RE: NOTHING TO SUBTRACT

I have had time to review Tony's comment and the snippet from his paper-which looks
very interesting, by the way. As expected, the maths appear to be fine. But I'm afraid the
model has got the bull by the wrong end. Here I repeat the critical passage from Tony's
paper:

"Finally, the budget deficit is financed by issue of outside money and issue of
government bonds, in proportions (?, 1-?) respectively:"

Hence, his model applies to households, firms, nonsovereign governments (state or local
governments, or governments operating on a gold standard). But the model has no
relevance for sovereign governments that issue and float the currency.

Let us back up and try to get the bull by the horns. I will start with a very simple
explication, very much along the lines suggested by Tony. I am going to leave to the side
all the complex coordination between the treasury and central bank. This coordination
causes much consternation but has been recently covered in a JPKE article co-written
with Bell. Here, we will only focus on the government and nongovernment sectors.
Alternatively, you could call the first the "issuer of the currency" and the second the
"user of the currency". Note that unlike Tony's model, I will leave out equity, but I will
distinguish between "bank money" and "government money" or HPM.

Let us presume government spends by issuing checks or by directly crediting bank
accounts of nongovernment agents. It buys goods or services, or it makes social welfare
payments. No nongovernmental agent is forced to take government checks or to accept
direct credits to their bank accounts-all sales or social welfare receipts are purely
voluntary. Those who receive checks must deposit them into their bank accounts before
they can write private checks (thereby "spending" the proceeds received from
government). Hence, when all is said and done, government spending results in an
increase of bank deposits held by the nongovernment sector. Of course, the government
simultaneously credits the banks with HPM reserves.

For the nonbank nongovernment agents, it is difficult to conceive of "excessive holdings"
of HPM. Government spending increases their deposit accounts ("bank money"). Only if
they wish to have some HPM will they go to the bank to withdraw some cash. If
somehow they end up with more than they desire, they deposit it into their bank account.
The "portfolio preferences" of the public are irrelevant here because they are
automatically and continuously accommodated by the banks. (Note that there is no trick
involved in presuming government does not spend by issuing currency. Even if
government spent by delivering a truckload of coins, the nonbank nongovernment agents
would take any excess to the banks.) No matter what the portfolio preferences of the
nonbank nongovernment happen to be, government can spend by proffering checks or
crediting accounts. No welfare recipients turn down the checks simply because their
mattresses are already stuffed full of cash; no bomb producers turn down government
offers to purchase simply because cash registers are already filled with 100s.

What about the nongovernment banks? Any bank receiving a reserve credit can choose to
keep it, or can offer it in the overnight market. If banks in the aggregate have all the
reserves they want, then any offer of reserves in the overnight market will meet with a
zero bid condition. The overnight rate must be zero. Hence, we see that the "portfolio
preferences" of nongovernment banks do matter-if they preferred to hold more HPM,
the overnight rate would be positive as offers would be met with positive bids. If they
want less HPM, the overnight rate is bid toward zero. But note that the only impact their
portfolio preferences have is on the overnight rate-their preferences have no impact on
the ability of the government to "finance" its deficit.

Now the question is whether government can set, hit, and maintain any particular
overnight rate target. Tony would like government to hit zero. Easy enough. Leave excess
(undesired) reserves in the banking system. Go home early, job well done.

What if a Volcker-run Fed wants 100%? Easy enough. Pay 100% on reserve deposits (Oz
system); or, announce a target of 100%, then drain excess reserves by selling government
bonds (US system). Might the nongovernment banking system's portfolio preferences be
interest rate sensitive? Perhaps. In that case, the amount of excess reserves that need to be
drained through government bond sales might differ depending on the target rate chosen.
This has no impact on government's ability to "finance" its spending, although the
resulting outstanding government debt stock might vary depending on the target rate
chosen. If government does not like that, it can always choose NOT to sell any bonds and
simply pay interest on reserves.

Of course, if bonds are sold in an open market that allows nonbanks to bid, then
nongovernment nonbank portfolio preferences enter because they will go into
determining the split of ownership of government bonds between banks and nonbanks.
When nonbanks buy them, bank reserves are debited, hence, reserves are drained just the
same. So bond sales always drain reserves, but nonbank preferences affect the allocation
of bonds between banks and nonbanks. If government does not like that, it can choose
NOT to sell bonds to the nonbank agents.

Now that we've got the bull by the right end, let us return to Tony's request for "some
formal demonstration" that government deficits have some self correcting mechanism. I
do think that it is worthwhile to posit some sort of "portfolio preference" for the nonbank
nongovernment sector, but not in the manner pursued by Tony. Let us presume that
nonbank nongovernment agents link their consumption to their net income (income less
taxes) and to their net financial wealth. Let us further presume, as Tony does, that these
agents view "outside" net financial wealth somewhat differently from "inside" net
financial wealth-with outside net financial wealth having a larger impact on
consumption. In any case, since all inside financial wealth cancels at the aggregate level,
increases to inside financial wealth impact consumption through differential marginal
propensities to consume. By contrast, increases to outside financial wealth will increase
consumption even if all marginal propensities to consume are the same.

A government deficit increases income and thus consumption through the multiplier
effect (as in Tony's model); it also increases net outside financial wealth and hence
consumption. If taxes are highly pro-cyclical and government spending is highly counter-
cyclical, then it is easy to see that a slow-down of private spending will generate deficits,
restoring income and increasing net outside financial wealth. Once we add expectations
and whirlwinds of optimism to the mix, it should be easy to see why the deficit will be
self-limiting. If automatic stabilizers are weak, the process will be slow-a downturn
may take years to depress income and employment sufficiently to get the deficit to the
required level to raise income and net outside financial wealth to the point that
expectations are restored. (Japan is a real world example.)

Warren and I like to think about a "desired net saving" position for the nongovernment
sector. When fiscal policy is too tight, the actual net saving position achieved can be
lower than the desired position, setting off deflationary pressures as nongovernment
spending is reduced. With automatic stabilizers, the government's deficit will rise until
actual net saving rises to equality with desired net saving. When fiscal policy is too loose,
net saving exceeds desired, generating more spending and reducing the deficit. Desired
net saving positions, in turn, might be cyclical-in a boom they may well go negative.
The government's budget can even be driven into surplus (USA during Goldilocks). In a
slump, desired net saving positions rise-raising the government deficit required to
reverse the slump. The strength of automatic stabilizers not only determines the
quickness with which the budget reacts, but also affects the ultimate size of the correction
required through impacts on expectations.

One can treat all this through portfolio preferences and varying multipliers. In any case it
is all very simple and clear enough that one supposes maths would be superfluous. But if
one begins with a Government Budget Constraint and portfolio preferences that constrain
government's ability to spend then all the maths in the world won't help to develop an
understanding of the fundamental difference between an issuer and a user of the currency.
Received on Thu Feb 20 03:26:14 2003