Modern Monetary Theory (MMT) makes clear that the spending of a currency-issuing government necessarily precedes tax payments and bond sales to non-government. This has important implications. It means that a currency-issuing government does not – and cannot – require revenue prior to spending. It means that it is government spending that makes possible the payment of taxes and non-government purchase of bonds, not the other way round. And it means that when a government requires itself to match deficits with bond sales to non-government, it does so voluntarily, and on terms that are entirely at its discretion. In short, the constraints on a currency-issuing government are not financial in nature. The constraints on the spending of a currency-issuing government are properly understood in terms of real resources and the capacity of the economy to meet demand for higher output at stable prices. A currency-issuing government always has the capacity to purchase whatever is for sale in the currency of issue (which, of course, is not the same thing as saying that government should always do this). Yet, the manner in which fiscal and monetary operations are usually conducted can create an appearance that things are somehow otherwise. So, why is it that government spending necessarily comes first? In view of its implications, this is an important matter to grasp.
It helps to begin with what MMT depicts as the ‘general’ case and then proceed to complications that arise in particular or ‘special’ cases. To be clear, whatever case is considered, it is always higher levels of government that authorize government spending and taxing measures. For example, in the US, spending measures are authorized by Congress and the President. Once the spending measures have been passed into law, it becomes the legal obligation of the government’s designated agent (or agents) to ensure that the measures are carried out. Higher levels of government may impose constraints on the way lower levels of government (including the government’s designated agents) conduct their operations. For our purposes, the most relevant lower levels of government are the fiscal authority (the “Treasury”) and the monetary authority (the “central bank”).
The general case can be regarded as a situation in which higher levels of government refrain from imposing technically unnecessary constraints on the way in which fiscal policy is conducted. Particular cases are then distinguished by the set of operational constraints that higher levels of government choose to impose on the Treasury and central bank beyond those that are technically necessary for the authorized spending actually to occur. When it comes to fiscal policy, three technically unnecessary operational constraints have often applied in practice:
A. The central bank is not allowed to incur an overdraft. Typically, the central bank must maintain a net financial balance of zero at all times.
B. The Treasury is not allowed to incur an overdraft. This means, in practice, that the Treasury must match fiscal deficits with bond issues.
C. The Treasury is forbidden from selling bonds directly to the central bank. The Treasury must instead offer bonds to the private sector.
In many countries, all these constraints have often been in force at the same time. In what follows, four cases are distinguished: the general case and three special cases. The general case involves no technically unnecessary operational constraints. The first special case applies A alone. The second special case applies both A and B. The third special case applies all of A, B and C.
As we will see, no matter what the institutional detail – whether the system is characterized by the general or a particular case – it always remains true that the spending of a currency-issuing government precedes tax payments and bond sales to non-government. This is not a matter of choice but of logical and technical necessity. The challenge is to understand why.
The general case
The general case is the simplest to describe. Once higher levels of government have authorized spending measures, the Treasury instructs the central bank to mark up the reserve accounts of banks at which spending recipients have accounts, and the bank accounts of spending recipients are marked up by the same amounts.
The newly created reserves are a form of ‘government money’ (currency being the other form). In adding new reserves, government spending creates government money.
The newly created reserves are a liability of government (specifically, the central bank) and an asset of non-government (specifically, the banks). The newly created bank deposits are a liability of one segment of non-government (the banks) and an asset of another segment of non-government (the spending recipients). Net financial assets – non-government financial assets minus non-government financial liabilities – rise by the amount of the government spending.
In aggregate, net financial assets are the sum of reserves, currency on issue and outstanding government bonds (though bonds are not needed in the general case). All other financial assets held by non-government have offsetting liabilities within the non-government sector, and so net to zero for non-government as a whole. In contrast, the extra reserves created by government spending add to the net financial assets held by non-government, because the corresponding liability is held in the government sector (reserves are a liability of the central bank).
In marking up reserve accounts as directed by the Treasury, the central bank creates the reserves ex nihilo (“out of nothing”). There is nothing unusual about this. All financial liabilities are – and can only be – created ex nihilo by their issuers. Banks create deposits ex nihilo when they lend. Individuals create liabilities ex nihilo when they issue personal IOUs. Ex nihilo is the only way a financial liability can ever be issued.
As just described, the central bank creates a form of government money simply by marking up reserve accounts. Government money is a liability of government to non-government, just as bank deposits (bank money) are a liability of banks to deposit holders, and just as a personal IOU is a liability of its issuer to the holder of the IOU. Like any issuer of an IOU, the government is obliged (i.e. liable) to accept back its IOUs as payment to it. The currency-issuing government commits to accepting its own IOUs back again in payment of taxes, fees, fines and other charges it has imposed.
Since taxes only settle in government money, government money must be created before taxes can be paid. Government spending creates the money that is required for tax settlement.
Payments to government reverse the effects of government spending. Whereas government spending creates government money, tax payments destroy it. For example, a taxpayer with an account at a bank can instruct the bank to mark down the account. The bank, in turn, will instruct the central bank to mark down a reserve account. Reserves are eliminated from the system as the ultimate result of taxation. Or if the taxpayer has an account at a building society, the taxpayer will instruct the building society to mark down the account, the building society will instruct the bank at which it holds an account to mark down an account, and the bank will instruct the central bank to mark down a reserve account. Again, reserves are eliminated from the system. In rare instances, a taxpayer might pay with currency rather than going through the banking system. But, in all cases, the tax payment will ultimately result in a deletion of government money from the economy (either reserves or currency).
Whenever the government spends more into the economy than it taxes out of the economy, the government’s net financial liabilities rise and the net financial assets of non-government rise by the same amount. This is always true, irrespective of the operational methods adopted by the Treasury and central bank. The only difference, under an alternative institutional arrangement, will be in how the net financial assets (government liabilities) are composed. In the general case, it is the central bank’s net financial liabilities that are affected. This is clear because, in issuing reserves, the central bank incurs a new liability without acquiring a new asset or eliminating an existing liability. In contrast, the Treasury’s financial balance, if it makes sense to speak of the Treasury’s financial balance in the general case, is unaffected. The Treasury, in the general case, is not directly engaging in financial operations. It simply directs the central bank to carry out the various spending measures that have been authorized by higher levels of government.
Needless to say, the central bank’s net financial balance holds no implications for its financial capacity to carry out the instructions delivered to it by the Treasury. Since reserves are created ex nihilo, the financial capacity of the central bank to issue new liabilities is unaffected by the extent to which it has issued such liabilities in the past.
What matters in terms of the economics – and what will continue to matter, no matter what set of operational constraints are entertained – are the substantive effects of the government’s policy measures. Relevant economic issues include the impact of government policy on the total level of spending, considered in relation to the supply capacity of the economy to meet demand at stable prices, along with the distributional and compositional impacts of policy. In contrast, the financial capacity of a currency-issuing government, being unlimited, is never a valid consideration.
Special case 1 – central bank prohibited from incurring an overdraft
For whatever reason, higher levels of government might require the central bank to maintain a zero net financial balance at all times. Under this operational constraint, it is no longer permissible for the central bank to alter the level of reserves in the system without at the same time engaging in offsetting operations to neutralize the net effect on its balance sheet. An offsetting operation might be for the central bank to require an asset in exchange for newly issued reserves so that its liabilities and assets change by the same amount. Or, alternatively, an offsetting operation might entail the central bank reducing one of its other liabilities, to negate the impact on its net financial position of the new reserve issue.
One way to handle the situation is to define into existence a liability of the central bank to some other part of government. That way, whenever the central bank issues reserves without requiring something from non-government in return (as occurs when the government spends), it can offset the effect on its own balance sheet by reducing its liability to this other part of government.
The establishment of a ‘Treasury account’ fulfills this purpose. The Treasury account, as the name suggests, serves as the Treasury’s account at the central bank. The balance on the Treasury account is a liability of the central bank and an asset of the Treasury. The central bank is now liable to the Treasury to the extent that there is a balance on the Treasury account. Of course, for government as a whole, the Treasury account is both an asset and a liability, and so sums to zero. It is simply an internal government accounting arrangement that has no effect on non-government. It is a way of recording – and keeping track of – government spending and taxing activities. But for the central bank specifically, the Treasury account is a liability. So, when the central bank issues new reserves as part of the government’s spending operations (which, viewed in isolation, adds to the central bank’s liabilities), the central bank can now offset the effect on its net financial position by marking down the Treasury account (which, viewed in isolation, reduces the central bank’s liabilities).
The introduction of the Treasury account does nothing to alter the economic effects of government spending. The economic effects, as always, will be in the government spending itself, including impacts on demand – both direct and indirect – and the utilization of real resources.
The introduction of the Treasury account also does nothing to alter the financial effects of government spending so far as non-government is concerned. As always, government spending will add net financial assets. Or, to say the same thing in a different way, the net financial liabilities of government will increase by the amount of the spending. All that changes is the entity within government that bears the change in net financial position. In the general case, it is the central bank’s net position that changes: reserves are issued without any operation to offset the net impact on the central bank’s balance sheet. In special case 1, it is the Treasury’s net position that changes: the reserve issue is offset, on the central bank’s balance sheet, by a debiting of the Treasury account.
The balance on the Treasury account is a wash for government as a whole, and equally irrelevant to non-government. The account’s existence, however, does enable the central bank and Treasury to carry out the spending measures passed into law by higher levels of government without violating the constraint (of special case 1) that forbids the central bank from incurring an overdraft. In effect, the Treasury incurs overdrafts when necessary so that the central bank need not do so.
Under the constraints of special case 1, government spending is accompanied by two basic steps. First, the Treasury instructs the central bank to mark down the Treasury account by the amount of the authorized spending and to mark up the reserve accounts of the banks at which spending recipients have accounts. Second, the central bank duly marks up reserve accounts, as it has been instructed to do, and directs the banks to credit the accounts of spending recipients.
Tax payments, as always, simply undo the effects of government spending. In special case 1, when reserves are deleted in final settlement of taxes, the central bank marks up the Treasury account. The central bank’s overall position is unaffected (reserve accounts are marked down and the Treasury account is marked up). The Treasury has a higher balance in its account at the central bank.
The higher balance on the Treasury account is of no consequence to non-government. The Treasury account is a liability of the central bank, not a liability of non-government. What is of consequence to non-government is the reduction in net financial assets that results from taxation. The reduction in net financial assets is reflected at the aggregate level in the depletion of reserves held within the banking system.
The fact that the balance on the Treasury account is not a liability of government, but merely a liability of one arm of government to another, is significant. It means that the balance on the Treasury account is not government money. Money is always a liability of its issuer. This is as true of bank money (bank deposits) and personal IOUs as it is of government money. Since the Treasury account is not a liability of government, but merely an intragovernmental accounting arrangement that nets to zero within the government sector, the balance in the Treasury account cannot be government money. Since only government money can extinguish tax obligations, it is government money (in practice, reserves) that must be spent into existence when the government spends. Otherwise, the financial obligations imposed by government could never be met. For this reason, the balance recorded on the Treasury account is not what the government spends when it spends. A currency-issuing government spends in its own money, which it creates ex nihilo every time it spends, through the marking up of reserve accounts.
Whereas the balance on the Treasury account nets to zero for the government sector as a whole – and so cannot possibly constitute government money – the same is not true of reserves. Reserves are a liability of government to non-government. This is clear because the accounting offset for reserves is in the non-government sector. Reserves, unlike the balance on the Treasury account, are a liability of government and an asset of non-government. For the very reason that reserves are a liability of government, they can serve as a form of government money.
The point is this. The spending of a currency-issuing government necessarily entails the creation of government money. If this were not the case, non-government’s financial obligations to government – which only settle in government money – could never be extinguished. Even government lending is insufficient for the purpose, because a loan from government to non-government to enable a tax to be paid would simply replace a tax liability with a loan liability. The liability created by the imposition of the tax cannot be eradicated until government actually spends its money into existence. And when it spends, a currency-issuing government cannot spend out of a pre-existing fund of government money. No such pre-existing fund exists. Just as it is logically impossible for the balance on the Treasury account to serve as government money, it is likewise impossible for any other intragovernmental accounting arrangement to serve as government money. The money that is needed to extinguish the financial obligations imposed by a currency-issuing government can only come from government spending.
Therefore, whether the government spends by marking up reserve accounts and directing banks appropriately (general case) or spends by marking up reserve accounts, marking down the Treasury account and directing banks appropriately (special case 1), the substance of government spending and any economic impacts are the same.
Special case 2 – Treasury must match government deficits with bond sales
When government spends more than it taxes on a more or less ongoing basis – which will occur if, as is typical of most nations, non-government opts to spend less than its income – the government’s net financial liabilities will rise over time in lockstep with the rise in net financial assets accumulated by non-government. This is simply a matter of accounting. The situation creates no financial difficulty for a currency-issuing government, since such a government is the source of the currency in which non-government is accumulating financial assets. But, as has been observed, the composition of government liabilities will differ, depending on the operational approach the Treasury and central bank are permitted to adopt. In the general case, ongoing government deficits translate into central bank overdrafts. The central bank operates, in these circumstances, with a negative net financial position. In special case 1, higher levels of government deem it inappropriate for the central bank to incur an overdraft. The way around this, under the constraints of special case 1, is for the negative net financial position to be shifted to the Treasury. In special case 1, ongoing government deficits entail the Treasury operating with a negative balance on the Treasury account. Now, in special case 2, higher levels of government – for whatever reason – decide to forbid overdrafts on the Treasury account as well.
This additional restriction on fiscal operations means that the Treasury and central bank must arrive at some other way of ensuring – as is their legal obligation – that the spending measures authorized by higher levels of government are in fact carried out. What is needed, to get around the imposed operational constraints, is some mechanism for enabling the government to spend in excess of taxes – if and when this is the decision made at higher levels of government – without either the central bank or Treasury incurring an overdraft.
A solution is to introduce another accounting construct, the ‘government bond’. Under special case 2, the Treasury, to avoid recording a negative balance on the Treasury account, can ‘borrow’ from the central bank by issuing bonds which will serve as an asset of the central bank and liability of the Treasury. Whenever the Treasury and central bank are required by law to spend in excess of the balance currently showing on the Treasury account, the Treasury will sell bonds to the central bank. In exchange, the central bank will pay for the bonds by marking up the Treasury account. Any interest that the Treasury might commit to paying on the bond will be returned to the Treasury, since the central bank is required by law to return its profits back to the Treasury.
The arrangement, of course, nets to zero for government as a whole. The central bank has a new asset (the bonds) that is exactly offset by the change in its liabilities (the additional balance on the Treasury account). The Treasury, by issuing a new liability (the bonds), has an offsetting asset (a higher balance on the Treasury account). In this way, both the Treasury and central bank comply with the operational constraints of special case 2. The central bank is able to maintain a zero net financial balance. The Treasury is able to avoid a negative balance on its account at the central bank.
Government spending, under the operational constraints of special case 2, is accompanied by three steps whenever the amount of spending exceeds the balance on the Treasury account. First, the Treasury sells new bonds to the central bank, the central bank paying for the bonds by marking up the Treasury account. Second, the Treasury instructs the central bank to mark down the Treasury account and mark up the reserve accounts of the banks at which spending recipients have accounts. Third, the central bank instructs the banks to credit the bank accounts of spending recipients.
In the general case, neither the Treasury account nor bonds are required. In special case 1, the Treasury account is required, but not bonds. In special case 2, operations involving both the Treasury account and bonds are required in the event that the government authorizes spending in excess of the balance on the Treasury account.
Needless to say, the changes mean nothing from the perspective of non-government. As always, the government spending adds net financial assets and results in new bank deposits for spending recipients. As in the general case and special case 1, the extra net financial assets register at the aggregate level as extra reserves, while the actual direct beneficiaries are the spending recipients. The banks’ additional assets (the reserves) are offset by additional bank liabilities (the deposits).
Special case 3 – Treasury not permitted to sell bonds to the central bank
In special case 3, higher levels of government forbid the Treasury from selling government bonds directly to the central bank. Three operational constraints are now in place: the central bank cannot incur an overdraft; the Treasury must maintain a nonnegative balance on the Treasury account; and the Treasury must offer its new bond issues to non-government rather than selling the bonds directly to the central bank.
Under this set of constraints, if the Treasury is legally obligated to spend more than the balance on its account at the central bank, it must sell bonds to non-government. Typically, this is done through an auction in which bidders offering to accept the lowest yields are allocated the bonds. Mostly, the bidding is done by officially designated primary dealers.
A question that arises is whether the constraints of special case 3 mean that bonds now precede government spending? Despite surface appearances, the answer is no. Consider the regime of special case 3 at its inception. The first Treasury auction – like all Treasury auctions – can only settle in reserves. The reserves needed for settlement of the first Treasury auction can only exist because of past government spending. Consider, also, the situation once some time has elapsed, such that the new regime is up and running and non-government has acquired bonds issued at various Treasury auctions. At this point, the central bank may decide to utilize bonds in its conduct of monetary policy, using open market operations to control the short-term interest rate. Whenever settlement of a Treasury auction threatens to dislodge the short-term interest rate from target, the central bank will need to advance reserves to non-government before the Treasury auction can proceed. In advancing reserves, the central bank will require already existing government bonds as collateral (as part of a repurchase agreement). But, of course, the government bonds that are required as collateral only exist because of past deficit spending. Thus, while superficially it may appear – under an interest-setting monetary policy regime – that it is actually government lending rather than government spending that comes first, this impression is misleading. The liquidity management operations of the central bank are predicated on a prior existence of government bonds, and these bonds only exist because in the past the government has spent reserves into existence that could then be used to settle the bond purchases.
Another question to consider is whether the operational constraints of special case 3 undermine the Treasury and central bank’s control over the terms on which government spending occurs? The answer here is also no. In relation to the Treasury auction, broadly speaking there are two possible scenarios. If it is clear that non-government is prepared to purchase bonds on terms consistent with the government’s policy objectives, the outcome of the Treasury auction will obviously be satisfactory to government. If, on the contrary, the cutoff bid is likely to exceed the rate of interest that government is prepared to pay, the central bank will simply purchase – or commit to purchasing – bonds in secondary markets. By doing so, the central bank exerts upward pressure on bond prices (and so downward pressure on bond yields). Bond dealers know that by purchasing bonds at the auction on somewhat more favorable terms than the central bank is offering in secondary markets, there is an opportunity to sell the bonds to the central bank for a capital gain. (Whether bond dealers should be presented with such gifts is a valid concern, but irrelevant to the point at hand, which is that the Treasury and central bank do not lose control of the terms on which bonds are issued. The gift to bond dealers is a manifestation of the decision by higher levels of government to impose the operational constraints of special case 3.)
Put simply, the central bank always has the capacity to dictate the terms on which government bonds are issued. For instance, if the government’s desired rate of interest on government bonds is 0.25 percent, the central bank can set the rate of interest on reserves to a level lower than this (e.g. zero percent) and, by committing to buy bonds in the secondary market at a certain price, signal to participants in the auction that any bonds acquired at the auction on slightly more favorable terms can subsequently be sold to the central bank for a capital gain. If government so determines, the Treasury auction will succeed, no problem, even when the bonds offer negative interest. In that event, the central bank would set a negative interest rate on reserves (e.g. -1 percent). This would motivate a ready market for newly issued government bonds that offered better than the negative interest rate applying to reserves. This would essentially entail a tax on reserves (and bonds if the bond yields were also forced negative). Purchasing bonds at a negative rate (e.g. -0.5 percent) would be a way to avoid the still more negative interest charge on reserves. Even if banks attempted to purchase currency from the central bank to avoid interest charges on reserves (assuming the charges exceeded the storage costs of holding currency), the central bank could impose comparable charges on the supply of currency.
So, under the operational constraints of special case 3, the Treasury and central bank ensure that any deficit spending authorized by higher levels of government can be carried out by including an additional couple of steps in their operations. Whereas, in special case 2, the central bank purchases bonds directly from Treasury, now the central bank acquires bonds, if and when necessary, indirectly via secondary markets. When government authorizes spending in excess of the balance on the Treasury account, the Treasury auctions off bonds to non-government. Meanwhile, the central bank exerts whatever pressure on the auction’s outcome it deems necessary through secondary market operations. As the issuer of reserves, the central bank has an unlimited capacity to buy up previously issued bonds.
The aggregate impact of government spending is the same in special case 3 as in all other cases: there is an increase in net financial assets and a corresponding increase in the government’s net financial liabilities. There is a change, though, in the composition of net financial assets (and the government’s net financial liabilities). Rather than the extra net financial assets being held as reserves, they are switched for bonds.
The economic effects of government net spending remain the same under the constraints of special case 3 if it is assumed that the interest rate applying to bonds is equivalent to the interest rate that would apply to reserves under the other cases. In particular, the various operational regimes are equivalent in terms of inflation risk. Purchasing a bond in no way prevents spending. Any bondholder wishing to spend need only sell the bond to someone else or use the bond as high-quality collateral to obtain a bank loan. In any event, the decision to save (which is simply the decision not to spend) is prior to the decision over how to allocate savings. If there were no bond issue, the would-be bond purchaser, in all likelihood, would simply look for a different saving vehicle. Nor does simply leaving reserves in the system – as in the general case and special cases 1 and 2 – in any way encourage spending. So long as reserves exist, they never leave the banking system. Reserves are used for transactions between banks and for transactions between banks and the central bank. Reserves are never – and can never be – lent out to the non-bank public. (Strictly speaking, if the interest rate applying to bonds in special case 3 happened to be more attractive than the rate applying to reserves in the other cases, then the practice of matching fiscal deficits with bond sales would, if anything, be more expansionary than other practices, not less, as frequently supposed, because of the possibility that more consumption will be induced out of the interest income accruing to bondholders.)
Four cases have been considered. In the general case, higher levels of government do not impose any technically unnecessary constraints on Treasury and central bank operations. In the remaining special cases, various technically unnecessary constraints are imposed. These turn out to be much ado about nothing, merely altering the workaround that the Treasury and central bank must adopt to carry out the spending and taxing measures authorized by higher levels of government.
In the general case, the Treasury and central bank are unencumbered by technically unnecessary constraints. The spending authorized by higher levels of government is conducted simply by marking up reserve accounts and bank accounts. The effect is to add net financial assets, in the form of reserves. The extra net financial assets are a liability of government which in the general case is borne by the central bank.
In special case 1, the central bank is prohibited from incurring an overdraft. To get around this, a Treasury account is defined into existence as an asset of the Treasury and liability of the central bank. When spending occurs, the central bank marks up reserve accounts as always but also marks down the Treasury account to neutralize the impacts on its own balance sheet. The impact on net financial assets is the same as in the general case. All that changes is the composition of government liabilities. The net financial position of the Treasury, rather than the central bank, is now affected.
In special case 2, the Treasury is prohibited from incurring a negative balance on its account at the central bank. To get around this, the Treasury sells bonds to the central bank whenever government spending exceeds the existing balance on the Treasury account. The central bank pays for the bonds by crediting the Treasury account. As in special case 1, the Treasury rather than the central bank carries a negative net financial position, but now the Treasury’s liabilities are in the form of bonds rather than a negative account at the central bank.
In special case 3, the Treasury is forbidden from selling bonds directly to the central bank. The workaround, in this case, is for the central bank to acquire bonds indirectly through secondary market operations.
No matter how higher levels of government circumscribe the operations of the Treasury and central bank, it always remains true that the spending of a currency-issuing government precedes both tax payments and bond sales to non-government. Since taxes and non-government bond purchases only settle in government money, government money must be created before non-government can pay taxes or acquire government bonds. This is true even when the Treasury is required to sell bonds to non-government. The reserves needed for settlement of Treasury auctions, and any bonds required by the central bank as collateral when advancing reserves to non-government to ensure smooth functioning of the monetary system throughout the auction process, only exist because of past net spending by the currency-issuing government.