In the classical model of monetary theory, central banks are able to affect real economic activity through control of interest rates. In practice, it is thought that control over short-term rates such as the cash and federal-funds rates is potent due to a mechanism known as the ‘lending channel’.
According to the lending channel model, central bank control of reserves affects the supply of deposit financing available to banks. In turn, this allows the commercial banks to greatly expand their lending capacities.
Whilst this mechanism is widely known by economists, the assumptions that underlie it are not so familiar. In particular there are 3 critical requirements for the model. (I use vocabulary corresponding to a contraction in reserves)
- banks can do stuff. The neutrality postulate of money breaks down in the short term. Price adjustments are imperfect such that that nominal changes have large bearings on real outcomes.
- Demand for banks. Firms in the economy are reliant on the efficiency of banks such that bank loans and public finance via bonds are not perfect substitutes. In other words, the Modigliani-Miller theorem breaks down in a particular way implying firms cannot offset decreased credit from commercial banks with public credit.
- Supply of bank credit is affected by the Central Bank. Commercial banks do not insulate their lending actions from central bank operations. This is interpreted as a reluctance from the banking sector to switch from deposit funding to other forms of finance such as commercial paper and equity which are less reserve-intensive.
(These assumptions were identified by Kashyap and Stein (1994))
Both empirically and intuitively conditions 1) and 2) are likely valid, but the 3rd is not self-evidently sound. In fact, it relies on the assumption that for commercial banks debt instruments such as commercial paper are not good substitutes for reserves. The substitutability between these two instruments is thus the focus of this post.
Luckily, the extreme monetary conditions of this decade have provided a clearer view of the situation which I hope will become the topic of more analysis going into the future.
To see how extreme monetary conditions can give us a clear view of the fault lines in condition 3, we must first look at how Quantitative Easing (QE) worked in the U.S. I will not give the full mechanistic exposition (though at some point I probably should since it is a frequently asked question), but rather will attempt to explain some stylistic elements which shed light on the Monetary Transmission mechanism.
When the Federal Reserve embarked on QE3, the most expansive and concentrated of the QE iterations, it underwent a series of huge purchases of long term government debt (to the tune of $80 billion per month). In exchange for these purchases of government debt it credited the equivalent amount of reserves into the banking sector. In addition, as a precaution to stop runaway credit growth, the Fed paid 0.25% interest on ‘excess reserves’ (the reserves that banks held above their 10% requirement).
The goal of course was to reduce long term rates via the purchase of long term assets. But there is something else here. Now that the Fed was paying 0.25% interest on excess reserves, reserves themselves had become a highly liquid asset that in all but name resembled short-term bonds issued by the Fed. Hence, what we would expect to see is not just a series of purchases that pull down the yield curve on the long end; but rather a series of purchases balanced by sales which should push up the yield curve on the short end. In other words, this operation should not have just compressed the term structure of interest, but actually twisted it. One can think of it as an increase of short term debt supply as well as an increase in long term debt demand.
Nevertheless, throughout QE3 there was a persistent fact that came to the surprise of quite a few monetary economists. Namely, the yield curve in the US did not actually twist, but in fact was compressed in many markets- contrary to expectations. Commercial paper markets for example in commercial banks fell quite dramatically over the period 2011-13. Thus, on balance there was likely a reduction in the issuance of commercial paper by financial institutions who instead financed operations through the deposit account as availability of reserves increased. Thus, the new reserves served simultaneously as debt issuances by the Fed but to some extent also capitalised bank balance sheets just as commercial paper would in other contexts. Remarkably, the increased issuance of debt by the Fed via QE was offset by decreased issuance of commercial paper by banks to fund their assets.
If this analysis is right, and there are no other factors at work, (even if there are I think the scope of QE3 would be enough to swamp most other financial forces) the conclusion of such an unexpected repression of the yield curve is that financial institutions may perceive commercial paper issuance as a substitute for deposit funding. This gives a view into the financing of assets which actively undermines the idea that banks do NOT insulate their balance sheets from central bank operations. The lending channel may be less potent than we think.