Kaminskia on Free Banking

Izabella Kaminska wrote a defense of the Bank of England on her blog at the Financial Times . It was apparently motivated by the new "positive money" proposal to fully separate money from debt and banking along with those who would like to see bitcoins replace fiat currency.   (I am pretty dismissive of both of those groups as well, and might defend central banking against them, despite my free banking sympathies.)

 However, she made a passing criticism of free banking:

As an aside, it’s worth pointing out that the Scottish period of free banking that preceded the great inflation — often touted by free-banking enthusiasts as an excellent example of how the free-banking model is inherently stable — revealed how cartels and monopolies could be used to stabilise the currency. In fact, it was only because the Scottish banks were so good at forging oligopolistic cartels that happily restricted competition that the Scottish free-banking period proved so stable in the first place (defeating the pro free-banking argument altogether).

The link is to a blog called   "SOCIAL DEMOCRACY FOR THE 21ST CENTURY: A POST KEYNESIAN PERSPECTIVE," which is not a title that generates much credibility with me.   Worse, when you try to find out the author of the blog, there is nothing.

However, the linked post at least cites Goodhart (1987,) which seriously critiqued free banking on a theoretical level.   Most of the rest of the post was pretty weak   The various episodes described as "free banking" were shown to have included at least some some government regulation of banking.   

George Selgin responded to Kaminska in the comments, and I thought he was much too harsh.    I admit that this is the frying pan calling the kettle black and that I am suggesting that Seglin do as  I say and not as I do, but at first pass, playing the role of the patient professor might have been a better place to start.

Still, I agree with Selgin that Kaminska's version of 19th century British monetary history was more than a bit skewed.    She certainly seemed to suggest that Peel's Act was a necessary corrective to inflation caused by the uncontrolled issue of paper currency by "country" banks in England.   Of course, it also spelled the end of the Scottish system of private note issue as well.   But really?  Does anyone think that Peel's Act was necessary or even sensible?   100% marginal reserve requirements for hand-to-hand currency?  Why?

Kaminska appeared to have replied to Selgin on her blog--Dizzynomics.    She didn't refer to him by name, and if Selgin's comment on her original article was too harsh, her reply was absurd   Free banking advocates (Selgin?) seem to be "reason and logic deniers."    Apparently, the key element of reason and logic they deny is:

But the main issue I have with them is that they appear to have no understanding or appreciation of the cyclicality of systems or the fact that whenever we’ve had free-banking systems they’ve resulted in chaos or alternatively co-beneficial collusion to the point the system is not free by the standard definition of free.

I don't think the "cyclicality of systems," is a principle of reason or logic.   The claim about "chaos" is hyperbole at best.

Kaminska writes on in a way that suggests that "free banking" means the monetary institutions of anarcho-capitalism.   Since many of the leading lights of "free banking" lean in that direction, and even more so those they have influenced, I can imagine that one can find people who will argue that all potential bank misbehavior can be handled by private arbitration.   While that may or may not be true, what I count as "free banking" hardly  requires that the banking industry be singled out for fully-privatized law enforcement.

Perhaps more interesting is her claim that more-or-less successful free banking systems are not truly "free" but rather instances of collusive oligopoly.

Unlike many, if not most, free bankers, I see free banking as a series of reforms rather than conceiving of a banking system without any government intervention.

One key reform, and one that superficially seems radical, is the private issue of redeemable hand-to-hand currency.   In my view, redeemability is pretty much all that is necessary to keep a competitive banking system in line.   By that I mean,each bank's market share in the issue of currency will reflect the preferences of those using the currency and the total amount of currency issued will reflect the preferences of the public to use currency relative to deposits.

Now, this constraint works though a clearing system.   And while there have been private clearinghouses that have handled the task quite well, this particular reform is consistent with having a central bank handle the clearing system--like the Federal Reserve.   The way I see it, whether or not it is desirable to privatize the clearing system or leave it in the hands of a government-run "bank" is a separate question from whether or not redeemable privately-issued hand-to-hand currency creates special problems.

A second reform, which I had thought was no longer controversial, is branch versus unit banking.   In the U.S., there were many regulations aimed at protecting unit banks.   These would be single office banks.   To bring it back to Kaminska's history, some of the regulations of the "country banks" in 19th century England were aimed at keeping them small.   One strength of some of the "free banking" systems, including both Scotland and Canada, was widespread branching.

The key reason why branch banking is superior to unit banking is geographical diversification.   To the degree that a unit bank receives deposits locally and makes local loans, when the local economy is troubled, the bank will have trouble collecting on loans exactly when its depositors are in need of their funds.   In the U.S., a key rationale for deposit insurance was to overcome this weakness.

However, the branch banking system does make private currency more feasible.   While thousands of unit banks might each issue private currency, and the currencies might each do just fine in their local community, the experience of the U.S. "free banking" era suggests that that they were less than suitable for a national currency.

But really, unit banking was not all that suitable for promoting a payments system by check either.   And so a system of banks,each with extensive branching, not only provides geographical diversification, it also allows both privately-issued hand-to-hand currency and checkable deposits to form a national payments system.

Now, is a nationally branched system an "oligopoly?"    In fact, the unit banking system was more about protecting local monopoly than creating competition by having a large number of banks.   In practice,a system made up of a smaller number of banks, able to open and close branches in various localities, has proven more competitive than a system of many small banks each tied to a particular locality.

Would a U.S. banking structure of 20 large banks all with branches nearly everywhere be "oligopolistic?"   Well, I suppose it doesn't meet the neo-classical definition of perfect competition.   But that is hardly a reasonable standard for real world competition.

There are more reforms that are associated with free banking.   Some are no-brainers like ending reserve requirements or allowing free entry.   Others are much more challenging, like ending deposit insurance and capital requirements.  But rather than treat them separately, I will focus on what I consider the point where the "oligopoly" and "collusion" charge is most likely to stick, and that is the clearinghouse.

This points to Goodhart's criticism of free banking.   Consider a clearinghouse organized as a private club. (Say each bank owning stock in the clearinghouse in proportion to capital?)    For the payments system to work well for the nonbanking public, each bank needs to accept banknotes and checks drawn on other banks at par, depositing them at the clearinghouse, cancelling offsetting claims, and settling up net clearing balances.

To start with, by accepting each other's items for deposit at par, the members of the clearinghouse serve as creditors to each other.   Goodhart, following Timberlake's study of U.S. practice in the 19th century, argues that the clearinghouse serves as lender of last resort.  This expands the club's role as creditor. It is important that the clearinghouse have information about its members in order to determine if they are good credit risks, but since the member banks are all competing with one another, they will not want to share that information.   QED, the Bank of England should exist.

Since Goodhart's argument was closely tied to arguments about banks being subject to runs, I didn't find it convincing    The solution to runs is an option clause.   They were banned long ago, because governments understood that it served as an alternative to holding reserves.   Encouraging banks to hold (gold) reserves was a key policy goal at the time.

Still, limiting membership to the clearinghouse would be an obvious mechanism to provide for a barrier to entry.   That each member is a creditor to the others provides a plausible rationale.   Perhaps membership could be limited to "sound" banks as proved by their unwillingness to pay more than a "fair"  interest rate on deposits or charge less than a "fair" interest rate on loans?

All this shows is that in a world without anti-trust, a clearinghouse association could be an avenue for collusion.  But we live in a world with anti-trust, so abuse of clearinghouse rules to organize and enforce a bank cartel would be illegal.   Unless, of course, free banking is taken to mean that the banking industry must be singled out for an exemption to anti-trust law.

However, I do think there is something very special about the clearinghouse.   It is what turns a variety of financial instruments into a medium of exchange.   Of course, it isn't exactly homogeneous, but retailers typically accept payments by check or electronic equivalent and deposit the funds in their own banks.   Banknotes, privately-issued hand-to-hand currency, can smoothly fit into the system.

The clearing system is very effective in limiting each bank to a market share determined by the preferences of depositors.   The same should be true of banknotes, even though it has been many years since redeemable banknotes have had much circulation.   The problem is that the aggregate quantity of money of all types is limited to the demand to hold money by some kind of response to macroeconomic disequilibrium.   This response is closely tied to the determination of the nominal anchor for the system.

Historical free banking systems were small open economies with monetary liabilities tied to gold--an international money.   As repeated by the anonymous Post-Keynesian cited by Kaminska, Goodhart's point that much of the redeemability by these free banking systems were with credit instruments drawn on a major money center should be no surprise.   What other routine calls for redemption would exist other than a demand for foreign exchange?  And while the quantity of those securities/reserves would not be fixed, neither would the quantity of bars or even gold coins from the point of view of the banking system of a small open economy in a  gold standard world.

What we think of as monetary theory these days would best apply to the entire portion of the world using the gold standard.   The price level depends on the supply and demand for gold.   While gold strikes might be inflationary, the major source of macroeconomic disruption would be shifts in the demand for gold.   It really doesn't matter if it is due to finance or fashion.   A shortage of gold requires a higher relative price for gold and so a lower price level.   Until prices and wages adjust, real output will be depressed.  Further, it is hard to see how liquidating debts based upon higher expectations of nominal income can be anything but disruptive and painful.   The notion that by having a "free banking system," some portion of the gold standard world could be insulated from these problems is implausible.

According to the wiki for free banking, cited by the anonymous post-Keynesian, an alternative to a free banking system tied to gold is one tied to a fixed quantity of fiat currency.   Selgin, in his Theory of Free Banking, describes such a scenario.    While I think the market process he describes, where nominal income tends to be stabilized, is instructive, there is something very problematic about a system where the nominal anchor depends entirely upon the demand for an asset solely held by members of  a private club--the clearinghouse.   Changes in the settlement rules at the clearinghouse could have a major impact on macroeconomic conditions.

Greenfield and Yeager's Black-Fama-Hall payments system was a type of free banking.   The constraint on the banking system was indirect convertibility.   I always argued that indirect convertibility would have its primary impact at the clearinghouse--as a practical matter, no one else would have any reason to do anything other than spend money.  In fact, I have always thought that a rule requiring indirect convertibility at the clearinghouse would be sufficient to constrain the banking system.

Practical considerations eventually convinced most of us thinking about this sort of system that indirect convertibility would need to involve some kind of index futures contract.   While Greenfield and Yeager aimed at stabilizing the price level, and Kevin Dowd has continued to promote free banking tied to that nominal anchor, I think slow steady growth of nominal income (NGDP) is a better approach.    Index futures convertibility provides tremendous flexibility regarding the nominal anchor.

I certainly don't think that having the nominal anchor determined by a private club of banks is a sensible monetary regime, no more than giving clearinghouses the right to vary the price of gold would have been sensible in a historical free banking system.   And that is where I think free banking theory is today.   What rules should be imposed on the clearing system to keep the total quantity of money created by the banking system consistent with the nominal anchor?

And to bring this back to Kaminska, the anonymous Post-Keynesian, and Goodhart, at some fundamental level, I grant that a desirable free banking order will require a clearinghouse with appropriate rules.   Collusion?   Well, I don't think that such rules should be anti-competitive, but they certainly involve constraining the banking system--what might broadly be described as regulation.

As for the historical record--I don't think that the free banking portions of the world were especially chaotic or the source of instability under the gold standard.   And the actual behavior of both the proto and actual central banks of the gold standard era most certainly caused massive macroeconomic disruption.    

Hummel on Quantitative Easing

Jeff Hummel wrote a nice article for Reason.   Most of it isn't new and is similar to his chapter from Boom and Bust Banking.

His basic argument is that Bernanke used quantitative easing to bail out the banking system to maintain the flow of credit.   This policy involved directing credit to where the Fed felt it was most needed.

This is in contrast with a more traditional monetarist approach, which Market Monetarists have emphasized.   The reason to expand the quantity of money is to prevent (or reverse) decreases in spending on output.

Hummel's emphasizes some policies (much criticized by Market Monetarists) that appear counter-productive if the goal of quantitative easing was to prevent (or reverse) decreases in spending on output.   First, the Fed undertook open market sales of Treasury bills to at least partly offset the expansion of loans to banks.   And then the Fed introduced interest on reserves--a clearly contractionary policy.

What Hummel doesn't mention and a concern that many Market Monetarists have emphasized over the last five years, is the temporary nature of the quantitative easing.   A temporary increase in the monetary base should have little effect on spending on output.   However, it should be able to support particular credit markets.

For example, suppose a central bank is committed to an inflation target.   Investors refuse to buy new issues of mortgage backed securities and sell off existing holdings.   The central bank buys the mortgage backed securities, but insists that this is only temporary.   If inflation starts to rise, it will sell off some sort of assets  or perhaps raise interest on reserves.   Any expansion in base money or broader measures of the quantity of money is temporary.   For the most part, the demand to hold this additional money expands to meet the supply.   There is little or no inflationary effect.

Perhaps more troubling, this logic appears to apply even more strongly to a central bank with a nominal GDP level target.   Suppose that nominal GDP is on target.   Further suppose that the central bank decides to support the President's plan for poor people to have homes, and so begins to buy mortgage backed securities.   The quantity of money expands, but people hold the additional money balances--they believe that the expansion is temporary.   There is little or no impact on total spending on output.   This certainly appears to create an opportunity for malinvestment.

Reserve Currency Status

What are the benefits of providing a reserve currency?

I was recently reading a post by David Glasner, where he explained that the benefit is seignorage.   Glasner was mostly responding to a proposal for replacing "the dollar" with gold.  

If we think of replacing gold with the dollar, and see the dollar as being paper hand-to-hand currency, then we can imagine foreign central banks holding stacks for $100 bills in their vaults, just as they once kept bars of gold.  

The U.S. government, then, can print up these dollar bills, and so creates a flow of revenue much as the gold mining industry received a flow of revenue from its output of money.  

But how realistic is this?

Doesn't the use of the dollar as a reserve currency really mean that various foreigners, including foreign central banks, purchase dollar-denominated bonds?   At first pass, then, I would see this as creating a ready market for low interest rate U.S. government debt.   Assuming the U.S. has a national debt, then it can be financed at lower interest rates than otherwise.

Presumably, this benefit spills over to private borrowers as well.   U.S. borrowers of all sorts can borrow in terms of our own unit of account at lower interest rates.  

To the degree that this results in a larger banking system, the demand for reserves is somewhat higher.   And further, there may even be some increase in the demand for U.S. hand-to-hand currency.   Certainly, the conventional wisdom is that there are substantial holdings of U.S. dollar notes in foreign countries.

But most of the "high finance" related to the role of reserve currency is not based upon sacks full of currency.  

Of course, I have become a "seignorage" skeptic anyway.   If there is a strong commitment to some nominal anchor other than the quantity of base money, then seeing base money as a type "paper gold" is a fallacy.   Monetary liabilities are rather a type of short term debt.    Those parts of it that are issued at a zero nominal interest rate, like tangible hand-to-hand currency represent at zero interest loan.   From this perspective, to the degree serving as reserve currency increases the demand for base money rather than other sorts of dollar-denominated debt, simply means more borrowing at lower interest rates.

Monetary Policy and Fiscal Policy

I have always been skeptical regarding "necessary" relationships between monetary policy and fiscal policy.   Most recently, these relationships supposedly play a key role distinguishing the neo-Fisherite and neo-Wicksellian approach to interest rate targeting.

The neo-Wicksellian approach to lowering the inflation rate is to raise the target for the interest rate.   Of course, this is only a tentative adjustment that might soon require a reduction in the interest rate to keep the inflation rate from falling too low.

The neo-Fisherite view is that the way to lower the inflation rate is to lower the target for the interest rate.

Which is correct supposedly depends on assumptions about fiscal policy in the long run.

In my view, if the "target" for the nominal interest rate is something close to a target for the growth rate for the quantity of money, then a lower target for the nominal interest rate will result in in a lower growth rate for the quantity of money   This will result in lower inflation.   If this is expected, then the lower inflation will lower the equilibrium nominal interest rate.  This approach doesn't necessarily involve the central bank hitting the target for the nominal interest rate consistently.

There is an alternative neo-Fisherite process that I find problematic.  If the expected future price level is tied down, then a fixed target for the nominal interest rate can be self-stabilizing.   In that situation, if the real interest rate is too high to clear markets, the price level falls.   If the expected future price level is given, then the expected inflation rate rises.   This lowers the real interest rate.  The price level level falls enough so that the expected inflation rate rises enough for the real interest rate to fall enough to clear markets.

If the price level falls too far, given the expected future price level, expected inflation will rise too much and the real interest rates will fall too low, which causes prices to rise.    The price level should gradually rise to the expected level.   The actual inflation rate should equal the expected inflation rate.

The neo-Fisherite result follows because if the nominal interest rate is increased, then it immediately would raise the real interest rate.   The price level would fall.   And then when it is low enough that higher expected inflation lowers the real interest rate back to the level needed to clear markets again, we now  have higher inflation for the price level to return to the expected level.

Unfortunately, it is not at all obvious what ties down the future price level under this system.   And that is where we get these fiscal theories.   As for using it to explain actual performance?    Really?   People supposedly have an expectation of the future price level?

Anyway, I reject the assumption that excessively high budget deficits must lead to inflationary default.   I realize that it is a possibility, but I consider it inferior to explicit default on the national debt.   The monetary constitution should not allow for inflationary default.   It is default either way, and inflationary default of government debt just creates a massive externality, causing the simultaneous inflationary default of private debt.

As for the notion that deficits must be sufficiently large to generate sufficient government debt for the central bank to purchase, this is simply based upon the assumption that the central bank must purchase government debt.    Perhaps the most extreme version of this framing is the "bills only" doctrine.   That is the view that the central bank should solely purchase short term government debt.

That might be a nice policy if there is sufficient short term government debt, but when the demand for base money outstrips the amount of short term government debt, then the obvious answer is for the central bank to expand its horizon and purchase something else.   After five years of the Fed purchasing mortgage backed securities, it is hard to understand why anyone would consider the amount of short term government debt outstanding to be a constraint on monetary policy.

If government were sufficiently frugal that the national debt falls, perhaps even to zero, does that require that the nominal anchor be changed?  Must there now be a deflationary policy, down to zero?

How about having the central bank purchase private debt?   Isn't there a long history of insisting that central banks should solely purchase private debt?    Real bills?

Don't like the central bank picking and choosing between borrowers?   Privatize the issue of hand-to-hand currency and end reserve requirements.

Demand for reserves still too high?    Make the sole asset of the central bank overdrafts to banks with adverse clearings.   Charge high interest rates on those overdrafts and pay low, maybe negative, interest on reserve deposits.   In other words, use the corridor system.      (Woodford's neo-Wicksellian world.)

Suppose we lived in a world with no government debt and an evolved gold standard.   Banks issue hand-to-hand currency and deposits.   The banks deposit gold at the clearinghouse to settle net clearing balances.    Want to stabilize the price level, inflation, or better yet, a growth path of nominal GDP?   Vary the price of gold--somewhat like Fisher's compensated dollar.

Or better yet, let the price of gold be determined by the market and make the monetary liabilities redeemable with index futures contracts on the nominal anchor.

Of course, the current monetary order does include a key role for government-issued hand-to-hand currency.  Reserve balances are huge and are at least quasi-government debt.

However, models that determine the current price level based upon rational expectations about what must happen to the quantity of base money in the distant (infinite?) future, is making an assumption about what systems will exist in the future.   I am not sure that it is really rational to assume that monetary institutions will remain the same in the distant future.

Of course, as a long-time money reformer, perhaps that is wishful thinking.

Neo-Fisherites

The neo-Fisherite view is that a higher target for the interest rate will result in higher inflation, and a lower target for the interest rate will result in lower inflation.

The Fed has kept its target interest rate at close to .02% for some years now, and it has been saying that they will stay there for some time.   The inflation rate has remained a bit low during this period.

So, evidence tells us that a low target for the interest rate results in low inflation.

This follows from a very simple theory--the Fisher relationship.   The nominal interest rate is equal to the real interest rate plus the inflation rate.

             .
R = r + P

                                   .          
And so, obviously, P = R - r.

If we have the Fed follow a rule of targeting the nominal interest rate, which is realistic, and the real interest rate depends on some kind of real economic factors independent of monetary policy, then a higher target for the interest rate will generate a higher inflation rate, and a lower target for the interest rate will generate lower inflation.

While it is only a model, we can test it.   And the low target for the nominal interest rate set by the Fed recently has generated low inflation.   The model predicts the data.

Rowe especially, but also Sumner have responded to this nonsense.

From a monetarist perspective, the Fisher relationship holds in the long run.   If the quantity of money grows more quickly, then the inflation rate rises.   For any given nominal interest rate, this reduces the real rate.   This benefits debtors and injures creditors.   The demand for credit rises and the supply falls, raising the nominal interest rate, and shifting the real interest rate back towards its initial value.  With all sorts of somewhat implausible assumptions, the real interest rate returns exactly to its initial value, and so the nominal interest rate is now equal to the initial real interest rate plus the new, higher inflation rate.   All of these implausible assumptions are necessary for "super neutrality" to hold, which means that the real economy is independent of the growth rate of the the quantity of money.

Suppose the central bank had the following rule for the growth rate of the money supply:

 .                           .       .
M     =  R* - rn + yp - V
                                                                                                                      .
 Where R* is   target for the nominal interest rate, rn is the natural interest rate,
 .                                                            .
M is the growth rate of the money supply, yp is the growth rate of potential output, and
 .
V is the growth rate of velocity.
                                                                   .         .       .     .
The inflation rate in the long run is P  = M + V - yp

So, by substitution:
 .
P   =    R*  - rn

This rule requires that when the target for the nominal interest rate rises, the central bank raises the growth rate of the quantity of money and so the inflation rate rises.

If one assumes that the natural interest rate is equal to the growth rate of potential output and the growth rate of velocity is zero, then the rule for the money supply is:
 .        
M   = R*

If the target for the nominal interest rate is just taken as fixed, then the economy is stable.   It is just a fixed quantity of money rule.   Sure, inflation will change with productivity shocks and inflation and real output will change with shifts in velocity,, but as long as prices and wages adjust to surpluses and shortages as they ought, then real output should adjust to potential in the long run.

If velocity growth, potential output growth, or the natural interest rate change, then the appropriate growth rate in the money supply will change--if the goal is really to keep nominal interest rates fixed?

And why would that be?   Right.. optimal quantity of money.

The nominal interest rate needs to be zero, more or less, so that there is no opportunity cost from holding hand-to-hand currency.

And if R* is zero, and the natural interest rate is equal to the growth rate of potential output and velocity is constant, then the quantity of money should be held constant!

Of course, actual central banks are not targeting the nominal interest rate in this way..   They adjust the nominal interest rate to target inflation and unemployment.   And they raise their target rate to slow inflation and raise unemployment and lower their target rate to raise inflation and lower unemployment.   And the way their trading desk raises interest rates is to slow money supply growth and the way they lower interest rates is accelerating money growth.

And so, it is difficult to see what the neo-Fisherite theory can tell us about what happened from 2008 to 2014.   It is rather a proposal for how monetary policy ought to operate.   With the only plausible goal being a zero nominal interest rate and so a deflation rate equal to the real natural interest rate.   And the only rationale for the goal is to allow for the optimal use of zero nominal interest rate currency.

Pascal Salin's Confusion: Inflation or Money Supply Targeting

Pascal Salin critiques Market Monetarism on the grounds that nominal GDP is not a good target.   However, his argument is confused.    His argument appears to be that a quantity of money rule would lead to stable inflation.   And then he correctly recognizes that stable nominal GDP growth is inconsistent with stable inflation when supply-side factors causes changes in real GDP growth.

And then, he becomes confused.   Is he criticizing Market Monetarism as being inferior to inflation targeting?   Or is he criticizing Market Monetarism as being inferior to targeting some measure of the quantity of money?

Or does he just have no idea what Market Monetarists propose?  Why would he suggest that Market Monetarists favor raising money growth to raise inflation on the grounds that this will dampen or reverse a slowdown in real GDP growth due to supply side factors?

In  truth, a quantity of money rule does not lead to stable inflation when supply-side factors influence real GDP growth.   A quantity of money rule has the exact same consequence as a nominal GDP target in that circumstance.   Given velocity, a constant growth rate of the quantity of money leaves nominal GDP on a stable growth path.   With both rules, a slow down in real GDP growth due to supply-side factors results in higher inflation.

For inflation to remain stable in the face of a slow down in productivity growth due to supply side factors, the quantity of money must grow more slowly as well.   This is exactly the policy required to target inflation.

Market Monetarists argue that slowing money growth when productivity slows due to supply-side factors is unwise in most circumstances.   In general, it is better that nominal incomes continue to grow at a stable rate, even if final goods prices rise at a faster rate.     Factor prices, like wages, are more stable under quantity of money and nominal GDP level targeting than under a rule targeting inflation in final goods prices.

Since Market Monetarists advocate  a rule rather than a discretionary monetary policy, we would argue that a stable growth path for nominal GDP, which is the same thing in this circumstance as a stable growth path for the quantity of money, is the least bad rule.   Yes, there could be some supply-side shocks where allowing nominal GDP to vary would have better consequences, and an omniscient and benevolent central banker could do better than targeting the quantity of money or nominal GDP.  But we don't have such a central banker.

But somehow Salin has Market Monetarists proposing to accelerate money growth to cause extra inflation to try to offset the adverse productivity shock.    Who knows where that comes from?

So what is the difference between a quantity of money rule and a nominal GDP target?   It is the response to shifts in velocity.   Market Monetarists believe that the quantity of money should shift in inverse proportion to any shift in velocity.   For the most part, this is equivalent to saying that the quantity of money should adjust to accommodate changes in the demand to hold money.

Of course, a quantity of money rule does not allow a change in the quantity of money due to a change in velocity.   The quantity of money does not shift in response to changes in the demand to hold money.   Rather inflation of final goods prices and factor price like wages change until the real quantity of money adjusts to the demand to hold it.   Inflation slows or shifts to a lower growth path, and so real output can be maintained (or recover) despite the lower velocity.

Market Monetarists are fully symmetrical on this matter.   We favor restricted money growth when velocity rises, so that inflation does not increase.

Interestingly, the Market Monetarist view of the proper response of policy to shifts in velocity is similar to that of inflation targeting.   The only real difference is that Market Monetarists favor a level target--that is a growth path for nominal GDP.   Inflation targets a growth rate.

Salin describes the following scenario.   The money supply is growing 3%.   Real GDP is shrinking 2%, so the inflation rate is 5%.    Market Monetarists supposedly would respond to this scenario by having nominal GDP grow 5%.   According to Salin, we would anticipate that this would cause real GDP to grow more rapidly (or shrink less,) but he insists that the actual impact would be inflation of 7%.

Well, where did this 3% money growth rate come from?

The 5% nominal GDP target comes from a scenario where the quantity of money is growing 5% and has been for some time.   Real GDP usually grows at 3%, resulting in 2% inflation.   (This is the high end of Milton Friedman's proposal for a money supply rule.)    Unfortunately, disastrous supply-side policies result in real GDP shrinking 2% a year.   The result would be 7% inflation.   This would be true whether the money supply target was 5% or the nominal GDP target was 5%.    (We can certainly hope that these policies solely lower the growth path of real GDP, so that after a run up in the price level, the inflation rate returns to something like 2%.  )

Now, if the growth rate of the money supply had been 3% for some time, then the natural target for nominal GDP growth would also be 3%.    Real GDP is usually growing at 3%, and the price level is stable.   And then, we have this disastrous productivity shock, and real GDP begins shrinking 2%.   The inflation rate is 5%.   This is the same result if the quantity of money continued to be targeted at 3% or nominal GDP is targeted at 3%.

Again, we can hope that the adverse policies shift real output to a lower growth path, and then it resumes growing.    If real GDP growth permanently slowed down, perhaps to 2%, then after the run up in the price level, inflation would stabilize at 1%.

By the way, Market Monetarists would propose fixing this problem by improved supply-side policies.

An inflation target is somewhat different.   To keep inflation stable in the face of real output shrinking 2%, the quantity of money would need to shrink roughly 2%.     Market Monetarists think that this would almost always be a horrible policy.

Salin argues that if the money supply remains on a constant growth path, velocity will settle down.  From the point of view of Market Monetarists, at first approximation, this would imply that keeping nominal GDP on a stable growth path would require a stable growth path for the quantity of money.   We certainly have no problem with that.

One reason Salin claims that velocity fluctuates is due to changes in inflation expectations.  However, a nominal GDP target and a quantity of money target generates the same inflation expectations--leaving aside possible changes in velocity.   To the degree that changes in the quantity of money can offset changes in velocity and accommodate changes in the demand to hold money, it will reduce fluctuations in inflation and so tend to stabilize inflation expectations and velocity compared to a quantity of money rule.

How do we pick a target for nominal GDP growth?   How do you pick a target for the growth rate in the quantity of money?   What you do is anticipate the growth rate in real potential output and add to it the inflation rate desired.   I go with zero.   And so, that results in 3% growth rate for nominal GDP.   And that, of course, is the low end for Friedman's proposal for the growth rate of the quantity of money.

Many Market Monetarists go with the high end of Friedman's proposal, which is consistent with adding the 3% trend growth rate in real output to the more or less arbitrary 2% inflation target that has been adopted by many central banks.   The resulting 5% nominal GDP growth rate happens to be very close to the actual trend growth path of nominal GDP during the Great Moderation.   A 5% nominal GDP growth rate seems pretty consistent with the 2% inflation target in the long run.

Salin accuses Market Monetarists of being a sort of new Keynesian.   This is because Market Monetarists supposedly believe that raising the inflation rate will reduce unemployment.   Salin explains to us that this can at best work in the short run.

I think it is fair to say that most Market Monetarists are especially interested in avoiding an increase in unemployment due to a slowdown in spending on output.  While this will also tend to cause some disinflation, what happens to final goods prices isn't our prime concern.   Our view is that by returning spending on output to its trend growth path, there will be a more prompt recovery in output as well as a more prompt reduction in unemployment.   That the disinflation might be simultaneously reversed is of little concern.   Yes, there might be some reflation along with the recovery in output.   And yes, it is all a short run phenomenon.

Suppose the quantity of money was on a 3% growth path, but banking troubles caused the quantity of money to fall 10%.   An advocate of a money growth rule would be compelled to support a rapid reversal, with the quantity of money rising roughly 13% to return to its previous growth path.

Now, would that temporary decrease in the quantity of money be associated with a sharp increase in unemployment?   Would real output fall?   Would there be some disinflation if not outright deflation?

And when the quantity of money recovers, wouldn't the result be a more prompt recovery of real output and reduction in unemployment?   And wouldn't any disinflation or deflation be reversed?

So, in some sense, there would be temporarily higher inflation associated with a reduction in unemployment.   That is all Market Monetarists really have in mind.

The only difference between Market Monetarists and Traditional Monetarists along these lines is that Market Monetarists favor institutions that cause the quantity of money to accommodate shifts in the demand to hold money, or more exactly, that offset shifts in velocity.  

Selgin on Keynes and Quantitative Easing

George Selgin favorably quotes Keynes,

"Rising output and rising incomes will suffer a set-back sooner or later if the quantity of money is rigidly fixed. Some people seem to infer from this that output and income can be raised by increasing the quantity of money. But this is like trying to get fat by buying a larger belt. In the United States to-day your belt is plenty big enough for your belly. It is a most misleading thing to stress the quantity of money, which is only a limiting factor, rather than the volume of expenditure, which is the operative factor."

Pushing on a string?

In comments, Selgin expands:

"The problem with the argument that, so long as spending appears deficient, more monetary expansion is called for, is precisely that it overlooks the possibility being emphasized here, namely, that conditions are such that banks simply aren't inclined to lend more, for whatever reason. In that case, some deeper problem must first be addressed before monetary expansion can serve any useful purpose; and then, if it is addressed, the expansion may prove unnecessary because spending revives without it."

While I cannot speak for all market monetarists, my support for quantitative easing was not about encouraging banks to lend.  Rather, it was about increasing the quantity of money enough to return spending on output to a higher growth path despite a large decrease in velocity.

If banks are willing to lend, then any increase in base money will be multiplied, so that the increase in base money necessary for the needed increase in the quantity of money would be smaller.

If banks are not willing to make bank loans but are only willing to hold other sorts of securities, then the impact of any increase in base money on the quantity of money is little effected if at all.

However, if banks are not willing to extend any sort of credit, either through loans or purchases of securities, then increasing base money only increases the total quantity of money dollar-for-dollar.   Worse, to the degree the Fed makes purchases of assets from banks, there might be no increase in the quantity of money at all.

However, whatever the reason for this sort of behavior by banks, this simply means that a larger increase in base money is necessary to generate the appropriate increase in the quantity of money.

What is the target for nominal GDP?   Divide by velocity to get the appropriate target for the quantity of money.  Divide by the money multiplier to get the appropriate target for base money.   And keep in mind that neither the money multiplier nor velocity are constant.

If there is feedback mechanism where open market purchases lower the money multiplier or velocity, or both, then this means that base money needs to be larger than otherwise.

While I can imagine a variety of feedback mechanisms where increases in base money cause decreases in the money multiplier or increases in the quantity of money cause decreases in velocity, the most plausible candidate for that sort of process is the expectation that an increase in base money is temporary.  

In my view, the purpose of quantitative easing is to increase spending on output.   By far the best way to do that is to explicitly target spending on output.  

In my view, the problem was not too little base money or even interest on reserves or tighter regulation of bank capital.   The problem was the Federal Reserve's vague inflation/unemployment rate target.    We are committed to 2% inflation, but we forecast inflation to be lower, but we will let it rise above 2 percent in the medium run and we would like to see unemployment fall, so we will maintain monetary accommodation of near zero short rates and we will purchase longer term assets to lower long term interest rates too.

The slow recovery was primarily due to a growth rate target--inflation is the growth rate of the price level.   A level target would have created a better bounce back.   However, a price level target would have been even more disastrous in 2007.   The supposed benefit of an inflation target is that the increase in the price level due to a supply shock can be ignored.   The Federal Reserve failed to do that for worry that inflation expectations would become unanchored.   A price level target would have required that response and exacerbated it.

Still, if the Fed had a nominal GDP level target, I would favor as much "quantitative easing" as necessary to hit the target.    

Yes, I would certainly propose reducing the interest rate on reserves to zero as well.   And if the Fed approached the statuatory limit on what it can buy, then I would propose negative interest on reserves and even privatizing currency before expanding what the Federal Reserve can buy.