Does Central Banking Lending to Government Count as Debt?

In the two previous blog pieces, we looked at the argument put forward by “monetary populists” that the central bank (CB) should finance fiscal stimulus by making direct grants to government. In this piece, we look at an alternative proposition put forward by other monetary populists, who argue that even when the CB lends money to government (rather than granting it), it is in effect providing debt-free funding. Their argument is that government borrowings from the CB you do not count as debt, and it is only when government borrows from the private sector that it incurs true debt.

In this spirit, the most vocal proponent of ‘people’s QE’ proposal in the UK, Richard Murphy (2015), has asserted that government bonds held by the CB have effectively been cancelled, and should not be counted when measuring debt. Why? Because the CB being part of government, a loan from the CB to government is like someone lending him/herself money. The fact that the government’s interest payments to the CB are simply repaid to it underlines this.

This view that government bonds held by the CB don’t count as debt – what we might call the ‘effectively cancelled’ proposition – is an old staple of monetary populism.

In accounting terms, the reason why government debt as conventionally measured does not subtract CB-held bonds is simple. Universally, the debt measure which governments use for fiscal policy purposes is based on a definition of ‘government’ which excludes government banks (‘public financial corporations’), the most important of which is the CB.[1]

What the proponents of the ‘effectively cancelled’ proposition want is an accounting change. As Murphy (2015) puts it, ‘in any proper accounting system that produced a single set of accounts for government … debt that was repurchased [by the central bank] would have been considered to be cancelled’. The proposal is, in other words, that the central bank be ‘consolidated’ (in a certain fashion[2]) with general government when debt is measured. In consolidated accounts, debt owed by one internal unit (in this case, the Treasury) to another internal unit (in this case, the CB) would not count. In the UK, this would reduce the government’s (2015) debt of £1.5 trillion by approximately one-quarter. In the US, the $13.1 trillion (end-2015) federal debt would be cut by 19 percent.[3]

Changing accounting practice is fine, but only if the debt measure it produces is more meaningful. However, consolidating central bank debt with government would in fact produce a much less meaningful debt measure.

In the first place, it would give the illusion that the government’s financial position improves if government borrows money from the CB rather than from the public. However, a moment’s thought (along the lines of the discussion in the last blog piece) will make it clear that what benefits the government is not the fact that it is borrowing from the CB rather than the public, but the fact that the CB is creating new money.[4] This is something which the proponents of the ‘effectively cancelled’ proposition do not understand. They sincerely but wrongly believe that the simple act of measuring debt properly, by subtracting CB holdings of government bonds, ‘undermines the case for growth-retarding austerity’ (Duncan, 2015).

In the second place, the measure of government debt which they favour is one which would fluctuate whenever the CB uses open market operations to implement monetary policy. Debt would fall whenever the CB purchased government bonds in order to relax monetary conditions, and would increase whenever the CB sold government bonds to tighten monetary conditions. The consequences of this could be particularly large in the aftermath of QE. If one deducts today from measured public debt the huge stock of government bonds acquired by CBs as part of QE, one will have to effectively add this amount back when the CBs unwind QE by selling the bonds. Only to the extent to which the QE monetary expansion turns out to be permanent would the debt-reducing effect be permanent.[5]

The ‘consolidated’ debt measure would therefore constitute a type of hybrid indicator of fiscal policy and monetary policy, and would be much less useful as a fiscal policy indicator than the conventional measure.

The proposition that CB loans to government amount to the cancellation of public debt is therefore incorrect, as is also the companion notion that government debt is correctly measured by netting off the CB’s holdings of government bonds. Moreover, as outlined in the last two blog pieces, it is an illusion to believe that the fiscal position of government can be improved if it avoids borrowing by tapping grants from the CB. The potential benefits of a self-financing fiscal stimulus supported by accommodating monetary policy are entirely unrelated to whether the package is funded by borrowing or CB grants.

It is therefore a mistake to confuse the case for a Keynesian alternative to austerity policies with notions derived from monetary populism.



[1] Many countries use a measure of ‘general government’ debt, based on the IMF GFS definition of government which excludes not only government banks, but also other government-owned enterprises (‘public non-financial corporations’). In the UK, the standard measure is ‘public sector net debt (PSND)’ – recently renamed ‘PSND excluding government banks’ – is different in that it includes public non-financial corporations. But it also excludes the Bank of England.

[2] Implicitly, he is calling for consolidation in which the CB’s ‘liability’ for base money (currency on issue and money held in the accounts of commercial banks at the CB) is not counted. In the official balance sheets of CBs, base money is treated as a liability. If the accounts of the CB and government are consolidated in the conventional manner (as, for example, in the UK’s Whole of Government accounts), these liabilities would be counted and this would offset the impact of consolidation in reducing consolidated public debt. However, this is a point of accounting pedantry rather than a substantial critique of the ‘effectively cancelled’ proposition. As Buiter (2014) has pointed out, base money is not in any real sense a liability of the central bank, and it is perfectly reasonable to ignore it when consolidating government and the central bank.

[3] As measured by the principal debt measure used by the government (‘debt held by the public’).

[4] Direct CB lending to the government creates new money, whereas government borrowing from the public does not. If the government borrowing is accompanied by the same degree of monetary expansion (by the CB purchasing bonds from the public), government enjoys an increased flow of income from the CB which is equal to the interest it has to pay the public on the newly issued bonds.

[5] A variant of the ‘effectively cancelled’ proposition put forward by McCulley and Pozsar (2013: 2-23) holds that if the CB declares that some part of the monetary expansion is intended to be permanent, it is reasonable  to deduct that from the (consolidated) measure of government debt. But the credibility of such a declaration must necessarily be doubtful, given the potential ‘time consistency’ problem.

Comments are closed.