سیاست های پولی و وام دهی بانک ها در منطقه یورو: آیا یک کانال بازار سهام و یا یک کانال نرخ بهره وجود دارد؟

In this paper I compare a traditional demand oriented model of bank lending with its focus on short-term interest rates in the money market, to a non-traditional capital budgeting model of bank lending based on movements in share valuations for the Euro area. Using non-nested hypothesis tests, omitted variables tests, and Granger Causality tests, I reject the traditional demand oriented model of bank lending and fail to reject the capital budgeting model of bank lending for Monetary Financial Institutions (MFI's) in the Euro area. Even though Europe is a bank-based financial system, it appears the stock market plays a key role in the lending decisions and allocation of resources in Europe. One possible policy implication of this research is that the central bank should try and stabilize stock prices in order to achieve their goal of stabilizing bank lending and the economy

Banks as financial intermediaries play an important role in the financial system and the real economy of countries. As the institution whose deposit liabilities represent an important component of the medium of exchange, they are well-positioned to reduce the information asymmetries that naturally arise in the transfer of resources from household savers to investing firms in a decentralized market economy. Accordingly there is a large amount of cross-sectional empirical research in banking that documents the importance of bank screening and monitoring of small and medium-sized firms where the real investments and investment returns of these firms are particularly opaque.1 Furthermore there is evidence that indicates the benefits of bank screening and monitoring go beyond small and medium-sized firms. Large firms having access to external capital markets also benefit from bank screening and monitoring. When a bank grants a new loan, or, extends (or fails to extend) an existing loan to a firm, that piece of information sends a strong signal to the capital market that is reflected in the market valuation of the firm's outstanding securities.2 In addition to allocating financial resources across firms in different sectors of the economy at a point in time, there are other important questions concerning the role of bank lending in the supply of finance over time. More particularly, does aggregate bank lending amplify or dampen the business cycle? How does monetary policy affect bank lending and the economy? As for the first question the preponderance of academic research in this area suggests that bank lending does amplify the business cycle by changing the budget constraints of firms and households.3 The run-up in bank lending prior to the world-wide Great Recession starting in 2007 is but the most recent case prompting Basle 3 to implement a countercyclical capital buffer. As for the second question the traditional view of monetary policy is that it works through an interest rate channel and balance sheet channel for both borrowers and lenders. According to this traditional view an expansionary monetary policy has the central bank purchasing financial assets in the money market that in turn reduces short-term interest rates and increases bank reserves both of which should in principle increase bank lending through supply and demand channels. In this paper I revisit the nexus between financial markets and bank loan finance in the Euro area. Why the Euro area? I choose Europe because Europe is generally regarded as having a bank-based financial system built on the substructure of a civil law code (Levine, 1998). This presumably makes Europe's bank oriented financial system different than the market oriented financial systems in the U.S. even though monetary policy in both have traditionally been carried out in the short-term money market. The main question raised in this paper is: What assets and markets should the central bank target when implementing a monetary policy designed to stabilize bank lending and ultimately real output and employment? Should they operate primarily in the short-term money market to influence short-term interest rates as they have traditionally done but without much success in the current Great Recession, or, should they operate directly in some other market? The market we will suggest below is not the long-term government bond market nor the market for mortgage backed securities liked the Federal Reserve has recently chosen in its “Operation Twist” and QE3, but instead the stock market. Why the stock market? Casual observation reveals that while short-term interest rates in the Euro area (and also the U.S.) have been historically low during the 2008–2012 period indicating that loan finance was quite cheap, it is also the case that bank stock prices and stock prices in general have been very low indicating that the required yield on bank equity is quite high and the expected returns on business loans are quite low.4 Conditions in the money market imply an abundance of available bank loan finance through the credit channel while conditions in the stock market indicate a small amount of available bank loan finance. To resolve these two conflicting conditions it would be useful to find a statistical technique to make a comparison as to which set of assets and markets have a stronger effect on bank lending; the short-term interest rate that emerges from the interaction of the supply of and demand for funds in the short-term money market, or, the valuations that emerge from the stock market. To be more specific, this paper compares two reduced-form specifications for aggregate investments in private loans by monetary financial institutions (MFI's) in the Euro area. One specification takes the view that MFI lending is an investment decision that involves a comparison between the required yield of bank equity investors to the expected internal rate of return on loans. This view emphasizes the capital budgeting aspect of bank lending (we use the term banks and MFI's interchangeably) and the role of the market valuations of bank stocks reflecting the equity cost of capital for banks, controlling for other relevant factors. These other relevant factors – e.g., expected cash flows from the investment projects financed with bank loans – will be captured with the market valuations of stocks in general. In this specification bank and non-bank share prices incorporate all the relevant information that is needed for the capital budgeting version of the lending decision including the many bank specific control variables (e.g., size, liquidity, capital requirements, loan loss reserves, monetary policy, etc.) used in many supply/demand analyses of bank lending. Certain rare exogenous events like the reunification of Germany in 1990 and the 2001 attack on the financial district in New York City will be accommodated in the regression analysis with dummy variables. The alternative and more conventional specification looks directly to the interest rate channel and the market for bank loans, and for institutional reasons somewhat unique to Europe focuses attention on the demand side of the market. This approach to bank lending emphasizes the importance of such variables as a measure of aggregate income like GDP as a measure of the ability of firms and households to support loan finance, and the interest rate charged on bank loans. This specification will also include time dummy variables for the reunification of Germany in 1990 and the attack on the New York financial district in 2001. These two specifications do not necessarily reflect different views regarding the supply/demand framework for analyzing bank lending in the Eurozone. The question for us is specification; namely, whether the stock market with its many eyes can see more clearly the underlying supply and demand factors determining the volume of new loans to the private sector, or, whether these factors can be better observed from the outside indirectly by economists. We compare these two specifications of bank lending using “non-nested hypothesis tests,” “omitted variables tests,” and Granger causality tests. To preview our results we find that we cannot reject the hypothesis that changing stock market valuations determine bank lending in the Euro area whereas we can reject the hypothesis that the traditional demand factors of short-term interest rates in the bank loan market and a measure of household and firm income (like GDP) determine bank lending. Furthermore we find that while our two measures of share prices Granger cause Euro area bank lending, short-term lending rates and GDP does not Granger cause bank lending. These results for Europe are similar to those found in Krainer (2009) for bank lending in the U.S. In this sense Europe and the U.S. may not be as different as previously supposed. A possible policy implication of these results is that the central bank might consider imposing transaction taxes and margin requirements on stock investors, and/or carrying out some open market operations in a portfolio of bank shares as well as a diversified portfolio of equities in the stock market. These regression tests are presented in Section II. Section III summarizes the statistical results, discusses some possible policy implications, and indicates possible directions for future research.

3.1. Summary In this study we compare two non-nested and parsimonious reduced-form specifications on the determinants of inter-temporal bank lending in the Euro area. The traditional view of bank lending in Europe focuses attention on the demand for loans by borrowers in the bank loan market. This traditional view is presented as the H2 reduced-form specification of bank lending where these demand variables are approximated with GDP (as a proxy for expected income and the ability to repay loans) and short-term interest rates, R, as a measure of the cost of loan finance. The policy implication of H2 is one where the monetary authority targets short-term money market interest rates that get arbitraged out to bank lending rates that are relevant for the spending decisions of households and firms. In this way the central bank attempt to stabilize that part of aggregate demand financed with bank loans. This is the approach currently followed by the ECB and central banks in other financially developed countries. Previous empirical studies and the one carried out here were unable to reject the H2 specification of bank lending in Europe when that was the only hypothesis on the table. This paper proposed an alternative specification of bank lending indicated by H1 based on capital budgeting theory and stock market valuations. We then proceeded to compare H1 to the more traditional H2 hypothesis. The view taken here was that bank investments in private loans, like investments in any asset undertaken by firms in general, have to meet a cost of capital hurdle. That cost of capital hurdle in this paper was approximated by the market valuation of bank equity shares. To complete the capital budgeting analysis we include a general stock price index to proxy for the present value of the expected future cash flows of business borrowers and the wealth of household borrowers. Changes in general stock valuations reflect changes in expected cash flows, the IRR on the assets of bank borrowing customers, and all other information relevant for the bank lending decision. These two reduced form specifications of bank lending were then compared using non-nested hypothesis tests, omitted variables tests, and Granger causality tests. The non-nested hypothesis tests in Table 1 indicated that we were able to reject the more traditional demand oriented bank lending specification in H2, but unable to reject the stock market/capital budgeting specification of bank lending in H1. Omitted variables and Granger causality tests reinforced this conclusion. The causality tests indicated that the two share price variables in H1 Granger caused MFI investments in private loans, but that investments in private loans did not Granger cause the two share price variables. For H2 we rejected the hypothesis that GDP and interest rates on loans Granger cause MFI lending. Finally for the omitted variables test GDP and interest rates on loans were found not to be omitted variables in the H1 specification of bank lending, but the two share valuations were found to be omitted variables in the H2 specification. Of course this is not to say that there isn't some other potential and as yet unknown third specification, H3, of bank investments in private loans that could beat H1. If a non-nested H3 does arise sometime in the future, it should then be compared to H1. However it seems doubtful that this unknown H3 would be devoid of an important role for equity valuations. In closing there has been much research in the legal, finance, and corporate governance literature indicating that there are important differences between the financial systems in Europe and the U.S., and the way corporate investments are financed in those countries. The former is classified by this research as a bank-based financial system and the latter a market oriented financial system. In both financial systems capital budgeting theory tells us that real corporate investment should respond to changes in expected real corporate cash flows and the cost of capital. The job of the public corporation is to generate a rate of return on their assets that is at least equal to the required rate of return of their equity investors. It should be no different for banks. It would therefore seem that if capital budgeting rules can evaluate the merits of tangible investments by nonfinancial companies, they could in principle be used to evaluate the merits of investments in paper assets like private loans by financial companies. Under these conditions bank lending should then respond to both changes in the market valuations of bank stocks reflecting their cost of capital, and the market value of stocks in general reflecting the change in expected cash flows on real investment projects of bank loan customers. Our research indicates that even though much of the financing of firms passes through banks in Europe, they are in fact guided by the stock market when it comes to determining their investments in private loans. However, it must still be remembered that this is only one time series study of bank lending based on an informational efficient stock market. Hopefully future research using panel data on both borrowing firms and lending banks across the Euro area countries (along the lines of the Jimenez et al. (2012) study of Spain) can shed more light on the link between the stock market and bank lending. 3.2. Policy implications and directions for future research If there is a stock market channel for bank lending in Europe, what implications might this have for the conduct of monetary policy? Bank lending decisions changes the budget constraint of the private sector and can therefore influence the demand for real output. In this way any fluctuations in bank lending can potentially amplify fluctuations in real economic activity. If this is the case the central bank might want to consider ways of stabilizing share prices. Selective controls are one possible way to achieve this goal within the present regulatory framework. One selective control mechanism would be for the central bank to impose margin requirements on all stock investors, both individuals and institutions. A second selective control mechanism would be to vary transaction taxes on stock purchases and sales. The margin requirements and transactions taxes could be varied over different stages of the stock market cycle; rising when stock prices are rising above some measure of intrinsic value, and falling when stock prices are falling below their intrinsic value. Of course borrowing abroad and/or foreign investors could prove to be an obstacle. Moreover economists in general eschew selective controls as interfering with the allocation of resources preferring instead monetary and fiscal instruments that affect all categories of economic output.12 Still another way of stabilizing stock prices is for the central bank to carry out some open market transactions in a set of well diversified (i.e., portfolios that minimizes non-systematic risk) index mutual funds containing stocks of all financial and nonfinancial companies. The idea would be to change the level of share prices but leave the structure of relative share prices unchanged. Central bank purchases and sales of equities are not a particularly new idea having been previously suggested in Tobin and Brainard, 1977 and Fischer and Merton, 1984, and Krainer, 2003, Krainer, 2013a and Krainer, 2013b. Moreover actual central bank purchases of equity shares and other risky assets are also not without precedent. The monetary authorities of Hong Kong, Japan, and South Korea are examples of cases where governments have purchased equities to prop up a sagging stock market during a crisis. Hong Kong presents a particularly important case study of central bank intervention in the equity market. The background for Hong Kong's intervention was the Asian financial and economic crisis in 1997/98 during which short selling in the spot and future markets by foreign hedge funds drove the Hang Seng stock index down by 40 percent over the period June 1997–June 1998 while the economy was in the midst of a severe recession (Goodhart and Dai, 2003). Starting in August 1998 the Hong Kong monetary authority in 10 separate trading days purchased 118 billion HK dollars of equities in the Hang Seng index. The end result was that this investment soared to 200 billion HK dollars at the end of 1999 yielding a capital gain of 69 percent for the government. More importantly this policy was credited with pulling the Hong Kong economy out of the recession at the time. A policy of conducting open market operations repeatedly in equity markets might appear to be drastic to some and even irresponsible to others. And yet drastic but innovative monetary and fiscal policies were implemented during the Great Crisis. For example in the Great Crisis Japan, Europe, the U.K., and the U.S. governments have made substantial investments in risky assets such as ETF's, REIT's, mortgage-backed securities, and equity investments in individual financial enterprises and even non-financial enterprises experiencing financial difficulties where there was a great deal of unsystematic risk. These investments by governments were made in the hope of stabilizing the financial system and the economy during the crisis. In general these non-traditional policy interventions were thought to be successful in preventing a further collapse in the financial system and contributing to the economic recovery.13 In Europe and the U.S. the unconventional monetary policies have for the most part been restricted to the debt markets. If these non-traditional policies in the debt markets were deemed to be successful in the crisis, it would seem that a policy of stabilizing bank loans by stabilizing shares valuations through margin requirements, transaction taxes, and central bank open market operations in equities in non-crisis periods is one that merits further consideration by researchers and policymakers. The above mentioned unconventional policies including those implemented by Hong Kong, Japan, and South Korea were one-off policies designed to address a specific crisis. Once the crisis passed, the unconventional policy usually lapsed. One important question then is how a monetary authority would implement a more continuous policy of open market operations in equity markets. One way is for the central bank to target several broad based stock indexes and then invest in mutual funds that track the targeted indexes.14 When would the monetary authority buy and sell the index mutual funds that it is tracking? One possibility is to set upper and lower filter bands centered on some measure of long-run intrinsic value for each stock index. For example, suppose that the long-run median price-earnings ratio for a broad-based index of stocks was 12. The central bank might then set a lower price-earnings band of 8 at which they would buy the mutual fund(s) tracking the index, and sell when the upper band of 16 was reached. The (±) 4 change in the P/E ratio bands would presumably reflect transitory variations in the earnings of the firms in the index. Of course these numbers are for illustrative purposes only. Obviously much research would have to be done to set the upper and lower bands that would trigger central bank sales and purchases of the mutual fund(s) tracking the targeted indexes. Would a policy of central bank buying and selling equities with the view of keeping a targeted index of equity prices within some upper and lower band be feasible? Do central banks have the resources to achieve the goal of determining the level of share prices? Historically the most frequent central bank intervention has been in the foreign exchange market. Their success in that market is mixed at best (Sarno and Taylor, 2001). For equity markets the example of Hong Kong in 1998 would indicate that central bank purchases can drive share prices up to some desired level. But what about preventing stock prices from rising above some predetermined price-earnings ratio band? We have less evidence on this side of the filter band. One thing seems certain and that is central banks would be under enormous political pressure not to intervene in the equity market and drive share prices down to a level between some upper and lower filter band. Besides political pressure markets would also have to believe that the central bank is credibly committed to reducing volatility in the stock market. Once that credibility is attained rational private investors transacting in markets would keep stock valuations between the two filter bands. Open market operations in equities on a continuous basis are not part of the standard policy toolkit of central banks. This suggests that there could be problems both in implementing this policy and unforeseen and unintended consequences were the policy to be implemented by central banks. We will conclude by briefly discussing some of the more obvious problems and propose them for future research. One potential problem would be whether the central bank in its attempt to restrict fluctuations in the level of stock prices would induce an increase in the volatility of relative prices of the stocks within the targeted stock index. To the extent it did it would create additional risk for individual firms within the index thereby reducing production and investment. The evidence while limited suggests that this would not be a problem. Parsley and Popper (2004) find that changes in the level of stock prices are positively associated with changes in the volatility in the cross-section of relative stock prices. To the extent open market operations of a central bank achieved its goal of reducing the magnitude of the volatility of the targeted stock price index, it would also reduce the volatility of relative stock prices within the index. This would be a direct benefit to borrowing firms in addition to the benefit of smoothing bank lending. However more research validating the Parsley and Popper research is needed. A second potential problem centers on the inclusiveness of the targeted index. No actual stock index covers all the companies whose shares are publicly owned and traded. This problem is particularly acute in Europe where a larger share of GDP is generated by non-listed companies compared to the UK and the US. The problem is that central bank open market operations in listed companies within a targeted index may result in a distortion in the cost of capital and the allocation of investment between companies within the index and those outside the index. If the companies outside the index are publicly traded in the market, then rational traders would arbitrage any discrepancy in share prices between the two due solely to central bank open market operations. The problem is more severe for companies outside the index that are privately owned. Rational arbitrage can play no role here. However to the extent these privately owned companies look to publicly traded firms whose returns are highly correlated with their own; the cost of capital signal for investment decisions may not be distorted between the two as a result of central bank open market operations in the targeted index. This is an empirical question and one that deserves more study. Finally, will central bank open market operations in the stock market discourage the so-called ‘smart money’ from trading? There always is the risk that a central bank could without notice change the upper and lower filter bands posing an additional risk to smart traders. If it did then more trading would be in the hands of the uninformed or ‘noise’ traders who would have a larger impact on relative share prices. If this would occur then the relative prices of shares would not necessarily reflect all the relevant information pertaining to their intrinsic valuation. The end result here is that real investment might be inefficiently allocated across the different sectors in the economy. How important if at all this might be should be the subject of future research. These caveats remind us that the case for central bank open market operations in equity markets is far from complete. While it is premature to recommend this policy change on the basis of one statistical study, it is not premature to call for further research in this important area.