Foreign Banks and the Bank Lending Channel

We provide new evidence on the effects of monetary policy on loans using bank-level data on 429 banks in CEE economies between 1998 and 2012. Only domestic banks adjust their loans to changes in monetary policy. This is driven by the supply side as deposits in foreign banks do not react to monetary policy, hence the bank lending channel is only active in domestic banks. Contrary to conventional wisdom, foreign banks do not have to rely on parent banks’ funding to insulate against monetary policy shocks. Indeed, we find parent bank and its country characteristics are irrelevant for subsidiary’s lending.


Introduction
Financial liberalization has led to an increased integration of financial markets over the last 30 years. The emerging and developing countries, however, entered this process with under-capitalized banks. In result, large shares of the financial sector in these countries are controlled by subsidiaries of foreign banks. In this paper we investigate the implications of foreign ownership for the monetary policy transmission through bank loans.
On the one hand, monetary policy may influence the supply of loans in line with the bank-lending channel hypothesis. It postulates that a monetary policy contraction shrinks both banks' reserves and deposits. If there is imperfect substitutability between different types of bank assets, the supply of loans is reduced. On the other hand, there is the conventional interest rate channel under which monetary tightening adversely affects the demand for credit. After a monetary policy tightening a higher real interest rate and decreased output growth expectations result in a weaker demand for bank credit. A relative decrease in bank lending occurs only from the demand side, without any change in the supply side. While regulation can influence supply of credit, imposing constraints on demand is a difficult task. Therefore, identifying which channels interest rate affects credit through and whether the transmission works analogously in foreign and domestic banks is an important policy question.
We explore the consequences of an asymmetric financial integration in the particular area of the Central and Eastern Europe (CEE). Banks dominate the financial structure of the CEE economies and the most of these banks are majority foreign-owned following a period of rapid increase in foreign penetration of the banking sector in the early and mid 1990s. We collect data on credit, deposits, ownership and other bank-level characteristics for 429 banks in ten CEE countries in the years 1998-2012. 1 We make three contributions to the empirical literature on the monetary policy transmission via banks.
First, we show that lending contracts (expands) after local monetary tightening (loosening) only in domestic banks, while lending in foreign banks does not. Second, we investigate whether this difference is demand or supply driven. We show the response of insured deposits to monetary policy is absent in foreign banks.
They rely less on insured deposits and enjoy lower cost of financing. Hence, we establish that the difference in loans behavior derives from the difference in the bank lending channel. Third, we explicitly account for characteristics of parent-banks and find that they do not matter for the dynamics of credit of their subsidiaries. We also show that the impact of monetary policy on loans (and their different behavior in domestic and foreign banks) is only present in tranquil times, being insignificant during the global financial crisis of 2008-10. The results stand out as a robust finding after a battery of additional tests.
The research on the bank lending channel was pioneered by Bernanke and Blinder (1992) on aggregate and Kashyap and Stein (2000) on bank-level data. The relationship between bank ownership and the growth of credit has been receiving an increased interest in the literature since Peek and Rosengren (1997), who show that Japanese-owned banks in the US contracted their lending in a response to the slump in the Japanese stock market.
The two most closely related papers to our study are Jimenez, Ongena, Peydro, and Saurina (2012) and Wu, Luca, and Jeon (2011). The former undertakes a laborious exercise to merge credit applications data with the bank balance sheet data to distinguish between demand and supply effects. We investigate whether the difference in credit response to monetary policy between domestic and foreign banks is demand or supply driven. A demand channel would imply differences in the composition of portfolio of loans and lower variation of credit in foreign banks in reaction to GDP and inflation. A supply driven channel would imply differences in composition and costs of funding. Our data confirm a supply-driven credit channel. 2 The latter provides evidence of a weaker response of lending to monetary policy in foreign banks and conclude that it is due to their access to an internal capital market. This hypothesis proposes that a foreign bank can easily obtain surplus liquidity from either the parent bank or other subsidiaries within the financial conglomerate (Campello, 2002, de Haas andvan Lelyveld, 2010). Foreign banks operations are thus less dependent on macro conditions in the host country and more dependent on macro conditions in the country of a parent bank. In Wu et al. (2011) sample (Central and Eastern Europe, Latin America and South-East Asia economies) this effect operates during local banking crises and not during tranquil times. In our sample we find somewhat different results, that monetary policy only affects bank lending in tranquil times, that it 2 It is worth noting that bank credit may also react to monetary policy via a "balance sheet channel". According to this channel, an increase in interest rates affects firms' creditworthiness negatively. A drop in banks' credit after a monetary tightening is then independent of credit demand (interest rate channel) and credit supply (bank lending channel). Although this channel is regarded as a supply-side channel, it can only be detected looking at firms' balance sheets, as in the cited study. affects only domestic and not foreign banks and that parent bank and parent bank country characteristics are irrelevant for lending of a subsidiary.
The CEE transition economies are a natural field for empirical studies of foreign banks behavior, as they exhibit variation in foreign banks penetration both in cross-section and over time. As of 2009 the share of foreign banks in the total assets of the banking sectors in the CEE economies was greater than 80 percent. In other European Union member states this number stood at 25 percent (own calculations based on Claessens and Van Horen (2014)). Foreign banks in CEE economies are found to be more cost-effective (Bonin, Hasan, andWachtel, 2005, Fries andTaci, 2005) and more profitable (Claessens, Demirguc-Kunt, andHuizinga, 2001, Havrylchyk andJurzyk, 2011). We find that after controlling for bank-level characteristics, the effect of foreign ownership is still significant for the response of credit to monetary policy. De Haas and van Lelyveld (2006) find a positive relationship between foreign banks and the private sector credit growth and that conditions in the parent bank country matter for foreign banks' growth of credit. Aydin (2008) further confirms that credit growth is higher in foreign banks, but that conditions in the parent bank country do not matter for the growth of credit.
In the literature investigating the role of foreign banks during crises it is found (Allen, Jackowicz, Kowalewski, and Koz lowski (2017), De Haas and van Lelyveld (2006), Popov and Udell (2012)) that they stabilize credit in times of a crisis. Ongena, Peydro, and van Horen (2015) use matched bank-firm data in CEE and Turkey to explore the consequences of the Lehman failure. Adams-Kane, Caballero, and Lim (2017) use global sample with matched owner-subsidiary data to study responses of foreign banks to home country crises. We complement those studies by examining the role of monetary policy during a global crisis.
The remainder of the paper is organized as follows. Section 2 lays out the data sources and the empirical procedure. We present the first (benchmark) result and robustness checks in Section 3. Section 4 disentangles demand from supply channels and studies the role of parent banks. Last section concludes.

Data and Methodology
We use bank-level balance sheet and macroeconomic data. We acquired bank-level data from Bankscope, which comprises a large number of standardized, comparable indicators at annual frequency and has been used extensively in the related literature. We use unconsolidated data on banks in ten CEE countries.
In Section 3 we investigate whether bank credit in foreign and domestic banks reacts differently to changes in monetary policy. Our dependent variable ∆Net Loans is thus the real percentage growth of net loans.
The first key independent variable, ∆MP, is the yearly difference of the nominal average central bank repo rate in a host country. It captures changes in monetary policy stance in the host country. The data was sourced from the Eurostat and host country central banks' websites. Other key explanatory variables are the ownership dummies. The identification of type of ownership required some laborious data collection as Bankscope only provides information on the bank owner in the most recent year. On top of Bankscope we used ownership data provided by Claessens and Van Horen (2014). For the banks not covered there we had to resort to individual banks' websites and financial reports to uncover any changes in ownership.
We distinguish three ownership categories: foreign, private-owned domestic and state-owned domestic banks. Although the focus of the paper is on the foreign-domestic distinction, we include state ownership for the sake of consistency with previous literature that found government banks submit to political pressures around election years or cater different clients (see eg. Micco and Panizza (2006)). The foreign ownership dummy Foreign takes value 1 if at least 50% of bank capital is owned by foreign entities. The state ownership dummy State is defined analogously based on the share of domestic state-related entities. These variables capture potentially different management practices, business objectives, know-how and an ease in accessing additional sources of capital and are, by construction, mutually exclusive. Whenever we refer to domestic banks we mean state domestic and private domestic banks together. When state and foreign ownership are controlled for, we call the reference group private domestic banks. We introduce four bank-level control variables. Size measures bank's prominence in the host country banking sector as a share of bank's total assets in all banks' assets in a given country in a given year.
Profitability is defined as a ratio of operating profit over total assets, Capitalization as a ratio of total equity over total assets and Liquidity as a ratio of liquid assets over total assets. All original variables are denominated in local currencies. We incorporate two macroeconomic controls: the real GDP growth rate to control for cyclical variation in bank balance sheets and proxy for sector-wide demand effects and Inflation measured by percentage growth of the consumer price index. These bank-level and macro controls constitute a standard set used in literature on bank-level credit. 3 We account for non-standard circumstances of the global financial crisis by introducing a time dummy Crisis that equals 1 for years 2008-2010. Next, to control for valuation effects (of foreign currency denominated loans) due to exchange rates volatility, we define ∆EUR and ∆CHF as yearly changes in the average exchange rate against the euro and the Swiss franc. Finally, some countries in our sample gave up their monetary policy autonomy in pursuit of adopting the euro. Therefore, we introduce a dummy variable No independent MP that equals one if for the majority of days in the year the country either used the euro or had its currency pegged. In those countries we can rule out any feedback between bank credit and monetary policy conduct. The details of construction of all variables are relegated to the Appendix 6.1.
The outcome of the data collection process is a panel of 3951 bank-year observations for 429 banks after removal of outliers. We construct analogous bank-level characteristics for parent banks of foreign banks identifying 95 parent banks yielding 1861 bank-year observations on subsidiary level. This is a reasonably high number, given that not every bank in our sample has a foreign owner and many banks have the same owner, dispersed ownership or non-bank owners.
In Section 4 we assess the competing hypotheses about the drivers of different credit responses in foreign and domestic banks. On the one hand, credit supply might be different. Domestic banks can enjoy competitive edge through better information, while foreign banks could benefit from better risk management and screening, access to modern technologies (Berger et al., 2000, Havrylchyk andJurzyk, 2011). Domestic and foreign banks may differ in their portfolio of loans, the structure of funding and its costs. On the other hand, credit demand might be different. Foreign and domestic banks can service different customers, e.g. foreign banks may profit from customer relationships of their parent institutions. Therefore we construct additional variables to study the behavior of deposits, portfolio and funding compositions and funding costs.
∆Consumer Deposits (∆Bank Deposits) is a real percentage growth of net consumer (bank) deposits used to investigate differentials in the dynamics of bank liabilities. Next, to investigate differences in funding structure and funding costs we construct variables Wholesale funding and Retail funding as ratios of wholesale and retail funding to bank's total assets, Interest Expenses as a ratio of total interest expenses to total assets. Finally, to investigate differences in loan portfolio structure we define Loans to Banks, and Commercial Loans as ratios of the corresponding balance sheet variables to bank's total assets.
In Table 9 we report unconditional correlations between bank-level variables in foreign, private domestic and state banks. Our final sample with identified ownership covers on average 97.25% of the volume of net loans reported in Bankscope. Unconditional averages are reported in Table 10 and tests of differences among ownership categories are reported in Table 11. Foreign banks are on average larger, less liquid and have lower capitalization than private domestic banks. State banks are the largest and have the largest share of liquid assets. Foreign banks rely more on wholesale funding and less on retail funding than domestic banks.
The growth of credit and deposits is slower in foreign banks.

Monetary Policy, Ownership and Bank Credit
In this section we estimate different versions of the following regression using data on bank i in country j in year t: where Owner i,t is a vector of ownership dummies, Bank it are the bank-level controls and Economy j,t are macro controls. It is well recognized (Adams-Kane et al., 2017, Claessens and Van Horen, 2014, Gambacorta, 2005, Wu et al., 2011 that the presence of bank-specific controls and lagged dependent variable induces an endogeneity problem. Thus, our method of choice is the system-GMM approach based on Arellano and Bond (1991). In this estimation we allow the dependent variable to be potentially auto-correlated and contemporary bank controls to be endogenous. However, as system-GMM approach often falls into the excessive use of instruments problem, for each regression we report the Hansen test to assess the validity of instruments.

Cross Validation
We start by building intuition about effects of bank-level characteristics on credit growth. We estimate equation (1) without the monetary policy variable ∆MP to cross-validate our data with other studies that focused on CEE countries but abstracted from monetary policy. The results of the estimation have been relegated to Table 12 in the Appendix. We formally test for auto-correlation of the dependent variable. We reject the null that AR(1) coefficient is equal to zero, but fail to reject the null that the AR(2) coefficient is equal to zero. This validates our specification with one lag dependent variable and the use of a system-GMM estimation. Credit in larger, more liquid and less profitable banks grows at a slower pace. We don't find significant effects of bank capitalization. Inflation has a negative, and GDP growth has a positive impact on credit growth. We don't find different trends in credit growth among foreign, state and domestic banks as in Allen et al. (2017)

Benchmark Results
We estimate the model of the real rate of growth of loans in (1). If the regression coefficient β ∆MP is significant and negative we can expect the bank lending channel to be at work. We introduce interactions of monetary policy instrument with foreign and government ownership dummies to see if credit growth trends are heterogenous. We distinguish responses to monetary policy in tranquil times and during the global financial crisis. Table 1 presents results of the benchmark model.
In column (1) we find that a 1 percentage point increase in the monetary policy rate slows down the growth of loans at the bank level by 3.16 percentage points. This effect is weaker in foreign banks, as β FGN ×∆MP is significantly larger than zero. Given that the estimates of β ∆MP and β FGN×∆MP are close in absolute term we run a Wald test to check whether a net reaction of foreign banks credit is significantly different from zero. The null is that β ∆MP + β FGN×∆MP = 0. We cannot reject the null, thus we find there is no evidence of credit of foreign banks responding to monetary policy at all. 4 The estimate of β FGN is not statistically different from zero, hence, we conclude that foreign banks do not have different lending policies compared to domestic private banks per se. Notes: The dependent variable is the real rate of growth of net loans at the bank level. The sample is 429 banks in 10 CEE countries in years 1998-2012. Details of all variables construction and data sources are described in the Appendix. All columns report system-GMM estimates. Estimates for other macro and bank controls are suppressed.
In column (2) we separate private-owned and state-owned domestic banks. There is no qualitative (and only marginal quantitative) difference in our findings. Again, there is no evidence to reject the null hypothesis Finally, in columns (3) and (4), we account for the global financial crisis of 2008-10 and so we interact the Crisis dummy with the monetary policy indicator (we don't introduce the dummy itself as it is a linear combination of year fixed effects). Results in column (3) are obtained without ownership dummies.
An increase in the monetary policy rate by 1 percentage point reduces the growth of credit by about 1.8 percentage points in tranquil times. Also, there is evidence that monetary policy does not have any effect on the growth of credit during the global crisis as we cannot reject the null that β ∆MP + β Crisis×∆MP = 0.
Once we differentiate between foreign and domestic banks we find a strikingly small difference of estimates of β ∆MP and β FGN×∆MP between columns (1) and (4)

Robustness
To test the stability of our results we run several robustness checks. We control for possible endogeneity of the main variable of interest, the ownership structure and the possibility of changes in credit dynamics around the time of take-over. We also allow for interactions of monetary policy with other than ownership bank-level characteristics. Next, we examine the role of the monetary independence. We also check whether the observed difference in lending response to monetary policy can be attributed to foreign-currency loans.
For each scenario we report the system-GMM estimates of the variables of interest. The results of the robustness exercises are presented in Table 2.
Ownership Endogeneity. Our results may suffer from possible endogeneity of the take-over decision by a foreign investor. Firstly, the timing of a take-over may be determined by the previous performance which can be correlated with the past credit growth. Secondly, bank-specific characteristics may change abruptly in the wake of a take-over or the growth of credit may change. Therefore, to confirm the robustness of our benchmark results we exclude the take-over observations, that is, the bank-year observations where Foreign dummy changes from 0 to 1. 5 The results are presented in column (1) of Table 2. The benchmark results are robust. Bank-level Heterogeneity It is possible that the ownership variable is strongly correlated with other bank-level characteristics and hence the results may be a reflection of underlying bank-level heterogeneity.
Thus, we introduce interactions of ∆M P with bank-level characteristics (as in Gambacorta (2005)   does not provide currency composition of assets at the bank level, we can only test the hypothesis indirectly. 6 We have also tried including the interactions one by one. Results are quantitatively similar. 7 In Hungary, for example, up to 1/3 of total loans are denominated in foreign currencies according to Ongena et al. (2017).
To do so, we follow the second finding in Ongena et al. (2017), that changes in foreign monetary conditions affect bank lending more in foreign than in domestic currency. Changes in foreign versus domestic monetary conditions are (apart from special cases) reflected in the change of the nominal exchange rate in the long run. We introduce changes in the exchange rate of the local currency against euro (∆EUR) and Swiss franc (∆CHF) plus their interactions with foreign dummy to measure impact of the exchange rate fluctuations on foreign vs domestic banks.
The results including the euro exchange rate are presented in column (4) and including the Swiss franc in column (5)

What Drives the Difference?
Having established the robustness of the different responses of credit to monetary policy in foreign and domestic banks, we now turn to determining the causes of this difference. The key challenge is that different responses of credit to monetary policy can be due to either supply or demand factors.
On the one hand, different responses of credit could be driven by differences in demand. If we could observe differences in the loans portfolio composition between domestic and foreign banks, this would point towards a demand driven channel. Also, if credit in foreign banks varies less not only in response to monetary policy, but also to GDP and inflation (when taking inflation as proxy for economic instability), this would also point towards a demand driven channel.
On the other hand, the different responses of credit could be driven by differences in credit supply. The supply driven bank lending channel originates from a fall in deposits as a response to monetary contraction.
There are two cases to be considered here. Either there is a difference in the response of deposits to monetary policy in domestic and foreign banks or the deposits respond identically but foreign banks mitigate the Robust standard errors in parentheses *p < 0.05, ** p < 0.01 Notes: The dependent variable (expressed in percentage points) is: the ratio of commercial loans to total assets in columns 1 and 2 and the ratio of loans to banks to total assets in columns 3 and 4. The sample is 429 banks in 10 CEE countries in years 1998-2012. Details of all variables construction and data sources are described in the Appendix. All columns report OLS estimates. Estimates for other macro and bank controls are suppressed.
liquidity constraint better.
Intuitively, there is less scope for heterogeneity in terms of a deposit contract than in a debt contract (absence of costly state verification or adverse selection among depositors). In the first case, if a different response of deposits is detected, one can associate it with differences among depositors. This amounts to a deposit market segmentation between domestic and foreign banks.
In the second case, a different response of credit can be explained by foreign banks' access to internal markets, that is trading with a parent bank (or its other subsidiaries) at favorable terms. Also, even in the absence of internal markets, foreign banks could enjoy lesser costs of private information due to superior reputation inherited after the parent bank. For example, foreign ownership could result in more efficient operations or better risk management which allows foreign banks to borrow on financial markets at preferable terms.

Demand Factors
In this subsection we test whether differences in credit stem from differences in demand for credit. First, we ask if there are differences in a portfolio of loans between foreign and domestic banks. We verify this by estimating the following regression by OLS, including lagged bank-level controls to avoid endogeneity: The results are reported in Table 3. The dependent variable is either the ratio of commercial loans to total assets (columns 1 and 2) or loans to banks to total assets (columns 3 and 4), both expressed in percentage points. We find that once bank-level characteristics are controlled for, foreign ownership does not matter for the differences in the ratio of commercial loans to total assets. There is a statistically significant, positive estimate of β State in column (4) but otherwise the data do not provide evidence for portfolio of loans being significantly different between domestic and foreign banks. However, the conclusion should be taken with a grain of salt. There is a data limitation issue on commercial loans reflected in a drop in the number of observations. 9 Second, we ask if the are differences in the reaction of credit demand to macro shocks. If we could observe a significant estimate of β Foreign×GDP , we would conclude that foreign banks and domestic banks lend to firms in different sectors. Next, high inflation episodes in CEE countries correspond to spikes in the overall instability and uncertainty. Significant estimate of β Foreign×Inflation would be another proof of differences in the demand side. We re-estimate equation (1) with additional interaction terms and report results in Table 4. Only state banks' loans respond differently to demand shocks. This is consistent with state banks extending credit to the public sector which is varying less over the business cycle. There is no evidence that credit in foreign banks responds differently to demand shocks. Thus, we find no suggestive evidence of a credit demand heterogeneity between domestic and foreign banks. Robust errors in parentheses * p < 0.05, **p < 0.01 Notes: The dependent variable is the real rate of growth of net loans at the bank level. The sample is 429 banks in 10 CEE countries in years 1998-2012. Details of all variables construction and data sources are described in the Appendix. All columns report system-GMM estimates. Estimates for other macro and bank controls are suppressed.

Supply Factors
In this subsection we test whether differences in credit stem from differences in credit supply. In other words, if bank lending channel operates differently in foreign and domestic banks. First, we check if deposits respond differently to changes in monetary policy. We estimate three versions of the following equation: without ownership dummies, with Foreign dummy only and with both dummies. The dependent variable is either consumer deposits (insured) or deposits from banks (uninsured). The results are reported in Table Robust standard errors in parentheses * p < 0.05, ** p < 0.01 Notes: The dependent variable is: the real rate of growth of consumer deposits at the bank level in columns 1 to 3 and deposits from banks and financial entities at the bank level in columns 4 to 6. The sample is 429 banks in 10 CEE countries in years 1998-2012. Details of all variables construction and data sources are described in the Appendix. All columns report system-GMM estimates. Estimated coefficient for macro and non-ownership bank controls are suppressed.

10
The effect of monetary policy on the growth of consumer deposits is present in private domestic banks but not in foreign banks. There also is no effect of monetary policy on the growth of deposits from banks in either domestic or foreign banks. We conclude the reaction of bank credit to monetary policy in domestic banks is supply driven. The bank lending channel is at work only in domestic banks. Therefore, the data point towards the first case discussed above, that credit reacts differently to monetary policy because deposits do as well. Foreign banks hence do not need to rely on internal capital markets to mitigate the effects of monetary policy shocks, which is further investigated and confirmed in the next subsection.
Deposits from banks do not react to monetary policy in either domestic or foreign banks, while deposits from consumers in domestic banks do. A natural question arises: do foreign and domestic banks rely to a different extent on those two funding sources? We verify this by estimating the following regression by OLS, 10 The number of observations drop as there are missing observations for the dependent variable.
including lagged bank-level controls to avoid endogeneity: We report the results in Table 6. There are significant differences in the structure of funding and its costs between foreign and domestic banks. The share of consumer deposits in foreign banks' funding is on average 5 percentage points lower than that in all domestic banks and 10 percentage points lower than in private domestic banks. Correspondingly, foreign banks rely more on deposits from other banks. Interest expenses to total assets are 0.3 percentage point lower for foreign banks which corresponds to a one-tenth reduction in these costs (as the average interest expense to total assets in the sample is 3 percent). The signs of coefficients of bank-level controls are in line with the intuition. Larger, more liquid and better capitalized banks face lower costs which constitutes proxy evidence of frictions in the financial market.
Higher ratio of deposits from banks and lower cost of funding in foreign banks may, but do not have to, be a result of their access to internal capital markets. If they did, one should expect that foreign banks would also extend more loans to banks, which is not what can be found in Table 3. An alternative explanation could hold that foreign banks enjoy reputation premium or manage their operations more efficiently.

The Role of Parent Banks
In the last exercise we examine the likelihood that an internal capital market is responsible for the dynamics of credit in foreign banks and its lack of response to local monetary policy. There are two dimensions along which the effects of internal market should manifest in the data. Firstly, parent-bank characteristics and secondly, parent-bank country macroeconomic conditions should influence the credit of a subsidiary bank.
We have identified and collected the data for 95 parent banks yielding 1861 bank-year observations. This is a reasonably high number, given that not every bank in our sample has a foreign owner, that many foreign banks have the same owner or have dispersed ownership or non-bank owners. On top of that we collected the data on the rate of growth of GDP per capita and monetary policy short term interest rates in 22 countries that we identified as foreign banks' home countries for the years 1998-2012. We estimate regression (1) solely on data on foreign banks, introducing parent bank controls and parent bank macro controls. Standard errors in parentheses * p < 0.05, ** p < 0.01 Notes: The dependent variable (expressed in percentage points) is: deposits from banks and financial entities to total assets (wholesale funding) in columns 1 and 2, total consumer deposits to total assets (retail funding) in columns 3 and 4 and interest expenses to total assets (funding costs) in columns 5 and 6. The sample is 429 banks in 10 CEE countries in years 1998-2012. Details of all variables construction and data sources are described in the Appendix. All columns report OLS estimates. Estimated coefficient for macro and non-ownership bank controls are suppressed.
We find that neither parent bank country monetary policy nor GDP growth affect lending of a foreign bank subsidiary. Hence we find no support for the home country hypothesis which states that foreign banks may transmit economic shocks from abroad. We find no evidence of parent bank characteristics having any effect on the dynamics of credit at the subsidiary level. Hence, the data also do not provide decisive evidence of the relevance of the internal capital market. This result should be taken with grain of salt, as our sample includes only parent bank and not all members of a capital group. Nevertheless, in the light of our results, we find the alternative explanation of efficiency and reputation premium more convincing.

Conclusions
In this study we provide new evidence on the effects of monetary policy on bank loans dynamics. We find that: only domestic banks adjust their loans to changes in host country's monetary policy and that the reaction of domestic banks credit to monetary policy runs through the supply side, the bank lending The internal capital market hypothesis proposes that a foreign bank can easily obtain surplus liquidity from either the parent bank or other subsidiaries within the financial conglomerate. If this is the case, foreign banks operations are less dependent on macro conditions in the host country and more dependent on macro conditions in the home country, compared to domestic banks. As a consequence, on aggregate, the higher is foreign penetration of the banking sector, the less effective is the bank lending channel of monetary policy.
If the linkages with the parent-bank influence the response of supply of loans in foreign banks then there is value in aligning monetary and macro-prudential policies internationally. In this study however, we find that foreign parent bank and parent bank country characteristics are irrelevant for lending at a subsidiary level.