Balance Sheets Have to Balance
The first point to address: why do central banks need to lend money in the first place? The answer is straightforward: balance sheets need to balance.
Currency in circulation (e.g. dollar bills) are liabilities of the central bank. The central bank does not give those notes away for free; they need to receive something in return.
If the currency were pegged to another instrument (gold, another currency) and backed 100% by the other instrument, then the answer is straightforward: they need to get the backing instrument in exchange for the notes.
For a fiat currency issuer, there is no backing instrument. So the central bank needs to get something from the private banks that want notes. That something is usually bonds, or the central bank can make loans to the banks against collateral (discounting, or rediscounting).
Double Layer of Credit Protection
Central banks are banks to banks. Although they are currently a giant employment programme for Ph.D. economists, they do have large staffs that can do credit analysis, monitor loans, and work out defaults. As such, the normal preference is to lend against collateral presented by banks. This gives them two layers of protection: the loan is actually to the bank, and if the bank fails, only then do they worry about the underlying loan. Since they are supposed to be regulating banks, central banks should be able to lend to them with some confidence that they understand the business.
Admittedly, central banks were forced to dabble in private securities in the aftermath of the Financial Crisis. However, they quite often had asset managers managed the portfolios, or else they were hoping that diversification will bail them out (like many credit investors).
One of the side effects of the central bank lending against a class of securities is that this effectively turns those securities into money-like instruments: an important class of money market participants can always get government money at a known haircut with those instruments. The implication is that the collateral decisions can influence money market behaviour.
The Post-War Framework
Central banks loaded up their balance sheets with central government debt in World War II, and the changes in the economics profession kept it that way in peacetime. As Minsky noted, by having the central bank lend against government debt, the central bank was no longer tied to lending to the private sector. It just majestically determined the money/bond allocation for private portfolios, and was freed to be a benevolent central planner that set the money supply to achieve optimal outcomes.
The end result of this shift was that central banks were staffed with Ph.D.'s who spun out fancy mathematical models, and who had no idea what the financial sector was up to in 2008. Those models turned out to be useless, and so policymakers were forced to dig up copies of Lombard Street and bring in private firms to manage portfolios of toxic securities.
Although I imagine it might not be universal, central bankers are now aware of these issues. (Outside central banks, learning might be slower. Remarkably, I had a recent online discussion with an anonymous mainstream economist who insisted that the Fed set LIBOR, and that the LIBOR-Fed Funds spread was zero.) Although I am far removed from the financial information pipelines, I think there is no doubt that central banks provided a far more competent response to 2020 than 2008.
Blowing Up Monetary Policy Models
Having central banks operating in the credit markets blows up models, particularly neoclassical models. DSGE models are arbitrage-free pricing models, which are never going to realistically deal with defaults.
Anyone wanted to apply quantitative thinking to these interventions are going to have a hard time. The reality is that the only useful quantitative metric will be prices (or more accurately, spreads). However, credit spreads will be the result of interventions, and there will be no useful link to size of interventions -- which quantity theory obsessed economists will focus on.
The other issue is that central bankers seem to be highly sensitive to free market enthusiasts screaming about "central banks destroying price discovery." If they followed a sensible operating procedure -- announce target spreads that they will operate at, it is clear to everyone that they are engaging in a price-fixing operation. Since that will offend their neoliberal sensibilities, they will put the focus on the size of the operations -- which may tell us nothing about price.
Supply and Demand Stories
One of the common features of economists is that they love supply and demand stories when it comes to fixed income markets. This includes some post-Keynesian economists, who otherwise sneer at supply and demand diagrams.
There is no doubt that if a large investor makes large orders, prices can immediately move. The question is: by what amount, and how persistent will they be? In practice, persistence is a lot smaller than economists' stories suggest.
Let us look at a market in a particular risky instrument. If we look at the current owners and potential buyers, we could theoretically infer supply and demand curves based on the spread.
The working assumption that economists like is that there is a wide distribution of views about the instrument, and so the supply and demand curves would imply smooth movements of the spread if a large actor (like the central bank) wades into the market.
However, if the bulk of trading activity is done by a relatively small number of entities with similar valuation metrics for the security, then the effect can be tiny. The portfolio managers just reallocate to fungible instruments, and since "everyone" has similar valuations, any price changes would be fleeting once the squeeze on dealers' inventories has passed.
If we are discussing the level of rates, we can see why supply/demand effects were generally weak in response to QE (with the exception discussed later). Unlike economists, fixed income investors generally take mathematical finance seriously, and understand how rate expectations work. The dispersion of views about near-run rate trajectories were not wide enough that central bank purchases were enough to move them. Duration is fungible, and so central bankers taking some out of the market did not change things that much.
(The exception is in ultra-long maturities, where the private sector cannot credibly supply duration, and there are captive buyers courtesy of liability matching. Central bankers can squeeze 30-years, and even 10-years where private bond issuance is not as vibrant as the USD market.)
If we turn to credit, the fair value of spreads is the sum of default risk and liquidity premia. In aggregate, credit is fungible (since a lot of investors just buy the index and pray for diversification), and so it would take extremely massive interventions to move the median spread. However, it is possible to move the spread on individual credits, since there might be a wider dispersion of views regarding default risk.
Either Go Big, or Target
Either central banks need to launch massive interventions to reduce aggregate spreads, or they do targeted interventions. Massive interventions would be highly dangerous politically, so I would expect more in the way of targeted interventions -- which have already taken place.
Unfortunately, we have no way of calibrating the interventions. For the credits being bought are widely perceived as safe, then the purchases will just force a minor rebalancing of portfolios, and the purchases would have no real effect. It is only the purchases that save borrowers that really matter. Since we will have no way of knowing which is which, any aggregate credit purchase amounts will be mixing the two categories of lending. This means that a hypothetical intervention that is twice the dollar the amount of another could actually have less of an effect if it is targeted in the wrong sectors.
Sub-Sovereign Lending the Path of Least Resistance
The central bank bailouts of bankers in 2008 led to political blowback across the political spectrum. Central bankers bailing out well-connected individuals could easily lead to another swell of populist discontent -- even if the actual deals are done in above-board fashion.
The cleanest solution to central bank lending is to lend to sub-sovereigns (or sovereigns, in the case of the ECB). We already saw provincial bank repo operations in Canada in 2020, and state and local lending facilities are a major concern in the United States. Such lending has the advantage of being targeted, yet not being seen as being tied to individuals like CEOs.
The other alternative is to buy riskier assets on a portfolio basis (e.g., ETF's). At best, this just fuels asset price bubbles. However, the link from financial asset bubbles and the real economy is tenuous; we have had sub-par real growth since the 1990s, despite a series of bubbles. In any event, given the fungibility of risk assets, the analysis is going to be nothing more than glorified dual y-axis charting -- which works until it doesn't.
A Semi-Automatic Stabiliser
No matter how large the central bank credit interventions get, they are likely to be defensive in character. They will keep particular entities solvent, allowing them to keep their current pattern of spending intact.
This will prevent drops to activity. But it is much less clear whether this will lead to any acceleration in activity. The post-1990s trend has been one of sluggish growth, with the main stimulant being housing market bubbles. There is no reason to believe that these central bank activities will change that tendency.
Analysis is Going to be Qualitative
Regardless of the path chosen, the only sensible line of analysis will be largely qualitative. What sectors is the central bank lending to? What will be the effect of the lending? Changes to the size of the central bank balance sheet are not going to add any useful information, other than providing fodder for fans of dual y-axis charts.
(c) Brian Romanchuk 2020