Bitcoin's bootstraps

by Paul Conrad

When we talk about bitcoin, one thing we need to ask ourselves is this: can worthless things circulate and be accepted in trade? If so, how? And can this state of affairs continue indefinitely?

An intrinsically useless, unbacked, and costless fiat object might be accepted in trade, but only if it already has a positive price. A history of positive prices will generate sufficient expectations among potential acceptors that they will be able to trade that object on tomorrow. But how might our fiat object earn a positive price to begin with? If we reply that early adopters expected it to be widely accepted by others in trade, how did these early adopters ever form these expectations if that object didn't already have a positive price? We're dealing with a problem of circularity. There is no way to "break into" a dynamic that might generate a positive value for a fiat object. So logically, worthless things cannot trade in the market at a positive value.

However, fiat objects like dollars and yen do seem to have a positive value. Two types of economists, Austrians and MMTers, recognize the circularity dilemma that emerges when trying to explain the positive price of a useless fiat object. Both solve the circularity problem in different ways.

Austrians say that when early adopters first acquired the fiat object, it was not yet intrinsically useless, unbacked, or costless. Thanks to its original commodity nature, or perhaps its status as a backed financial asset, it already traded at a positive price. Even if that character is lost, the object suddenly becoming a fiat one, it may still be widely accepted in trade on the basis of people's memory of its pre-fiat price. Thus the circle can be broken into, and worthless bits of paper can legitimately have a positive value in trade. This is Ludwig von Mises's famous regression theorem.

MMTers solve the circularity problem by bringing in the tax authority. As long as some agency like the government imposes an obligation on people to pay taxes with these fiat objects, that will be enough to drive their positive value.

I should point out that I don't think we actually face a circularity problem with modern central banknotes since they aren't worthless bits of paper but rather exist as a liability of their issuer. But we do run into the problem with bitcoin. Here we have an unbacked, intrinsically useless, stateless fiat object trading at $950 or so, not to mention a legion of copycat coins trading at various positive prices. [1]

Austrians are all over the board on bitcoin. Because their solution to the circularity problem is to invoke the legacy commodity value of a fiat object, bitcoin poses some theoretical hurdles for them since it is by no means clear whether bitcoin ever had an original commodity value. Bob Murphy for one argues here that bitcoin may have earned its first foothold thanks to non-pecuniary ideological reasons. However, there seems to be no consensus among Austrians on that point. MMTers seem to genuinely dislike bitcoin since their preferred tax obligation story can't bear the load of explaining bitcoin's price. Here is L. Randall Wray who says that bitcoin is a test of the "infinite regress view of money", then gleefully points to its falling price as evidence that the taxed backed theory is the dominant theory (it later rebounded).

Let's move on from MMTers and Austrians. George Selgin recently came up with an interesting way to explain how bitcoin might have earned its all important original positive price:

Records show that a just a few persons took part in most early Bitcoin transfers, and especially in the larger-volume ones. My guess is that they all knew each other, and that those trades were more-or-less fictitious, with large values being traded and then traded back again, with the intent of enhancing the prominence of the positive-value equilibrium by drawing attention away from the much larger set of inactive Bitcoin markets. Bitcoin’s inventors, I’m now almost certain, were making conspicuous leaps onto their own bandwagon, so as to encourage others to do so, whether to express themselves or to profit by doing so. In short, a clever marketing strategy, including a little strategic sleight-of-hand, can substitute for history in putting a positive sign on the expected value of an otherwise useless potential exchange medium.

Here we have neat way to break into the circle. Have a group of insiders trade the fiat object amongst each other in order to generate an artificial history of positive prices, at which point outsiders will be willing to accept it in trade based on the expectation that others will repurchase it from them later.

Making "conspicuous leaps onto one's own bandwagon," as Selgin calls it, is a well worn tactic. In stock markets, the term wash trading refers to the illegal practice whereby an individual or group of schemers trade an illiquid, often worthless, stock back and forth among different accounts. The goal is to give the illusion of activity, thereby attracting innocent traders who would otherwise pass up the stock. A more colourful term for this is "painting the tape", which refers to the old ticker tape of yore.

Another way to paint the tape is to high close a stock. Using this technique, a trader or group of traders will buy a stock in the closing seconds of the day, pushing its price up. Since media outlets tend to focus on a stock's daily closing price, and stock charts depend on the daily close, high closing may be a cost effective strategy for traders to create and benefit from the positive price momentum that news of a high closing price engenders.

Auction markets, say in livestock or art, are sometimes populated with confederates—those who work in conjunction with a seller to provide fictitious bids so as to drive some object's price, say a dubious piece of abstract art, or a lame horse, far higher than it would otherwise be worth. Should the confederate's bid be the only bid, the worst that happens is that the schemers get their own painting or horse back, upon which they can try the same trick over again in the next auction. If their bidding excites someone else to add a bid, then they've succeeded in earning something for nothing.

In any case, all of these techniques can push a worthless object's price above zero, at which point that object may have generated enough of a history of positive prices that it will be valued by enough outsiders that it will join the mass of non-fiat objects in circulation. From nothing, our worthless item it has pulled itself up by its own bootstraps.

Which explains bitcoin's incredible volatility. A bootstrapped object can just as easily let go of its own straps and fall back to zero. Without some real use or backing, there's nothing to catch it on the way to $0. And at $0, there's no guarantee of re-bootstrapping bitcoin back to some positive price. As such, Bitcoin users justifiably expect incredible returns from bitcoin holdings in order to bear the risk of a zero-value equilibrium. Expected hyperdeflation is the carrot that must be proffered up for risky cryptocoins to be held. When those expectations of price appreciation aren't met, a large crash in the current price (relative to its future expected price) is necessary in order to tempt the next crop of speculators to hold it again. Thus bitcoin's pattern of incredible rises, or hyperdeflation, followed by 50% flash crashes, followed by the next round of hyperdeflation.

So if unbacked, useless, and costless objects can be imbued with a positive price via Selgin's painting-the-tape story, why isn't everyone doing it? But they are! Attracted by the potential for large gains, plenty of people are creating alt-coins, as I wrote here and here. In theory, their combined greediness should have the effect of swamping the market with fiat objects, driving their price towards the cost of production. The idea here is similar to the Somali shilling story, in which continual counterfeiting of old fiat shilling notes drove their price down to the cost of production, namely the costs of paper, printing, and shipment.

This hasn't happened yet with bitcoin, which is hovering at around $950. In my old post Milton Friedman and the mania in "copy-paste" cryptocoins, I hypothesized that the seeming inability of competitors to drive bitcoin prices down had something to do with the unassailable benefits that bitcoin enjoys as being the first mover, including superior security and liquidity. Tyler Cowen has some interesting thoughts on this. Bitcoin has a market cap of about $20 billion. As long as Bitcoin's entrenched advantages are so supreme that it would cost $20 billion to create a competitor, then there's no profit in tackling its niche. Cowen, however, thinks that the cost of mimicking bitcoin is far less than this. Rather than being in equilibrium, the cryptocurrency market is currently working itself via a process of "supply-side arbitrage" to a new equilibrium at which bitcoin will be worth far less.

On this same topic, Nick Rowe suggests that a BackedCoin might be one of the competitors capable of carrying of this feat. I agree with Cowen and Rowe —that's why I mostly sold out of bitcoin last year, and why I plan to eventually sell my litecoin. Of course, I'm the dummy who sold BTC back at $100, so my opinions should be taken with a grain of salt.

Where will the competition come from? Robert Sams makes a good argument for why bitcoin knock offs like litecoin, sexcoin, etc., though costless to produce, can't easily compete with bitcoin itself. The mining power that goes into maintaining the integrity of the various blockchains is in scarce supply. Merchants will always congregate to the blockchain with the most security, since that will be the coin that guarantees that the threat of double-spending is the smallest. While clones can be created with a few key strokes, good security can't be bought. Thus bitcoin's price can't be competed down to $0ish by alt-coins.

I think I buy Sams's point. However, he couches his argument within the existing universe of bitcoin and its clones. I'd make the argument that the crypto phenomena through which "supply-side arbitrage" will be carried out could be something entirely different than bitcoin, say Ripple or something we haven't yet seen. Ripple for one isn't constrained by the supply of existing mining power, or hashing, since the Ripple blockchain is maintained via consensus, not by hashing miners. Is this type of security cheaper? I'm no techie, so I won't speculate. But it is something different. And though it may take a while, at some point new and different will also be cheaper.

Another bonus of the Ripple system is that the crypto currency it creates are not bootstrapped assets, they are redeemable IOUs (let's not confuse Ripple IOUs and XRP!). In Rowe's UnbackedCoin vs BackedCoin world, Ripple IOUs are the equivalent of BackedCoin. It is their backing that should protect the exchange value of Ripple IOU from the threat of competition. This very same backing frees them from the hyperdeflation-crash-hyperdeflation patten that bootstrapped coins tend to display, stability being a desirable feature among  those who want to hold an inventory of media of exchange. As long as Ripple IOUs are just as transferable & secure as bitcoin and other alt-coins, this stability will be the edge that pushes them above the crypto competition.

So in sum, worthless assets can be kickstarted into circulation, say by a group of confederates who paint the tape in a way to attract outsiders. The riskiness of these bootstrapped assets requires that they yield incredibly high returns, or constant price appreciation. However, this state of affairs can't last forever since others will be eager to issue their own competing fiat objects, including superior non-volatile competitors. If I'm right, in the future bitcoin will be a smaller part of the cryptocoin world than it it now, whereas stable-value non-bootsrapped crypto assets, like Ripple IOUs, will be a larger part of that world.

---

[1] Bitcoin may not be entirely intrinsically worthless. I have floated the idea before that bitcoin has commodity value as a symbol of geek cred.

How much are Warren Mosler's business cards worth?

While it somewhat lacked in structure, a good time was had by everyone in the live chat section of the big Warren Mosler vs Bob Murphy debate last night. You can see a replay of the debate here.

One of Mosler's running themes (and which Murphy turned into a running gag) was the idea that he (Mosler) could give his business card a positive price by requiring members of the audience to hand over said card to Mosler's bouncer should they wish to leave the room unmolested. Voilà, modern fiat money. In order to get his fiat cards out into the audience, presumably Mosler would have had to spend them into existence by purchasing stuff from the the audience. Only then would audience members be able to afford to leave the room without being subjected to the bouncer's whims.

In the MMT literature, this is called twintopt.* A state imposes an obligation on its citizenry to pay a tax, and then dictates what good or item (the "twintopt") will be sufficient to discharge that debt. Whatever is made into twintopt, be it US dollars, coupons, or business cards, they will now be valued and circulate. This is the tax-drives-money, or chartal theory of money.

I don't doubt that the theory could be true in practice. For instance, I describe in this post how McDonald's Corporation could give colourful bits of paper coupons a positive value by imposing an obligation on the burger-eating community to pay for all Big Macs with these coupons. No doubt Mosler's business cards could also earn a positive value if he required people to submit cards as an exit fee. Rather than chartalism, I prefer to call this the coupon theory of money in order to underscore that it has nothing to do with the state. Corporations and individuals can issue "chartal" coupon media of exchange just as easily as the State can.

But Mosler's business card analogy doesn't make it as far as US dollars. To see why, let's go the reverse direction and remove twinopt status.

Having been converted by Murphy, let's say that Mosler's bouncer becomes a libertarian and moves to New Hampshire.  With twinopt no longer being enforced, audience members can now leave the room without having to post a business card as payment and the market value of Mosler's cards will quickly fall back to 0. So far the chartal tax-drives-money theory of money is borne out perfectly.

Next let's migrate this idea over to the modern US monetary system. Say that the US government announces that it will no longer accept US paper dollars or US dollar-denominated cheques/deposits as payment for taxes and instead will require Mosler-issued business cards. What happens to the value of the US dollar? According to the taxes-drives-money theory, the US dollar, which no longer serves as twintopt, should rapidly become valueless, just as Mosler's business cards became valueless after the desertion of his bouncer.

Here's what actually happens. Since US paper currency and US-denominated deposits & reserves are now useless for the set of transactions involving the US government, individuals, banks, and corporations will all seek to simultaneously reduce their inventories of these instruments. After all, they're a less effective medium of exchange.

This disinclination to hold dollars will in turn drive prices up. In order to ensure that prices don't rise above their inflation target, the Federal Reserve will suck up and destroy all unwanted currency and reserves. The Fed does so by selling assets. It has gold, bonds, foreign exchange, economics books, and other things in its vaults which it will continuously use to purchase dollars until the urge to divest dollars has been quenched. Thus the run on US dollars set off by their lack of tax acceptability is just as quickly matched by a decreased supply.

Mosler's business cards will probably be fairly liquid thanks to their government prop. The interesting thing is that despite no longer being required to pay taxes, the dollar will continue to have the same value as before.  And while it will be useless in government transactions, the private sector will probably continue to use US paper currency and US dollar reserves among each other as a handy medium of exchange. Something other than taxes drives the value of US money, it would seem.

New-fangled metallism provides an explanation. I say new-fangled because it's not old-fashioned metallic backing per se that helps pin down the dollar. Rather, what does the pinning is the general financial assets that the central bank holds (MBS, bonds, stocks, gold, whatever), combined with the bank's threat to use those assets to ensure price stability. Taxes certainly contribute to the dynamic, but they're not in the driver's seat.

---

*Specifically, Understanding Modern Money by L. Randall Wray

The Social Contrivance of Money... a bit contrived?

An Old Man and his Grandson, circa 1490. Domenico Ghirlandaio

The title of this post comes from this Paul Samuelson paper. Paul Krugman mentioned the paper last week, as did a few others, including Bob Murphy and Garett Jones. Nick Rowe has brought it up a few times (here for instance). So I trudged through it. Here are a few quick thoughts.

The setup goes something like this. Samuelson comes up with a science fiction world in which there's a young generation and an old generation. There are no durable goods. So if the current generation can't save, how can it prepare for a retirement in which it can hardly produce anything? Retirement will be "brutish".

On the other hand, what if they print "oblongs of paper" and "officially through the state, or unofficially through custom, make a grand consensus on the use of these greenbacks as a money of exchange." Then the current generation will be able to acquire some of these paper bits and exchange them in their old age to the young for consumption. The world is now a better place, thanks to the social contrivance of fiat money. [Note: by fiat, I mean intrinsically useless and valueless].

This is an interesting model. But just because the model says that the social contrivance of fiat money is an optimal solution doesn't mean that modern central bank money, which by all means appears to be like the fiat money in Samuelson's model, is likewise a social contrivance.

Samuelson never addresses the incredible degree of inter-generational coordination that would be required to arrive at a social contrivance that dictates that useless objects must be accepted. Why would future generations accept bits of paper if they don't know from the start that the next generation won't laugh at it? Think of the incredible amount of government resources that would be required to enforce the "grand consensus" or, absent force, the quantity of storytelling that would be necessary to instill the customs necessary for fiat to always be accepted.

Samuelson's "social contrivance" is in many ways similar to the idea of a Walrasian auctioneer. Both mechanisms do herculean amounts of work. But outside of an overlapping generations model, this work would be tremendously costly. Now I don't doubt that we could arrive at a social contrivance of fiat money if we really put the resources into it, but why bother? If we want items that are capable of storing wealth over time, we can skimp on the tremendous effort of running a fiat scheme by using things that already have some intrinsic value. A modern central bank that issues well-backed credit will do, as will commodities like gold. The intrinsic value of a commodity or credit item will adequately secure the confidence of the various generations that any exchange they participate in will be fair and quid pro quo. Intrinsic value thereby performs the same function as Samuelson's social contrivance… but at far less cost.

The other thing I noticed about Samuelson's model is that he misappropriates the word "money".  Money is commonly considered to be a highly liquid object, but in Samuelson's model his green bits of paper only get passed off once every generation. For the rest of time they are simply held. The bits of paper that restore his science fiction world to health could equally be called guano, social security, or whatever. In short, his model explains the possibility of fiat stores of value, but not fiat means-of-exchange.

So to sum up, using Samuelson's model of fiat money in order to explain modern central bank money doesn't follow. Social contrivances are expensive. Far easier to secure each generations' trust and participation by setting up a central bank that issues intrinsically valuable and well-backed notes and deposits. A central bank therefore spares society from spending the mass quantity of resources that a socially contrived fiat item would require.  Modern central bank liabilities may look like the socially-contrived green bits of paper in Samuelson's paper, but they aren't.

Questions for Bob Murphy and other Austrians on the inevitability of the bust

David Glasner had some recent posts (here and here) on Ludwig von Mises and Austrian Business Cycle Theory (ABCT). Bob Murphy pushed back here with a good rebuttal. But David's general point still stands: what necessarily forces a central bank that has adopted the practice of lending at a rate below the natural rate to ever cease this practice? Why does there have to be an inevitable bust?

I consider myself an Austrian in that one of my favorite economists is Carl Menger. I've also written a thing or two for the Mises Institute, my most recent being on Menger and Leon Walras and how the two would have differed on the phenomenon of high frequency trading. On the other hand, when it comes to macroeconomics, I remain a business cycle agnostic. I'm willing to be converted though. All you've got to do is answer a few questions of mine.

Say a central bank decides to reduce the rate at which it lends below the natural rate. Businesses can come to it for cheap loans -- and they do. Mises points out that as long as this differential exists it'll eventually lead to a "crack-up boom". The currency enters hyperinflation stage and, at its peak, people either turn to barter or dollarization occurs. Alarmed at this prospect, the central bank will probably increase rates in order to stave off the crack up.

But say the central bank and the currency users exists on a small island far from everywhere so dollarization can't happen. Say also that the police force is vigilant about preventing people from bartering. As a result, the currency issued by the central bank continues to be used, even during hyperinflation. The inevitable flight from money — the crack-up boom — can't occur. The currency perpetually falls.

So having assumed the crack-up boom away, why should the setting of market rates below the natural rate inevitably end in a bust? Sure, in the interim there might be a redirection of capital towards projects that are only profitable at low rates. In this context, a sudden increase in rates by the central bank back up to the natural rate might show some of these projects to be unprofitable. You've got a bust of sorts. But our central bank, released from the disciplining threat of a crack-up boom, steadfastly refuses to raise rates.

So if rates can be kept perpetually too low, and a crack-up boom can be averted, what causes the bust?

To start off, one explanation for a bust occurring is that when rates are kept too low, excess resources are allocated to interest-sensitive distant projects and not enough to less interest-sensitive near-term projects. At some point there's a realization that not enough capital has been allocated to present needs and all those future projects suddenly collapse in value. Thus a bust. What causes this sudden epiphany? As David Glasner asks, are workers dying of starvation?  It can't be higher interest rates that render these projects unprofitable since, as I've already pointed out, the central bank keeps rates permanently low.

Even if capital begins to flow into projects that are only profitable at low rates, wouldn't the prices of materials required by those projects be bid up relative to other prices, thereby putting a quick end to the profitability of these distant projects? Wouldn't the relative prices of material required for near term projects fall, thereby increasing the profitability of near term projects? How can any significant capital misallocation proceed given these rapid relative price adjustments?

If you can answer all my questions, then you'll have successfully converted me.

Do credit-induced asset price bubbles show up in GDP?

Having read Larry White's book on free banking (pdf) and a number of George Selgin's papers I consider myself to be an advocate of free banking. That being said, I can't help but wonder about a few of George's recent points in his post on Intermediate Spending Booms, the most recent in a series of posts that trains a critical eye on market monetarists. Here is George:

But in seeking to free monetary theory and policy from the Keynesian overemphasis on interest rates, the Market Monetarists tend to downplay the extent to which central banks can cause or aggravate unsustainable asset price movements by means of policies that drive interest rates away from their "natural" values. Such distortions can be significant even when they don’t involve exceptionally rapid growth in nominal income, because measures of nominal income, including nominal GDP, do not measure financial activity or activity at early stages of production.

George is saying that nominal GDP might not properly capture the effects of a central bank setting its rates below the natural rate because it doesn't measure a few key variables, namely financial activity and early-stage investment projects. This sounds somewhat like a rechristening of the classic Austrian complaint against traditional measures of inflation. Here, for instance, is an article by Bob Murphy talking about how relatively tepid changes in CPI might mask credit-induced asset bubbles.

As I pointed out in my comment on George's blog, GDP calculations include investment, presumably much of which is in the early stages of production. GDP also includes inventories. Bill Woolsey describes this better than I can.

As for changes in financial activity due to excess credit, I think GDP should capture it pretty quickly. The next little bit is just a paraphrasing of Fritz Machlup's The Stock Market, Credit, and Capital (pdf). Machlup wrote it to counter claims that the stock market was capable of "tying up capital". It's a great read.

Fritz Machlup

Say artificially low interest rates convince a speculator to borrow from a bank in order to fund the purchase of a stock. When the transaction is completed the speculator owns the stock. On the opposite side of the transaction, the seller of the stock has been freed of her position and owns cash. She in turn can do two  things with this cash. First, she might buy goods. This will immediately show up in GDP. Alternatively, she can buy another stock, in which case a third person now owns the cash. This third person can in turn either buy goods, which registers in GDP, or purchase new stock from a fourth person. This fourth person can.... you get the point. The process proceeds fairly quickly until someone in the chain purchases a good, thereby allowing GDP to capture the effect of excess credit.*

Another factor limiting the ability of long chains of stock transaction to tie up capital is that the longer the chain continues, the more likely stock prices are to be bid up. At higher prices, firms are more willing to finance themselves by issuing new shares since their cost of capital has fallen. This is because they can raise more today than the day before while issuing the same amount of shares. When firms issue new shares they drain the purchasing power originally created by excess credit creation out of the market and invest it in new capital. This allows GDP to ultimately capture the effect.

So a decline in market rates below the natural rate will result in more credit, and this credit could very well be used to purchase stocks, and this will put upward pressure on prices. But just as quickly as credit is used to buy stocks, purchasing power is released from the stock market as the sellers of stocks use the proceeds to buy real goods. GDP measures will capture the effect. I think this process happens fairly quickly given the agileness of financial markets. Maybe George thinks these chains can persist for some time.

One interesting side note. Say that the purchasing power created by excess lending exits the stock market when someone in the chain purchases used goods, say an old couch. Second hand goods transactions are not included in GDP calculations. So in this case, the effects of excess credit might not show up in GDP. It's for this reason that Nick Rowe  prefers the value of total transactions ( P x T) to nominal GDP (P x Y) as his choice indicator. Incidentally, George too invokes the idea that measures of transactions might be better indicators of monetary conditions than income measures:

When interest rates are below their natural levels, spending is re-directed toward those earlier stages of production, causing total nominal spending (Fisher’s P x T) to expand more than measured nominal income (P x y) 

It would seem then that market monetarists like Nick Rowe and Selgin do have some things in common.

*There is a third thing that can be done with the cash. It can be held. But if people do this, then the issuing bank never issued excess credit in the first place. Sufficient demand already existed in the economy for bank liquidity and this is expressed by the fact that people willingly decide to hold those newly created deposits.

The world of monetary affairs in the 1920s and 30s: a complex affair

Bob Murphy asked whether Lionel Robbins was right in saying that central bank policy in the late 1920s and early 1930s was a complete reversal of traditional central bank doctrine.

A blogger named Lord Keynes, (perhaps the ghost of Keynes? ), takes exception to this idea, noting somewhat dramatically that Murphy is "dead wrong" and "utterly absurd".

These sorts of us vs. them dramatics would be best left to the likes of professional sports casters (economic history is not a competition), but I'm going to look past the silly theatrics so as to delve into what is a very interesting issue.

The nub of the debate, in my view at least, boils down to the definition of traditional central banking doctrine. My conclusion, which I'll get around to explaining, is that compared to the 1800s, central bank policy between 1929-1932 was probably a complete reversal. But compared to Fed policy through most of the 1920s, the Fed's policy during the Depression was similar. One major categorical difference between 1923 and 1932 was the broad powers provided by the Glass Steagall Act. More on that later.

First, what is it that Lionel Robbins said? Here is a snippet:

Now in the pre-war business depression a very clear policy had been developed to deal with this situation. The maxim adopted by central banks for dealing with financial crises was to discount freely on good security, but to keep the rate of discount high. (Robbins, The Great Depression (pdf), Pg 72)

What Robbins is enunciating here is basically the classical advice of Walter Bagehot. In a crisis, central bankers should accept all collateral in return for loans, but should only do so at a penalty rate. (The latter is what Robbins is referring to when he says that a central bank must "keep the rate of discount high"). By penalty rate, Bagehot meant that central bank should only lend at some interest rate above the current market rate. In modern parlance, this is like saying that the central bank must force borrowers to take a haircut on their collateral.

A Bagehotian central bank is an extremely passive one. As a pure lender, it must wait for banks to come to it, as opposed to a buying central bank which can initiate transactions on its own behalf. Furthermore, a Bagehotian central bank can't influence banks to borrow at its discount window by lowering its own rate below the market's, since it must lend at a penalty rate. It can drop its rates to some low level, but only after the market has done so.

Walter Bagehot

The sort of central banking that Robbins would contrast to the traditional Bagehot view is the post-WWI central banking practice that first arose in the US in the early 1920s. How was it different? Here, for instance, Allan Meltzer describes a New York Fed governor Benjamin Strong's position in 1921 on lending at a penalty rate:

He [Strong] believed the correct policy was to lend freely at a penalty rate, and he again cited Bagehot's rule. In July he continued to favor a penalty rate in principle, but recognized that the principle had to give way. He told [Montagu] Norman that money conditions "hardly justified.... making a further reduction." There were other considerations, however that made classical methods "not always the wisest," and he added, there were "political considerations brought about by the change of administration". History of the Federal Reserve, Vol 1, pg 125-126

By the early 1920s, due to a combination of factors, some of them political, others technical, the Fed had effectively stopped discounting at a penalty rate.

At the same time the Fed had begun to use open market operations in monetary policy, not just discounts. Daniel Kuehn touches on this here. Open market operations were not entirely unknown to Bagehotian central banks. But they were a sideshow to discount window lending. The thinking behind the Fed's unveiling of open market operations can be found in the Fed's 1923 Annual Report, in which the Riefler-Burgess doctrine (a term coined by Meltzer) was first described.

The idea behind Riefler-Burgess was that if the Fed embarked on a campaign of open market operations, member banks would have more reserves and could borrow less at the discount window. This would lower their borrowing costs and allow them to expand their lending, thereby buoying the economy. Rather than constantly varying discount rates, often a politically unpopular option, the Fed could simply engage in an activist open market sales/purchases program so as drive banks to, or pull banks away from the discount window, and thereby slow or spur on the economy. As Meltzer points out, discount rates now played second fiddle to open market operations:

The Riefler-Burgess doctrine is ambivalent about the role of the discount rate. At most, it has a supporting role; at worst, it has little supplementary effect. Strong, who used the doctrine as a guide to policy, was ambivalent about the independent effect of discount reate changes. ibid. Pg 264

The result of all this was that by 1923, the Federal Reserve was no longer Bagehotian. Rather than a passive wallflower that stepped in only rarely, it actively participated at all points in an economy's trajectory.

The upshot is that when Lionel Robbins describes the Federal Reserve open market purchases of $410m from October 1929 to December 1930 as breaking with tradition (pg. 73), he is right... since the tradition he is referring to is Bagehot's and not the more recent 1920s Riefler-Burgess tradition. At the same time, Lord Keynes isn't wrong to point out that Fed policy in the Great Depression was similar to Fed policy during the 1920s, insofar as they both relied heavily on open market operations.

 As Lord Keynes points out in his post, there is certainly a quantitative difference between Fed policy in 1923 and 1932. See the chart below of Fed government bond holdings and bankers' acceptances, the former of which shows a terrific spike in 1932. In this post, I called the 1932 event QE0. The Fed was permitted to buy bankers' acceptances in the open market, but usually just set a fixed rate and let sellers come to it.

There is an important qualitative difference between 1923 and 1932. The difference is this. Before 1932, all Federal Reserve notes had to be backed 40% by gold and 60% by “eligible paper”. The latter was made up primarily of commercial paper. The passage of the Glass Steagall Act in February 1932 allowed government securities to be eligible, and therefore dramatically increased the ability of the Fed to engage in QE with government debt. Prior to then, the government could only engage in QE with government debt to the extent that it already had excess gold and eligible paper. It was constrained. The effects of this constraint, and its removal, is best seen in the chart by the orange line, which shows the Fed's government bond-to-gold ratio. Having never risen above 20%, the passage of Glass Steagall allowed the ratio to rise to  60% in 1934.

In a nutshell, the Fed could never have done what it did in 1932 in 1923.

QE-zero

Bob Murphy asks if central bank actions taken during the early 1930s might be considered "unprecedented". In the comments I pointed out that during that era an early form of QE was tried. I'm not referring here to the famous 1933 Roosevelt purchases of gold that market monetarists often point to. For instance, see David Glasner here, David Beckworth here, and Scott Sumner here. Scott also has a very interesting paper on the 1933 gold purchasing program (pdf). No, I was referring to the 1932 treasury purchasing program.

I'm going to replicate the simple graphical analysis that market monetarists use in order to look at the 1932 episode. See this post by Lars Christensen, for example, who overlays important monetary events (QE1, QE2, LTRO) over the S&P500.

Here is the context. Prior to 1932, the Federal Reserve system was significantly limited in its ability to embark on large purchases of government securities. This was because of strict backing laws in the Federal Reserve Act that limited eligible backing assets to gold and assets accepted as collateral for Fed discount loans, primarily commercial paper. In effect, the Reserve banks could only purchase government debt to the extent that there was already excess gold and discounted assets on the Reserve bank balance sheets.

This limitation was removed with the passage of the Glass Steagall Act of February 1932, which allowed the Fed to include government debt as backing for notes and deposits. Almost immediately the Federal Reserve began a large scale asset purchasing program that increased the system's government debt portfolio from $743 million at the end of February 1932 to $1413m by May. The program, which I'll call QE0, continued at a slower rate after May, eventually hitting a peak just above $1800m by the end of July 1932. I overlay this on the Dow Jones Industrial Average.

The second chart extends the time frame to include 1933, putting QE0 on a scale with the Roosevelt devaluation.

Gavyn Davies, who has treaded this path before, notes that Milton Friedman and Anna Schwartz declared QE0 to be a success. In their Monetary History of the United States, the two drew attention to the conjunction of QE0 with a lull in bank failures and a "tapering off of the in the decline in the stock of money". They point to the bottoming of industrial production in August, five months after QE0 started, as a sign of its success. In his History of the Federal Reserve, Allan Meltzer also strikes a note of optimism when he discusses QE0, noting many of the same improvements in data that Friedman and Schwartz point to. Meltzer writes that "it seems likely that had purchases continued, the collapse of the monetary system during the winter of 1933 might have been avoided" and notes the rise in stock prices beginning in July as evidence.

But no market monetarist would agree with Friedman and Schwartz's analysis, since the new breed of monetarists take asset prices as the best indication of monetary stance. Scott Sumner points out here, for instance, that US equity markets had one of their fastest two day rallies in history as President Hoover met with Congressional leaders to begin work on Glass Steagall. All good, then, for the market monetarist stance, who like to see rising market prices coincide with easy monetary policy at the zero lower bound. Unfortunately for them that was the end of the rally. Markets continued falling to new lows even as QE0 accelerated. Scott Sumner indeed notes that "In many respects, the period from April to July 1932 was the worst three months of the entire Depression. Commodity prices continued to fall, and both stock prices and industrial production reached their Depression lows in July." Oddly enough, only with the end of the QE0 did stock prices begin to rise again, as the first chart shows, which runs contra to market monetarist thinking.

No wonder then that market monetarists prefer to look at the second chart. In 1933, the conjunction of increases in stock prices with various monetary events, including the departure of the dollar from gold convertibility and Roosevelt's gold purchase plan, is quite striking. This cozy relationship is no doubt the main reason that market monetarists prefer to point to 1933 rather than QE0 for evidence of monetary policy effectiveness at the zero lower bound.

QE0's seeming failure might seem to confirm Murray Rothbard's view that the huge increase in the money supply engendered by QE0 "endangered public confidence in the government's ability to maintain the dollar on the gold standard," leading to a loss of confidence on the part of foreigners who drew out gold, and on the part of Americans who converted deposits into notes. This turned an intended inflation into an unintended deflation. The aboves is also Peter Temin's view, who points out that the purchases reduced confidence, the resulting gold outflow nullifying QE0's potential for expansion.

My reading of Scott Sumner is that the 1932 purchasing program was rendered ineffective because of growing expectations that the dollar would float, leading to gold ouflows and an ensuing general panic in equity markets. In meting out blame for this panic, Sumner emphasizes the role of Congress in engendering uncertainty rather than the Fed's QE0 program. Once the dollar panic was alleviated and the hoarding instinct of foreign central banks and the private sector satiated, markets began their rise in the latter half of 1932.

Hsieh and Romer (pdf), on the other hand, use data on dollar forward rates to show that traders were not particularly worried about a dollar devaluation. If H&R are right, then one can only conclude that there was no dollar crisis, leaving market monetarists with no corresponding event to blame for counterbalancing the inflationary effects of QE0. So QE0, it would seem, was irrelevant -- a non-event. Scott talks about Hsieh and Romer's paper here. It all seems rather tortured to me, and leads me to (somewhat dismally) conclude that one can probably get a set of historical events to say almost anything one wants it to say. This is not a criticism of Scott, but one of economics in general.

All of this leads to current discussion of QE3. The New Keynesians point to the ineffectiveness of QE itself at the zero lower bound. For instance, see Simon Wren Lewis. This view is inherited from John Maynard Keynes who, it would seem, got it from his observations of the failure of QE0 in 1932. Here is Keynes in Chapter 15 of the General Theory:

There is the possibility, for the reasons discussed above, that, after the rate of interest has fallen to a certain level, liquidity-preference may become virtually absolute in the sense that almost everyone prefers cash to holding a debt which yields so low a rate of interest. In this event the monetary authority would have lost effective control over the rate of interest. But whilst this limiting case might become practically important in future, I know of no example of it hitherto. Indeed, owing to the unwillingness of most monetary authorities to deal boldly in debts of long term, there has not been much opportunity for a test. Moreover, if such a situation were to arise, it would mean that the public authority itself could borrow through the banking system on an unlimited scale at a nominal rate of interest.

The most striking examples of a complete breakdown of stability in the rate of interest, due to the liquidity function flattening out in one direction or the other, have occurred in very abnormal circumstances. In Russia and Central Europe after the war a currency crisis or flight from the currency was experienced, when no one could be induced to retain holdings either of money or of debts on any terms whatever, and even a high and rising rate of interest was unable to keep pace with the marginal efficiency of capital (especially of stocks of liquid goods) under the influence of the expectation of an ever greater fall in the value of money; whilst in the United States at certain dates in 1932 there was a crisis of the opposite kind — a financial crisis or crisis of liquidation, when scarcely anyone could be induced to part with holdings of money on any reasonable terms.

The market monetarists, of course, believe in the effectiveness of QE, although announcing a nominal target would greatly improve a purchase program's effectiveness.

This is what Nick Rowe means when he says that there are two types of economists (HT Bob Murphy). There are those who think monetary policy is useless at the zero lower bound, and those who don't. I wonder how much of the divergence between these two traditions has its origins in the data generated by the separate 1932 and 1933 monetary events. If you focused on the latter, you became a monetary policy believer, if you focused on the former you stopped believing.

Other posts on the efficacy of QE or lack thereof:

Stephen Williamson (here and here), Bruegel blog, Richard Serlin, Miles Kimball (here and here), Paul Krugman (here and here), James Hamilton, John Taylor, John Cochrane, Michael Woodford (pdf), and Simon Wren Lewis.

The natural rate of interest and the own-rate argument

The Austrian vs Keynesian end of the blogosphere often battle over the existence of a natural rate of interest. The Keynesian side typically points to Piero Sraffa's argument that there are many natural rates of interest, or own-rates, and therefore an Austrian sort of monolithic natural rate of interest simply doesn't exist.

Over the last few weeks I've participated in the comments here at Jonathan Finegold Catalan's blog and here at Daniel Kuehn's blog. Here is an older comment in this vein on "Lord Keynes" blog. Bob Murphy also has a paper (pdf) on this subject and has commented on the above blogs on this subject.

Debt, generations, savings, and economic categorization or the "Borges Problem"

I didn't comment much on the great debt debate, stirred up a Krugman post called Debt Is (Mostly) Money We Owe to Ourselves, but followed it quite closely.

Nick Rowe taught me (here, here, here, and here), and Bob Murphy clarified (here, here, here, here, here, here, here, and here), that present generations can indeed take resources from future generations via debt issuance.

I also learnt via Daniel Kuehn here and here that if you use a very unintuitive definition of "generations", than this is not the case. Basically, you can swap the meanings of terms to argue your way out of a tight spot.

My comment is from a Murphy post:

I’ve learnt that the method by which one aggregates individuals into groups, and the labels that one attaches to such groups, can have an important influence on a debate’s ability to reach resolution. If people are aggregating differently, and using non-standard words for their categories, then the debate will degenerate into shouting matches.

In a comment on a post called Why "saving" should be abolished, Nick describes this as the Borges Problem, which I rather like. Says Nick,

Let me first do one general response:

 There are lots of different ways we can divide up the world into categories see Borges on "animals" http://en.wikipedia.org/wiki/Celestial_Emporium_of_Benevolent_Knowledge%27s_Taxonomy

 Which would be the most useful way to divide up income, and define saving?

 Which of these 3 definitions of desired saving is the most useful?

Nick later on:

Notice also that the recent debate about the burden of the debt was also an example of the "Borges Problem". Do we divide the future up into time periods or into cohorts? We get very different results depending on how we categorise the world. And sometimes the categories we use are chosen by someone long ago who had a totally different purpose and/or a totally different theory to ours. Our way of seeing the world gets distorted by the dead hand of historical ways of seeing.

Yes! I did notice that. It caused me a lot of confusion. Nick also notes that the solution is to choose the most useful categorization out of all possible options, and proceeds to advocate a different category to which we should attach the word "savings". Interesting stuff. I'm not sure how Nick proposes we solve for "usefulness" though. Isn't the fact that almost everyone uses the same term for a given categorization a good enough claim for usefulness?

Here's another Rowe comment on Kuehn's blog which is relevant:

Put it another way: there's more than one way to aggregate. We shouldn't let our theories of what is happening in the world be determined by the choices made by long-dead National Income Accountants.

Anyways, in my comment on Nick's savings post, I proposed a more useful (at least to me) Borgian response to the categorization problem. Instead of categorizing the world on the basis of flows, categorize it as a series of balance sheets, or stocks. The result is that consumption, investment, and savings are all attached to entirely different bins (and more intuitive ones, to me at least) than in a world composed in terms of flows:

Nick, I agree with you that the conversation on debt was mainly about categorizations and the lack of standardized terms associated with categorizations. That made it very frustrating to follow.

So I am all in favor of standardizing terms, as you advocate in this post. 

I noticed you originally introduced C and I as flows and A and M as stocks. Then when you brought in the individual's economy, you introduced not a stock of antique furniture, but a flow of antiques, and not a stock of money, but a flow of money. Presumably you did this to preserve stock flow consistency.

The idea of a flow of antiques or money is very unintuitive to me. Why not go the other way? Not flows of consumption and investment, but stocks? Thus you have and individual's goods C, I, A, and M, which are all stocks. Sum them all up and you have S (the noun form of S, not the verb). This S can rise or fall. As a solution to the Borgian categorization problem, this configuration makes more intuitive sense to me.

And later:

N: "but if we think of income as a flow, then thinking of C and I as stocks is going to create problems."

 Me: You start out with the C and I that you have produced in your stock of assets, hold this C and I until you find someone who'll exchange for them with the M they have in their stock of assets. Now they are holding C and I and you are holding M. So here income isn't a flow, it's just a trade, an instantaneous swap of assets held in a portfolio.

 How much of economics is taken up by definitional debates and confusion? You'd think there would be a universal set of definitions for economic terms somewhere so these issues don't pop up. When I read William Hutt's books I'm always pleased because he uses his first chapter to explicitly define every term he'll be using.

and once more *phew*:

N: "Will those trades all take place in an instant, with some buying and some selling a stock of antiques? Or will those trades happen slowly over time, as people buy or sell a flow of antiques, and slowly get back to their long run desired stocks? That depends. If antiques are a small part of your wealth, and the market is frictionless with all antiques identical and so zero search costs (obviously not, for antiques). Each person would instantly buy or sell a stock of antiques to get back to his personal desired stock. Otherwise, there will be a flow of trades. If antiques are a large fraction of your wealth, you may only buy and sell slowly, in a flow."

 me: Ok, thinking in a world with stocks, (an infinite series of balance sheets), trades still happen in an instant, even if you introduce search costs. You hold the antique on your balance sheet until you don't. The antique is in your hand up until the moment it enters the hand of the buyer.

 Introducing frictions means that someone can have the intention of selling that antique and will need to incur costs to search out someone to trade. But it doesn't mean the process must be a conceptualized as a flow. Rather, the intention of selling an antique just moves the antique to a different part of an individual's balance sheet. It continues to lie in the asset column of their balance sheet, but is moved from long-term assets to current or liquid assets. Introducing search costs means that instead of an interval of two balance sheets before a swap occurring, the interval is some number larger than two.

My rough final thoughts are that thinking in terms of stocks, not flows, introduces a number of important categories that flow-based economics ignores because it is focused on flows. The most important of these is a stock of non-durable consumption goods. In flow-based economics, it's always been odd to me that factories produce, and we instantaneously use up, consumption goods.

A stock based world also is terribly confusing way to go about things, because the word savings in a flow-based world is attached to a different category than that which it is attached to in a stock based world, much like how in the Great Debt debate the word "our children" can be attached to either a period of time or a cohort.