During the peak of the crisis in autumn 2008, spreads in money markets rose sharply and volumes contracted, forcing many banks into difficulties with their standard liquidity management and refinancing strategies. Central banks reacted by taking deposits from some banks (via deposit facilities and excess reserve accounts) and lending directly to other banks (via various lending facilities) at much larger scale than in normal times. What some observers and policymakers thought to be a short-term measure has since become an important structural element of public policy during this long crisis. The lending facilities of the ECB, for instance, are still heavily used by many banks, including most troubled Irish banks and several Spanish savings banks.1 Arguably, this lending offers a crucial source of funds for financial institutions which, due to perceptions about their credit risk, would find it difficult to cover their refinancing needs in the market.Is this type of replacement of the money market trade just an unconventional tool of monetary policy? Is it large-scale long-horizon version of standard lending of last resort? Is it consistent with a strict price stability mandate? Does it have fiscal dimensions that would recommend counting with the explicit backing of governments and parliaments? In his analysis of the US experience, Goodfriend (2011) defends that several instances of unconventional monetary policy should be more properly called “credit policy” so as to explicitly acknowledge their potentially significant fiscal implications.
In a recently published paper (Bruche and Suarez 2010), we examine a model in which concerns about counterparty risk can create money market freezes. Our analysis is based on the presence of deposit insurance and regionally segmented bank markets. The analysis provides a clear rationale for a government supported replacement of the malfunctioning money market, but also makes clear that the institution performing such a replacement is very likely to be exposed to credit losses. If it is a central bank using extended deposit and lending facilities, such policy will indeed be closer to “credit policy” than to “monetary policy.”2 The implication is that the corresponding central bank should take action with the explicit budgetary backing (and perhaps a proper delegated mandate) from the relevant fiscal authorities.
A hypothetical relationship between Germany and Spain
To explain the basic idea of the paper, let us consider the hypothetical situation of two countries with asymmetric net financial positions – we shall label them Germany and Spain, just to make it easier to refer to them. Suppose that German banks have their deposit base mainly in Germany, give loans mainly to German companies, and that Germans like to save. Meanwhile, Spanish banks have their deposit base mainly in Spain, give loans mainly to Spanish companies, and that Spaniards like to spend. Suppose, then, that in relation to the quantity of retail loans that they can profitably make, German banks have too much deposit funding, whereas Spanish banks are in the opposite situation.3
Obviously, the Spanish companies (or Spanish banks) would like to access some of those avid German savers. However, assume that because Spanish banks know a lot more about Spanish companies, German banks do not want to lend to Spanish companies directly. Also, assume that at least in the short run, it would be very hard for a Spanish bank to attract a significant number of German savers directly. Of course, as long as both economies are going well and money markets are functioning, these things do not matter: Spanish banks can simply borrow the excess funding of the German banks in money markets. The German banks are happy with charging a close-to zero spread since they know that the probability of default of Spanish banks is negligible. If the German banks compete with each other, then one would expect the German banks to lend to the Spanish banks essentially at the German deposit rate, meaning that for both Spanish and German banks, the marginal cost of funds is the German deposit rate. Indeed this is a bank-intermediated version of the equalisation of the marginal cost of funds that corresponds to a situation of perfectly integrated capital markets.
Everyone is happy, until a crisis hits. And with it, the probability of default of all companies, and hence of all banks, suddenly becomes “significant”. To keep things simple, assume that there is no asymmetric information about the true probability of default, and that the probability of default is the same for all Spanish and German companies and banks (either of these assumptions could be relaxed without changing the essence of the results). Germans banks will now demand a high spread to compensate for potential losses from lending to Spanish banks.
In this situation, the marginal cost of funds for Spanish banks has increased substantially. Spanish banks react by raising their deposit rates, trying to attract more deposits. In the short run, though, they do not attract German savers and so do only succeed in triggering a deposit war in Spain, driving up deposit rates up to the point the marginal cost of deposits is equal to the cost of raising funds in the money market.
What happens next?
What happens next depends on whether or not Spanish and German deposits are protected by a deposit insurance scheme. If not, both Spanish and German depositors should demand a spread from their banks – equivalent to the one demanded from Spanish banks in the money market. That means that the marginal cost of funding for German banks (the German deposit rate) and for Spanish banks (the money market rate) remain aligned, reflecting the higher credit risk in both economies. As long as the total amount of deposits in Germany and Spain does not change much, the reallocation of funds across countries, via the money markets, will not be much affected by the rise in default risk.
If deposits are insured (as they typically are, in reality), the implications are very different. For their marginal funding, the Spanish banks have to pay a high spread in money markets, whereas the German banks keep using cheap, insured German deposits. There is now a wedge between the marginal cost of funds in the two countries. Ultimately, because of the increased cost of funds vis-à-vis the Germans, the Spanish banks scale back their lending, reduce their demand for funds, and some of the funds previously flowing from German depositors to Spanish firms stay in Germany, and are lent to (marginally) less productive German firms. Money market volumes decrease.
How big would this effect be? Well, it depends on the marginal productivity of capital in the two countries, and the size of the spread and hence the size of the wedge between the funding costs of the banks in each country. In our paper, we show that, when model parameters are calibrated according to standard macroeconomic practice, spreads of 100 to 200 basis points can easily cause a large fall in money market volume (as large as 75%) or even a complete freeze.
Of course, the fact that less funds are invested in Spain depresses economic activity there. The larger amount of funds kept in Germany (albeit being invested in slightly less productive activities at the margin) do actually imply an increase in economic activity in Germany (unless there are some extra linkages whereby the lower activity in Spain hurts Germany, e.g. via trade).
Can a central bank alleviate the situation? A central bank willing to expand its balance sheet can allow German banks to deposit funds at a certain rate and use the funds to lend to Spanish banks at a higher rate. Standard deposit and lending facilities could be used for this. The difference between the rates applied in each facility is a spread. If this spread is equal to the money market spread that would prevail in the absence of intervention, then the central bank does not change the underlying asymmetry in banks’ funding costs across the two countries, and hence does not alleviate the freeze. In contrast, setting a spread lower than the market spread (such as under the flat-rate full-allotment facilities put in place by the ECB) may alleviate the implications of the freeze.4 Intuitively, this “credit policy” may be welfare improving because it offsets the effect of a friction (German deposit insurance) that, in a context with non-negligible default risk, prevents funds flowing from the region where they are more abundant (Germany) to the region where they are more productive (Spain).
However, such credit policy is likely to generate expected losses to the institution that acts as an intermediary between the lending and the borrowing banks (i.e. the central bank in the example). This is indeed the case if the market spread reflects the credit risk of the borrowing banks.5 The failure of one or more Spanish banks might cause losses to the central bank similar in nature to those that the failure of a German bank would cause to the German deposit insurance system. However, having such losses absorbed by a central bank (i.e. either by its net worth or by some ad hoc recapitalisation arranged after the event) may be problematic, especially if the central bank has a narrow price-stability mandate and/or does not want to erode its credibility in the fight against inflation (potentially damaged by the fear that it will end up monetarising a part of the losses).
While we strongly defend the use of credit policy as described above, we also think that it would be desirable to acknowledge its true nature in advance and to provision for the coverage of its potential costs either with an arrangement similar to a deposit insurance fund or against government budgets. Doing so and expanding central banks mandate to the preservation of financial intermediation in times of crisis will reduce the risk that central banks do not fully fulfil their crucial role as intermediaries of last resort in systemic crises.
Atkins, Ralph (2010), “Loans data boost European recovery hopes”, Financial Times, 26 November.
Bruche, Max, and Javier Suarez (2010), “Deposit Insurance and Money Market Freezes”, Journal of Monetary Economics, 57:45-46.
Farhi, Emmanuel and Jean Tirole (2011), “Collective Moral Hazard, Maturity Mismatch, and Systemic Bailouts”, American Economic Review, forthcoming.
Gertler, Mark, and Nobuhiro Kiyotaki (2009), “Financial Intermediation and Credit Policy in Business Cycle Analysis”, prepared for the Handbook of Monetary Economics.
Goodfriend, Marvin (2011), “Central Banking in the Credit Turmoil: An Assessment of Federal Reserve Practice”, Journal of Monetary Economics, forthcoming.
Heider, Florian, Marie Hoerova, and Cornelia Holthausen (2009), “Liquidity Hoarding and Interbank Market Spreads: The Role of Counterparty Risk”, European Banking Centre Discussion Paper 2009-11S.
1 “[...] banks in Portugal, Spain, Ireland and Greece account for about 60% of total ECB lending, in spite of those countries creating only about 18 per cent of eurozone gross domestic product” from Atkins (2010).
2 Other recent papers finding potential welfare improving effects associated with implementing an active credit policy when money markets freeze include Farhi and Tirole (2011) and Gertler and Kiyotaki (2009).
3 If you don't like the story about German scrooges and Spanish spendthrifts, you can suppose instead that Spain is an economy with better investment opportunities, at the margin, than the more developed Germany - the conclusions would be very similar.
4 In fact, unless stigma or similar self-exclusion incentives apply, deposit and lending facilities may completely crowd out borrowing and lending in the malfunctioning money market.
5 Central banks typically argue that the haircuts imposed on their collateralised loans are enough to protect them against potential credit losses. Yet some banks' revealed preference for central bank lending is evidence that the underlying terms are overall more advantageous than those that could be found in the market. There may be reasons, other than central banks’ assumption of some expected credit losses, that explain why this occurs. The pricing in the market may be affected by problems of adverse selection (like in Heider, Hoerova, and Holthausen, 2009) that perhaps a central bank can partly circumvent thanks to its access to confidential supervisory information about the banks. Market prices may also be too onerous due to over-reaction, panic, or incentive problems and institutional constraints at the level of the potential lending banks that imply high aversion to having dealings with a potentially problematic bank. Finally, central banks may enjoy especial collateral rights when involved in emergency liquidity assistance. Importantly, we do not need any of these additional ingredients for our results about the desirability of the described central bank policies.