Tuesday, March 19, 2013
Cullen Roche penned an article on moneyness last week. In it he hypothesized that bank deposits are more money-like than paper dollars. In this post I'm going to highlight some thorny problems in ranking moneyness. I'll get back to Cullen's observation at the end.
If an observer wants to rank items by their moneyness, or liquidity, they might choose to begin the task by evaluating data like bid-ask spreads and frequency distributions of various assets in trade. Assets with low spreads and high frequencies might appear near the top of the observer's moneyness list, and those with high spreads and low frequencies might appear at the bottom. But spreads and frequencies are the objective liquidity data of markets. What we want to know is how the market experiences and digests this objective data on the way to building its own unique moneyness ranking of assets. Moneyness, after all, is subjective.
The best way to find out what the market actually thinks about something is to see how it prices it. If we can strip out the value of all the other services provided by an asset, all that remains is the value that the market puts on an asset's moneyness. Once we have this price, we can proceed to rank each asset according to its moneyness.
While market prices are surely the best way to build moneyness rankings, the problem is that markets rarely provide the prices to facilitate the analysis. Most markets are all-or-none markets, meaning that traders have to buy an entire asset and all the services it provides. There is no independent moneyness market in which an asset's liquidity services can be sliced away from all the other services, thereby allowing those liquidity services to be bought, sold, and priced.
When markets reveal moneyness rankings
Despite the lack of an independent moneyness market, from time to time markets do give us accurate snapshots of relative moneyness. One case in which we can get an unambiguous ranking is in comparing the prices of a bank's savings deposits to its chequing deposits. From the point of view of a potential owner, the two instruments are similar along almost all axes. Chequing and savings deposits are liabilities of the same issuer, so their credit risk is identical. They are both perpetual debt instruments convertible on demand, so their term risk is identical. Finally, they are convertible into the same underlying asset, paper dollars, so their expected change in purchasing power is exactly the same.
Yet the two instruments have very different market prices. A savings deposit currently yields a return to its holder of around 1% a year whereas a chequing deposit costs its holder around 1% a year in fees.
This variation in return structures arises from the one set of properties that differentiates a savings deposit from a chequing deposit: the range of transactional services attached to each one. Chequing deposits offer more options. In Canada, for instance, we are limited in using savings deposits to pay bills, write cheques, or make debit card transactions.
Since the difference in returns between the two deposit types comes to ~2% a year, the marginal depositor is effectively placing a premium of 2% on the extra bit of services provided by a chequing account. Put differently, given a $100 investment, a marginal depositor is willing to lose $1 in chequing account fees and forego $1 in savings account interest in order to enjoy the superior monetary services of a chequing account. Chequing account moneyness is worth $2 per $100 on the margin—so says the market.
Another unambiguous example of the market providing a moneyness ranking comes from the stock market. While my old post on this topic goes into the topic in considerable depth, I'll give a quick recap. A certain company that trades on the Toronto Stock Exchange has issued two classes of shares, RET and RET-A. Like the chequing/savings account example, both shares have the same credit risk—they are ranked pari passu and pay the same dividend. Despite these similarities, RET-A shares have almost always traded at a premium to RET, as is currently the case. RET-A yields 7.46% whereas RET yields 7.58%, a difference of 0.12%. Why not just invest in lower-priced RET and enjoy an excess return of 0.12%?
The best explanation for the difference in yields is the varying liquidity of each share. As I pointed out in my old post, there is very little activity in RET. As such, RET's bid-ask spread is wider than the RET-A's spread. Investors are willing to forego RET's 0.12% extra return in return for the enjoyment of the superior monetary services of RET-A, which arise from its low bid-ask spreads and active market. This is an unambiguous market signal of moneyness ranking.
Difficulty of using relative yields to rank assets by moneyness
Ranking the moneyness of assets is difficult to do when we compare across different asset classes and issuers. For instance, 4-week treasury bills currently yield around 0.10%, far below the 0.25% yield on deposits held at the Federal Reserve. Stephen Williamson thinks that this gap can be explained by the fact that t-bills are more liquid than deposits held at the Federal Reserve. After all, you can use t-bills as collateral for repo, but you can't use Fed deposits as collateral.
Confusing matters somewhat is that unlike the savings/chequing deposits and RET/RET-A examples, deposits and t-bills have different issuers. The Federal Reserve issues deposits whereas the US Treasury issues t-bills. We often assume that these institutions are effectively consolidated, and therefore carry the same credit risk, but there is no guarantee that this must be the case. As I pointed out in this post, a central bank can be left flapping in the wind by an irresponsible parent government. The two instruments also have different terms and payout structures. T-bills mature after four weeks whereas reserves are perpetual instruments convertible into dollars on demand. The upshot is that the reason for the yield gap between these two instruments is difficult to determine since differing credit risk, term risk, and moneyness might all be reasonable explanations. Without more data, we can't rank these assets according to moneyness.
The same goes for Cullen's point about the superior moneyness of bank deposits relative to paper dollars. While Cullen could very well be right, only evidence in the form of market prices would be sufficient to verify his claim. Let's say a deposit yields 0.25% whereas paper, as always, yields 0%. It would be tempting to say that paper is more moneylike, since it compensates for its 0% return by providing superior monetary services.
Muddying the waters is the fact that both a bank deposit and a paper note have different issuers, and therefore possess unequal credit risks. After all, there's no guarantee that a bank deposit will always be worth the value of its underlying paper, as the Cyprus case shows us. Thus, if a deposit yields more than zero-yielding paper money, this could be due to a deposit's higher credit risk, not the superior moneyness of paper. Further complicating matters is that paper notes burden their holder with an extra set of costs, namely the risk of loss and theft. If a deposit yields less than a zero-yielding paper notes, this premium could be due to either a deposit's superior moneyness or the awkwardness of paper notes. Because we can never be sure which effect dominates, we can't definitively rank these two assets according to moneyness.
To solve our problem we need independent moneyness markets. These sorts of markets would allow households, traders, and economists to strip out the moneyness properties of an asset in order to price that property and rank it. Incidentally, that's the same conclusion I arrived at in last week's post about the fundamental valuation of equities. If we could strip out the liquidity value of equities, we could better analyze and price the real yield provided by a stock. But more on this topic later.