Ben Bernanke Portfolio Balance Effect

Question

I have a question based on reading some speeches and research papers, mainly by Ben Bernanke, on the portfolio balance effect. Here is Bernanke in a recent paper he authored where he describes the portfolio-balance effect as one of the two principal channels of large-scale asset purchases:

"The portfolio balance channel depends on the premise that securities are imperfect substitutes in investors’ portfolios, reflecting differences in liquidity, transactions costs, information, regulatory restrictions, and the like. Imperfect substitutability implies that changes in the net supply of a security affect asset prices and yields, as investors must be induced to rebalance their portfolios (Bonis et al., 2017). In principle, the two channels of QE can be distinguished by the fact that the signaling channel operates by affecting expectations of future policy rates while the portfolio balance channel works by changing term and risk premiums."- Bernanke (2017), in a paper that can be located here: https://www.brookings.edu/wp-content/uploads/2017/10/bernanke_rethinking_macro_final.pdf

I'm having some trouble here understanding the finance of this. In particular, how does this channel necessitate imperfect substitutability? I understand that the argument hinges on 'habitat preferences' -- certain investors prefer, say, a specific type of security -- but I haven't been able to put the pieces together.

As a potential example: let's say that we have an investor who will only deal with 5-year Treasury bonds due to either preference, risk appetitive, liquidity requirements, and so forth. The Federal Reserve then purchases 5-year Treasury bonds. Bernanke seems to suggest that the reason that reductions in the supply of these securities will increase their prices and reduce their yields is that investors don't view other securities as perfect substitutes. Is the argument perhaps that, if our investor considered 7-year bonds a perfect substitute, the increase in the price of the 5-year bond would cause him to sell off his supply of 5-years for a significant capital gain and purchase some other security? Or, for that matter, his portfolio would consist of such a wide array of different maturities that buying 5-year bonds wouldn't impact his behavior at all, as he would just replace it with any other bond.

I know this has something to do with elasticity of demand, but I just haven't been able to piece together the logistics of imperfect substitutability, hopefully, leading to a rebalancing of investor portfolios and a decline in interest rates across asset classes in a broad-based way.

I'd appreciate any insights people might have on this.

Answer 1

In standard economic theory, financial assets are viewed differently from standard goods and services because they are only valued for the underlying cash flow they are a claim on. More formally, standard economic theory treats financial assets and their underlying cash flow as fungible.

What this means is that if you have two financial assets (call them A and B) that are a claim on an identical cash flow, these two financial assets are perfect substitutes and should be worth the same. Continuing with this example, if you were to increase the supply of asset A, its price should not be impacted since its underlying cash flow is the same as asset B. This follows from a straightforward arbitrage argument. If the price of A goes up, then an arbitrageur could short A and go long B and make infinite money (vice versa).

So given the example above, changing the supply of a financial asset will only have a price impact, if the arbitrage mechanism can be inhibited. One way this can occur is if there are no perfect substitutes. In such a case it may not be possible to do the arbitrage trade, and hence there will be a price impact.

My explanation above is a bit hand-wavy and uses analogies. To see a more formal model of this see Vaynos and Villa (2021), "A Preferred-Habitat Model of the Term Structure of Interest Rates".

Answer 2

My argument is that he is just looking to find a justification for a policy choice he made, and so there is no reason to believe that his arguments are completely coherent.

We need to understand why he is making such arguments in the first place. If interest rates are purely the result of expectations (term premium equal to zero), the purchases themselves have no direct effect on yields, only as the result of signalling what the Fed will do (the first channel). However, the Fed could achieve the exact same effect by just saying that the economy stinks, and they will not raise rates for a really long time.

In order to find a justification for the decision to purchase bonds (instead of just talking), he needs to argue that the change in supply affects bond yields. The problem with that argument is that the front end of the yield curve does a pretty good job of tracking expectations, regardless of the state of supply and demand in the bond market. (One can look at how market participants use front end pricing as an estimate of what the central bank will do in the next few rate setting meetings across all the major markets as an empirical validation of that assertion.)

Therefore, we need some reason to believe that the short maturity bonds are somehow different from long maturity bonds, i.e., there is some form of segmentation. This way, one can hope that supply and demand affects long-term bond risk premia more than short-term premia (which would sound plausible to most bond market analysts). What matters is that the term structure of the term premia is changing due to the supply effect. One can come up with different verbal explanations for why term premia move, they will end up being indistinguishable in practice. That said, we need to ignore the possibility that other factor affect term premia, such as returns volatility (which would be suggested by portfolio theory).

The obvious issue with this argument is that Fed purchases are not the only factor affecting supply and demand in the bond market. Why are those other factors being ignored?