Saturday, April 9, 2016

ETFs as money?

Blair Ferguson. Source: Bank of Canada

Passive investing is eating Wall Street. According to 2015 Morningstar data, while actively managed mutual funds charge clients 1.08% of each dollar invested per year, passively managed funds levy just a third of that, 0.37%. As the public continues to rebalance out of mutual funds and into index ETFs, Wall Street firms simply won't be able to generate sufficient revenues to support the same number of analysts, salespeople, lawyers, journalists, and other assorted hangers-on. It could be a bloodbath.

Here is the very readable Eric Balchunas on the topic:

Any firm that faces declining profits due to narrowing margins can restore a degree of profitability by driving more business through its platform. In the case of Wall Street, that means arm-twisting investors into holding even more investment products. If Gordon Gecko can get Joe to hold $3000 worth of low margin ETFs then he'll be able to make just as much off Joe as he did when Joe held just $1000 in high margin mutual funds.

How to get Joe to hold more investments? One way would be to increase demand for ETFs by making them more money-like. Imagine it was possible to pay for a $2.00 coffee with $2.00 worth of SPDR S&P 500 ETF units. Say that this payment could occur instantaneously, just like a credit card transaction, and at very low cost. The coffee shop could either keep the ETF units as an investment, pass the units off as small change to the next customer, use them to buy coffee beans from a supplier, or exchange them for a less risky asset.

In this scenario, ETF units would look similar to shares of a money market mutual fund (MMMF). An MMMF invests client money in corporate and government debt instruments and provides clients with debit and cheque payments services. When someone pays using an MMMF cheque or debit card, actual MMMF shares are not being exchanged. Rather, the shares are first liquidated and then the payment is routed through the regular payments system.

Rather than copying MMMFs and integrating ETFs into the existing payment system, one idea would be to embed an ETF in a blockchain, a distributed digital ledger. Each ETF unit would be divisible into thousandths and capable of being transferred to anyone with the appropriate wallet in just a few moments. One interesting model is BitShares, provider of the world's first distributed fully-backed tracking funds (bitUSD tracks the U.S. dollar, bitGold tracks gold, BitCNY the yuan, and more). I wrote about bitUSD here. Efforts to put conventional securities on permissioned blockchains for the purpose of clearing and settlement are also a step in this direction.

Were ETFs were to become a decent medium of exchange, people like Joe would be willing to skimp on competing liquid instruments like cash and bank deposits and hold more ETFs. If so, investment managers would be invading the turf of bankers who have, until now, succeeded in monopolizing the business of converting illiquid assets into money (apart from money market mutual fund managers, who have tried but are flagging).

Taken to the extreme, the complete displacement of deposits as money by ETFs would get us to something called narrow banking. Right now, bankers lend new deposits into existence. Should ETFs become the only means of payment, there would be no demand for deposits and bankers would have to raise money in the form of ETFs prior to making a loan. Bank runs would no longer exist. Unlike deposits, which provide fixed convertibility, ETF prices float, thus accommodating sudden drops in demand. In other words, an ETF manager will always have just enough assets to back each ETF unit.

Two problems might emerge. Money is very much like an insurance policy—we want to know that it will be there when we run into problems. While stocks and bonds are attractive relative to cash and deposits because they provide superior returns over the long term, in the short term they are volatile and thus do not make for trustworthy money. If we need to patch a leaky roof, and the market just crashed, we may not have ETF units enough on hand. Cash, however, is usually an economy's most stable asset. So while liquid ETFs might reduce the demand for deposits, its hard to imagine them displacing traditional banking products entirely.

A fungibility problem would also arise. Bank deposits are homogeneous. Because all banks accept each others deposits at par and these deposits are all backed by government deposit insurance, we can be sure that one deposit is as good as another. And that homogeneity, or fungibility, means that deposits are a great way to do business. ETFs, on the other hand, are heterogeneous. They trade at different prices and follow different indexes. Shopkeepers will have to pause and evaluate each ETF unit that customers offer them. And that slows down monetary exchange—not a good thing.

Somewhat mitigating ETF's fungibility problem is the fact that the biggest ETFs track the same indexes. On the equity side, three of the six largest ETFs track the S&P 500 while on the bond side, the two largest ETFs track the Barclays Capital U.S. Aggregate Bond Index. Rather than just any ETF becoming generally accepted media of exchange, the market might select those that track the two or three most popular indexes.

In writing this post, I risk being accused of blockchain magical thinking—distributed ledgers haven't yet proven themselves in the real world. All sorts of traditions and laws would have to be upended to bring the world of securities onto a distributed ledger. Nevertheless, it'll be interesting to see how the fund management industry manages to squirm out of what will only become an increasingly tighter spot. Making ETFs more liquid is an option, though surely not the only one.

PS. Here is the blogosphere's own Tyler Cowen (along with Randall Kroszner) on "mutual fund banking" in 1990:
In contrast to traditional banks, depository institutions organized upon the mutual fund principle cannot fail if the value of their assets declines. Since the liabilities of the mutual fund bank are precisely claims to the underlying assets, changes in value are represented immediately in a change in the price of the deposit shares. The run-inducing incentive to withdraw funds at par before the bank renders its liabilities illiquid by closing vanishes with the possibility of non-par clearing. In effect, there would be a continuous (or, say, daily) “marking to market” of the assets and liabilities. Such a system obviates the need for much of the regulation long associated with a debt-based, fractional reserve system, as the equity-nature of the liabilities eliminates the sources of instability associated with traditional banking institutions.
With the rise of ETFs and blockchain technology, the modern version of mutual fund banking would be something like distributed ETF banking described in the above post,


  1. I think it's much easier to get this form of money as a new offering by banks, because they are already plugged in to the payments infrastructure. You have an earlier post noting that banks do naked shorting of dollars. All we need is to get them to start shorting economic indices other than the consumer price index. Most importantly we need them to provide the market index. It shouldn't be too hard to get them interested: by providing the liquidity services you describe above, they get a cheap source of funding, that is also safe (relative to dollar funding).

    1. Good point. So why is it that banks don't currently offer equity deposits?

      "... to start shorting economic indices other than the consumer price index."

      That's an interesting way to describe it.

    2. If we think in terms of Perry Mehrling's hierarchy of money, then some important pieces of monetary architecture are missing. At some point, the US decided to issue a uniform currency that is used by banks as the base. The deposits we use for payment purposes are a derivative of this underlying base. The central bank acts as lender of last resort in order to prevent runs.

      In order to extend the banking model to ETFs, we need a similar type of cooperation between banks and a monetary authority that issues a base unit. As for this completely displacing dollar deposits, I think that it is unlikely. If we consider Tobin's mutual fund separation theorem, if people have access to both risk-free debt and the market portfolio, they will hold some of each, and have little need to shop around for other investments (unless if they are professional investors who can generate alpha).

  2. "Bank runs would no longer exist."

    1. That phrase struck me, too. A run on banks is ridiculous in the post metal-backed-money world of central-database-money.

    2. I'm not sure what you're implying, David. Are you saying that a sudden drop in share price has all the same implications as a flight out of an instrument that is convertible at a fixed price?

      ALex, the bank runs sentence refers to private banks that offer fixed price convertibility.

    3. "Are you saying that a sudden drop in share price has all the same implications as a flight out of an instrument that is convertible at a fixed price?"

      I think the "markets are better than banks" crowd is really missing something. Yes, it sucks to be a depositor and due to a liquidity crisis get 95 cents on the dollar when your bank fails. But it also sucks to be in the market when there's a liquidity event and to be invested in a fund that forced to sell into the event (b/c of derivatives collateral calls or exits of co-investors) and therefore to realize liquidity-driven losses. What's to prevent the long-run reaction to such losses from being worse than the reaction to bank runs.

      Liquidity based losses are a problem for markets just as they are for banks.

      Working on a paper on this ...


    4. I've seen some commentary on this problem, comparing "liquidity spirals" in markets to bank runs. It's a good reason to expand the banking model to cover index investing, because an expanded backstop role for the monetary authority would enhance financial stability. By "banking" I mean: monetary finance of illiquid investments, and the extension is to add a new monetary unit based on an index of market performance.

  3. "Should ETFs become the only means of payment, there would be no demand for deposits and bankers would have to raise money in the form of ETFs prior to making a loan."

    That's a curious vision - and doubtful as a necessary implication, I think.

    "Raising money" means there is a liability.

    And whatever that liability is (e.g. a "deposit liability" payable in ETF), it can just as easily be created "endogenously" - i.e. with the loan resulting in a corresponding liability as a matter of associated internal bank bookkeeping.

    And any problem associated with a potential risk management mismatch as between the characteristics of loan and its associated liability is present regardless of whether the liability is created exogenously or endogenously in association with the loan.

    The specification of the liability can be up to the bank advancing the loan, just as it is now. And (the medium of account for) bank reserves must be consistent with all this of course in order to allow for interbank transfers of loan proceeds.

    The issue of whether commercial bank (deposit) liabilities are exogenous or endogenous with respect to commercial bank loans is separate, and depends on institutional organization for clearing operations and permissible liability creation - as between a central bank and its commercial banks.

    1. JKH, maybe you can back up a bit. Are you disagreeing with the idea that universal usage of ETFs as money would lead to narrow banking?

    2. I should have said that it would be possible for banks to offer transferable/checkable demand deposits denominated in ETF money – i.e. deposits created by loans. I see no technical impediment to this.

      A regulatory/institutional design restriction could prohibit this, and require some form of matched funding such as term deposits, debt, or equity – which I think is what you suggest and where the idea of narrow banking falls in. But the liabilities and equity created would presumably still be denominated in ETF money.

      Given the operational realities of matching up loans and funding, it seems there would still have to be a central bank LLR function for ETF money. It would be possible for a bank to have a loan commitment to advance funds and fail to find matching funds on the same date. An LLR function would advance ETF funds to a capital-solvent bank until it could find those funds. I don’t see how such situations could be precluded. This also reveals a banking system balance sheet expansion effect, which although not in the form of demand deposits created by loans, is still inherently endogenous.

  4. The comment about runs is almost certainly wrong. ETFs would be susceptible to runs for at least two reasons: anytime getting out first is advantageous, there will be potential pressure to run. A couple months ago, Nicola Cetorelli, Fernando Duarte, and Thomas Eisenbach (NYFed) wrote about the potential for runs on mutual funds.

    Second, ETFs don't actually liquidate assets as people sell. To keep trading costs low, they use an intermediary and trade "warehouse receipts." About a month ago, Eric Balchunas wrote an article about that issue in Bloomberg. So that's another point of vulnerability.

    1. "Nicola Cetorelli, Fernando Duarte, and Thomas Eisenbach (NYFed) wrote about the potential for runs on mutual funds."

      Yes, but this post is about ETFs. They are different beasts from mutual funds.

  5. John Cochrane also posted something about using stock index funds as a medium of exchange (that had a variable exchange rate to the unit of account eg USD) (sorry if you are all linking to it all the time and I haven't kept up).
    It seems to me that a good way to have a run-proof lending system would be to have the time-to-maturity of the lenders' liabilities always matched to that of the loans they make and then to package those liabilities as bond ETFs. Those bond ETFs could then be held by savers. Instead of having a risk of bank runs, savers would have to put up with somewhat variable prices for those bond ETFs -but it would be a run-free, robust, system never needing bailouts.

  6. Or, "money as ETF" may be the answer. The central bank ought to simply be an enormous ETF, issuing money as needed to keep NAV in line at a single fixed rate of interest (equivalent to LR NGDP growth, say 5%).

    Otherwise, ETF as money is exactly like the old "banknote" system of bank issued currency.

    1. Interesting idea. ETFs are kept in line by authorized participants who conduct arbitrage, either submitting underlying stock in return for units or asking for redemption of units in kind. Would a central bank ETF that targets 5% NGDP growth use have an equivalent redemption mechanism?

    2. Well, it's just a rough thought, but couldn't a base-money-creating-unit (a share of a special purpose central bank vehicle) be the new exchange traded fund asset? (It's an asset, because the right to create base money is quite valuable.) ETF units would deliver future base money at a 5% annual pace, like a dividend.

      If anybody needed base money liquidity, it would have to buy these ETF units to receive the flow of newly created reserves. Base money liability, ETF asset. So (units+capital=base money, with a 5% annual pace of growth). If there were a liquidity run (the base money yield decline sharply), then the CB issues new units to get back to the 5% yield by issuing more units & possibly a faster base money payout.

      Also, float 49% of the central bank's equity for monitoring and additional arbitrage possibilities (controlling interest publicly owned at the Treasury).

      CB balance sheets almost never shrink, so the redemption mechanism is almost moot, but delivering base money back to the CB would receive a below-par priced base money unit (so creating a higher base money yield).