In “What are Taxes For?”, Paul Krugman chooses to mention Modern Monetary Theory (MMT) again, though only in passing on this occasion. Krugman’s post follows recent consideration of MMT by Brad DeLong, Nick Rowe and Steve Randy Waldman. Much of the disagreement expressed by these economists appears to stem from different conceptualizations of monetary operations, but implicit assumptions over long-run employment determination and the nature of interest also seem to contribute to the differences in perspective. In this post, I want to explore these points a little, using Krugman’s brief remarks as a springboard for the discussion.
Operational aspects of government net spending
Krugman’s remarks seem to reflect an understanding of monetary operations that, from an MMT perspective, is inaccurate. Bill Mitchell has already provided a response at billy blog, as has Tom Hickey here. I haven’t had a chance to check everywhere yet, so apologize if I’ve missed others.
So taxes are, first and foremost, about paying for what the government buys (duh). It’s true that they can also affect aggregate demand, and that may be something you want to do. But that really is a secondary issue.
From an MMT perspective, it is operationally impossible for taxes to “pay for” (in the sense of providing initial finance for) the spending of a sovereign currency issuer. The act of government spending results in a net increase in financial assets held by non-government. The government subtracts back out some of the net financial assets it has already created through taxation.
If this point is all that is preventing Krugman from seeing the MMT viewpoint, simply reading Warren Mosler and Mathew Forstater’s A General Analytical Framework for the Analysis of Currencies and Other Commodities would presumably clarify the matter.
For a currency issuer, taxes are not first and foremost (in fact, not at all) about providing initial finance for government spending but are about ensuring a demand for the currency. If there were no tax obligation, there would be no imperative for non-government to get hold of the currency and little reason for non-government to part with real goods or services in exchange for seemingly worthless pieces of paper or credits in bank accounts. In short, without the tax obligation, there would be no compelling reason for non-government to transact with government. Once a tax-driven demand for the currency is established, taxes can then function as a moderator of demand (but never a source of initial finance).
A closer look at spending operations
Consider what happens, in the US context, when the Treasury auctions debt supposedly to “finance” its spending. The auction can only proceed if there are sufficient reserves in the system or non-government owns enough government bonds with which to obtain the necessary reserves. If reserves are inadequate, the Fed needs to do a ‘reserve add’ as part of a repurchase agreement with primary dealers. Otherwise, there will be no way for non-government to participate in the Treasury auction.
In exchange for newly issued reserves, primary dealers are required to supply collateral in the form of government bonds. The bonds they supply as collateral only exist because of a previous round of government spending. When the bonds are purchased at the Treasury auction, reserve accounts are debited and the Treasury’s account at the Fed is credited.
It is worth pausing, here, to reflect on the nature of reserves on the one hand, and the balance in the Treasury account at the Fed on the other.
Reserves are a liability of the consolidated government sector and a form of ‘government money’ (or high-powered money, defined as currency plus reserves). Since reserves are a form of government money, the debiting of reserve accounts as a result of the Treasury auction constitutes a destruction of government money.
In contrast, the funds credited to the Treasury’s account at the Fed as a result of the Treasury auction are not a liability of the consolidated government sector. They are merely an intragovernmental liability of the Fed to the Treasury. Since the balance in the Treasury’s account at the Fed is not a liability of government to non-government, it cannot be considered government money.
This means, as a matter of logic, that the bond sale at the Treasury auction does not provide government with the money it spends. To the contrary, the Treasury auction results in a destruction of government money, not a financing of government. (For an academic treatment of this point, see Stephanie Kelton (Bell), ‘Do Taxes and Bonds Finance Government Spending?’, Journal of Economic Issues, 2000, vol. 34, pp. 603-20.)
Moreover, from inception, such a destruction of government money is only possible because of the prior reserve add conducted by the Fed. Reserves must be created before they can be destroyed. Nothing can be destroyed before it is created. The reserve add is a creation, ex nihilo, of government money.
When, in the current round of government spending, the government actually spends, it spends by creating government money (specifically, reserves). To conduct the actual spending that has been authorized by Congress, the Treasury instructs the Fed to draw down its account at the Fed. This instruction, in itself, has no net effect on government liabilities. It is simply a reduction in the Fed’s liability to the Treasury.
It is only the Treasury’s subsequent instruction to credit the reserve accounts of the banks of spending recipients that results in the issuance of new government liabilities. Specifically, by crediting reserve accounts, the Fed creates government money, ex nihilo as always, and instructs banks to credit the accounts of spending recipients.
In short, government spending is always self-financing. It occurs through the creation, ex nihilo, of reserves. Taxing (and bond sales) are simply the reverse operation – a reserve drain, which constitutes a destruction of government money. Reserves can only be drained (destroyed) if they have previously been added (created).
On the basis of these considerations, it is clear that the practice of maintaining a Treasury account at the Fed is for the purposes of internal accounting and record keeping. It helps to keep track of government spending, taxing and bond sales, but has nothing to do with providing initial finance for government spending.
Government spending is always conducted through a creation of government money, irrespective of whether the newly created funds are converted into bonds or left as reserves. (A more in-depth treatment of government spending operations is provided in Exercising Currency Sovereignty Under Self-Imposed Constraints).
Appropriate level of government net spending
Upon correctly noting that a currency user such as a state government in a federal system must finance its spending, Krugman writes:
Does the same thing hold true for the federal government? Well, the feds have the Fed, which can print money. But there are constraints on that, too — they’re not as sharp as the constraints on governments that can’t print money, but too much reliance on the printing press leads to unacceptable inflation.
The reference to “printing money” is misleading in that it seems to imply that some government spending creates money whereas other government spending does not. In reality, all government spending creates government money in the form of reserves. There is no other way that government spending can be conducted. Whether the newly created government money is left in the form of reserves or converted into bonds has no macroeconomic implications other than indirect effects associated with the different interest rates paid on reserves and bonds. Functionally, there is no difference between leaving the newly created funds in the form of reserves and paying the policy rate on reserves or converting the newly created funds into short-term bonds and paying the same policy rate on bonds.
Leaving this point to one side, the operative words in the quoted passage are “too much reliance”. Nobody is arguing for excessive government net spending. Rather, MMT economists argue that the fiscal deficit should be consistent with full employment and price stability. (See Krugman and Galbraith on Deficits.)
When the government spends on real goods and services, the act of spending is at the same time a transferral of real resources to the public sector. This transfer of real resources is independent of tax measures (provided there is already a tax-driven demand for the currency) and occurs even if government tax revenues remain constant. The question then becomes, will this transfer of real resources be inflationary? After all, some resources that might have been purchased by the private sector have been purchased by the government.
The answer, of course, as both MMT economists and Krugman would agree, is that the spending will only be inflationary if the additional demand exceeds the capacity of the economy to respond at current prices. At full employment and full capacity, the transfer of real resources to the public sector would leave less resources available for purchase by non-government. If there was no extra taxing alongside the additional expenditure to subtract private spending power, inflation would occur.
But if, instead, as is currently the case, there is unemployment and excess capacity, it is more likely that the additional government expenditure will simply induce greater output at current prices. Under competitive conditions, firms with excess capacity will tend to respond in this way. The transfer of some resources to the public sector, in these circumstances, will not prevent non-government from obtaining what it demands.
Now, as noted, Krugman would agree with this. He has consistently argued for stimulus to make up for weak private demand, and also recognizes the fallacy of composition in expecting wage reductions to generate employment, although in neoclassical fashion he links this to a so-called liquidity trap in which interest rates have reached a lower bound.
Controversy over long-run employment determination
So what is the source of disagreement when it comes to the long-run effects of fiscal deficits? On one level, it may just be a matter of Krugman having a different view of the monetary operations. But there are also other fundamental differences between MMT and neoclassical theory that might be holding back agreement.
Most notably, as has recently been emphasized by John T. Harvey, the presence or otherwise of a long-run automatic tendency to full employment is a central point of contention. It remains true to this day that neoclassical economists implicitly suppose, in the absence of ‘rigidities’, a long-run tendency for the economy to gravitate to the level of real output associated with full employment (currently interpreted as the level of output associated with the NAIRU). In this view, the price mechanism will eventually ensure full employment irrespective of non-government net saving decisions. In particular, it is implied that once the economy escapes the liquidity trap, private investment will respond to price signals and real interest rates in such a way as to ensure this gravitation.
Now, if the economy were really predestined to return to full employment in the long run, then any fiscal deficit that was sufficient to sustain full employment would merely be delivering the same employment outcome that would eventually be achieved without fiscal stimulus. Since attaining full employment via a fiscal deficit would result in a higher share of public demand in total demand, there would be less room for private investment and a need for higher taxes to avoid inflation.
This supposed automatic tendency to full employment is essentially Say’s Law for the long run, in which neoclassical causation goes from (planned) saving to investment, leakages to injections and supply to demand, even though neoclassicals do acknowledge Keynesian causation (the reverse) in the short run. In this view, full employment is ultimately guaranteed irrespective of fiscal policy, which will only influence prices..
But if, in reality, there is no automatic tendency to full employment, then fiscal policy matters for output and employment over any time frame, and it becomes clear that there is unemployment now and quite possibly unemployment later.
In contrast to neoclassical theory, MMT (in common with Post Keynesian theory and the surplus approach) rejects the notion of an automatic tendency to full employment, including in the long run. In this view, it is a fallacy of composition to think that the price mechanism can guarantee a particular level of output and employment, even with fully flexible real wages and interest rates. This was the position of Keynes and Kalecki. Their arguments did not depend on wage or price stickiness. Further, the capital debates made clear that real interest rates cannot be relied upon to induce private investment consistent with full employment, quite apart from any liquidity trap.
Instead, in the MMT view, it is non-government net saving behavior that determines the appropriate level of the fiscal deficit. This behavior, in turn, primarily depends on political, institutional and distributive factors, not interest rates. From a Keynesian perspective, interest rates can influence the form in which savings are held but not the level of saving in any direct or predictable way. Institutional factors include properties of the welfare system – e.g. the presence or absence of universal pensions, unemployment benefits, publicly provided health care services, etc. – which can all be expected to influence the desire to net save. Different propensities to spend among various income groups mean that distribution will also clearly impact on net saving desires. There are no grounds for expecting interest-rate movements to induce a net saving desire consistent with a particular deficit/GDP or debt/GDP ratio.
Over the long run, different societies will exhibit different saving behaviors, and the same society can exhibit different behavior under different institutional settings and political circumstances. On the demand side, what matters for employment and price stability is the size of the fiscal deficit in relation to non-government net saving behavior. If, as MMT economists suppose, there is no guarantee of full employment in the absence of appropriate fiscal policy, it makes no sense to claim discretionary fiscal actions today inevitably require reversal tomorrow. The appropriate level of taxes will vary over time with movements in private-sector demand and net saving behavior.
Controversy over interest
Related to the neoclassical notion of an automatic tendency to full employment is a particular conception of interest rates. For neoclassicals, the interplay of real factors (productivity and thrift) is conceived to determine the real rate of interest in the long run. Excess saving, by bringing about lower real interest rates, will induce stronger private investment. MMT economists, in contrast, point out that there is no mechanism in a modern monetary system akin to a ‘loanable funds market’ determining the real rate of interest on the basis of productivity (marginal product of ‘capital’, well known to be an invalid construct) and thrift. Rather, interest rates are a distributive variable and ultimately subject to policy control (see Scott Fullwiler’s Interest Rates and Fiscal Sustainability, or for an informal discussion Interest, Money and Crisis.)
Modern Monetary Theorists recognize that the short-term nominal interest rate is set exogenously as a matter of policy. Irrespective of the chosen interest-rate target, demand-pull inflation can result only if total demand (both public and private) is excessive relative to the capacity of the real economy to respond at current prices. Since, in this view, inflation is controllable through the fiscal stance, and the short-term interest rate is set exogenously, the real short-term rate of interest is ultimately controllable by policymakers, not markets or real factors.
Rather than having a natural determination, in the MMT view the choice of interest rate is above all a distributive (political) decision. It concerns the degree to which income is to be transferred from producers (real economy) to non-producers (savers).