Wednesday, March 14, 2012

MMT Redux

I knew if I stuck my hat on a pole above the trenches, the MMT missiles would come flying.  Specifically, I hear two general arguments whizzing over my head:

1. Loans are not made out of reserves, silly!  Loans are simply made, and if the bank finds itself short of reserves at the end of the day it borrows them from the overnight market.  Lending and the quantity of reserves at any moment in time are decoupled.

2. The CB targets an overnight interest rate.  If an over- or undersupply of reserves puts pressure on that rate, the CB injects or soaks up reserves to maintain its peg.  Implication: the Treasury can borrow as much as it pleases, as long as the CB purchases whatever proportion of the bond issuance it needs in order to maintain its peg.  Notional measurements of “the money supply” have no independent significance.


1. It is a fair criticism to say that textbook presentations of fractional reserve banking, including my own, are counterfactual; convenient exposition is, in this case, at odds with observed reality.  It should be borne in mind, however, that the money multiplier model, like conventional supply-and-demand models, operates at an aggregate level and is not truly microfounded.  (In S&D, prices are supposed to equilibrate in a perfectly competitive model, in spite of the definition of perfect competition as a state in which no agent has the ability to influence the market price.)  The money multiplier really specifies a limit at which further loans need to be backed by an infusion of new reserves.  (The MM has been getting fuzzier of late due to changes in financial institutions and instruments, not to mention international capital flows, but we’ll leave that aside.)  If banks were always fully lent, the CB would have sole control over “the” money supply through control over the monetary base: that would be one corner solution.  But banks are not fully lent at all times.  If banks were never fully lent, the CB would have no influence at all on monetary aggregates except through its ability to stimulate or discourage lending, but this is another implausible corner solution.  I take the middle road.

2. If the CB targets a nominal interest rate, it runs the risk of the following scenario: increased borrowing by the Treasury increases inflationary expectations, which reduce the real interest rate, perhaps even below zero.  This leads to ever-reduced lending standards and overheating (including bubblish activity), validating those expectations, disastrously.  Or the CB can target a real rate, but if inflationary expectations rise it is compelled to increase its nominal peg, dumping an increasing share of bonds on the market.  This looks like, and is, contractionary monetary policy.  The notion that a CB can passively accommodate any and all fiscal deficits strikes me as very strange.  To put it differently, there are two dubious corner solutions, one in which any exercise of fiscal expansion is completely vitiated by offsetting changes in inflation, and another in which there are no such changes.  I’ll take the old fashioned Keynesian view that the proportion of fiscal expansion absorbed by inflation roughly increases as the output gap decreases, a sensible—and empirically validated—middle position.


ATR said...

I’d like to take a shot at this, because it’s intellectually stimulating, but I am not the authority. I am sure Dr. Fullwiler will follow up with you, and he is. Also, I’m not 100% following your thoughts here, partially due to language choice, but you and MMT might be talking past each other and could potentially be in close alignment. I’ll just tackle 1) for now.

“The money multiplier really specifies a limit at which further loans need to be backed by an infusion of new reserves.”

I’d say it is more directly the reserve requirement that specifies that. (In undergrad, the money multiplier was taught to me as a broader theory explaining a causal relationship from reserves to an expansion of loans and deposits.)

(continued in next post due to character limit)

ATR said...

(continued from above)

“If banks were always fully lent, the CB would have sole control over “the” money supply through control over the monetary base: that would be one corner solution. But banks are not fully lent at all times. If banks were never fully lent, the CB would have no influence at all on monetary aggregates except through its ability to stimulate or discourage lending, but this is another implausible corner solution. I take the middle road.”

Let’s think this through a bit more. When you say a bank is not “fully lent,” what I take you to mean is that at the bank’s current level of reserves, it would be able to make additional loans without surpassing the reserve requirement. But think about that for a second. That means the bank has excess reserves. As Dr. Fullwiler has made clear, banks have no use for reserves beyond settling payments and meeting reserve requirements. So what that bank will do is try to get rid of those excess reserves in the interbank market for some sort of return at least as high as the interest rate that those reserves would otherwise be earning. Prior to IOR, banks would loan their excess reserves away at the FFR, and the Fed would have to make sure there weren’t too many reserves flooding the interbank market such that downward pressure was being placed the FFR. If this was happening, as it would be in your scenario, the Fed would drain excess reserves to maintain the target FFR. In a theoretical model, there would then be no more excess reserves; in the real world, banks desire a small buffer of excess reserves, but the Fed is still targeting an interest rate and is making sure reserves aren’t having consequences on changing the FFR from the target. And yes, indeed, this is what banks would do prior to IOR. Go to FRED and you’ll see excess reserves were practically nonexistent; that’s necessarily the case if the Fed is targeting a FFR without IOR. With IOR, the FFR becomes the IOR as long as the Fed makes sure banks have sufficient reserves for their needs.

In either case, this is a story about targeting interest rates, not the monetary base. The Fed is accommodating private sector activity when it adjusts the money supply with the goal of maintaining their target interest rate. As Dr. Fullwiler has said, “that's what it means to set an interest rate target.”

That all said, let me repost something Dr. Fullwiler wrote in the earlier thread, which makes clear that there are instances where causation can go from MB to aggregates. But this is different than when we are thinking about lending. Perhaps this clarification achieves the “middle road" answer you are truly seeking.

“Now, it could be that the point that the MB and monetary aggs aren't "decoupled" (and, again, we just say the causation is backwards, not necessarily decoupled, though it could be) isn't necessarily based on lending. For instance, in QE the FEd raises both the qty of reserve balances and deposits simultaneously when it buys securities from non-banks, so obviously there is a direct relationship in that case. That is, I just want to stress that the MMT point (which is the same as the Horizontalist point of Moore, as Nathan pointed out) is about the monetary base not causing lending; there are instances in which monetary aggregates and MB can be directly related and causation can go from MB to the aggregates.”

TheIllusionist said...

"If the CB targets a nominal interest rate, it runs the risk of the following scenario: increased borrowing by the Treasury increases inflationary expectations, which reduce the real interest rate, perhaps even below zero."

Assuming that inflation expectations actually generate inflation. This is pretty up in the air. It is also highly context dependent.

"This leads to ever-reduced lending standards and overheating (including bubblish activity), validating those expectations, disastrously."

Again, heavily context dependent.

"The notion that a CB can passively accommodate any and all fiscal deficits strikes me as very strange."

Of course it can. BUT this may lead to inflation. That is an entirely different thing altogether.

This is a central point of MMT. We must carefully distinguish between what a CB can do and what a CB ought to do. A CB can accommodate any level of deficits but in some cases this would lead to inflation.

The reason we need to separate what they can and ought to do is due to the context of the situation.

Take a concrete example: Japan have been running deficits for years and the CB has been accommodating. There has been nearly no inflation. This is because of the specific situation that Japan was in at this point in time. The same policy during the Clinton boom years would, ceteris paribus, have likely led to inflation.

This was the point I kept making in the piece written in contrast to Dean Baker's position. When people try to defend the upward sloping LM curve they will ALWAYS end up defending a particular type of CB policy, given a particular 'ideal' environment. These are thus normative, not positive judgements. Or to put it differently: defenses of the upward sloping LM are NORMATIVE not positive defenses.

When you say 'middle position' you're implicitly talking about policy at a given point in time under presupposed circumstances (NAIRU employment etc.). This might be the correct position if we agree that this is the normal case (I disagree), but still this is policy. So, when you take a 'middle position' you're describing policy stance, not the really existing operative framework.

A good economist should be able to tell the difference and explain it to students. Otherwise their heads will be filled with loanable funds silliness and they will find it difficult to argue for fiscal expansion when it might be a good idea.

Put differently: the upward-sloping LM curve contains normative assumptions about how an economy 'should' be operating (full employment/labour market equilibrium etc.) that do not usually apply in reality.

Finally, I would like to point back to Neil's comment on previous discussion: John Hicks came to realise that ISLM was not a good representation of reality and published papers on it. While it was not the loanable funds model that he focused on, he was raising similar criticisms as I do here and in my original piece.


ATR said...

On 2), MMT is practically aligned with your old Keynesian proposition. MMT says deficit spending can certainly be inflationary, specifically when the economy is running near full capacity. MMT definitely does not say fiscal deficits cannot impact inflation; it says precisely the opposite. Many people misinterpret MMT as saying 'deficits do not matter' however.

" The notion that a CB can passively accommodate any and all fiscal deficits strikes me as very strange."

I am a bit confused on what you're trying to say here. I think we agree that the CB targets an interest rate. That means, in that world, it always targets an interest rate. So what does not accommodating a fiscal deficit mean? Whatever the Treasury is spending, the Fed will continue to target an interest rate. Maybe it will decide to change its target interest rate in response to the economic environment, but it's targeting an interest rate...

Don Levit said...

Is the money multiplier the same as money velocity?
Money velocity is now below 1.6, which means, to me, that a lot of the excess reserves are staying as excess reserves. There is not much activity accruing between banks and individuals/companies.
If true, could a significant portion of the excess reserves be placed in the stock market, thus goosing the averages?
Don Levit

ATR said...
This comment has been removed by the author.
TheIllusionist said...

@ Levitt

Money multiplier is a slightly crude construction that doesn't take velocity into account. Back to the old Fisher identity:


Money x Velocity = Price x Quantity

Money multiplier on affects the 'M' or 'Money' variable in the equation.

So, reserves sitting idle do not have to do with velocity. They have to do with the fact that the money multiplier has supposedly 'broken down'.

MMT claims that there's no such thing as a money multiplier, so nothing has broken down at all. We say there is no money multiplier, only a 'credit divisor'. (I.e. The amount of reserves 'pulled' in after the fact to back up lending. Credit divisor is what makes it look like there is a money multiplier where there is none).


TheIllusionist said...

"If banks were always fully lent, the CB would have sole control over “the” money supply through control over the monetary base: that would be one corner solution."

Also, problems with this. If a bank is fully lent, they can go to the discount window. Again, they will have to price the increased interest payment into any loans being made, but if they were profitable they could make them.

Rate at discount window is set off the overnight/Fed funds rate. So, we're back to the point: Fed set overnight rate and let demand for loans adjust.

ATR said...

Phil, you're jumping the gun on the discount window. Without IOR, if the Fed wants the FFR at 2%, and banks need more reserves because they want to lend more, then they'll inject reserves by buying tsys. That'll keep the FFR at 2%. Going to the discount window at 3% won't be necessary. That's standard ops, and it's how it always worked before IOR. Banks were always 'fully lent' - otherwise there would be downward pressure on the FFR due to excess reserves.

ATR said...

Peter if you are actually interested and have the time, I would suggest engaging the actual literature.

Here are some MMT papers on banking that I recommend:

STF said...

Hello Peter,

Appreciate you engaging again. Very quickly since I'm unfortunately too tied up to write more at the moment and possibly for the next several days.

Regarding your points,

1. The point, as I suggested in the previous post comments, is that MMT and others associated with the Post Keynesian school argue that the money multiplier has causation backwards. Further, because the central bank is targeting an interest rate, it MUST accommodate banks' demand for reserve balalnces related to reserve requirements and payment settlement needs. To do otherwise would mean not achieving its target rate. This not a controversial point even in neoclassical literature on central bank operations; central banks simply do not control credit creation through a money multiplier process or reserve requirements aside from gold standard or other "special case" regimes (China). Regarding being "lent up," not enough time to respond completely, but I'll just say that I find the concept to be largely theoretical and not very applicable, as central bank trading desks are well-known to attempt to estimate banks desired excess reserve holdings at the target rate and accommodate (not to mention that in many countries there aren't RR at all).

2. There is no suggestion in MMT that the CB can't or doesn't control where it sets its target rate, only that to achieve whatever target rate it is setting it must accommodate banks demand for balances. It obviously alters its target rate based on macro conditions, as people like you think it should do. This has nothing at all to do with the argument we are making. This is the difference between operational tactics of monetary policy and the strategy of monetary policy. The MMT point you are critiquing is about tactics, but you are talking about strategy, which is not unimplortant but not relevant to the point we are making. In the process of accommodating banks' demand for balances, the CB will adjust its target rate and will obviously do so in order in a way related to its goals to slow/increase bank credit expansion, etc.

Scott Fullwiler

Tom Hickey said...

The money multiplier is an accounting residual, the effect of the reserve requirement on banks. As Scott notes, lending is not affected by quantity but by price. The cb as monopolist sets the price (interest rate) and lets quantity float.

So the causation is not quantity but price.The cb uses the price (interest rate) to influence the ratio of saving and borrowing due to the spread that the banks charge customers by adjusting the banks's cost of making loans up or down. Regrettably, not only do most textbooks have the causation reversed, but also the Fed's own information does too.

I was under this mistaken impression for some time, and I am grateful to the MMT economists for setting me straight on this.

TheIllusionist said...


Perhaps you'd find this to be more convincing if it came straight from the horse's mouth. How about a Vice President of the ECB?

"Central bank reserves are held by banks and are not part of money held by the non-financial sector, hence not, per se, an inflationary type of liquidity. There is no acceptable theory linking in a necessary way the monetary base created by central banks to inflation. Nevertheless, it is argued by some that financial institutions would be free to instantly transform their loans from the central bank into credit to the non-financial sector. This fits into the old theoretical view about the credit multiplier according to which the sequence of money creation goes from the primary liquidity created by central banks to total money supply created by banks via their credit decisions. In reality the sequence works more in the opposite direction with banks taking first their credit decisions and then looking for the necessary funding and reserves of central bank money. As Claudio Borio and Disyatat from the BIS put it: “In fact, the level of reserves hardly figures in banks’ lending decisions. The amount of credit outstanding is determined by banks’ willingness to supply loans, based on perceived risk-return trade-offs and by the demand for those loans.”7 In modern banking sectors, credit decisions precede the availability of reserves in the central bank. As Charles Goodhart pointedly argued, it would be more appropriate talking about a “Credit divisor” than about a “Credit multiplier”.8"

Peter Dorman said...

Thanks to the MMT brigade (or its rapid response unit) for these replies. I won't go into detail, except to say that I see little that contradicts a Keynesian perspective like mine. You can take this as warm and fuzzy or as a sign that MMT may not be something new under the sun, just a different ribbon on the same package.

Unlearningecon said...

'which reduce the real interest rate, perhaps even below zero. This leads to ever-reduced lending standards'

You know Keynes argued the opposite - that high interest rates push investors into riskier endeavours in search of higher returns, and low rates encourage sustainable investment?

ATR said...

Peter, just so you know, MMT comes out of the Post-Keynesian tradition.

In some senses it is an amalgam of many PK ideas. But the coherent whole and the way it is applied to analyzing the banking system and fiscal and monetary policy, is MMT's unique signature, in my opinion. Furthermore, if MMT is just the same old, then why is there so much debate and confusion around it? Why did we have difficulty coming to terms on some fairly basic banking system issues?

I can only speak as someone who had an econ/business background from undergrad, but the neoclassical econ I was taught there has some striking differences with MMT. The money multiplier is one example, although the rejection of that theory isn't unique to MMT.

Barkley Rosser said...


1) Not all Post Keynesians subscribe to MMT, even if some of its more fervent backers think that this should become the new litmus test for who is a PK and who is not.

2) As you noted, of course when CBs are targeting interest rates, money is endogenous. The issue for the harder core MMT group is that they argue that it is endogenous no matter what the CB is doing, something much more debateble. Was Milton Friedman really wrong that the Fed played a major role in turning a bad recession into the Great Depression by its contractionary monetary policy after 1929, particularly in bringing about the crash of 1931 that certainly did engender an endogenous collapse of M as banks failed around the world?

Unlearningecon said...

Btw, have you guys at Econopseak seen this?

Empirical link is strong, not sure how anyone could say Keen is wrong...

Unlearningecon said...

Also, you seem quite reasonable, so I'm not sure why you would cling to the 'MM' model instead of just endogenous money. I mean, MM predicted that:

(a) We could control the money supply and

(b) M1+ expansion would succeed M0 expansion

But neither were true. The response is:

(a) The multiplier/V are volatile, which just smacks of epicycles to me

(b) ???

STF said...

Hi Barkley,

Regarding your points,

1. Yes, I completely agree with you, and didn't mean to suggest otherwise.

2. There are many, many things that were happening in 1931 to bring down the money supply, one of them being that the US was still on the gold standard--note that the MMT/horizontalist view on cb operations is not intended to apply in its completeness there. Otherwise, the argument is that the CB cannot DIRECTLY target a monetary aggregate. Note that even in the 1979-1982 period, the Fed's own literature and that of Ann-Marie Meulendyke makes it clear that this was the case: the Fed was targeting reserves, but only indirectly over a 2-3 month period, not daily. The daily target was a range for the fed funds rate. That is how it must be--to directly target is still to set a de facto interest rate target at the rate the CB pays on reserve balances (in the case of more balances supplied than the banking system wants) or the CB's penalty rate (in the case of too few balances that send banks to the discount window for reserves to settle payments or meet reserve requirements). And note further that targeting too few balances threatens the stability of a payments system that is settling 10%-20% daily payment volume of annual GDP with reserve accounts in most countries. The exception is when the target rate is set equal to the rate paid on reserve balances or 0 (the lower de facto rates in providing too few balances) and engages in QE (which can then target directly reserves or an aggregate). In that case, though, as Japan and the US found out, it is quickly learned that Goodhart's Law prevails.

Scott Fullwiler

STF said...

"settling 10%-20% daily payment volume of annual GDP"

That is, settling daily payment volume equal to 10%-20% of annual GDP, according to the BIS's figures.

TheIllusionist said...


I think there is a profound difference between holding the MM view and thinking in terms of endogenous money. Someone posted above a link to Keen's excellent piece on the LSE website. At the heart of the debate (albeit hidden) is the loanable funds theory. Consider this:

"So can we ignore the level of private debt? No—because this “profound insight” is in fact a blind-spot about the role of banks and debt in a capitalist economy. Neoclassical economists treat banks as irrelevant to macroeconomics—which is why banks are not explicitly included in their models—and regard a loan as merely a transfer from a saver (or “patient agent”) to a borrower (or “impatient agent”), as in Krugman’s “New Keynesian” model of our current crisis:

In what follows, we begin by setting out a flexible-price endowment model in which “impatient” agents borrow from “patient” agents, but are subject to a debt limit. (Krugman and Eggertsson 2010, p. 3)

With that model of lending, a change in the level of debt has no inherent macroeconomic impact: the lender’s spending power goes down, the borrower’s goes up, and the two changes roughly cancel each other out.

However, in the real world, banks lend to non-bank agents, giving them spending power without reducing the spending power of other non-bank agents. The difference between the neoclassical model of lending and the real world is easily illustrated using transaction tables. Figure 2 illustrates the neoclassical model (with an implicit banking sector): in this world, a change in the level of debt has no macroeconomic implications."

The loanable funds framework is false and has very large implications for how we do macroeconomics. I think economists avoid the importance of the distinction at their own peril.