and Zhangb (2017) provides a significant
contribution to this topic in two ways; first it presents new data on the shift of domestic credit between four categories (e.g. home mortgages, consumer credit, bank loans to non-bank financials, and loans to non-financial business) and secondly it tests the economic conditions under which the allocation occurs. Their dataset shows trends in the shift of credit away from traditional non-financial business lending. The econometric analysis on key macroeconomic variables indicates that the debt shift away from business lending is more prevalent in advanced economies with higher government investments, presence of foreign banks and trade. The contrary holds true for emerging economies, where more government spending translates to a lower share of business credit allocation. Subsequent research by the same group of authors Bezemer, Samarina and Zhanga (2020) explore the influence of mortgage lending on business
negative in emerging and developing economies. The results regarding credit allocation away from non-financial business in these articles focus on domestic characteristics which influence allocation patterns. There is a clear trend that emerges where advanced economies react differently from emerging economies. In a similar vein the paper by Gozgor (2018) also investigates determinants of domestic credit allocation. Using OLS fixed effect models they decompose their analysis into emerging and non-emerging developing economies creating a panel of 64 countries from 1984 to 2016. Omitting the influence of global factors from their analysis they find evidence of macro-economic factors (e.g. money supply and income) having a positive effect on domestic credit growth.
Even with a balanced current account however capital movement can have an effect on credit dynamics (Borio and Disyatat, 2011; Gourinchas, 2012; Obstfeld, 2012). Gross capital inflows absorbed by banks can create an alternative funding source. This will lead however to a change in the balance sheet composition of domestic banks when absorbed by the non-financial business. This mechanism is especially relevant in economies with limited investment opportunities (Rodrik and Subramanian, 2009). Presbitero and Rabellotti (2016) attributes these rational investment decisions to the increased credit availability which pressures domestic banks to lend to informational opaque borrowers, which are often households. Samarina and Bezemer (2016) are the first to my knowledge to provide a framework highlighting the mechanism through which the change in balance sheet composition occurs. It has received support from other researchers such as Sugimoto and Enya (2017); Gumata and Ndou (2019), which have used this mechanism with some modification to conduct further research on this topic. This substation would also assume that banks will not lower their lending requirements, which increases the scarcity in viable investment projects when competing with foreign credit. Thus, as indicated by Samarina and Bezemer (2016) and shown in Figure 1 banks will decrease their lending to the Non-Financial businesses.
non-bank sectors in South Africa. The research by Samarina and Bezemer (2016) broadens the geographical scope and finds a negative relationship between capital flowing into the non-bank business sector and domestic credit allocation to non-financial businesses. Using fixed effects and GMM regression techniques they find that these results are weaker for countries with more investment opportunities (36 countries from 1990-2011). The research by Sugimoto and Enya (2017) is comparable to that of Samarina and Bezemer (2016) as it researches the effects of capital inflow on the domestic credit allocation with household credit as the main variable of interest. They further distinguish between the type of capital moving in. Their results are however in line with that of Samarina and Bezemer (2016), where every type of credit inflow causes an allocation of domestic credit away from firms, increasing household credit, using OLS-fixed effects regression techniques for 33 economies from 2000-2016.
2.3 Regional Patterns
Literature highlighting international regional patterns is scarce as most literature which seeks regional trends do so within the border of a country (Meslier, Sauviat and Yuan, 2020; Samolyk, 1994). However, a research by Yaltaa and Yaltab (2018) regarding the bias in credit agencies does take on such an approach. Here they find statistically significant regional 3
patterns which are in favor of countries that are economically, geo-politically and culturally aligned. Literature on regional patterns concerning the effect of capital inflow on domestic credit allocation is non-existent, to my knowledge. However, the literature explored does suggest the existence of regional heterogeneity. The research by Gumata and Ndou (2019) finds results for Africa which are contrary to those presented by (Sugimoto and Enya, 2017; Samarina and Bezemer, 2016). Even within the research by (Sugimoto and Enya, 2017; Samarina and Bezemer, 2016) authors highlight differences patterns along the lines of developed and developing countries, which tend to be geographically concentrated. As regional economic integration increases in popularity (e.g. free trade areas; custom unions; common markets and economic unions) a unified approach becomes more relevant in dealing with the negative effects of capital inflow.
3 Regions recognized within (Yaltaa and Yaltab, 2018) research: EU15, MENA, Latin America,
2.4 Hypothesis
As indicated by the literature surveyed capital inflow and increased credit allocation towards households can have detrimental effects on a domestic economy. As capital movement increases firms will substitute domestic funding by foreign ones forcing domestic banks to recalibrate their balance sheet towards household credit (Samarina and Bezemer, 2016). Such substitution will make firms susceptible to sudden stops and the international business cycle OECD (2011), creating instability in the domestic economy. The reallocation of domestic credit itself goes against the process of creative destruction presented by Schumpeter (Kaat, 2016), creating a divergence in financial and economic development (Bezemer, Samarina and Zhanga, 2020). Such issues are not only important to deal with from economic stability and
growth perspective but also from an inequality perspective as owners of asset benefit from the accumulation of capital which follow from capital inflow and credit growth (Piketty, 2014). The aforementioned elements have been extensively researched as presented by the literate, however a gap still exists. As indicated by OECD (2011) decisions regarding capital movement should not be taken unilaterally as countries negatively affected by such actions may seek retaliation. Thus, they suggest international cooperation in such respect. To my knowledge my paper will thus be the first to tackle this subject, e.g. looking at the effects of capital movement on domestic credit allocation from a regional perspective. In particular my paper adds the nexus of the substitution effect of capital inflow and the different types of inflows along regional lines. As highlighted in Figure 1 the current pattern of credit allocation to the Non-Financial sectors is globally negative. This perspective seeks to provide policy makers with an overview of capital inflows which have a detrimental effect on their respective domestic economies and allows for regional cooperation. Thus, from the discussion above materialize the following hypothesis which I will test empirically:
Hypothesis 1: all capital inflow types have a negative effect on bank credit allocation to the Non-Financial Business
3. Methodology
The main inspiration for my research is the papers by (Samarina and Bezemer, 2016). It empirically investigates the relationship between credit reallocation and capital inflow. This seminal paper on the topic of credit allocation by Samarina and Bezemer (2016) looks at the aggregated effects of capital into particular domestic sectors and employs non-financial business credit as their main variables of interest, citing their interest in the substitution effects capital inflow creates in an investment constraint economy. My research will take this paper as a starting point but will make two changes, first exploring the ways in which different types of capital movement have impact on the allocation of domestic credit and two exploring regional heterogeneity which may exist in this allocation effect. It thus seeks to answer the question:
“What are the regional effects of capital inflow on domestic credit allocation?”.
As presented by Samarina and Bezemer (2016) foreign capital can be allocated towards the bank and non-bank sectors in a country. If the latter occurs domestic firms will substitute domestic funding by foreign funding, thus decreasing the demand for credit from the local financial market. This substitution effect makes domestic firms more susceptible to foreign funding shocks and increases the effect of sudden stops in funding (OECD, 2011). To remain viable banks will increase household lending activities increasing household credit for consumption purposes creating asset price booms (Samarina and Bezemer, 2016; OECD, 2011). This substitution effect does not happen because of a change in funding of domestic banks but rather because of a change in direct funding source by local non-financial firms (Samarina and Bezemer, 2016) (Figure 3). The literature is limited however in that it doesn’t indicate which shape such a substitution takes. As banks are often procyclical such a shape may very well be monotonic as extending credit to households will increase the need for
0% 20% 40% 60% 80% 100%
Before Inflow After Inflow
Non-Financial Business
Domestic Credit Foreign Credit
0% 20% 40% 60% 80% 100%
Before Inflow After Inflow
Bank Balance Sheet
Household Credit Non-Finacial Business Credit
additional credit, as long as the influx of capital continues. Hardie and Howarth (2013) attributes this cycle to the procyclicality of market prices.
Most of the papers consulted deal with the border effect of capital inflow and domestic credit and economic growth (Gozgor, 2018; Cuestas and Staehr, 2017; Kaat, 2016; Aizenman, Jinjarak and Park, 2013; Igan and Tan, 2015 ). A subset of these papers are in line with my research, investigating the allocation of credit empirically using Fixed-effect and Dynamic models (Samarina and Bezemer, 2016; Sugimoto and Enya, 2017). Estimating this relationship is complex, therefore in-addition to the variables of interest these authors also employ additional control variables to make the results more efficient. Using a base Fixed effects model is standard as surveyed in the literature. The literature also considers a dynamic model (GMM model) by Areliano and Bover (1995); Blundell and Bond (1998) to deal with data-points which tend to be jointly determined. The system-GMM model combines the base equation in levels with the equation in first differences. The endogenous variables are instrumented by their lags in the first-difference equation and by first differences in the level equation. In all specifications, one or two lags are used as instruments, where the number of instruments is smaller than the number of countries. To test for consistency of estimates and validity of instruments, the Hansen test of overidentifying restrictions is performed, along with tests for first- and second-order autocorrelation of the residuals to test the efficiency of the estimates. In this paper I follow the methodology presented by Samarina and Bezemer (2016) and (Furceri, Guichard, & Rusticelli, 2012). The paper by Samarina & Bezemer (2016) forms the basis for the allocation mechanism of domestic credit, where the main variables of interest are non-financial business credit coming from banks and the inflow of capital. The latter is consulted for techniques on how to deal with the regional patterns/ perspective. The paper by Furceri et. al (2012) investigates regional bias in credit rating agencies.
phenomenons. Research by (Aizenman, Jinjarak, & Park, 2013) (Igan & Tan, 2015 ) (Davis, Mack, Phoa, & Vandenabeele, 2016) attributed this relationship to the interaction of capital flows with the current account. (Igan & Tan, 2015 ) describes capital flows as the mirror of the current account and highlights that capital flows are heterogeneous in nature and lumping them together would not be sensible. FDI is fundamentally different from portfolio investments and other inflows, as it provides direct ownership and creates knowledge spill-overs which are prone to job-creations (Aizenman, Jinjarak, & Park, 2013). Equity investments and Other Inflows both have elements which places them closer to debt-like instruments on a spectrum, the latter more than the former. Equity investments for example require dividend pay-outs to shareholders creating a future outflow at a later moment. Other inflows which have more debt-like components will create outflows in the forms of interest and principal payments also creating future outflows (International Monetary Fund , 2014). In reality all these types of capital movement can have a negative effect on the current account and create future outflows. However, the type of outflow determines the nature and the delay of the outflows. When the current account is in balance the marginal effect of the rising private sector debt level is rather smaller (Davis, Mack, Phoa, & Vandenabeele, 2016). Thus, all these flows will interact with the current account at a varying degree and fuel the need to understand the effects of capital movement at a disaggregated level.
As indicated by the literature I employ gross inflows, because net investment position can hide important allocation mechanisms which net inflows can omit (Lane and McQuade, 2014). To deal with the empirical challenges present I employ a fixed effect regression, which follows regression equation 1.
CRBit = α + β1CRD0it + β2 FLOWSit-1 + + γ Xit + δRegioni + ui + ωt. + εit (1)
Where CRBit is the share of credit from the bank sector to the private non-financial business of
country i in period t. CRD0it controls for the initial financial development of a country proxied
by the total credit at the beginning of the period. FLOWSit-1 is the 1 period lag of the array of
macroeconomic factors (income levels and GDP growth) and financial factors (credit market regulation and bank leverage). The former controls for the initial level of financial development along with the growth of the economy which absorbs the credit created. Bezemer, Samarina and Zhangb (2017) suggest that it is sensible to control for these measures as there are strong
linkages between income level and economic structure. The latter controls for factors which determine the rate at which credit can be created. Deregulation will impact the allocation of credit, increasing its scope creating financial innovation (originate-to-distribute lending) that facilitated household lending (Bezemer, Samarina and Zhangb, 2017). The research surveyed
employed additional control variables to those mentioned above but failed to find significance in their contribution to credit allocation, thus I only include those mentioned hitherto. The
δRegioni is the array of regional dummies which allow for the observation of heterogeneity
between the different regions. I further control for unobserved country specific fixed effects with ui. The last two terms in the regression equation are ωt. and εit, which represent time fixed
4. Data
4.1 Data description
This paper uses annual data from 5 datasets covering 2000-2017 for 43 countries. The selection of countries and the time period covered is guided by the scarcity of data availability on credit, where large gaps in the series exist before 2000. The main variable of interest in this paper is domestic bank credit extended to non-financial firms and the different types of capital inflows which influence its allocation. Data regarding domestic bank credit is taken from the Bank of International Settlement (BIS) database. Domestic credit is reported here by the borrowing sector, lending sector and borrower’s country. One shortcoming of this database is that data on the lending sector is presented at a high level; either by the bank sector or all sectors, where the former is only available for credit to the private non-financial sector. Another approach to collecting domestic credit allocation data would have been to follow Samarina and Bezemer (2016) where they compiled data from Central Bank balance sheets. This approach however brought about uniformity concern as data is reported at the discretion of that central bank and may have some measurement errors/ discrepancies when harmonized, thus I proceeded with the data from BIS. The second database I use the International Monetary Fund (IMF) statistics on International Investments Positions (BPM6) for capital inflow. The IMF reports micro-level data on each country and year. Disaggregated capital inflow data is found on the liability balance side of the financial accounts for each country. The remaining databases I consult are the World Banks; World development Indicator and Global Financial Development statistics and Fraser Institute; Financial Freedom database
4.1.1 Main Variables
Bank Non-financial Business credit.
data. Thus, subtracting total credit to household (directly reported in BIS) from bank credit to the private sector (directly reported in BIS) I obtain the portion of bank credit which is allocated to the non-financial sector. I construct the variable: Bank Credit to the non-financial sector by dividing the residual by the total bank credit provided to the private non-financial sector to get the proportion of bank credit which is extended to non-financial firms. The variable Bank Credit to the financial sector is thus the percentage of credit which is allocated to non-financial firms.
Capital inflows
As aforementioned data on capital inflows are presented in the IMF statistics as a liability on the financial account. The financial account of a country’s balance of payment, as explained by Cypher and Dietz (2009) records some of the most important flows across national borders. Here the inflow of foreign currency increases investment potential for current and future consumption. Capital inflows are often debt instruments which need to be repaid and, thus itself creating out-flows at a later moment, affecting the current account (Cypher and Dietz, 2009). The IMF statistics on capital inflows report the type of inflows in Millions of US dollars for each country. For my analysis I followed Samarina and Bezemer (2016) and analysed these inflows as a percentage of GDP. I proceed by dividing the data on capital movements from the IMF BoP database by the GDP figures from the World Bank Database, for the respective country and year. Each type of capital flow is described by the IMF in their “Balance of Payment Manual” as following:
Direct investment “Direct investments (DI) is a category of cross border investment associated with a resident in one economy having control or a significant degree of influence on the management of a company that is resident in another economy. In addition to funds, direct investors may supply additional contributions such as expertise, innovation, technology, management, and marketing. As well as the equity that gives rise to control or influence, DI also includes investment associated with that relationship, including investment in indirectly influenced or controlled companies, investment in fellow enterprises, debt, and reverse investment” (International Monetary Fund , 2014).
securities, and similar instruments normally traded in the financial markets” (International
Monetary Fund , 2014).
Other Investments “Other investments represent other equity; currency and deposits; loans; insurance, pension and standardized guarantee schemes; trade credit and advances; other accounts receivable/payable-other; and special drawing rights (SDRs)” (International
Monetary Fund , 2014). 4.1.2 Control variables
My choice of control variables is dictated by the literature, namely total credit to GDP, income levels, GDP growth, Credit Market regulation and Bank Leverage. Total credit to GDP is taken from the BIS database (BIS, 2020). Income levels (log GDP per capita) is defined in the World Bank statistics as the sum of gross value added by all residents divided by midyear population and reported in current U.S. dollars (The World Bank, n.d.). GDP growth is also taken from the World Bank statistics where it is reported as the annual growth rate of GDP at market prices based on constant local currency (The World Bank, n.d.). The control variable for credit market regulation is an index from the Fraser institute. This metric comprises 3 subcomponents: 1) ownership of banks, 2) private sector credit and 3) interest rate controls/ negative real interest rates. A higher value in the index is indicative of less regulation (Gwartney, Lawson and Hall, 2017). The last control variable is bank leverage taken from the Global Financial Development database. Here the ratio of bank credit to bank deposits determines the level of leverage in a particular country.
4.1.3 Regional Dummies
The choice of countries within each region is guided by the Bank Non-financial Business credit variable provided by BIS. Here I initially divide the country sample into 7 regions4, namely
Africa, Asia, Europe, Middle/ South America, Middle East, North America and Oceania, which are based on the United Nations classification. However, upon closer inspection it can be observed that for the region of Africa there is only one country available. I therefore exclude that region from my analysis and proceed with 6 regions.
4 Countries within each region – guided by (BIS, 2020) data availability on Domestic Credit – Europe [Argentina,
4.2 Trends in domestic credit and capital flows
The final dataset covers of 43 countries and 6 regions5 from 2000 to 2017. The descriptive
statistics for the variables discussed in section 4.1 are presented in Table 1. Our dependent variable, Bank credit to the Non-Financial Business has a mean of 45,17% over all observed countries and year. Table 2 presents the correlation matrix for the variables of interest and the control variables, which measures the direction and strength of the linear relationship between the variables. Direct and Portfolio Investments seem to be highly related (0,816), suggesting that these variables interact positively with each other. GDP per capita and Total credit also exhibit the same relationship but are less strongly correlated (0,749). I additionally employ a variance inflation factor test to inspect the presence of multicollinearity. The results show no indication for potential problems as the VIF score is 2.31 (Appendix A).
Table 1 – Descriptive Statistics
N mean sd min max
Credit to Non-Financial Business (%Bank
Credit) 692 0.452 0.198 0.00738 0.944
Total credit (%GDP) 772 194.2 84.19 41.80 444.3
Direct Investment (%GDP) 717 0.156 0.859 -0.158 13.01
Portfolio Investment (%GDP) 694 0.137 0.782 -2.270 8.434
Other Investments (%GDP) 716 0.0639 0.330 -0.857 4.228
GDP per capita (log) 774 9.830 1.088 6.094 11.69
GDP growth (annual %) 774 2.975 3.217 -10.89 25.16
Credit market regulations 764 8.756 1.181 3.670 10
Bank Leverage (%) 743 118.3 56.73 17.79 367.1
Source: BIS, IMF and World Bank Statistics, Source: Author’s calculation.
Table 2 – Correlation Matrix
(1) (2) (3) (4) (5) (6) (7) (8) (9)
(1) Bank Credit to
non-financial Business 1 (2) Direct Investment -0.062 1 (3) Portfolio Investment -0.057 0.816 1 (4) Other Investments -0.001 0.356 0.569 1 (5) Total Credit -0.413 0.296 0.265 0.205 1 (6) GDP per capita -0.564 0.242 0.250 0.230 0.749 1 (7) GDP growth 0.347 0.018 0.060 -0.021 -0.326 -0.325 1
(8) Credit market regulations -0.403 0.090 0.084 0.097 0.281 0.504 -0.021 1
(9) Bank Leverage -0.117 -0.227 -0.214 -0.168 0.150 0.167 -0.084 0.174 1
Source: BIS, IMF and World Bank Statistics, Source: Author’s calculation.
As can be observed in Figure 3, from a regional perspective credit extended to the Non-Financial Businesses by Banks are heterogeneous, across regions. Plotting this relationship against the level of GDP shows that there is a negative relationship between the level of GDP and Non-financial Business credit provided by banks (Figure 4) a result that was also found in (Aizenman, Jinjarak and Park, 2013; Enya and Shinkaia, 2018; Calderón and Kubota, 2012).
Plotting the capital movements against the proportion of bank credit allocated to the Non-Financial sector by region also highlights heterogeneous patterns which is the basis for my research. Along regional lines the relationship between Non-Financial bank credit allocation and inflows show idiosyncratic patterns. Figure 5, 6 and 7 present these relationships where a fitted line indicates that for Europe, Middle/ South America and Oceania the relationship between the capital inflows and Non-Financial bank credit allocation is negative. For the other regions the fitted line is less pronounced. Although these analyses are elementary, they do provide some evidence on what can be expected from the empirical analysis in section 5.
Figure 4 – Non-financial Business credit by Banks and log of GDP per capita. Source: Source: Author’s calculation from BIS and World Bank Data
Figure 5 - scatterplot of Non-financial business credit by Banks and log of Direct Investment Inflows, by region. Source: Source: Author’s calculation from BIS and IMP BoP
5. Empirical analysis
I start my empirical analysis by following the fixed effect model discussed in section 3 as presented by Samarina & Bezemer (2016) paper. Applying this technique to the different capital flows types individually and jointly I estimate their interaction with the main variable of interest. As noted by Samarina & Bezemer (2016) the inclusion of the total credit, a proxy for the initial level of financial development has the potential to be an endogenous regressor. This would imply that E(CRD0it
ε
it) ≠ 0, e.g. the correlation between the error term of totalcredit and independent variable tends to move together. Thus, I test my model with and without
this variable and observe no substantial change in the efficiency of the point estimate, F-test and the R2. Appendix B presents the results of the model including this variable and Table 3
presents the subsequent model excluding this variable, which will be the main results I use for my analysis.
Table 3 – The effect of Capital Inflows on Domestic Credit Allocation, Fixed-Effects Model
V VI VII VIII
Direct Investment (% of GDP) (lag) 0.376*** 0.383***
(0.054) (0.038)
Portfolio Investment (%GDP) (lag) -0.081* -0.177***
(0.042) (0.064)
Other Investments (%GDP) (lag) 0.051 0.047
(0.049) (0.067)
GDP per capita (log) -0.078** -0.079** -0.078** -0.087**
(0.036) (0.038) (0.035) (0.038)
GDP growth (annual %) 0.005*** 0.005** 0.005*** 0.005***
(0.002) (0.002) (0.002) (0.002)
Credit market regulations -0.000 -0.001 -0.001 -0.001
(0.009) (0.009) (0.008) (0.009) Bank Leverage (%) -0.000 -0.000 -0.000 -0.000 (0.000) (0.000) (0.000) (0.000) Europe -0.007 -0.015 -0.020 0.016 (0.099) (0.104) (0.095) (0.104) Middle/South-America 0.208*** 0.194*** 0.183*** 0.202*** (0.036) (0.039) (0.035) (0.039) Middle-East 0.259*** 0.250*** 0.241*** 0.264*** (0.039) (0.037) (0.038) (0.046) North-America -0.245*** -0.248*** -0.300*** -0.107 (0.089) (0.090) (0.085) (0.096) Oceania 0.075 0.051 0.075 0.081 (0.071) (0.072) (0.069) (0.078)
Europe x Direct Investment (% of GDP) (lag) -0.377*** -0.381***
(0.054) (0.037)
Middle/South-America x Direct Investment (% of GDP) (lag) -0.456 -0.442
(0.452) (0.445)
Middle-East x Direct Investment (% of GDP) (lag) -0.831 -0.705
(0.709) (1.173)
North-America x Direct Investment (% of GDP) (lag) -0.794*** -1.922***
(0.116) (0.184)
Oceania x Direct Investment (% of GDP) (lag) -0.030 -0.089
(0.122) (0.428)
Europe x Portfolio Investment (%GDP) (lag) 0.091** 0.184***
(0.043) (0.064)
Middle/South-America x Portfolio Investment (%GDP) (lag) -0.150 0.291
(0.871) (0.806)
Middle-East x Portfolio Investment (%GDP) (lag) -0.606* -0.480*
(0.360) (0.257)
North-America x Portfolio Investment (%GDP) (lag) -1.222*** -3.548***
(0.312) (0.513)
Oceania x Portfolio Investment (%GDP) (lag) 0.554*** 0.569**
Table 3 – (Continued)
Europe x Other Investments (%GDP) (lag) -0.025 -0.020
(0.050) (0.067)
Middle/South-America x Other Investments (%GDP) (lag) 0.968*** 1.012***
(0.192) (0.361)
Middle-East x Other Investments (%GDP) (lag) -0.118 0.125
(0.391) (0.171)
North-America x Other Investments (%GDP) (lag) 2.101*** -0.929
(0.409) (0.656)
Oceania x Other Investments (%GDP) (lag) -0.691 -0.431
(0.749) (1.056) Constant 1.147*** 1.169*** 1.168*** 1.229*** (0.278) (0.292) (0.270) (0.290) Observations 586 575 586 575 Number of country 39 38 39 38 R-squared Within 0.478 0.479 0.481 0.508 Between 1.000 1.000 1.000 1.000 Overall 0.901 0.894 0.902 0.903
Country FE YES YES YES YES
Year FE YES YES YES YES
Notes: The dependent variable is the share of domestic bank credit to the non-financial sector in all bank credit. The Table reports coefficient estimates with F-test in parentheses, robust standard error clustered by country. ***p < 0.01, **p < 0.05, *p < 0.1. Constant, regional- and time (not shown) dummies. Source: Author’s calculation
5.1 Empirical results 5.1.1 Capital Inflow effects
employ different datasets causing divergence in results. The third reason might be, not taking accounting of the sector in which the capital flows. Samarina & Bezemer (2016) make a distinction between capital flowing into the bank sector and non-bank sector, which is not considered within my research.
In column VIII the capital inflows are tested jointly, here the predictive power of the control variables and regional dummies remain mostly unchained but portfolio investments gains predictive power becoming significant at a 1% level. Calderón and Kubota (2012) find contrasting results to mine with respect to Portfolio Investment inflows, where they found them to have a positive impact on credit booms. According to the paper presented by Furceri, Guichard and Rusticelli (2012) which investigates the relationship between capital movement and domestic credit, these results are driven by the characteristic of the inflow. Capital which is more debt-like will have the most influence on domestic credit. In comparing my results to those in the existing literature I find that jointly, Direct investment and Portfolio Investment both have a significant effect on the substitution effect of bank credit towards businesses. A result which is partially supported by Samarina & Bezemer (2016). Nonetheless these results can be interpreted as follows, a one-unit increase (capital inflow %GDP) of Direct Investment, holding all else equal will increase bank credit allocation to the Non-Financial Businesses by 0,383%. Although the magnitude of this allocation is not large it does present a possible roadmap for policymakers in restricting/ loosening capital flow.
5.1.2 Regional effects
effect is however mitigated when countries have significant capital account restrictions (Davis et al., 2016). These restrictive measures can be an explanation for the significant and non-significant results found in the models. Nonetheless these results can be interpreted in the following way, bank credit to the private sector has the propensity to be allocated 0,202% more to Non-Financial Businesses for countries located in Middle/ South-America than those located in Asia6.
5.1.3 Regional Capital flow effects
In answering the research question (“What are the regional effects of capital inflow on
domestic credit allocation?”), it is vital to compute the joint effects of regional idiosyncrasies
and the heterogeneity of the different types of capital inflows. Interacting these two variables a pattern can be observed with which we can answer the question. Although not significant for every region and credit inflow type interacted the results indicate the presence of heterogeneous patterns in credit allocation for the capital inflow type and regions interacted. When Direct Investment capital flows into the economy for Europe and North America, a statistically significant negative effect is found when comparing to Asia (Table 3). The results can be interpreted as flowing, inflow of Direct investment in the model increases the proportion of bank credit allocation to Non-Financial business (0,383), which has the propensity to being lower in North America than Asia (-0,107 – not significant). Thus, holding all else constant, a one unit increase in the inflow of Direct Investment capital into North-America would further decrease the bank credit allocation to Non-Financial businesses by 1,922%. Statistically significant results along the same line is found in the full model (VIII) for Europe and North America at a 1% level of significance when the inflow is Direct Investment and Portfolio Investment. For Middle/ South-America a positive coefficient also at a 1% level of significance when the inflow type is Other Investment.
5.2 Robust Analysis
My first attempt at dealing with the endogeneity problem as mentioned before is to drop the Total credit variable and to lag the different Inflow types in order to mitigate concerns of endogenous regressor and reverse causality, respectively. To further control for the possibility of endogeneity, I employ the generalized method of moments (GMM) model. As it is possible that the allocation of bank credit to businesses is an endogenous choice, bringing about
6 Asia is not included as a regional dummy variable and thus serves as the base for comparison against all other
concerns of reverse causality. Because it is often difficult to get valid instruments to remove endogeneity bias statistically, the GMM estimator is an appropriate alternative approach because it generates instruments using both lagged dependent and explanatory variables (Zamore, Beisland and Mersland, 2019). This dynamic approach has been implements in many of the researched surveyed hitherto Samarina and Bezemer, (2016); Bezemer, Samarina and Zhangb (2017) ;Gozgor (2018) and Kaat (2016) and will be implemented to test the robustness
of the results found in Table 3.
Arellano and Bond (1991) suggest using the GMM difference estimation when employing the lagged dependent variable in estimations. This method removes the fixed effects by transforming the lagged values within the data in order to deal with endogeneity concerns. The difference GMM estimation is inefficient however when the variables are close to a random walk and makes difference GMM subject to sample bias when the time period is small (Roodman, 2009). Hence, I followed Roodman (2009) and Samarina & Bezemer (2016) in using the System GMM which improved efficiency by lagging difference and levels in the estimation.
Roodman (2009) provides a practical roadmap to implementing the system GMM by way of xtabond2 command in statistical programming. He cautions however that utilizing this command can weaken post estimations tests (Hauseman test) through creating a large number of instruments, causing false positives. I thus proceed in my analysis by collapsing the instrument matrix in order to limit the number of instruments in the model. I further proceed by applying a one-step system GMM as the two-step procedure offers little efficiency gais (Roodman, 2009). The lagged dependent variable is treated as predetermined, and all the control variables are assumed to be endogenous, except for the interaction terms, the regional and year dummies (Jha, 2019 ). The System GMM requires two specific tests to determine its efficiency: 1) the autocorrelation test [AR (1) and AR (2)] and 2) the test for overidentification restrictions [Hansen test] (Roodman, 2009). The former tests for second-order autocorrelation in the residuals from difference equations, while the latter tests the validity of the instrument set. If for both tests the p-value value is larger than 0.05, the null hypothesis can be rejected. This would indicate that you have no second-order autocorrelation and your instruments are exogenous and can therefore be used as instruments (Roodman, 2009).
the signs as observed in Table 3. In the full model only the dummy variable for Europe gains significance. The results observed in Table 4 indicate that the Fixed-Effects estimates should be interpreted with caution.
Table 4 – Extended model on the effect of Capital Inflows on Domestic Credit Allocation, System GMM Model
IX X XI XII
Direct Investment (% of GDP) (lag) 0.041 0.002
(0.026) (0.003)
Portfolio Investment (%GDP) (lag) 0.008 -0.004
(0.061) (0.008)
Other Investments (%GDP) (lag) 0.174 -0.009
(0.115) (0.020)
GDP per capita (log) -0.344** -0.088** -0.077** 0.004
(0.153) (0.036) (0.031) (0.204)
GDP growth (annual %) 0.014 0.017*** 0.028*** 0.015
(0.011) (0.006) (0.008) (0.016)
Credit market regulations 0.000 -0.031 -0.075*** 0.040
(0.049) (0.025) (0.027) (0.084) Bank Leverage (%) 0.000 0.000 0.000 0.001 (0.001) (0.000) (0.000) (0.001) Europe -0.744 -0.760 -0.887 -0.112*** (0.949) (1.064) (1.306) (0.034) Middle/South-America -1.076 -1.938 -0.972 -0.087 (1.438) (1.394) (1.295) (0.067) Middle-East 1.323 -0.708 0.861 0.137 (2.025) (3.096) (2.460) (0.117) North-America -3.155 -0.641 -3.820 -1.170 (3.626) (3.348) (5.423) (0.910) Oceania 0.735 -0.036 0.358 0.132 (1.164) (1.391) (1.666) (0.141)
Europe x Portfolio Investment (%GDP) (lag) 0.021 0.003
(0.024) (0.006)
Middle/South- x Portfolio Investment (%GDP) (lag) 1.672 0.098
(5.084) (0.976)
Middle-East x Portfolio Investment (%GDP) (lag) -15.921 -4.651
(16.148) (2.820)
North-America x Portfolio Investment (%GDP) (lag) 19.996 -4.322
(33.870) (7.387)
Oceania x Portfolio Investment (%GDP) (lag) -16.750 -7.206
(15.419) (4.725)
Europe x Portfolio Investment (%GDP) (lag) 0.090 0.003
(0.115) (0.004)
Middle/South- x Portfolio Investment (%GDP) (lag) -5.866 -1.201
(7.388) (0.886)
Middle-East x Portfolio Investment (%GDP) (lag) 14.474 -2.217
(22.930) (1.669)
North-America x Portfolio Investment (%GDP) (lag) 14.291 33.407
(29.243) (30.577)
Oceania x Portfolio Investment (%GDP) (lag) 0.571 -3.633
Table 4 – (Continued)
Europe x Other Investments (%GDP) (lag) -0.116 0.028
(0.128) (0.034)
Middle/South-America x Other Investments (%GDP) (lag) -6.227 -0.223
(6.838) (0.832)
Middle-East x Other Investments (%GDP) (lag) -11.687 -1.745
(15.227) (1.496)
North-America x Other Investments (%GDP) (lag) 71.965 27.953
(89.752) (29.714)
Oceania x Other Investments (%GDP) (lag) -7.980 2.570
(12.969) (2.507) Constant 95.989* 122.006** 119.304** 251.957*** (50.638) (52.600) (49.887) (31.648) Observations 596 586 596 586 Number of country 41 40 41 40
Hansen test p-value 0.461 0.214 0.464 0.00
AR(1) test p-value 0.128 0.341 0.058 0.551
AR(2) test p-value 0.164 0.940 0.884 0.902
6. Conclusion
6.1 ConclusionAs stated by Samarina and Bezemer (2016), since the 1990s domestic bank credit has been reallocated away from lending to non-financial business. This reallocation to more household credit causes lower private saving, slower economic growth, asset price increase, larger external imbalances (negative current account) and increased probability of crisis with a deeper and longer recession (Arcand, Berkes, and Panizza, 2012). A catalyst of this reallocation is capital inflow according to (Samarina and Bezemer, 2016). Capital inflows themselves also make domestic economies susceptible to external shock if it’s not allocated efficiently (OECD, 2011). Thus, these issues are important to deal with from an economic stability, growth and inequality perspective.
As indicated by OECD (2011) decisions regarding capital movement should not be taken unilaterally as countries negatively affected by such actions may seek retaliation. Thus, they suggest international cooperation in such respect. To my knowledge my paper will thus be the first to tackle this issue. Looking at the effects of capital movement on domestic credit allocation from a regional perspective. In particular my paper adds the nexus of the substitution effect of capital inflow and the different types of inflows along regional lines. This research will thus be aimed to answer the question: “What are the regional effects of capital inflow on
domestic credit allocation?”.
In conclusion fixed-effects estimates indicate that across regions diverging patterns can be observed especially for credit allocation behaviour of banks, especially for Europe, North America and Middle/South America, dependent on the inflow type. These results are not robust to System GMM estimates however and should be observed with caution.
6.2 Limitations and future research
Despite efforts to cub them this research has some limitations which should be taken into account. Most of the limitations arise from the data availability. Data on Credit in the BIS database for the region of Africa is limited, only being available for the country South Africa, which caused me to drop this region from my analysis. Furthermore, credit provided by banks is reported at an aggregate level in the BIS. This will lead researchers to make assumptions which can be erroneous.
7. Reference
Aizenman, J., Jinjarak, Y., & Park, D. (2013). Capital Flows and Economic Growth in the Era of Financial Integration and Crisis, 1990–2010. Open Economies Review volume, 371–396.
Arcand, J.-L., Berkes, E., & Panizza, U. (2012). Too Much Finance? IMF Working Paper. International Monetary Fund.
Areliano, M., & Bover, O. (1995). Another look at the instrumental variable estimation of error-components models. Journal of Econometrics, 29-51.
Arellano, M., & Bond, S. (1991). Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations. The Review of Economic
Studies, 277-297.
Bezemer, D. (2019). Money and its uses development of wealth.
Bezemer, D., Samarina, A., & Zhang, L. (2017). The shift in bank credit allocation: new data and new findings. DNB Working Papers(Working Paper No. 559), 1-59.
Bezemer, D., Samarina, A., & Zhang, L. (2020). Does mortgage lending impact business credit? Evidence from a new disaggregated bank credit data set. Journal of Banking
and Finance, 2-24.
BIS. (2020). Credit to the non-financial sector. Retrieved from
https://www.bis.org/statistics/totcredit.htm?m=6%7C380%7C669
Bloomberg. (2020, May 22). Next China: Ready to Retaliate. Retrieved May 2020, from Bloomberg News: https://www.bloomberg.com/news/newsletters/2020-05-22/next-china-ready-to-retaliate
Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 115—143.
Borio, C., & Disyatat, P. (2011). Global Imbalances and the Financial Crisis: Link or No Link? BIS Working Paper No. 346.
Brunnermeier, M. K. (2009). Deciphering the Liquidity and Credit Crunch 2007–2008 .
Journal of Economic Perspectives, 77-100.
Calderón, C., & Kubota, M. (2012). Gross Inflows Gone Wild: Gross Capital Inflows, Credit Booms and Crises. Finance and Private Sector Development. The World Bank, Office of the Chief Economist.
Cuestas, J. C., & Staehr, K. (2017). The Great Leveraging in the European crisis countries Domestic credit and net foreign liabilities The Great Leveraging . Journal of
Economic Studies, 895-910.
Cypher, J. M., & Dietz, J. L. (2009). Macroeconomic Equilibrium: The External Balance. In J. M. Cypher, The Process of Economic Development (pp. 497-499). London:
Routledge.
Davis, J. S., Mack, A., Phoa, W., & Vandenabeele, A. (2016). Credit booms, banking crises, and the current account. Journal of International Money and Finance, 360-377. Enya, M., & Shinkai, J.-i. (2014). The Impact of Capital Inflows on Asset Prices in East Asia.
Faculty of Economics and Management, Institute of Human and Social Sciences.
Kanazawa University.
Enya, M., & Shinkai, J.-i. (2018). Capital Inflows, Asset Prices, and Financial Systems in East Asia. Faculty of Economics and Management, Institute if Human and Social
Sciences. Kanazawa University.
Furceri, D., Guichard, S., & Rusticelli, E. (2012). The effect of episodes of large capital inflows on domestic credit. North American Journal of Economics and Finance, 325-344.
Gabor, D., & Vestergaard, J. (2016). Towards a theory of shadow money. Institute for New
Gourinchas, P.-O. (2012). Global Imbalances and Global Liquidity. Asia Economic Policy
Conference . Federal Reserve Bank of San Francisco.
Gozgor, G. (2018). Determinants of the domestic credits in developing economies: The role of political risks. Research in International Business and Finance, 430-443.
Grauwe, P. d. (2018). Economics of Monetary Union. New York : Oxford University Press. Gumata, N., & Ndou, E. (2019). Bank and Non-bank Capital Flows and The Sectorial
Reallocation of Credit Away from the Household Sector . In N. Gumata, & E. Ndou,
Capital Flows, Credit Markets and Growth in South Africa (pp. 195-205). Cham:
Palgrave Macmillan.
Gupta, P. (2016). Capital Flows and Central Banking: The Indian Experience. Policy
Research Working Paper 7569, pp. 1-46.
Gwartney, J., Lawson, R., & Hall, J. (2017). Economic Freedom of the World 2017 Annual
Report. Vancouver, B.C.: Fraser Institute.
Hardie, I., & Howarth, D. (2013). Framing Market-Based Banking and the Financial Crisis.
Market-based banking and the international financial crisis, 22-55
Igan, D., & Tan, Z. (2015 ). Capital Inflows, Credit Growth, and Financial Systems. IMF
Working Paper. International Monetary Fund.
International Monetary Fund . (2014). BPM6 CoMPilation Guide - the Balance of Payments
and InternatIonal Investment PosItIon manual. Washington, D.C. : Joint Bank-Fund
Library.
International Monetary Fund. (n.d.). What is the Central Bank policy rate? Retrieved from Data Help IMF: http://datahelp.imf.org/knowledgebase/articles/484375-what-is-the-central-bank-policy-rate
Jha, C. K. (2019 ). Financial reforms and corruption: Evidence using GMM estimation.
International Review of Economics and Finance, 66-78.
Kaat, D. M. (2016). International Capital Flows and the Allocation of Credit Across Firms.
Verein für Socialpolitik / German Economic Association.
King, R., & Levine, R. (1993). Finance and Growth: Schumpeter might be right. The
Quarterly Journal of Economics, 108, 717-737.
Lane, P. R., & McQuade, P. (2014). Domestic Credit Growth and International Capital Flows. The Scandinavian Journal of Economics, 218-252.
Meslier, C., Sauviat, A., & Yuan, D. (2020). Comparative advantages of regional versus national banks in alleviating SME’s financial constraints. Journal Pre-proof. Obstfeld, M. (2012). Does the Current Account Still Matter? American Economic Review,
102 (3), 1-23.
OECD. (2011, June). International capital flows: Structural reforms and experience with the
OECD Code of Liberalisation of Capital Movements. Retrieved May 2020, from
OECD: http://www.oecd.org/economy/48972216.pdf
Piketty, T. (2014). Capital in the 21st Century. Cambridge Massachusetts: The Belknap Press of Harvard University Press.
Presbitero, A. F., & Rabellotti, R. (2016). The Determinants of Firm Access to Credit in Latin America: Micro Characteristics and Market Structure. Economic Notes by Banca
Monte dei Paschi di Siena SpA, 445–472.
Rodrik, D., & Subramanian, A. (2009). Why did financial globalization disappoint? IMF Staff
Paper. 56, pp. 112–138.
Roodman, D. (2009). How to do xtabond2: An Introduction to Difference and System GMM in Stata. Stata Journal, 86–136.
Samarina, A., & Bezemer, D. (2016). Do capital flows change domestic credit allocation?
Journal of International Money and Finance, 98-121.
Storper, M. (2018). Separate worlds? Explaining the current wave of regional economic polarization. Journal of Economic Geography, 18, 247–270.
Sugimoto, K., & Enya, M. (2017). Global Liquidity and Reallocation of Domestic Credit.
Risk Management of Global Liquidity (pp. 1-25). Modern Monetary Economics
Summer Institute.
The World Bank. (n.d.). Metadata Glossary. Retrieved from The World Bank - DataBank : https://databank.worldbank.org/metadataglossary/all/series
Torres-Reyna, O. (2007, Dec). Panel Data Analysis Fixed and Random Effects using Stata. Retrieved from Pinceton University - Data & Statistical Services:
https://dss.princeton.edu/training/
Yaltaa, A. T., & Yaltab, A. Y. (2018). Are credit rating agencies regionally biased? Economic
Systems, 682-694.
Appendix A – Test for multicollinearity
Variable VIF 1/VIF
Portfolio Investment 4.16 0.240197 Direct Investment 3.34 0.299532 GDP per capita 3.07 0.325910 Total Credit 2.49 0.402124 Other Investments 1.61 0.619814
Credit market regulations 1.44 0.694115
GDP growth 1.21 0.827812
Bank Leverage 1.16 0.859686
Mean VIF 2.31
Appendix B – Fixed Effects Model
I II III IV
Direct Investment (% of GDP) (lag) 0.445*** 0.450***
(0.092) (0.084)
Portfolio Investment (%GDP) (lag) -0.108* -0.228***
(0.061) (0.075)
Other Investments (%GDP) (lag) 0.051 0.050
(0.051) (0.074)
Total Credit (%GDP) -0.000 -0.000 -0.000 -0.000
(0.000) (0.000) (0.000) (0.000)
GDP per capita (log) -0.101** -0.094** -0.088** -0.107**
(0.044) (0.046) (0.044) (0.045)
GDP growth (annual %) 0.004** 0.004** 0.004** 0.004***
(0.002) (0.002) (0.002) (0.002)
Credit market regulations -0.002 -0.002 -0.002 -0.003
(0.009) (0.009) (0.009) (0.009) Bank Leverage (%) 0.000 0.000 -0.000 0.000 (0.000) (0.000) (0.000) (0.000) Europe 0.099 0.053 0.030 0.111 (0.147) (0.158) (0.152) (0.151) Middle/South-America 0.213*** 0.193*** 0.184*** 0.204*** (0.036) (0.037) (0.035) (0.038) Middle-East 0.264*** 0.250*** 0.241*** 0.265*** (0.041) (0.036) (0.039) (0.047) North-America -0.146 -0.184 -0.253* -0.018 (0.134) (0.142) (0.140) (0.140) Oceania 0.143 0.095 0.106 0.140 (0.096) (0.101) (0.100) (0.104)
Europe x Direct Investment (% of GDP) (lag) -0.443*** -0.446***
(0.091) (0.082)
Middle/South-America x Direct Investment (% of GDP) (lag) -0.535 -0.512
(0.483) (0.489)
Middle-East x Direct Investment (% of GDP) (lag) -0.871 -0.826
(0.718) (1.167)
North-America x Direct Investment (% of GDP) (lag) -0.857*** -1.954***
(0.144) (0.198)
Oceania x Direct Investment (% of GDP) (lag) -0.056 -0.098
(0.132) (0.412)
Europe x Portfolio Investment (%GDP) (lag) 0.118* 0.234***
(0.063) (0.076)
Middle/South- x Portfolio Investment (%GDP) (lag) -0.269 0.125
(0.896) (0.842)
Middle-East x Portfolio Investment (%GDP) (lag) -0.518 -0.367
(0.385) (0.276)
North-America x Portfolio Investment (%GDP) (lag) -1.103*** -3.420***
(0.344) (0.515)
Oceania x Portfolio Investment (%GDP) (lag) 0.574*** 0.639**
(0.140) (0.287)
(Continued)
Europe x Other Investments (%GDP) (lag) -0.027 -0.025
(0.050) (0.073)
Middle/South-America x Other Investments (%GDP) (lag) 0.967*** 0.925***
(0.212) (0.340)
Middle-East x Other Investments (%GDP) (lag) -0.087 0.202
(0.440) (0.208)
North-America x Other Investments (%GDP) (lag) 2.016*** -1.051
(0.431) (0.655)
Oceania x Other Investments (%GDP) (lag) -0.657 -0.339
(0.743) (1.049) Constant 1.371*** 1.318*** 1.274*** 1.429*** (0.354) (0.387) (0.371) (0.368) Observations 586 575 586 575 Number of country 39 38 39 38 R-squared Within 0.486 0.483 0.483 0.514 Between 1.000 1.000 1.000 1.000 Overall 0.902 0.895 0.902 0.902
Country FE YES YES YES YES
Year FE YES YES YES YES
