Friday, July 24, 2015

Know thyself... or carry a wallet

Of all the axioms of utility theory, the completeness axiom is perhaps the most questionable.  
- Robert Aumann, Nobel Prize Winner

One of the reasons you keep a well-stocked wallet in your pocket is because you don't know very much about yourself. Know thyself, as the Greeks say, and you can skimp on the amount of media-of-exchange you keep on hand.

Greater self-awareness leads to a cleaner "mapping out" of an individual's tastes and the preferred timetable for the enjoyment of those tastes. For instance, a moment of self reflection might lead you to conclude that pistachio ice cream at 8:31 PM next Friday is the best possible state of the world. If a complete set of futures markets exists, you can purchase a futures contract that is time stamped to deliver pistachio ice cream at 8:31 PM Friday, guaranteeing ahead of time that your tastes will be satisfied.

The problem is that introspection is difficult. We simply don't have the time, knowledge, or energy to sketch out a full timetable of carefully-delineated tastes and preferences. Even if we are blessed with a full range of futures markets, missing preferences prevent us from making use of these contracts.

Instead of committing ahead of time to satisfying taste A rather than tastes B, C or D at 8:31 PM Friday, an individual may prefer to remain non-committal. They can act on this preference by buying a broad range of option contracts that allow them to satisfy tastes A through D over a fuzzier time period, say Friday evening-ish. At the last minute they'll exercise just one of these many options while allowing the others to expire worthless. This sort of last minute off-the-cuff gauging of preferences allows for direct appeal to the mind's current state. This is surely a far more accurate way to get what one wants than trying to imagine what tastes will be like a week from now and locking that decision in by buying the relevant futures contract.

The problem is that the real world is bedeviled by not only missing preferences but also missing markets. Options on future consumption don't exist. Try buying a range of options exercisable between 6 and 10 PM Friday on twenty different flavours of ice cream.

There's an alternative. People can mimic an option buying strategy by allocating a portion of their portfolio to 'monetary assets,' those assets which are more liquid, stable, and cheaper to store than regular assets. The ability of a monetary asset to act as a good store of value up until the final act of acquiring a consumption good means that its owner needn't worry about lacking sufficient purchasing power to satisfy any of tastes A to D. And the liquidity of these monetary assets means that they needn't worry about being unable to swap for whatever consumption good they feel will satisfy their needs. So by holding a monetary asset, an individual has effectively bought themselves an option to satisfy a whole range of tastes at any point on Friday night. This is hassle-free flexibility.

Options aren't free. In financial markets, for instance, traders must pay a premium to secure an option. Likewise for liquidity. By holding monetary assets, individuals gain more flexibility surrounding the satisfaction of their tastes but give up potential returns. After all, a chequing deposit is more liquid than a term deposit, but a term deposit—which serves no monetary purposes—offers a superior capital gain.

So on the margin, people always measure the cost of becoming a bit more self aware against the drawbacks of holding monetary assets. If there is some low-hanging introspective fruit to be harvested, it may be worthwhile to spend a few minutes in reflection if this allows for a subsequent shift in wealth from liquid low-return assets (like chequing deposits) into illiquid high-return assets (like term deposits). On the other hand, if it is desirable to remain fluid and non-committal about tastes and the timetable for achieving them, cash and liquid securities are a means to buy this flexibility.


Here's the punchline.

Economists often (though not always) specify that individuals have a complete set of preferences. This means that the cast of characters that populate economic models come outfitted with fully specified sets of tastes and timetables for their enjoyment. There is no room for self-doubt, waffling, or vacillation. Nor do the people in these models need to spend any time or energy on introspection. Self-knowledge is free.

This no doubt makes economic models mathematically tractable. In the world outside of these models, however, our desire to hold liquidity is motivated by the fact that we are not fully self aware. Our tastes and timetables for realizing them are frequently left empty, usually because introspection is costly, inaccurate, and slow. Liquid media of exchange are an ideal way to stay flexible and uninformed about future tastes. By choosing to assume perfect self-knowledge, economists rule out at the outset some very important reasons people have for holding liquid media of exchange. With the 2008 credit crisis having illustrated the importance of liquidity factors, this seems like an unfortunate assumption to make.

Friday, July 17, 2015


The whippersnappers who work in the cryptocurrency domain are moving incredibly fast.

As I've been saying for a while, assets like bitcoin (or stocks) are unlikely to become popular as exchange media; they're just too damn volatile relative to incumbent fiat currencies. There's a new game in town though: stablecoin. These tokens are similar to bitcoin, but instead of bobbing wildly they have a fixed exchange rate to some other asset, say the U.S. dollar or gold.

Now this is a promising idea. If a crypto-asset can perfectly mimic a U.S. dollar deposit's purchasing power and risk profile, and do so at less cost than a bank, then the monopoly that banks currently maintain in the realm of electronic payments is in trouble. Rather than owning a Bank of America deposit, consumers may prefer to hold an equivalent stablecoin that performs all the same functions while saving on storage and transaction fees. To compete, banks will either have to bribe customers with higher interest rates on deposits, thus putting a crimp in their earnings, or go extinct.

Let's look at these stablecoin options more closely.

Type A: One foot in the legacy banking sector, one foot out

The unifying principle behind each type of stablecoin is the presence of some sort of backing, or security. Bitcoin, by way of comparison, is not backed. Stablecoin backing is typically achieved in two ways. With type A stablecoin, an organization creates a distributed ledger of tokens while maintaining a 1:1 reserve of dollars at a traditional bank. Owners of the tokens can cash out whenever they want into bank dollars at the stipulated rate, thus ensuring that the peg to the dollar holds. Until then, the tokens can be used as a stable medium of exchange. Examples of this are Tether and Ripple U.S dollar IOUs.

Could stablecoin be a bank killer?

We can think of a bank as enjoying stock and flow benefits from its deposit base. The existence of a stock of deposits provides it with a cost of funding advantage while the flow of those deposits from person to person generates fees.

Type A stablecoin pose no threat to the stock benefits that banks enjoy. After all, each stablecoin is always backed by an equivalent bank deposit held in reserve. If people want more stablecoin, the deposit base will have to grow, and that makes traditional bankers happy.

The flow benefits, however, are where the fireworks start. At the outset, people who receive stablecoin--through lack of familiarity--will probably choose to quickly cash out into good old fashioned deposits. But if stablecoin provides an extra range of services relative to deposits, rather than "kicking" back into the bank deposit layer, more people may choose to keep their liquid capital in the overlying stablecoin layer. Merchants will have more incentives to accept stablecoin, only adding to the snowball effect. Once all transactions are routed through the stablecoin layer, underlying deposits will have become entirely inert. While banks will continue to harvest the same stock benefits that they did before, they'll have effectively yielded up all the flow benefits to the upstarts.

So while Type A stablecoin doesn't kill banks, it certainly knocks them down a few wrungs.

By constructing a new layer on top of the deposit layer, stablecoin pioneers would be cribbing off the same playbook that bankers have been using since the profession emerged. Centuries ago, the first bank deposit layer was built on top of an original base money layer. Base money consisted then of gold and silver coin, but in more recent times it morphed into central bank banknotes and deposits. Because bank deposits inherited the price stability of base money (thanks to the promise to redeem in base money), and were highly convenient, bankers succeeded in driving transactions out of the base coinage layer and into the deposit layer. That's why gold and silver rarely appeared in circulation in the 19th century, being confined mostly to vaults. Perhaps one day stablecoin innovators will succeed in confining bank deposits to the "vault" in favour of mass stablecoin circulation. If this sort of displacement hadn't already been done before, I'd be more skeptical.*

Type B: Both feet out of the banking sector

More ambitious are type B stablecoin, which try to liberate themselves entirely from the traditional banking layer. Rather than using old-fashioned bank deposits as backing, a pre-existing issue of distributed digital tokens is used to secure the stablecoin's value.

As an example, take bitShares, a brand of bitcoin-like unbacked tokens. These tokens are every bit as volatile as bitcoin, up 10% one day and down 10% the next. Here's a chart. So far nothing new here, there are literally hundreds of bitcoin look-alikes.

The unique idea is to turn volatile water into stable wine by requiring that a varying amount of bitShares be used to back a second type of token, bitUSD. A bitUSD is a digital token that promises to provide its owner with a U.S. dollar-equivalent return. As long as each bitUSD is secured by, say, $3 worth of bitShares, the owner of one bitUSD will be able to cash out (into one U.S. dollar worth of bitShares) whenever they want and the peg to the U.S. dollar will hold.**

My understanding is that bitUSD, which debuted last year, is coming close to consistently hitting its peg. If bitUSD were to catch on as an alternative transactions layer, banks would lose not only their flow benefits but also stock benefits. After all, a bitUSD-branded stablecoin is not linked to an underlying deposit. We're talking complete devastation of the banking industry.

The system has some warts, however. If the market price of bitShares starts to fall, the scheme requires that more collateral in the form of bitShares be stumped up by the issuer of a bitUSD. This makes sense, it protects the peg. But what if the value of bitShares falls so much that the total market capitalization of bitShares is insufficient to back the total issue of bitUSD? At that point, bitUSD "breaks the buck." A bitUSD will be only worth something like 60 cents, or 30 cents, or 0 cents. Breaking the buck is what a U.S. money market mutual fund is said to do when it can't guarantee its one-to-one peg with the U.S. dollar.   

I'm skeptical of type B stablecoin for this very reason. Cryptocoin like bitcoin and bitShares are plagued by the zero problem; a price of nothing is just as good as a price of $100. They thus make awful backing assets, and any stablecoin that uses them as security has effectively yoked itself to the mast of the Titanic. A breaking of the buck isn't just probable, it is inevitable. Stability is an illusion. Maybe I'd get a bit more bullish on type B stablecoin if there emerged a brand that used digital backing assets not subject to the zero problem.

Anyways, keep your eye on these developments. Like I say, the young whippersnappers who are working on these projects aren't slowing down.

*In principle, type A stablecoin ideas are very similar to m-Pesa and Paypal. Both of these services construct new banking layers, but keep one leg back in the the existing banking infrastructure by ensuring that each Paypal or m-Pesa deposit is fully backed by deposits held at an underlying brick & mortar bank. See Izabella Kaminska, for instance, on m-Pesa.
 ** For those who like central bank analogies, this is an example of indirect convertibility, whereby a central bank sets market price of its liabilities in terms of, say, a bundle of goods, but only offers redemption in varying amounts of gold. See Woolsey and Yeager.  
*** Another working examples of Type B stablecoin is NuBits. Conceptual versions include Robert Sam's Seignorage Shares, the eDollar, and Vitalik Buterin's Schellingcoin.

Friday, July 10, 2015

There won't be a drachma-induced recovery

I catch both Lars Christensen and Brad DeLong making the claim that an introduction of the drachma will work wonders for Greece. Lars, for instance, says that:
However, Grexit will also remove the monetary straitjacket, which has had caused an enormous amount of economic hardship in Greece since 2008. The removal of this straitjacket will cause a significant easing of Greek monetary conditions, which in my view very likely will cause a sharp rise in nominal GDP in Greece in the coming years.
I hate to rain on the party, but even if a drachma is introduced and it collapses in value there won't necessarily be a drachma-induced recovery. Greece is currently in a straitjacket because its monetary standard -- the system for measuring and conveying economic value -- is a euro standard. Think of the euro as being akin to the metric system, a standard for measuring weights and distances, or dots per inch, a standard for measuring print resolution. An introduction of drachmas banknotes into circulation is simply not a sufficient condition to create the sort of effects that Brad and Lars want. To get their recovery, Brad and Lars need an all out monetary standard switch. But as long a Greek prices continue to be expressed in euros, drachmas will simply swim within the existing euro standard fish bowl. All sorts of mountains must be crossed before the penultimate switch of standards. This isn't "snap of the fingers" territory.

Drachma as another bitcoin

To better understand the destiny of a new drachma, its nice to have an example. Lucky for us, we can find one in the recent emergence of one of the world's newest currencies, bitcoin. Now Lars assumes that a collapse in the drachma will have all sorts of beneficial effects on the Greek economy. But does the world economy roar when bitcoin plunges? No, and here's why.

While merchants accept bitcoin as payment, they haven't accepted it as a standard. Sticker prices continue to be set in units like dollars, yen, pounds, etc., Bitcoin swims within the existing fiat standard. To accommodate those who want to pay in bitcoin, merchants will typically use the last-second bitcoin-to-dollar exchange rate (taken from foreign exchange markets) as the basis for the bitcoin price of their goods. Which means that as bitcoin plunges in value, the amount of bitcoin that retailers ask for their stuff is immediately adjusted upwards by a concomitant amount.

This is important because implicit in Brad and Lars's drachma recovery story is a certain degree of price stickiness. As the fundamental value of the standard unit plunges, sticker prices are slow to adjust upwards. Knowing that sticker prices will at some point start to catch up, people spend their currency now before it loses value, thus stoking an economic boom as merchants' inventories are drawn down. This sticky price effect is entirely lacking in bitcoin. Given a sickening collapse in bitcoin, those who own the stuff have no reason to spend it on before it loses value. After all, the amount of bitcoin that retailers require is adjusted every second, so prices in terms of bitcoin don't stay sticky. A bitcoin collapse therefore has no real effects on the world economy.

Applying this lesson to Greece, there's no guarantee that a decline in the drachma will boost the Greek economy. The appearance in circulation of drachma banknotes needn't mean that sticker prices will be set in drachmas. To determine how many drachmas Greeks must pay for an item, retailers may simply refer to its euro sticker prices and convert that amount into drachma by glancing at the last-second drachma-to-euro exchange rate. If so, then just as bitcoin prices are not sticky, neither will drachma prices. Without the requisite stickiness, a collapse in drachmas will not have the real effects that Lars and Brad want. Only a collapse in the euro, the monetary standard, will harness the sticky price effect the two implicitly invoke. But that effectively means Greece remains in its monetary straitjacket, despite having debuted a drachma.

Hurdles to switching

I've talked about the network externalities involved in switching standards before. Take an incumbent standard and a new standard. Even if the quality gap between the novel standard and the inferior incumbent is quite high, the costs of coordinating everyone onto the new standard may be too onerous for adoption to occur. Hysteresis, or lock in, is the result.

Switching to a drachma standard requires that a strong third party step in to overcome these network externalities. They need to punish, or credibly threaten to punish, those who refuse to comply. The Greek government, which has already demonstrated an inability to execute on basic tasks like tax collection, could very well lack the resources that are necessary to adequately perform the tasks of a third party.

Further militating against a drachma standard is its massive quality gap. A monetary standard should be as nuisance-free as possible. Merchants do not want to be adjusting their prices every day, and customers want to know that the blender they saw on a store's shelves on Tuesday will be worth the same amount when they go back on Wednesday. If sticker prices must be adjusted hourly, or even by the minute, the amount of time and mental space that people must allocate to calculation and measurement will displace other more meaningful activities. Like bitcoin, the drachma would probably be a one of the world's most volatile currencies; the incumbent euro one of its steadiest. So rather than improving on the euro standard, a drachma standard would represent a regression in quality.

Given that a strong government must spend a significant amount of resources getting its population to adopt a better standard, it's hard to imagine that a weak government will ever be able to foist an inferior standard on its population without facing a backlash. So contrary to Lars and Brad, there is no guarantee that issuing drachmas offers an ultimate salvation. In the end, there's probably very little difference between a Greece that introduces a drachma or one that doesn't, since either way the incumbent euro standard will likely stick around.

Note: This is very similar to my previous post on the topic. I've introduced the bitcoin analogy, which I think helps drive home the point, and also brought the quality gap to the forefront. 

Nick Rowe comments here.

Tuesday, July 7, 2015

A Visa/MasterCard theory of recessions

Statistical agencies employ data collectors who walk up and down aisles with hand-held computers gathering sticker prices for things like frozen french fries and bicycles. The data they collect gets amalgamated into an index and passed on to central bankers who use it as the basis for rate change decisions. It seems simple enough, but what happens if the source material has been corrupted? Might central bankers be reacting to mere shadows on the wall?

Here's how prices might go bad. Start with the U.S. and Canadian payments systems. For each credit card payment, a North American merchant must pay 1-2% in fees to the card networks Visa and MasterCard. Retailers in both countries have very different strategies for coping with this burden. In the U.S., retailers are permitted to offload network fees onto customers by asking them to pay a surcharge on each credit card payment. Because Canadian retailers are prohibited from surcharging customers, they react by marking up every sticker price in their store by a percent or two, the extra margin they collect sufficient to cover the card network fees. (Canadian retailers almost never offer cash discounts.) For a more complete explanation, see here.

This arcane difference in payments habits has the potential to result in a divergent evolution of prices, vastly different monetary policies, and an uncoupling of North American economic growth.

Consider what happens when the Visa and MasterCard networks decide to offer North American customers a universal 5% cash back reward. The networks fund this bonus by requiring merchants to submit a 5% fee on each card transaction. U.S. retailers cope with this levy by boosting the surcharge they apply on each card transaction to 5% or so, forcing the card-paying customer to bear the cost. Those Americans who pay with cash i.e. banknotes continue to get the sticker price, which stays constant. Without the ability to surcharge, Canadian retailers cope by boosting sticker prices by 5%, thereby indirectly passing the costs of the cash-back bonus onto the customer.

Marching up and down the aisles, U.S price collectorsdon't notice a thing. As a result, the U.S. consumer price index stays the same and the Fed doesn't lift a thumb. Canadian price collectors, however, find that prices have risen. Upon reception of this data, the Bank of Canada anxiously raises rates. This is because Bank officials believe that the rapidly rising price level indicates that the economy-wide rate of return (the natural rate of interest) has risen above the Bank's market rate, breeding inflation. A rate hike is necessary to bring the two rates back in lane, thus choking off the incipient inflation that seems to be developing. The natural rate hasn't budged, of course. All that has changed is the card networks' fee policy. Instead of bringing the market rate of interest in line with the natural rate, the Bank of Canada has been fooled into moving the market rate of interest above the natural rate.

If the card networks again increase the cash back reward, say to 10%, Canadian prices will rise even more. U.S. prices stay flat. The Bank of Canada once again confuses the effects a new cash-back policy with a rise in the natural rate of interest, tightening while the Fed stays pat.

Assuming that prices react rapidly and fluidly to central bank policy, then the Bank of Canada's tightening will simply drive consumer prices back down and restore equality between the natural rate of interest and the market rate. But if changes in monetary policy have effects on the real economy, then we've got problems. It could be that certain prices are sticky downward so that markets can't clear, the result being inventories of goods going unsold. Or perhaps the fact that debts are nominally denominated creates a Fisherian debt deflation. If so, then the Bank of Canada could end up unnecessarily driving Canada into a recession. The Federal Reserve, reacting to the very same set of stimuli, does not.

Now of course I'm exaggerating things. In real life such large increases in cash-back policy are unlikely. Nevertheless, we have seen a progressive increase in network fees over the years, enough to have probably inspired Canadian retailers to ratchet up sticker prices. As a result, Canadian CPI may be slightly over-stating actual inflation. On the U.S. side, consider that American retailers only recently gained the power to surcharge credit cards. As U.S. retailers roll out surcharge policies and reduce sticker prices, CPI will be pressured down. This may fool Fed officials into believing that the economy is slowing and draw them into an unnecessary rate cut when in realtiy all that has changed is credit card pricing habits. One wonders if the monetary authorities take into account these arcane features of our payment system when they set their monetary policies.

This answers a question I asked more than a year ago.