Donnerstag, 24. April 2014

Mostly Model Tailoring against the Miraculous Money Treat of Modern Monetary Theory (MMT)

Cullen Roche has written a brillant in-depth "Critique of Modern Monetary Theory (MMT)" (Nov. 2013, unfortunately not paginated).
While he hopes that "this critique will be viewed as a constructive criticism and not an attack on MMT", at times he still does get pretty irritated with "doublespeak",  "sloppy ... terminology", "sloppy generalization" etc. of MMT-proponents. And in the context of how payments are effected through the VISA system he concludes that "[Stephanie] Kelton does not even have the most basic facts right."
So let as take a look at some other facts which MMTers assert (or which we can logically discern as facts behind their verbal constructs) and analyze, with the help of models, the truth value of those claims. 

In my blog post "MMT: Modern Monetary Theory - or Monstrous Mental Twisting?" I had analyzed several (shorter) MMT-papers, amongst them "MODERN MONEY THEORY: A RESPONSE TO CRITICS" by Scott Fullwiler, Stephanie Kelton and L. Randall Wray (2012).
In this paper, which I will now subject to a closer scrutiny, MMTers provide a comparatively palpable description of why they think that government bond sales must be preceded by an injection of currency (unnumbered p. 5, emphases added):
"Beginning with the simple or general case, consider a consolidated government (central bank plus treasury) running a deficit. The basic transactions could be listed as the following:
1A. The government’s spending credits bank accounts with reserve balances (HPM). These accounts are liabilities of the government/central bank.
2A. Banks credit the deposit accounts of the spending recipients. So, overall, the increased reserve balances have raised bank assets while the increased deposits have increased bank liabilities by the same amount. Further, because spending to the private sector is greater than taxes drawn from the private sector, the private sector’s net financial wealth has increased. [.....]
3A. Absent interest on reserve balances, the government/central bank issues bonds or offers time deposits to drain or otherwise replace the reserve balances (HPM) created by the deficit if they are not consistent with banks’ demand for reserve balances at the targeted interest rate. [.....]"
Let me translate this, by using (imaginary) figures into something even more seizable.


Status of system at the beginning of spending period 1 (SP1):
a) 10.000,- (Dollars, Euros or whatever) currency (reserves plus cash) in the economy (i. e. on bank accounts with the central bank - CB - and as cash in the banks and with the economic agents).
b) 100.000,- deposits in bank accounts .
c) -0- Government bonds out.
Further parameters
Government deficit 1.000,- per SP (say: a year).
Then (all else equal) at the end of the 1st spending period each of the above steps would yield the following result:
1A: Reserves 11.000,-
2A: Deposits 101.000,-
Now  the financial expansion is reverted as we continue (say: at the beginning of the 2nd spending period) with step
3A: government issues bonds worth 1.000,- and sells them to the public. Which means that the next step is
2A: Deposits back down to 100.000,- (bond purchases by bank customers reduce their deposits) and finally arrive at the level of reserves with the (original) step
1A: Reserves back down to 10.000,- (banks transfer 1.000,- reserves to the Treasury account with the fed)


Status of system at the beginning of SP2:
a) 10.000,- currency.
b) 100.000,- deposits in bank accounts .
c) 1.000,- bonds out.
Then we will see the same steps with exactly the same balances for reserves and deposits as in SP1. The only variable that changes is an increase of bond stocks in the private sector: At the beginning of SP3 = 2.000,-; SP 4 = 3.000,- and so on.
And of course the government debt (burden) grows accordingly.

So even  if we were willing to concede all the rest of MMT-mythology we could already state:
If the government runs a constant deficit each year, even under MMT-assumptions the net injection of money into the economy occurs only once. (Meaning that repeated deficits do not lead to a cumulative effect of a constantly growing currency volume).
(Naturally, if the deficit for any SP is higher, than in MMT-thinking extra net money would have to be injected into the economy.)

Of course there is no valid reason why a government would have to spend currency first and could only afterwards "drain" the reserves. Ónce upon a time, the currency (i. e. the 10.000,- of our model) may have gotten into the system by government spending (or CB-lending). But no extra money has to be injected in order to sell bonds,
If the bonds had been sold at the beginning of the SP, then (all else equal) each of the above steps would yield the following result:

SP1 start:1A: Reserves 9.000,-
2A: Deposits 99.000,-
3A: 1.000,- bonds out
SP1 end:
1A: Reserves 10.000,-
2A: Deposits 100.000,-
3A: 1.000,- bonds out

SP2 start:
1A: Reserves 9.000,-
2A: Deposits 99.000,-
3A: 2.000,- bonds out
SP2 end:
1A: Reserves 10.000,-
2A: Deposits 100.000,-
3A: 2.000,- bonds out
etc. (stock of bonds and govt. debt increasing as in previous example).

In their paper "Modern Money Theory 101: A Reply to Critics", MMTers Éric Tymoigne and L. Randall Wray (Nov. 2013) claim (p. 6, emphasis added):
"Another important logical conclusion is that the injection of government currency (through expenditures or advances) into the other sectors must occur before the destruction of the government currency (through tax enforcement and repayment of advances). In an economic system in which a sovereign government operates through its own monetary system, spending (or lending) must occur before taxing. In addition, taxes are not a funding source in that logic. They are part of the destruction of government currency, i.e. they return currency to the issuing government. ..... Within that logic, a fiscal deficit represents a net injection of currency that usually needs to be drained as explained in section 3."

As I have shown in my abovementioned blog entry "MMT: Modern Monetary Theory - or Monstrous Mental Twisting?" it doesn't make sense to claim that money gets destroyed if the Fed books currency (taxes or proceeds from bond sales) onto the Treasury account. And even if that assumption were made, it is irrelevant, because when the government sells bonds to the public, the purchasers pay with money already existing (i. e. not "printed" by the Fed for that purpose).
This can be demonstrated in a model which is based on the aforementioned paper "MODERN MONEY THEORY: A RESPONSE TO CRITICS" by Scott Fullwiler, Stephanie Kelton and L. Randall Wray. (I am using here, slightly modified, the text from my previous MMT blog post.) Again, I have enriched their still somewhat abstract presentation with some (imaginary) figures. (Emphases added)

"... now consider how operations are really done in the US—where the Treasury really does  hold accounts in both private banks and the Fed, but can write checks only on its account at the Fed. Further, the Fed is prohibited from buying Treasuries directly from the Treasury (and is not supposed to allow overdrafts on the Treasury’s account) and thus the Treasury must have a positive balance in its account at the Fed before it spends. Therefore, prior to spending, the Treasury must replenish its own account at the Fed either via balances collected from tax (and other) revenues or debt issuance to “the open market”. (p. 6) 
As a result [of self-imposed constraints], in the US case there are at least six transactions related to deficit operations, rather than three in the general case." (p. 6)
(For the 3-step "general case" cf. my discussion above.)
(1B, p.6/7): "The Fed undertakes repurchase agreement operations with primary dealers (in which the Fed purchases Treasury securities from primary dealers with a promise to buy them back on a specific date) to ensure sufficient reserve balances are circulating for settlement of the Treasury’s auction (which will debit reserve balances in bank accounts as the Treasury’s account is credited) while also achieving the Fed’s target rate."
Model: CB (Fed or other) purchases 1.000,- (of whatever currency) from banks (who promise to buy the bonds back later - "repo" sale)
Result: Bank reserves +1.000,-. Assumption for our model: "normal" level of reserves = 10.000,-. New total reserve level 11.000,-.
(2B, p. 7) "The Treasury’s auction settles as Treasury securities are exchanged for reserve balances, so bank reserve accounts are debited to credit the Treasury’s account, and dealer accounts at banks are debited."

Model: Banks purchase 1.000,- worth of bonds from the government and transfer 1.000,- HPM from their reserve accounts with the CB (Fed) to the Treasury account with the CB. 

Result: Reserves ./. 1.000,-; total bank reserves back down to 10.000,-.
(Note that procedures in step 1 and 2 have offset each other, as far as bank reserves go. However, the extra 1.000,- HPM created by the CB in step 1 is still out on the market: Only now on the government account. (MMT'ers will of course say that this money has been destroyed. But this is irrelevant for the effect on the real economy. Whether the extra 1.000,- are dead or dormant in the Treasury account, they can now be spent by the government. Which - regardless whether the constraints are technical or self-imposed - the government could not do before.)

(3B, p. 7) "The Treasury adds balances credited to its account from the auction settlement to tax and loan accounts. This credits the reserve accounts of the banks holding the credited tax and loan accounts."
Model: Treasury transfers 1.000,- into accounts with the banks. On the reserve level this means 1.000,- are transferred from Treasury account with CB to bank accounts with CB.
Result: Reserves + 1.000,-; total bank reserves back up again to 11.000,-.

(4B, p. 7) "The Fed’s repurchase agreement is reversed, as ..... primary dealer purchase Treasury securities back from the Fed."

Model: Banks (are obliged to) buy 1.000,- of (old) bonds back from Fed; money is taken from their reserve accounts with Fed.
Result: Reserves ./. 1.000,-; total bank reserves back down again to 10.000,-.
(5B, p. 7) "Prior to spending, the Treasury calls in balances from its tax and loan accounts at banks."
Model: 1.000,- HPM ("reserves") transferred from bank accounts with Fed to Treasury account with Fed.
Result: Reserves ./. 1.000,-; total bank reserves down  to 9.000,-.
(NOTE: Reserves have now fallen below their initial level of 10.000,-!)
(6B, p. 7) "Treasury deficit spends by debiting its account at the Fed, resulting in a credit to bank reserve accounts at the Fed and the bank accounts of spending recipients. This increases the net financial wealth of the private sector."
Model: Government spends 1.000,-; this spending increases bank reserves by 1.000,-.
Saying that government "deficit spends" is deceiving for anybody not familiar with MMT double Dutch. What they mean is that the money is newly created because in their world view the proceeds from the bond sales in Step 2 have been destroyed when booked into Treasury account. Under this definition the money that the government spends is, of course, "newly created". However, any such games of definition do not affect the economic effects of the operation.
The money that the government spent (Step 6) came from a reduction of reserves below the original level (Step 5). And NOT from the extra money which the Fed had, temporarily and for smoothing only, injected into the reserves. (At the time of govt. spending, the Fed had already withdrawn the extra reserves from the market: Step 4.)
In terms of the real economy, this is logical and necessary: Government is spending a 1.000,- share of the 10.000,- reserves (= HPM = purchasing power) that the banks and/or citizens have surrendered (with bond sales: temporarily, with taxation: for good) to the state. If the government had simply printed fresh purchasing power for itself and gone shopping with it, inflation might have ensued (depending on the degree of capacity utilization).
Result: Reserves + 1.000,-; total bank reserves back up again to 10.000,-.

What has our model demonstrated, in what respect has it falsified MMT-assertions?
First of all, in applying their own model we have disproven the thesis of Fullwiler/Kelton/Wray, that what they hold to be the "general case" (1A ff.) is in the end identical to the reality of government funding in what they call the "US-case (Step 1 ff.).

This also implies a confutation of the claim made by Tymoigne/Wray, according to which "a fiscal deficit represents a net injection of currency".

Equally wrong (if not worse) is the statement p. 27 Tymoigne/Wray that "The end result is exactly the same as if the central bank had bought directly from the Treasury." Obviously, the authors must have forgotten in a flash of a second what they had written in the very sentences preceding this statement (emphases added):
"if one wants to account for institutional aspects ....., one should account for all of them, namely those that constrain Treasury-Central Bank operations, and those that allow Treasury to bypass these constraints (Tymoigne 2013). For example, in the US special dealer banks always stand by to purchase treasuries and the Fed ensures there are sufficient reserves to do so by supplying them through temporary repos (a matched purchase of Treasury debt with a requirement that the seller must repurchase later). While the Fed is not in that case directly buying the new issue directly from the Treasury, it uses the open market purchase to buy an existing bond in order to provide reserves needed for a private bank to buy the new security." 
Well, a "repo" is a "repurchasing agreement" (Wikipedia). Economically speaking (i. e. legal aspects aside) with a "repo" the Fed only lends currency to the primary dealers, who have to pay it back (i. e. buy back the - old - bonds) in a relatively short time (if the description of Fullwiler/Kelton/Wray is correct).
If the Fed buys new bonds from the Treasury, there is no repo contract (where would the govt. get the money from to redeem its bonds?). Instead, the Fed would have to "print" currency and hand it over to the government. Which would have to pay it back only at maturity - maybe after a couple of years. 
So while the extra currency which the Fed "prints" and "lends" to the primary dealers is destroyed after a few days or weeks, it would be out in the economy for years if the Fed would buy the bonds for good. [Regardless whether directly from the Treasury, which is illegal, or indirectly from the market, which the Fed in fact is doing through its so-called quantitative easing (QE).]

True: In connection with the bond auction procedure, our model did assume a "net injection of currency". But this had already been removed by the CB before the Treasury spent the proceeds from the bond sales. Also, immediately before govt. spending the reserves were lower than at the outset (i. e. before the Fed injected money).
Actually, we have even enhanced our model in favour of MMT-assumptions. Fullwiler/Kelton/Wray themselves say (under 1B, p. 7, emphasis added):
"Note that the point here is not that the Fed necessarily engages in operations that are equal to or greater than the auction, but that the operations ensure that sufficient balances circulate such that the auction settles without the effective federal funds rate for the day moving above the target rate. This requires that the balances already in circulation plus those added via operations are sufficient to settle the auction and enable banks in the aggregate to end the day with their desired positions at the target rate largely equal to actual positions."
So assuming that enough reserves were already out in the system to meet the above requirements there would be no need whatsovever for the Fed to agree on repos of (old) bonds with the primary dealers.
Once again, phrasing our findings a little differently:
The money that the Treasury spent was not "drained" from the reserves by subsequent bond sales. Instead, the Treasury had first sold bonds. This has happened within the framework of a rather complicated market-procedure which included smoothing-operations by the Fed (Step 1 - Step 4).
However: These smoothing-operations have no lasting effects. The financial position of the fed was already back to start before any government spending took place.
Therefore, the removal of reserves from the banking system by the Treasury (just before spending the money) necessarily drove down the reserves below their original level.
Consequently, when the Treasury spent the money (Step 6) the reserves only were brought back up to their initial level of 10.000,- (i. e. prior to government bond sales and government spending the proceeds). Contrary to the phantoms of MMT imagination, the government could go and has gone shopping without extra money created by Fed.
This is not surprising at all if we conceive of government income (here including loans) as a share taken out of the citizens' purchasing power.
But for any normal reader who does not analyze the MMT-esotericist presentation of purely technical and transitory procedures it must look like government does and can only spend EXTRA money "printed" by the Fed. This, of course, is precisely the impression that the high priests of MMT want to convey to their believers and to an unsuspecting public.
It is the indispensable mindset behind the idea that government is not spending constrained, which they tout as their fabulous discovery.
Of course it is perfectly true that there are no financial spending restrictions for a government that prints its own money. But only on a purely technical level. This fact is simply included in the definition of "fiat" money.
And at the moment it is even true at the factual level: when the Fed purchases the government bonds for good from the banks, as it now does (up to a certain amount) by its so-called QE, then obviously the HPM supply (i. e. from a bank perspective: reserve supplies) must necessarily go up. But under normal circumstances the "self-imposed restrictions" prevent the Fed from monetizing government debt.
And these constraints exist for good reasons (more about this in the piece by Cullen Roche).
To obtain a framework what the debate on government funding is all about*, it may be useful to construct a theoretical framework which displays the -4- alternatives (A1 - A4) theoretically feasible (regardless of what effects - some potentially negative, others maybe benign - they may have on the real economy).
* (From what I've read in MMT-contributions and those of their opponents, I did not gain the impression that the debaters always had a clear grasp of what the others meant. And sometimes not even on what they themselves were referring to.)

The basic assumptions for all 4 varieties of the model are:

- 10.000,- currency (of whichever denomination, created some time in the past by government spending or CB lending) out in the economy prior to any government collecting and spending
- 2.000,- government spending each month (full amount of what is collected in advance or retroactively p. m. is spent by the end of the month)
- 1.000,- monthly taxation (alternatively at the beginning or at the end of the month)
- 1.000,- monthly bond sales (alternatively at the beginning or at the end of the month)

A1 (funding before spending, bond sales to the public):
  1. Beginning of month 1 (M1): Govt. collects 1.000,- in taxes and 1.000,- in bonds. Currency is down by 2.000,- to 8.000,-. Public holds 1.000,- bonds; govt. has 1.000,- debt.
  2. End of M1: Govt. has spent full 2.000,-; reserves are back up to the original 10.000,- (Note: currency fluctuates during the month between 8.000,- at the beginning and 10.000,- at the end when govt. has spent the full amount.)
  3. Beginning of M2: Same game (2.000,- collected by govt., currency depleted accordingly; 2.000,- spent, currency refilled to original level). Public now holds 2.000,- bonds, govt. has 2.000,- debt.
  4. ................... etc. Currency does not exceed the initial level.
  5. End of M12: Currency 10.000,-. Throughout the year, govt. has collected 12.000,- in taxes and 12.000,- in loans. Accordingly, it has spent 24.000,- Public holds 12.000,- bonds, govt. has 12.000,- debt.

A2 (final funding after spending*, bond sales to the public):
  1. Beginning M1: Govt. "prints" (i. e. govt. proper* borrows from CB) 2.000,-      [* The reason why I speak of "govt. proper" is to make a distinction between the administration (represented by the Treasury) and the CB. This is necessary here because MMT consolidates both. (Allegedly they do this for clarity and simplicity. In reality it is the indispensable logical basis for their claim that money gets destroyed when deposited in Treasury account with Fed).]
  2. End of M1: Govt. has spent 2.000,-; currency up to 12.000,-. (Note: currency fluctuates during the month between 10.000,- at the beginning and 12.000,- at the end when govt. has spent the full amount.)
  3. Turning point of M1/M2: Govt. collects 1.000,- in taxes and sells 1.000,- in bonds to public. Currency down 2.000,- to 10.000,-.
  4. Beginning of M2: Since the proceeds of funding at the turning point between months only serve to cover the advance from the CB from the previous month, govt. proper must borrow another 2.000,- from CB.
  5. End of M2: govt. spending during the month has again raised currency level up to 12.000,-.
  6. The same prodecure occurs every month.  Consequently, currency throughout the year stands at 12.000,- (up 2.000,- from initial level) at the end of each month. (But it fluctuates during each month between 10.000,- at the beginning and 12.000,- at the end when govt. has spent the full amount.) To put it differently: Reserves grow only in the 1st and remain at
  7. End of M12: Currency 12.000,-. Throughout the year, govt. has collected 12.000,- in taxes and 12.000,- in loans. (However, under this scenario only 10.000,- from the public and the missing 2.000,- from CB.) Accordingly, govt. has spent 24.000,- and is left with 12.000,- debt. Of this, the public holds 10.000,- and CB 2.000,- (which is why reserves are up by 2.000,-).
*[Of course funding must always take place BEFORE the money is spent. Therefore, whatever funding comes from the public but only AFTER spending, is temporarily pre-financed by the CB.]

A3 (funding before spending, bond sales to CB):
  1. Beginning of M1: Govt. collects 1.000,- in taxes and 1.000,- loan (duration???) from Fed. Currency down to 9.000,-. (Here, only taxation drains currency / reserves. Bonds in this case do not, because the money was "printed" by the Fed.)
  2. End of M1:govt. has spent 2.000,-; currency up 2.000,- to 11.000,-
  3. Beginning of M2: Govt. collects 1.000,- in taxes and 1.000,- loan from Fed. Stock of currency diminshed by 1.000,- to 10.000,-
  4. End of M2: Govt. has spent 2.000,-. Currency up 2.000,- to 12.000,-
  5. ........... And so on, i. e. currency grows at the end of every month by 1.000,-.
  6. At the end of M12, we have 10.000,- + 12.000,- = 22.000,- currency. Govt. has spent 24.000,- (same amount as in all the other varieties of our model). Fed holds 12.000 govt. bonds (or, because this debt could just as well be documented in a simple credit contract: "Fed has lent 12.000,- to govt.").

A4 (final funding after spending, bond sales to CB):
  1. Beginning of M1: Govt. borrows 2.000,- from CB 
  2. End of M1: Govt. has spent 2.000,-; currency 12.000,-
  3. Turning point of M1/M2: Govt. collects 1.000,- in taxes (and pays back the corresponding amount to CB lending). Currency 11.000,-
  4. Beginning of M2: Govt. borrows another 2.000,- from CB.
  5. End of M2: Govt. has spent 2.000,-; currency up 1.000,- to 12.000,-
  6. ............ And so on, i. e. currency grows at the end of every month by 1.000,-.
  7. At the end of M12, we have 10.000,- + 13.000,- = 23.000,- currency. Govt. has spent 24.000,-. Fed has lent 13.000,- to govt. (1.000,- net credit to govt. for each month plus a onetime advance of 1.000,- because taxes are collected only at end of month). 
And now, Ladies and Gentlemen: Have it your way! Make your choice what maximum amount of currency you want to have in the system at the end of the year:
  • 10.000,-
  • 12.000,-
  • 22.000,- or
  • 23.000,-.  

Of course, you can easily increase the amount in the following years. After all: It's only money!

(Decreasing is somewhat more difficult because, as we all know, for political greasing expenditure must keep increasing.)

What insight do we gain from this theoretical model in 4 variations?
That MMT-literature (but also the criticism of it) is utterly vague! It is by no means clear which of the four financing models for government expenses they refer to whenever MMTers say "government spending [or at least "injection" of currency] must take place before funding".
And most likely MMTers don't know themselves what exact situation they have in mind. At least this is the impression I glean from reading their pseudo-erudite mumble-jumble.
For one thing, speaking of "deficit spending" is void of meaning in any MMT-discourse without an explanation as to who finances the deficit (public or CB) and at what intervals in time any govt. collection of money from the public is being assumed.
I'm really anxious to see what concoctions MMTers will come up with to "prove" that my models are all wrong, how they have been misunderstood and that (their) reality is completely different.

Text as of Jan. 26, 2015

Keine Kommentare:

Kommentar veröffentlichen