Building bank reserves is not inflationary

Today I am working in Dubbo, which is in the western region of NSW and getting into the remote parts of the state. There is a great beauty to enjoy in remote Australia which often passes people by. My field trip is in relation to continuing work I am doing with indigenous communities in this region. I will report on this work in due course. But today’s blog continues the theme I developed yesterday on bank reserves. In yesterday’s blog – Building bank reserves will not expand credit – I examined the dynamics of bank reserves but left a few issues on hold because I ran out of time. One issue is the possible impact of expanding bank reserves on inflation. This is in part central to the mainstream hysteria at present about the likely legacies of the monetary policy response to the crisis. The conclusion is that everyone can relax – the only problem with the monetary policy response is that it will be ineffective and more fiscal policy effort is required.

This blog is intended to conclude the two-part series I started yesterday on bank reserves. It will also serve to clarify – by expressing in a different way – some of the essential principles of modern monetary theory (MMT) which are developed in Deficit spending 101 – Part 1 | Deficit spending 101 – Part 2 | and Deficit spending 101 – Part 3 – plus other blogs. Go to Debriefing 101 for a collection of blogs covering the first-principles of MMT.

In yesterday’s blog – Building bank reserves will not expand credit – I used a recent working paper from the Bank of International Settlements – Unconventional monetary policies: an appraisal – to explain some of the operational aspects of the monetary system which are at the core of MMT and which mainstream macroeconomics fails to depict in any coherent manner.

The specific context was Paul Krugman’s on-going policy suggestion that quantitative easing is required to stimulate lending. I showed that QE is largely incapable of stimulating lending and not surprisingly because lending is not reserve constrained. The conclusion was that Krugman clearly doesn’t understand the banking operations that link monetary policy to bank reserves. Developing this understanding is a core differentiating feature of MMT.

The use of the BIS research paper was deliberate because it demonstates that the banking professionals also understand key components of MMT even if they use language conventions that appear to place them in the mainstream discourse. To further demonstrate this I will continue to examine this BIS working paper in the context of the impact of bank reserves on interest rate policy and the potential for inflation.

In normal times (that is, not during this significant economic crisis) monetary policy is “defined exclusively in terms of a short term interest rate” where the central bank signals that it will set a “policy rate” – this is the central bank’s estimate of the rate that will achieve its policy goals. However, the central bank then has to engage in liquidity management operations:

… to help make that interest rate effective: they ensure that a market “reference rate”, typically an overnight rate, tracks the desired interest rate level closely. As such, liquidity management operations play a purely technical and supportive role.

To understand MMT it is important to link the idea of the liquidity management operations in the monetary policy domain to the impacts of government spending and taxation arising in the fiscal policy domain. If you read any mainstream macroeconomics textbook the linking of these two domains is not made in any coherent way. In fact, because the two domains are mis-specified in that paradigm, false conclusions are drawn (for example, they claim that fiscal deficits cause interest rates to rise).

When the government spends it credits bank accounts (or issues cheques which end up in the bank settlement system) and after all the transactions are completed the result is that bank reserves expand. Conversely, when the government collects taxes it debits bank accounts (or accepts cash/cheques) and this results in a $-for-$ decrease in bank reserves.

These are vertical transactions between the government and non-government sector and thus create or destroy bank reserves. Logically, when the government runs a fiscal deficit (spending is in excess of tax receipts), the impact on bank reserves is a net positive – reserves expand in net terms. Conversely, when the government runs a fiscal surplus (spending is less than taxation receipts), the impact on bank reserves is a net negative – reserves contract in net terms.

It is essential to understand these impacts because they allow you to then understand the way in which the liquidity management operations conducted by the central bank, though separate from fiscal policy, are intrinsically linked to these impacts.

The link is created not because the fiscal position adopted by the government “has to be financed”. That is the common error that mainstream macroeconomics textbooks make. A sovereign government that issues its own currency does not need to finance its spending. That is so obvious that it continually surprises me that commentators persist in asserting the opposite.

All the financial machinations (debt-issuance, taxation) that might look on the surface to be “financing operations” are in fact nothing of the sort. The following discussion will clarify the role that debt-issuance serves in the monetary system. The rhetoric of the government might lead one to conclude otherwise (that debt-issuance is a financing operation) but the hard reality of the daily liquidity management operations (once understood) reveal the truth – the separation of political stance from reality.

The BIS say that:

The fulcrum of the implementation of interest rate policy is the market for bank reserves. This is a peculiar market. By virtue of its monopoly over this asset, the central bank can set the quantity and the terms on which it is supplied at the margin. As such, the central bank is able to set the opportunity cost (“price”) of reserves, the overnight rate, to any particular level, simply because it could stand ready, if it so wished, to buy and sell unlimited amounts at the chosen price. This is the source of the credibility of the signal.

Crucially, the interest rate can be set quite independently of the amount of bank reserves in the system. The same amount of bank reserves can coexist with very different levels of interest rates; conversely, the same interest rate can coexist with different amounts of reserves. What is critical is how reserves are remunerated relative to the policy rate. We refer to this as the “decoupling principle”. It is a principle that has far-reaching implications for the rest of the analysis.

The signal they refer to is the announced policy or target rate. You will note that the central bank (as an integral component of the consolidated government sector with treasury) also does not have a revenue-constraint – it can buy and sell unlimited reserves at whatever price it chooses to set.

The decoupling principle was examined in yesterday’s blog – Building bank reserves will not expand credit. The following discussion clarifies the statement that “what is critical is how reserves are remunerated relative to the policy rate”.

The BIS use the following figure to motivate this discussion. It appears on Page 4 of their paper . It is a good way to understand how different reserve remuneration schemes operate in relation to monetary policy and the profit-seeking aims of the commercial banks.

There various parameters in the figure are defined as follows:

  • rp = the central bank monetary policy rate – which is the rate that the central bank believes will advance its policy goals.
  • ro = the overnight rate which is established by supply and demand of reserves in the interbank market, which allows the commercial banks to dispose of or acquire overnight reserves. In the case of a system-wide excess the transactions in the interbank market are unable to clear the excess. In this case, only a draining operation (selling government bonds) by the central bank will eliminate the excess. Similarly, when the is a system-wide shortage of reserves, only a central bank injection can make up the shortfall, irrespective of the positions held by the individual commercial banks.
  • rE = the rate that the central bank pays to the commercial bank for excess reserves left in the system overnight.
  • Rmin = the minimum amount of reserve balances required for settlement purposes. Banks keep this amount so that all daily transactions that require transfers of reserves between banks can be settled in an orderly manner.
  • R* = the amount of reserves in equilibrium. This is achieved when the banks have the minimum reserves required for settlement purposes and they are earning a competitive rate on any further reserves. Equilibrium implies that there is no further transactions in the interbank market.


How do we interpret this figure?

There are two schemes shown (Scheme 1 and Scheme 2). Scheme 1 refers to the normal situation where the central bank sets the support rate (rE) that it pays on overnight reserves below the policy rate (rp). In Australia, the support rate is typically 25 basis points below the target rate. In New Zealand, the RBNZ has set the rE = rE. Until recently, the Bank of Japan and the US Federal Reserve set rE = 0.

In the monetary system described by Figure 1 there are no reserve requirements with averaging provisions imposed. So according the BIS authors the:

… amount of reserves that banks need to hold overnight, Rmin, is determined entirely by banks’ settlement needs, including any precautionary element. This demand depends on the wholesale settlement arrangements in place, and is in effect independent of the interest rate.

Accordingly, the demand for excess reserves by banks (DD) is vertical (that is, invariant to the short-term (overnight) interest rate) because banks will desire to hold minimum reserves (Rmin) to ensure they can meet the demands of settlement each day. That level is not a function of the interest rate. The central bank has to ensure this minimum amount is always available irrespective of the remuneration approach it adopts with respect to excess reserves. If the central bank fails to provide this level of reserves then “significant volatility of the overnight interest rate” would result.

The BIS elaborate:

Any excess would drive it to the floor set by the remuneration on excess reserves (zero or the rate on any standing deposit facility), as banks seek to get rid of unwanted balances by lending in the overnight interbank market. Any shortfall would lead to potential settlement difficulties, driving the rate to unacceptably high levels or to the ceiling set by end-of-day lending facilities. Once the demand for bank reserves has been met, the central bank can set the overnight rate at whatever level it wishes by signalling the level of the interest rate it would like to see.

So how does the remuneration scheme matter? Under Scheme 1, profit-seeking banks will attempt to economise on excess reserves above the minimum required for settlement purposes because the “support rate” is below the overnight rate.

If there are excess reserves (greater than Rmin), the “existence of an:

… opportunity cost of reserve holdings (ro – rE) implies that when excess reserves exceed Rmin, banks will attempt to lend out this surplus. In so doing, they will drive the overnight rate down to rE. At this point the opportunity cost is eliminated.

So whenever there are reserves beyond the minimum required for settlement purposes, the banks will seek to lend these reserves out in the interbank market to get a return above rE (the support rate paid by the central bank). If rE = 0 then this competition for loans in the interbank market will drive the overnight rate to zero if left unchecked by the central bank.

The clear implication is that the central bank then loses control of its target short-term interest rate (rp > ro). In these instances the central bank has to intervene to choke of the interbank activity and it does this by offering the commercial banks an alternative near-equivalent asset to the bank reserves which earns a return commensurate with rp. This asset is a government bond. So debt-issuance is properly understood to be a means by which the central bank maintains control of its policy target interest rate.

Under Scheme 2, the central bank pays the commercial banks a return on excess reserves (above the minimum required for settlement purposes) equivalent to the policy rate. In this case:

…. there is no opportunity cost of holding excess reserves and banks will be indifferent about the amount of reserves they hold as long as the minimum for settlement purposes is satisfied.

So in this case there is no need to sell public debt to the commercial banks. By setting the support rate equal to the policy rate the central bank is offering the equivalent to the a government bond to the commercial banks. The payment short-circuits any need the commercial banks have to eliminate their exceess reserve holdings and they will happily sit with large stocks of reserves instead of seeking alternative interest-bearing assets (such as public debt).

The other important point to connect relates to the impact of fiscal policy on the reserves. We know that fiscal deficits create excess reserves. In this case, they create a dynamic in the liquidity (or cash) system whereby the interest rate is bid down via interbank market competition (as above). This forces the central bank to issue debt to the commercial banks (to drain the reserves) or pay a support rate equivalent to the policy rate if it wants to maintain that particular monetary policy stance.

Under these circumstances, the debt-issuance stops the interest rate from falling to the support rate in the face of budget deficits. Budget deficits per se do not put upward pressure on interest rates. The monetary operations associated with the liqiuidity management operations stop the interest rates from falling but that is a different matter altogether.

Once you understand all that you will also grasp why the relevant chapters in mainstream macroeconomics textbooks and most of the so-called informed commentary on fiscal and monetary policy is erroneous in the extreme.

Bank reserves and inflation

Later in the paper, the BIS authors examine the question: “Is financing with bank reserves uniquely inflationary?” Recall, in yesterday’s blog – Building bank reserves will not expand credit – we noted that Paul Krugman was advocating quantitative easing again as a desirable policy at present because it would promote expectations in the private sector of future inflation. He constructed QE as an expansion of bank reserves.

The inflationary expectations would, in turn, lead “savers and investors” to conclude that real interest rates, in an environment of zero or very low nominal interest rate, would become negative. This would according to Krugman stimulate the demand for bank loans and help kick-start the economy.

So an essential part of the argument is the build-up of bank reserves is inflationary.

The BIS authors say:

The proposition that highlights the inflationary consequences of financing via bank reserves is closely related to the first. If bank reserves do not contribute to additional lending and are close substitutes for short-term government debt, it is hard to see what the origin of the additional inflationary effects could be. The impact on aggregate demand, and hence inflation, would be very similar regardless of how the central bank chooses to fund balance sheet policy. For example, it is not clear how inflationary pressures could be more pronounced in a banking system that keeps its liquid assets in the form of overnight deposits at the central bank compared to one that holds one-week central bank or treasury bills.

The same would apply to concerns about the “monetisation” of government debt, whereby the central bank purchases government bonds either in the primary or secondary market. Here the issue is whether the financing of government expenditures through the creation of bank reserves, quite apart from the boost to aggregate demand associated with expansionary fiscal policy, would lead to inflation or not.

First, you can see that while the BIS authors understand the operations of the banking system they are not operating within the MMT paradigm. The use of terminology such as “financing medium” tells you that. The use of this terminology, while conventional among bankers such as this, is highly misleading. The central bank does not “finance” government spending by creating bank reserves. Bank reserves are created by public spending which, as we discussed above, present the central bank with some choices depending on its monetary policy stance. The monetary operations conducted by the central bank are not “financing” operations but rather they are correctly understood to be liquidity management operations.

Further, the idea that the central bank can monetise public spending and maintain a positive interest rate target (and pay a support rate below the policy rate) is impossible. If the central bank tried doing this, the competition in the interbank market as banks tried to shed their excess reserves would lead to the central bank losing control of its policy rate. It would have to sell public debt to drain the excess reserves or increase the support rate on excess reserves to the policy rate. You will not get an understanding of this sort of reasoning in any mainstream macroeconomics textbook or research article.

Second, in Building bank reserves will not expand credit – we showed that an expansion of bank reserves will not increase bank lending. The idea that it would is based on the flawed understanding that banks need reserves before they will lend. Categorically, they do not. Mainstream macroeconomics textbooks are completely wrong in that regard.

In this context, it is essential to understand that the analysis of inflation is related to the state of aggregate demand relative to productive capacity. Credit growth manifests as increased spending. In itself that is not inflationary. Nominal spending growth will stimulate real responses from firms – increased output and employment – if they have available productive capacity. Firms will be reluctant to respond to increased demand for their goods and services by increasing prices because it is expensive to do so (catalogues have to be revised etc) and they want to retain market share and fear that their competitors would not follow suit.

So generalised inflation (as opposed to price bubbles in specific asset classes) is unlikely to become an issue while there is available productive capacity. Even at times of high demand, firms typically have some spare capacity so that they can meet demand spikes. It is only when the economy has been running at high pressure for a substantial period of time that inflationary pressures become evident and government policy to restrain demand are required (including government spending cutbacks, tax rises etc).

Further, spending growth can push the expansion of productive capacity ahead of the nominal demand growth. Investment by firms in productive capacity is an example as is government spending on productive infrastructure (including human capital development). So not all spending closes the gap between nominal spending growth and available productive capacity.

The BIS authors then examine the inflation issue in the context of the remuneration rate offered to commercial banks by the central bank.

In the case where excess reserves are remunerated at a rate that is below the policy rate, their injection would push overnight rates down to the floor established by the remuneration rate on any deposit facility, possibly zero (scheme 1). This is tantamount to an easing of interest rate policy. As a result, any ensuing inflationary pressure can be largely attributed to the usual expansion of aggregate demand that accompanies such a move.

So it is not the monetary policy setting but rather the final spending in excess of productive capacity that causes inflation. However, this would depend on how sensitive aggregate demand is to interest rate movements. In maintstream economics, spending is assumed to be relatively sensitive to changes in interest rates which is why they advocate inflation targetting to discipline the inflation process.

In MMT, there is less confidence that changes in policy interest rates within normal ranges alter spending dramatically. One has to realise that there are two sides to consider. The cost of funds side – whereby an increase in the interest rate will possibly reduce the demand for funds (for a given expectation of revenue flow arising from the investment) – and the income side – whereby an increase in the interest rate provides a boost to fixed income recipients and may, in turn, boost spending. This would impact back on investment because investors would not only face higher costs of borrowing but would likely feel more confident about future income flows.

So these distributional complexities make it difficult to conclude unambiguously that interest rate changes are an effective way to manipulate aggregate demand. They also work more slowly and indirectly anyway and make it very hard to disentangle from other impacts on demand.

The point is that the sale of public debt in Scheme 1 will drain excess reserves and prevent the overnight rate falling below the policy rate. In that sense, there would be no inflationary impacts (irrespective of how sensitive aggregate demand is to changing rates).

In terms of Scheme 2 (support rate = policy rate), the BIS say that:

In the case where excess reserves are remunerated at the policy rate or interest rates are already at the zero lower bound, so that the opportunity cost of excess reserves is zero, their expansion would not affect overnight rates (scheme 2). To the extent that any additional impact on inflation existed, it would result mainly from the effect on aggregate demand of a flatter yield curve that the quasi-debt management operation may induce. For example, if the central bank were to inject reserves through the acquisition of long-term government bonds, the net impact on yields and inflation would not be dissimilar to the rebalancing of government financing from long to very short maturities. In fact, such an “operation twist” can be achieved by the fiscal authorities themselves.

So while this might sound complex, it is saying nothing more than changes to long-term rates may stimulate investment which, in turn, might drive nominal spending growth faster than the real capacity of the economy to absorb it.

In other words, any inflationary effect that might arise comes from the spending side and is not intrinsic to the way the central bank conducts its liquidity management operations.

The BIS authors also consider so-called monetisation;

More generally, inflationary concerns associated with monetisation should be largely attributed to the impact on aggregate demand via a fiscal policy that is accommodated by the monetary authorities, who refrain from raising rates. That is, it is not so much the financing of government spending per se (be it in the form of bank reserves or short-term sovereign paper) that is inflationary, but its accommodation at inappropriately low interest rates for a sustained period of time. Critically, these two aspects are generally not distinguished in policy debates because the prevailing paradigm has failed to distinguish interest rate from balance sheet policy. Given the pervasive assumption of a well behaved demand for bank reserves, one is seen as the dual of the other: more reserves imply lower interest rates. But, as we have stressed all along, this is not the case. And the decoupling of interest rate from balance sheet policy during the current crisis has simply confirmed this again.

See my comments above on monetisation. This could only occur if the policy rate was already at zero or the support rate was equal to the policy rate. In that case, a build up of reserves would accompany the net fiscal injection but as we have seen above – this has nothing intrinsically to do with what happens to spending growth in relation to the real capacity of the economy.

If aggregate demand is relatively insensitive to interest rate settings then this point has not relevance either. It is clear, however, that if interest rate changes do impact on spending such that low interest rates are more expansionary than higher interest rates, then a fiscal expansion and a zero interest rate policy will be stimulatory. The issue is not that this will be intrinsically inflationary as is asserted by the mainstream.

It just means that the extent of the fiscal injection that is required to achieve full capacity utilisation is reduced. The government always has the capacity to balance aggregate spending to match the capacity of the economy to absorb it.


I wrote this blog and yesterday’s blog to provide some further discussion of some of the main tenets of MMT. By using a BIS working paper as motivation I was able to use “their language” to demonstrate these essential underpinnings of MMT. Using different language sometimes helps overcome mindsets and broaden one’s understanding of a fairly complicated subject matter.

But what should be very obvious to anyone is that the mainstream macroeconomics textbook depiction of the monetary system and how fiscal policy uses it and impacts on it are totally erroneous. Students who only engage with these textbooks and lecturers that use them will leave their studies with a false impression of how the system functions.

Some of them get jobs in policy making areas of government and take this erroneous view of the system with them. It is no wonder that we get poor policy responses and ridiculous speeches written for the political leaders.

Others who graduate become journalists and you can then see the problem that arises from that.


I received an E-mail overnight from my mate Marshall Auerback who was recently visiting Newcastle and who writes for the New Deal blog published by the Roosevelt Institute. He thought the BIS research paper was an important contribution because it is written by non-MMT economists who clearly are operating within the mainstream paradigm (note their use of the financing nomenclature) but who understand the monetary operations that govern real world financial systems.

Marshall noted the quote that I provided from the BIS report:

In order for the funds (bank held reserves) to be inflationary they must be borrowed and spent.

His response was as follows and I thought worthy of quoting in its entirety. It shows I do get interesting E-mails (-:

Marshall writes:

That is a point I have tried to make clear with the Austrian brigade and the deficit doves, amongst others. There is no magical mechanism that reprices goods and services with any given change in the money stock. Money does not chase goods. That is mystification – money is not an agent that can take action on its own. Nor is there any such thing as “the market” – there are people behaving and acting within the strictures of market mechanims. People borrowing and spending money pay prices for goods and services (and assets, financial and tangible). If money (qua reserves) is created but not borrowed and spent there is simply no market mechanism that automatically and mystically changes all product prices. There may be a change in relative supplies, but there is no market where this automatically yields a change in relative prices except the product market themselves, where sellers make offering prices and buyers bid on goods and services.

The BIS has been one of the few official agencies willing to think and write and advocate clearly about financial balances and financial stability. My fellow Canadian William White’s legacy, carried on by Claudio Borio and others, is in no small part responsible for this. With this document you have surfaced, they now openly shatter some more mainstream macro illusions. Very big steps forward … but won’t stop Krugman from misfiring, and doesn’t mean they are yet willing to sign on to MMT … but they are getting much closer to the truth.

A good way to end and rush to the airport to catch my plane to … Dubbo. Hard place to get to from Newcastle – 3 hour train to Sydney airport then plane out west.

This entry was posted in Debriefing 101, Economics. Bookmark the permalink.

40 Responses to Building bank reserves is not inflationary

  1. Greg says:

    Great post Bill

    Now, just so I’m clear. I totally understand how building reserves, in and of itself, is not inflationary. The example is how the TARP funds went to reserves to “stimulate” lending and helped…………zip. You also pointed out that govt spending ends up increasing reserves. But government spending IS spending which can be inflationary…………..but its not BECAUSE it increases reserves, its simply because govt spending is demand which when excessive will be inflationary. Am I thinking straight here?

  2. Michael says:

    If I understood correctly, sale of public debt by the central bank drains excess reserves from the system (“The point is that the sale of public debt in Scheme 1 will drain excess reserves …”), but sale of public debt by the Treasury for deficit spending purposes increases excess reserves in the system (“When the government spends it credits bank accounts (or issues cheques which end up in the bank settlement system) and after all the transactions are completed the result is that bank reserves expand”). Is this right? Could you directly compare these and explain the differences and similarities more fully?

    Thanks. I like the direction you’re going, with more exposition of MMT and less diatribe against conventional theories and theorists.

  3. BruceMcF says:

    It IS an oddity that you can catch a train direct to Gunnedah / Moree / Narrabri, but Dubbo, which is a bit north of due east from Newcastle, requires you to catch the Sparkie to Sydney and fly … but according to The Google Machine, the rail line heading that way out of Muswellbrook hits Merriwa and just stops.

    If y’all ever get an inland rail expressway from Melbourne to Brisbane, which would includes the Dubbo junction, the bit of rough terrain that the Muswellbrook / Merriwa corridor already traverses would be the roughest bit between the Lower Hunter Valley and Dubbo … that is, the rail line gets over the rough bit and then stops, rather than connecting to the Dubbo junction which links to the main inland NSW junction at Parkes … and would be much less daunting than a replacement for the Murrurundi tunnel.

    I’d presume, if I was guessing, that’s because the easier terrain is through Dunedoo while the coal lines are in the rougher terrain around Ulan and Moolarben, so its easier for the coal companies to truck the coal to the railhead.

    Mind, while Australia has the natural resources, the human resources and much of the equipment required to do it, there is a thought to be a magical shortage of pieces of flexible plastic with QEII’s visage on them, so a project that would get as many trucks off the road as the inland rail expressway remains in the domain of “if y’all ever get”.

  4. RSJ says:

    By setting the support rate equal to the policy rate the central bank is offering the equivalent to the a government bond to the commercial banks. The payment short-circuits any need the commercial banks have to eliminate their exceess reserve holdings and they will happily sit with large stocks of reserves instead of seeking alternative interest-bearing assets (such as public debt).

    Hmmm — isn’t this pure evil? When the government sells bonds, it is selling them at auction, so no net financial assets are added to the private sector (the financial assets were added when the deficit spending occurred; the bond sales are just portfolio shifts).

    When the government pays interest on reserves, it is adding net money to the private sector — unlike bond sales. And this money goes to banks! This puts banks in a seignorage position, where they get to earn net money because the government deficit spends. Bond sales, however, do not put anyone into a seignorage position, because the NPV value of the income stream from the government to the private sector is offset by the cash flowing in the other direction when the bonds are sold.

    Long term, these seignorage profits go directly into bank capital. How will this succeed in limiting credit growth? If it does not succeed, then why pay interest?

  5. Ramanan says:


    Your suspicions are correct in my opinion. The central bank can in principle set the yield curve. The only policy which is consistent with public purpose is to set the overnight rate and the whole yield curve to zero.

  6. bill says:

    Dear RSJ

    Your point is definitely correct. So the alternative is to issue debt (and give the private sector a guaranteed flow of government income) or set the target (policy) rate to zero and let the competition between the banks to get rid of excess reserves keep it there.

    The latter option is – as you know – my individual preference. Then the longer maturity rates will just reflect risk.

    best wishes

  7. bill says:

    Dear Bruce

    Hope things are going well with you back in the US.

    Yes, we cannot have common sense infrastructure here because the government says we haven’t got enough money (those pieces of flexible plastic with QEII’s visage). As you point out … we have surplus labour (especially in the regional towns I have been to today), plenty of land and other resources, no shortage of tractors etc … but the feds haven’t enough money – and they might run out of it very soon.

    It is such a surreal sort of argument when all they would be doing is typing dollar amounts into bank spreadsheets (government spending!) – but everyone (but me and a few others) believe it … they actually think the feds cannot “afford” to build the infrastructure.

    The folly of human existence is exemplified in that belief. Meanwhile the indigenous communities out there live in below third world poverty without work or opportunity.

    best wishes

  8. bill says:

    Dear Greg

    You are thinking straight. Any spending can be inflationary if it drives nominal demand faster than the real capacity of the economy to produce output in response.

    best wishes

  9. RSJ

    I agree, like Bill, with your NPV statement regarding seigniorage, as it’s true by defiition. However, as we’ve discussed before, I disagree that this would necessarily be a transfer directly to bank capital. The rate paid on reserve balances is a money market rate, and the money market rate is just barely higher than the average cost of short-term bank liabilities (particularly given that previous Tsy holders would now be holding bank liabilities, and could be expected to move their funds to the banks offering the highest returns). I haven’t seen every bank business model, but I have yet to see one that makes enough profit to survive by holding money market assets. The addition to bank capital is net of costs of liabilities and net non-interest expenses, of course. The only entities that can really pull that off are the ones that can carry much higher leverage ratios than typical banks are allowed. And while I know that banks would still be making loans and thus would not be holding only reserves, the point is the profitability of the reserve portion of their assets would be much, much lower than you are suggesting.

    At any rate, if the Tsy/Fed are going to ignore Bill’s suggestion regarding a zero rate, then Warren’s proposal is to sell as many 3m Tsy’s at a fixed spread to the fed funds target as the public desires. That would drain most of the reserves anyway.


  10. pebird says:

    I understand that reserves are not inflationary (certainly in terms of the goods/services economy) – but what about the financial economy? Is it a coincidence that reserves exploded in the US and this was accompanied by a price bubble (short-term “inflation”) in stocks, commodities and real estate (at least stopping the sharp price decline)? If not, then what was the purpose of increasing reserves in the first place? If so, what is the process by which the reserves influence the price bubbles in certain asset classes?

  11. scepticus says:

    hi bill, a great couple of posts. good find that BIS document.

    what are your views on how excess reserves affect velocity and liquidity in financial markets. is this the main danger of excess reserves, rather than inflation?

  12. RSJ says:

    Hi Bill,

    So the alternative is to issue debt (and give the private sector a guaranteed flow of government income) or set the target (policy) rate to zero and let the competition between the banks to get rid of excess reserves keep it there.

    No, the government is selling the flow of income in a public auction. It is replacing one risk-free government liability (cash) with another (debt). No one is “getting” anything, unless you systematically believe that the result of the auction is to underbid for the security.

    I am not saying that we need to sell bonds. But there are fears that if banks funding costs are “too low”, then this will encourage credit-growth if the economy swings to an expansionary/inflationary environment. Certainly we are not in that environment now, and Japan has not been there for many years, but we might return to such an environment.

    Leaving aside whether these fears are founded or not, our understanding of inflation is poor, so if you need some fallback mechanism to limit credit growth in the case that these fears *are* founded, then the appropriate way to do this is to impose costs, not to pay interest. There are many ways that Government can impose costs on banks — think outside the box! Particularly as Government is the monopoly issuer of currency and can exercise strong regulatory control over banks.

    I am advocating for three principles:

    1. Net government fiscal spending should only be with the (non-financial) private sector. Not with banks or the investment community.
    2. To the degree that government engages in financial transactions vis-a-vis the investment community, it should only exchange one financial asset for another at market prices, to prevent an increase or decrease in value from occurring.
    3. To the degree that government engages in financial transactions with regulated banks, it should only *loan* money to banks at some interest rate. It should never pay interest to banks on money that they already hold. Regulation, taxes, and fees should be used in those cases when credit growth needs to be suppressed.

  13. RSJ says:

    Hi Scott,

    Yes, we’ve debated this before, but I was hoping to hear Bill’s position on this. The government should only *loan* money to banks at whatever (non-negative) rate it wants to charge. The government can always regulate banks in such a way as to be able to increase or decrease this rate whenever it wants.

    Paying money to banks makes me want to scream, and I am not alone.

    The rate paid on reserve balances is a money market rate, and the money market rate is just barely higher than the average cost of short-term bank liabilities (particularly given that previous Tsy holders would now be holding bank liabilities, and could be expected to move their funds to the banks offering the highest returns).

    Yes, it is a money market rate, but which market? The interbank market is not the same as the commercial paper market.

    Commercial MM rates are the open market rates, although generally commercial MM paper rates follow the CB target provided that the CB target is “reasonable”. Moreover they do not *stay* at the CB target rate, as both borrowers and lenders will adjust their capital structure to avoid paying too much or receiving too little, but such adjustments take time.

    If the real CB target rate was 10%, but the economy was growing at 2%, that would not mean that businesses would be able to pay 10%. They would eventually withdraw from the market and shift their capital structure so as to be less dependent on short term debt. This shift would bring rates down.

    If the CB target rate was 0% and inflation was 5%, that would not mean that investors would accept the low yield. Instead of lending to banks, investors would pool their surplus funds and purchase long term securities, issuing short term liabilities. There would be a divergence between the CB target and the commercial paper market, but all of that would also take time.

    So you need to distinguish between short-term liquidity spikes, and longer term yields, and also between bank costs and investor costs. Unless everyone has the same access to government-supplied funds, there will be a divergence between investor and bank costs.

    Historically, to my knowledge no one has ever (yet) tried to keep the CB target negative in real terms during a period of economic expansion. In the same way, we have rarely increased the CB target to be higher than expected nominal growth rates. When we have done this, there were divergences between the MM rate and CB rate, and greater divergences between the 3 month rate and the CB rate. Volcker’s hikes are the perfect example: the MM rates did not follow the CB target, but there was a 200+ basis point divergence between interbank rates and commercial paper rates, as well as a 400+ basis point divergence between the CB target and the 3 month yield. And even in those cases, those hikes were temporary as they could not be sustained over the long term (all the banks would have failed).

    I haven’t seen every bank business model, but I have yet to see one that makes enough profit to survive by holding money market assets.

    Of course, because *now* we sell bonds and do keep the CB target in line with all the other rates. But if we never sold bonds, then bank reserves would be 7 trillion dollars. In that scenario, interest paid on reserves would be significant if you were to try to “raise” the CB target to say, 4%, right?

    At any rate, if the Tsy/Fed are going to ignore Bill’s suggestion regarding a zero rate, then Warren’s proposal is to sell as many 3m Tsy’s at a fixed spread to the fed funds target as the public desires. That would drain most of the reserves anyway.

    You bet! It would be free money given to investors, as the 3 month yields are typically about 150 basis points lower than the CB target, and have gone as much as 500 basis points lower when the CB target was too high. That is a huge windfall offered to investors, and is a source of more screaming :) They would drain reserves dry and would force the CB to add more reserves at which point they would be drained again. This shifts the seignorage from banks to investors, but I am against all seignorage.

    Again, the government, if it is to buy or sell securities to investors, should only do so at market rates. That is the only way you can argue that no net assets are being added or removed from the economy, at least according to market expectations (which can fail, but equally in both directions).

    Let bank costs be set administratively in exchange for heavy regulation that ensures banks lend only the amounts we want for the public purpose we want. Investors should not get free returns from the government as a result of some ham-fisted attempt to limit credit growth.

  14. “But if we never sold bonds, then bank reserves would be 7 trillion dollars. In that scenario, interest paid on reserves would be significant if you were to try to “raise” the CB target to say, 4%, right?”


  15. Ramanan says:


    Random thoughts: You seem to have trouble with all the goodies going to the bank. If say the central bank is targeting an overnight rate of say 400bps and no bond issuance, the interest paid has to be 400bps necessarily. The banks do get interest on the reserves but they will be paying interest on their deposit liabilities as well. If they are paying less, then the central bank can issue 1-day T-bills which allows for a broader participation. When a household moves, say $1,000 of deposits from banks to these T-bills, deposits in the banking system go down by $1,000 and so do the bank reserves.

  16. RSJ says:


    “You seem to have trouble with all the goodies going to the bank. ”

    I am troubled by any net spending going towards banks or the financial markets generally. Government deficit spending should be for public goods, and spending in the financial markets should not result in any net value being added. If anything, money should flow from the financial markets to government via taxes on capital gains and fees imposed on bank lending. You can lower these fees when you want to promote credit growth, or raise them when you want to restrict such growth.

    If say the central bank is targeting an overnight rate of say 400bps and no bond issuance, the interest paid has to be 400bps necessarily.

    No, what the government is really trying to do here is to promote or restrict credit growth — that is the only way that bank lending “leaks” into areas such as inflation. If the government wants to restrict credit growth, it can tighten lending standards, it can change capital requirements, it can impose a fee of 400 basis points on all loans issued, etc. There are countless ways to accomplish your goal here, none of which include subsidizing either banks or investors by either paying interest on reserves or selling bonds at below-market prices.

    In general, you constrain credit growth and limit inflation by draining money from the private sector, not by adding money to the private sector. If you want to promote credit growth, you can unwind the above operations. In either case, no net assets should be transferred either to banks or to the private sector, unless this is part of a bailout in which risk capital is first zeroed out.


    You are right, the proceeds could be paid out as dividends, but why should they? The yield will be set by the non-bank cost of equity overall.

    Historically when the government provided below-market funding to banks, the proceeds were used by the banks to expand their asset base via off-balance sheet entities and/or paid out as bonuses. If you prevent SIVs and the like, then I guess it’s a toss up as to whether they are paid out as dividends, go to bank capital, or are paid out as bonuses. In either case, it is not only unfair but completely unnecessary to what you are trying to accomplish vis-a-vis inflation.

  17. Hi RSJ,

    1. Question: What are you assuming banks have on the liability side when they have (in our example) $ trillions in interest-bearing reserve balances? My point wasn’t about dividends, but rather that I don’t see how the bank is necessarily more profitable by interest on reserve balances. Asset turnover has definitely fallen compared to the cb draining all excess balances, as would profit margin unless somehow the spread on rbs over matching liabilities can be as high as it was for the bank overall when excess balances were drained. Sure, if banks can use SIVs and raise leverage via off-balance sheet entities, then I can see it. I’m curious how you are putting this together.

    2. Agree with you regarding perverse effect of interest pmt on agg demand (in your reply to Ramanan). This was the point I was trying to make regarding Japan several weeks ago, but I shouldn’t have assumed such a high rate as that took the discussion off to Minskyan instability instead of this point. But the effect on AD remains largely the same whether bonds are sold or rbs are interest bearing . . . the difference is maturity, as Tsy’s are time deposits at the cb, whereas rbs are overnight accounts. Interest on the national debt in the US has followed longer run averages in the fed funds vs. nominal GDP growth fairly consistently over time, even though the Tbill can deviate from the FF rate, particularly during times of stress (the average deviation for 1954-1979 is 38bp; for 1979-2000 it’s 72bp; and 2001-2009 it’s 25bp . . . not the 150bp you suggested).

    3. GREAT comments on Nick’s site.


  18. ParadigmShift says:

    Interesting discussion, though I admit to feeling out of my depth! I’d feel much happier if I understood the accounting mechanics involved in the payment of interest on reserves. I’ve read that (using mainstream expressions, please forgive me) CB remittances to the treasury are reduced by the amount of interest paid on reserves, which suggests to me that the net result on the central bank balance sheet is that the treasury balance at the CB is reduced by the same amount that reserve balances are increased – hence, the CB balance sheet does not immediately expand as a result of interest payment on reserves, and there is simply a liability swap (debit treasury account, credit reserve balances). Is this correct?

    Presumably, at some point the central bank would have to expand it’s balance sheet in order to enable reserve balances to grow in line with the interest paid on reserves, and so some government debt would be required, if only to prop up the CB balance sheet.

  19. Ramanan says:


    Will try to put in a slightly different language. This does not answer your question but is more of a discussion. The central bank can be considered – as far as accounting is concerned – like a corporate. It has financial statements. An important thing is that the profits are given to the Treasury. (In the US, some banks get dividends, not sure). I think this is very important to remember. As far as interest payments on reserves are concerned, the accounting is just similar to how banks pay interest on our accounts. However, as far as the actual frequency of payment of the dividends (profits) are concerned – whether it happens daily or weekly or quarterly- not sure.

    There is a document somewhere on the Fed’s sites which gives details on how this is to be done. Has details/guidelines on how to do it for other Federal Reserve banks. Has details on how to account for furniture as well. Cant find it! I would imagine it is accurate and does not make neoclassical mistakes. The thing about neoclassicals is that they are even legally incorrect :)

  20. ParadigmShift says:

    Thanks Ramanan

    So it’s incorrect to think of interest on govt debt held by the CB as “funding” payment of interest on reserves, as I suspected. Just more “out of paradigm” thinking? The CB simply creates a liability against itself (reserves) with no matching asset, with associated decrement to CB equity. This could presumably put the CB in a potential accounting loss making position if its income on its assets is insufficient to compensate. This brings me to ponder whether it really matters if the CB makes a paper loss, but maybe that’s another topic altogether.

  21. bill says:

    Dear ParadigmShift

    Unless the CB is truly private institutions (which they are not despite arrangements that might look like they are – say in the US) then they might make “spend” more than they “earn” continuously (your “accounting loss) and be able to sustain that forever. Any accounting conventions that look like they are having offsetting asset/liability entries are just smokescreens.

    best wishes

  22. ParadigmShift says:

    Thanks, this brings me to another question: If the equity position of the central bank is inconsequential, why all the mainstream fuss about PBOC’s treasury portfolio losing value, with sensationalist invocations of China’s ‘nuclear option’ etc? It seems to me that the book value of assets held by the CB are less relevant than the strategic value gained in their acquisition. Same goes for the alphabet soup of asset swaps engaged in by the Fed.


  23. RSJ says:

    Hi Scott, long comment — to busy to write a short one :)

    1. Question: What are you assuming banks have on the liability side when they have (in our example) $ trillions in interest-bearing reserve balances?

    Currently we have about 7 Trillion of debt held by the public, so if this amount was not drained, there would be 7 Trillion in deposits (as a result of fiscal spending). These would be shifted into different types of interest bearing accounts, but the majority of this would end up with commercial banks one way or another (e.g. if I transferred my deposit into a money market fund, then we can assume that this fund has some commercial deposit with a bank).

    So the current liabilities of commercial banks are:

    Interbank Liabilities: negligible
    Checkable deposits: .7 Trillion
    Time/Savings/Large Deposits: 6.6 Trillion
    Repos: .7 Trillion
    Credit Market Instruments: 1.7 Trillion (mostly bonds)
    Misc: 2.5 Trillion (Mostly holding company investments)

    And I think the new deposits would be something along these lines:

    Interbank Liabilities: negligible
    Checkable deposits: .7 Trillion
    Time/Savings/Large Deposits: 11.2 Trillion (6.6 + 7 Trillion – 1.7 – .7)
    Repos: negligible
    Credit Market Instruments: (negligible)
    Misc: 2.5 Trillion (Mostly holding company investments)

    Looking at commercial bank assets, the current assets are:

    Cash (Vault/Reserves): .8 Trillion
    Financial assets: 6.6 Trillion
    Loans: 6.6 Trillion

    And if banks were not allowed to hold financial assets, then bank assets would be:

    Cash (Vault/Reserves): 7.8 Trillion
    Loans: 6.6 Trillion

    This assumes bank equity (about 2.2 Trillion) remains the same, and that the policy does not result in an increase in bank loans (e.g. no asset bubbles).

    Of course banks hate holding reserves like this, particularly during boom times. Imagine if inflation is running at 5%, and land is increasing in price at, say 7%.

    And the funny thing about banks and money is that no matter what they buy with that money, it all comes back to the banks as deposits anyways. So they can just keep buying and buying more and more land, and no amount of regulation will prevent banks from creating straw borrowers and doing this, even if the public doesn’t desire to join in the land speculation (which it will).

    Obviously these are concerns valid only for a growing economy. If you are in a Japan-situation in which nominal GDP are now at 1992 levels, it’s a different matter. They are busy recovering from this exact problem.

    Suppose that banks can lever 12:1, and that the economy is growing at a nominal rate of 6%. In that case, you better have a situation where the bank costs of loaning out an additional dollar are at least 5.5% (this includes administrative costs, bonuses, loan loss reserves, fees paid to government, etc.)

    If those costs are only 2%, then banks will be earning a return of 12*(6%-2%) 48% for every dollar of capital committed, whereas the non-bank sector will be earning only 6%.

    This situation can result in one of two end games:

    1. Banks will evade regulations and compete with the non-bank sector in such a way as to buy up huge amounts of land, leading to an eventual banking collapse and rapid deflation, followed by another asset inflation, etc. A cyclical banking model with inflations followed by deflations, similar to U.S. history prior to WW2.

    2. Banks will evade regulations and compete with the non-bank sector in such a way as to buy up huge amount of land, but will not collapse — they will drive long term yields to a level so that 12*(G – 2%) = G, or G = 2.2%. So now you have pushed nominal growth down from 6% to 2.2%. The only fixed point here is zero.

    The difference between option 1 and option 2 is what happens when asset prices rise.

    Say we have nominal growth of 6%. With no asset bubbles, you would expect the earnings yield to also be 6%. So with median incomes at 50K, and assuming 1/4 of incomes service rent, that would be a median house price of about 200K — close enough for government work. Now enter the banks, with their below market funding costs. They start buying up/lending on land, which makes the land worth more, but does not increase the income available to pay rent or service debt by the same amount, so people either default (option 1) or this ratio rises (option 2).

    If the ratio rises, then more and more incomes are being siphoned off into borrowing and repayment, which does not contribute to GDP. The effect of this is to slow nominal GDP growth down, up until we reach the fixed point: G = L*(G-C), where C = marginal bank costs, and L = leverage allowed, and G is the nominal growth rate.

    In the same way, you do not want bank costs, C, to be such that G < L*(G – C), as in this case, banks can only lend money at an interest rate greater than the private sector can afford to pay, and this reduces total borrowing, slowing nominal GDP growth as well. Volcker put the economy into rapid disinflation by hiking up rates. So marginal bank costs should be equal to the expected nominal growth rate.

    Within the credit markets (which have their own bubbles), the mechanisms of buying and selling income streams ensures that this happens, although expected growth rates are not the same as actual growth rates. But within the banking sector, it is the responsibility of the government is to set these costs administratively. You can also be countercyclical to the credit markets (e.g. if they are too pessimistic, then you want to decrease bank costs).

    If you have a religious attachment to zero interbank rates, them you can impose other fees or costs to bring the total marginal costs up to G, but not higher than G.

    But the effect on AD remains largely the same whether bonds are sold or rbs are interest bearing .

    No, an investor must choose between making an investment in non-government debt and making some other investment. This ensures that he will not pay more or less for a bond that what he thinks he could make if no bond was offered for sale. No one is making any money from government interest payments that they would not have been able to make without these payments, on average. That is why yields fall to zero when conditions are bleak. Perhaps in a period of high inflation expectations you are right, in that yields are biased to be too high, and when expectations are on the low side, then yields are too low, so there may be some pro-cyclical aspects to this.

    When government sells assets to the private sector, those assets *replace* private sector assets, as investors liquidate non-goverment assets of equal value and use the proceeds to buy treasuries. Government debt is a portfolio shift only, and cannot be inflationary. The initial fiscal spending may be inflationary, but the debt issuance cannot.

    A good example is to take a look at the period 1952-1982, which shifting government assets were offset by shifts in non-government assets, but total debt was fixed:

    What about the banks?

    Regardless of whether banks loan money out or not, they cannot shed themselves of excess reserves. By paying interest, you are preventing only a negligible amount of loans from being made. Say the baking system is in a large overall excess reserve position, the government is paying 4% on reserves, and a borrower approaches you that you believe will pay 3%, the only way that you will turn down that borrower is if you estimate that the probability of your reserve position decreasing as a result of making the loan is more than 70%. With only a handful of large banks and netting effectively eliminating the vast majority of cash-transfers, the only rational thing to do is to grant the loan, earning 3% *and* also make 4% off the reserve when it comes back to you. Instead of making 3%, you earn 7%. When the government pays banks money, it increases yields, resulting in a positive net transfer to the private sector. This can be inflationary, but it will primarily result in asset inflation.

    Interest on the national debt in the US has followed longer run averages in the fed funds vs. nominal GDP growth fairly consistently over time, even though the Tbill can deviate from the FF rate,

    Sure, because we keep FedFunds reasonable, not because FedFunds is somehow forcing yields up or down. What you want to look for here are interventions — periods of unusual tightening (since we rarely have periods of unusually loose policy). Generally FF just follows the the three month and everything moves together. I guarantee you a FF rate of 10% now would not result in the 3 month going to 10%. It would remain at zero, agreed?

    On the other hand, FF at 0% in 1980 is a toss up — I think it would have effectively ended the inflation we saw, but at the price of banks owning everything. If we somehow kept the banks from owning everything, then I don’t think it would have pulled the 3 month yields down.

  24. Hi RSJ

    1. You’re assuming bank assets will earn the nominal rate of GDP growth on their assets, on average. Leaving aside loans and bonds, I don’t agree on this for interest on reserves = fed funds target, which is what the discussion here is about. Fed held the fed funds rate below nominal GDP growth for virtually the entire 1953-1979 period, for instance, and of course they did the same during the 1940s. In fact, the 1979-2000 period is the aberration . . . before and since they kept fed funds rate below nominal GDP growth, and both short-term and long-term Tsy’s did the same, on average. Regardless, though, the point is the profit margin on rbs, not the ave interest earned, and my contention is that the profit margin on the reserves will be lower than the profit margin on the rest of the bank’s assets since only a small % of the liabilities offsetting the rbs will be deposits (as you showed), whereas the non-deposit liabilities will earn close to the rate on rbs (they typically move with the target . . though not perfectly . . and there will be increased competition for them in this environment). This will reduce ROE and growth potential. And even if profit margin does increase, it will have to be a lot higher than previously to raise ROE because asset turnover is reduced (i.e., reserves will usually be among the lowest ave earning assets on bank balance sheets). You still haven’t demonstrated to my satisfaction that bank ROE will rise as a result of this change (not that I’m necessarily in favor of it), for what it’s worth.

    2. Agree with you that all bets are off (i.e., can be large divergences) regarding Tsy’s vs. fed funds rate when there are existing or anticipated financial crises.

    3. The point about interest on rbs versus interest on Tsy’s is that regardless, it’s a net financial asset created for the non-govt sector when the interest is paid. Not so if you’re holding a non-govt liability . . . those interest transfers just send income from one prvt sector agent to another, not net increase in financial assets.


  25. Ramanan says:


    I too wanted to point this out at various discussions with you and Scott has said it explicitly – Point 1 about GDP growth and interest rates. An economy can grow at 3% or whatever with govt yield at 0 and high quality corporate credit at say 2%.

  26. Wipaporn Pojthaveekiat says:

    Please correct me if I’m wrong. There are two ways building deficits are indirectly inflationary:
    (1) Lower interest rates spawn more lending. The lending leads to higher spot prices if supply doesn’t increase.
    (2) Government deficit spending is used to buy imports, the new reserves are exchanged on the FX markets lowering the demand for the currency. The weak currency raises prices in countries where most of what consumers and factories need is imported and bought with a foreign currency.

    #2 can be countered if demand for the domestic currency rises, perhaps by raising taxes?

    Best, Luknaam

  27. RSJ says:

    Hi Ramanan,

    An economy can grow at 3% or whatever with govt yield at 0 and high quality corporate credit at say 2%.

    Sure, but then equity will grow at 4%, or returns to privately held equity will grow at 5%, etc. In general, the “enterprise value” is what counts here, so you need to look at everything.

    Hi Scott,

    You’re assuming bank assets will earn the nominal rate of GDP growth on their assets, on average.

    No, I am only arguing that loans or financial assets (bank credit advanced) will yield this, not reserves or vault cash. I define leverage as bank credit advanced/bank equity — bank credit advanced excludes reserves and vault cash, as well as checkable deposits (e.g. from FoF L.109 data). Bank equity = total bank assets – total liabilities.

    Reserves will yield whatever the government pays, and nothing more. In my analysis, excess reserves are relevant in so far as they affect bank costs of extending credit. Paying interest on reserves lowers this cost, as banks cannot eliminate excess reserves by lending, but the proceeds from the interest payments can be used to absorb loan losses or offset other bank costs, causing those fully loaded costs to decrease.

    Fed held the fed funds rate below nominal GDP growth for virtually the entire 1953-1979 period

    Woah, I am not saying that FedFunds should equal the growth rate, but that the fully loaded bank costs, “C”, of which FedFunds is just one input, should be such that G = L*(G-C). So with a leverage of 12, and G = 6, you want FedFund costs + all other costs = 5.5. That necessarily means that FF is below the GDP growth rate, but how much below depends on leverage and other and other bank costs.

    I claim that in the earlier era you cite, banks operated as a large collection of smaller entities each with their own fixed costs, large staffing needs, and also tight regulation mean that leverage was lower. All of these factors compensated for the lower level of FF vis-a-vis GDP growth, to some degree. But having FF be, say 200 basis points lower than GDP growth is a lot different than having it be 600 basis points lower, which is what you are advocating for. Of course, in a deflationary era when asset prices are falling, it is fine to argue for ZIRP.

    Over the whole period, 1954-2007. Quarterly averages of FF were about 1.29% lower than the (quarterly) nominal GDP SAAR. From 1954-1970, FF averaged 2.55% less than GDP. From 1971-1987, it was 0.5% below GDP, and from 1988-2007 it was 0.64 below GDP. I certainly believe that there was a meaningful difference in bank costs and allowed leverage during those eras.

    I will try to dig up some more data about total bank expenses.

    But we can still do some analysis, even though the actual value of C is unknown. I’ve collected L (as defined above) from FoF data, together with effective fedfunds from Fred, and SAAR GDP data from BEA. Then, I computed L*(G-(FF+0.129)), together with the Freddie Mac Mortgage Index, measuring purchase prices of conventional mortgage houses from 1970. The data is very noisy, so I plotted 5 year averages here. So that at least suggests that whatever the “true” (and changing) nature of C, a combination of loose/tight regulation allowing, L to increase, together with loose/tight funding costs are correlated with rising/falling real estate prices. Of course in the last 10 years, the true nature of L was much higher due to off-balance sheet funding mechanisms that masked leverage. Also the Freddie Mac index does not measure Alt-A/subprime loans and is not asset weighted.

    You still haven’t demonstrated to my satisfaction that bank ROE will rise as a result of this change (not that I’m necessarily in favor of it), for what it’s worth.”

    My argument is a bit different than this. In the credit markets, the expected return on capital tends to converge to the expected cost of capital, but with banks, the cost of capital is dominated by an administratively set component, not a market component. e.g. FF + other bank fees and regulations. I don’t see how you can argue that keeping these costs say 600 basis points below GDP growth when historically this spread has been 129 basis points will not result in abnormally low funding costs — not unless you offset this with other fees.

    The net result of this process is that yields fall on real estate (or whatever banks lend against), which means that asset prices rise and land speculation increases. You can view this as the market trying to equilibrate returns, but it does so in a way to force G down towards the fixed point G = L(G-C).

    My concern then is with this downward pull, not bank ROE. Bank ROE will be very high during booms and very negative during the subsequent bust.

    The issue of paying interest on reserves is a separate issue. I claim that it will only exacerbate the problem, by lowering bank costs further, just as any income subsidy would. You should be trying to keep bank costs at the appropriate level, and you raise these costs by paying money to banks. It is easy to lower the costs, obviously, by lowering FF as needed.

    The point about interest on rbs versus interest on Tsy’s is that regardless, it’s a net financial asset created for the non-govt sector when the interest is paid.

    No, a net financial asset is not created by selling bonds. It is created by fiscal spending, but bond sales at market prices do not do this.

    If the fiscal spending occurred but no bonds were sold, the private sector would still experience an increase in financial assets due to the overall growth rate of the economy.

    When you are talking about action X causing an increase in financial assets, you need to assume that action X was not taken, and then count the assets in period 2 to see if they are less. I claim that the result of the count is the same whether bonds were sold or not, assuming the same level of fiscal spending occurred, and assuming that the markets are not consistently mispricing the bonds. The result of the count in period 2 is not the same when you are talking about bank reserve interest payments.

  28. RSJ says:

    I meant to say:

    “Then, I computed L*(G-(FF+0.0129))” — missing a “0″.

    Also, I meant to say “You should be trying to keep bank costs at the appropriate level, and you cannotraise these costs by paying money to banks.”, but forgot the “cannot”.

    I should not post while drinking. Please forgive the other typos.

  29. Ramanan says:

    Hi Paradigm Shift,

    Are you talking of PBoC’s holding of US Treasuries ? That case is different – this asset is not held in the domestic currency. There is a lot of talk on this but the important thing is that the US citizens have a better standard of living by importing Chinese stuff. The US government should support this though and could finance the trade deficit by running a JG program – basically higher government deficit and is actually sustainable. As Warren Mosler said somewhere, insisting on policies that make Chinese accumulate less USDs is like killing a duck which is laying golden eggs. The Chinese have been dumb enough to export so much and the US enjoys the products and the Chinese just get an electronic increase in the amount of US dollars they hold.

    I don’t know the balance sheet implications of this but its probably the Chinese realizing that they “got cheated” but because of their own export policies. This however won’t prevent them from exporting more and more because they just smell something wrong but don’t fully understand it in an MMT way.

  30. Ramanan says:

    Hi RSJ,

    Why so much digression ? We started on interests being paid on reserves and as Scott mentioned, 3m Treasuries can be sold at a particular rate in as much quantity as the private sector needs. I don’t see why so many complications such as Freddie Mac Index are needed in a discussion on reserves.

  31. Ramanan says:


    (Assuming no comments after 20:23, because I can’t see)

    Appreciate your perserverance – after some thought, I think I know what you want to say. (based upon your comments in other posts as well) You are thinking of a situation (simplied to a less complicated scenario) where

    The CB pays 400 bps interest on reserves to banks and no bond sales
    Banks hardly pay anything on deposits
    A reserve drain by the Treasury simply does not interest anyone from the non-banking sector because they want to invest elsewhere mainly corporate bonds and equities.
    The Fed can’t control corporate bond yields
    Households and production firms will end up buying a lot of equities and corporate bonds because they may have high returns like 7%
    In the end the banks are making interest on reserves for free.

    Yes, the only possible case then is to set the interest rates to 0! Make banks simple and prevent them from securitizing and “innovating”. Banks do not need capital to lend. Remove capital requirements. Allow bank to lend in unlimited quantities but regulate. Payment systems can be settled with the CB directly and there is no need for an interbank market either. Also remember that even though the central bank cannot control the corporate bond yields, high yields lead to higher interest payments. At least higher interest payments on bank loans lead to lower aggregate demand (somewhat a monetarist argument). This is cyclical and the JG program is a really nice stabilizer. Also since banks do not have a cost of capital or reserves, they wouldn’t mind lending at lower rates. Asset bubbles are really caused by poor regulation and interest rates cannot be blamed. Imagine when someone makes a leveraged bet he/she is betting on a sharp rise, it doesn’t matter if the borrowing rate is 2% or 5%. As far as equity bubbles are concerned, I will come back to you on this sometime. The US domestic private sector has been in deficit almost every year since 1997 and it is amazing that there was such a boom in the equity market. Good education can be very helpful.

  32. Tschäff says:

    Ramman: How are the Chinese being cheated? Assuming prices are stable, they’ll be able to buy quite a lot from the US in the future in exchange for buying less today. If they wanted to exchange their dollar denominated assets for assets in some other currency though, they’d cause a large devaluation of the dollar with respect to other currencies, and maybe probably loose on their investment in treasuries.

    I don’t know the answer to Luknaam’s question though.

  33. RSJ says:


    The Chinese are cheated because they have no say in the matter. There is a large amount of forced investment, in which local officials levy ad hoc taxes and fees on their subjects, using the proceeds to foreclose on land and open up government-run businesses. Many funny stories along the lines of government officials ordering people to buy the official cigarette of a certain city, or the official soap, as they own the company, or are investors in the company. Each local chinese government entity is a nexus of business relationships owned by that group of government officials, and the primary objective of government officials is to siphon off the maximum income stream from their population via locally-levied taxes, fees, and edicts. Western economists look at this and say “”The Chinese” choose to have a high savings rate. They are so forward thinking!”. Historians and others look at the tens of thousands of protests that occur each year and declare “The Chinese people are being squeezed out of consumption possibilities by a corrupt rentier class of government officials.”

    In the same way, there are many forced investment policies, for example the stories of the army forcing conscripts to purchase life insurance from the a business run by officials, or the practice of withholding a portion of wages as part of a forced re-investment plan.

    At the same time, the government controls the forex pegs without public input. And all of this comes at a cost of high inflation to sterilize the dollar inflows, and this inflation reduces real wages — another thing that people have no say over.

    In the same way, I’m curious as to how people calculate costs and benefits when saying the “U.S.” is benefitting from a current account deficit with China. If 9 people have their incomes cut by $1, but the tenth person buys $10 more worth of goods, then according to Ramanan’s calculus, this is a $1 net benefit, but according to my calculus, it is a net cost. So there are two problems with using such a simplistic definition of costs and benefits:

    1. Use of “asset-weighted” welfare functions, and so ignoring distributional issues or externalities. Trade with high wage nations boosts incomes and productivity, trade with low wage nations suppresses incomes, and suppresses overall efficiency, as products are created in a more resource intensive manner when the source of productivity is a wage-differential.

    2. Not thinking long term, but only instantaneously — e.g. if U.S. households have their incomes decreased via wage arbitrage, but still manage to maintain spending via debt, then this is considered to be a net consumption benefit. But long term, when households cut consumption in order to repay the debt, people do not connect the dots; they do not attribute the current lack of consumption to the original wage arbitrage, of which outsourcing to China is a contributing factor (but not the sole factor, or the largest factor).

    In my book, *both* nations are being cheated of sustainable growth and a more equitable income distribution. Just as trade is not a zero sum game, it can be a negative sum game when used to discipline wages as opposed to disciplining corporate earnings. Those are different types of trade — e.g. trade with Japan in which Japanese companies compete with U.S. companies in order to sell their wares to american consumers, and trade with China in which U.S. workers compete with Chinese workers in order to sell their labor to american businesses.

    So the type of trade determines whether it is a real benefit or a cost — and you need to look at the long term distribution of earnings, consumption, and capital formation in order to make care one way or another.

  34. Ramanan says:


    There are many scenarios to consider and I am not undermining the importance of exports at all. In fact in Asia, export led growth has had its positives like increasing salaries even for the first job etc. Also exports help a lot in getting oil (and we have to make cars which run on alternative fuels to reduce this dependence).

    However obsession with exports is really not the best way in a fiat monetary system to grow. The $2T which the Chinese have as US Treasuries are more than enough for China and the best strategy for them is to grow domestically. Think of the following two situations: (assuming $1 = ¥7)

    China exports $100m worth of toys to the US. The Chinese domestic private sector assets increases by ¥700m and PBoC accumulates $100m worth of US Treasuries. The other scenario is that the Chinese government spends ¥700m on some insfrastructure projects and the private sector assets increases by ¥700m. As far as the Chinese private sector net worth is concerned, in both the cases the increase is the same. The only difference is that in the former, China has $100m worth of US Treasuries and in the latter – better quality of life.

    The Chinese have used the export strategy to accumulate a lot of USD denominated assets but whats the use ? It can buy oil but think of the rate of imports vs rate of exports! Their target of (USD reserves/GDP) is high and the rate is also going up in general.

  35. ParadigmShift says:

    Hi Ramanan

    Yes, I was referring to the PBoC’s UST holdings. Even though they are denominated in a foreign currency, presumably they are still valued on the PBoC balance sheet in RMB? So any appreciation of the RMB/USD would increase their paper losses on their treasury holdings. I was just thinking aloud really, wondering why PBoC losses on their USTs would be of such concern to the Chinese, given that it seems to derive from a false analogy of CB v commercial bank capitalisation, much like the false analogy between government and household borrowing. Or are the PBoC balance sheet implications not the issue for the Chinese? I suspect that, in general, CB solvency issues are secondary to overarching strategic aims associated with the CB acquisition of various assets – in the Chinese case, this would be the aim of pegging the yuan to the dollar. As long as this primary goal is achieved, do the Chinese really care about the value of their US treasury holdings as much as the MSM suggests they do?

  36. ParadigmShift says:

    Also, I was of the understanding that China values it’s US Treasury holdings using historic cost accounting. So any market fluctuations in the treasury market have no effect on the PBoC balance sheet, until the treasury securities are actually sold?

  37. Ramanan says:

    Hi Paradigm Shift,

    If one wants to write a transaction flow matrix such as the one in the post Stock-flow consistent macro models and/or a balance sheet matrix in case of an open economy, it is better to use the market value for financial assets. The various transactions happening when an American buys some Chinese goods, end up finally with the PBoC increasing bank reserves of Chinese commercial banks in exchange for the dollars. (To target an overnight rate, they may issue “debt” to drain the reserves). It may then be simpler to value them in RMB. (Also in general, with market value of the exchange rate).

    At a less formal level, it is not really that important. The Chinese central bank is a government entity. I think because the Fed is has done a massive QE operation, the Chinese buy the US Treasuries at a low yield and will try to make all sorts of noises. (Note: QE causing low yields on government bond is also implied by Stock-flow consistent analysis though it doesn’t support crowding out) The fact that because of low yields, the market value of their current portfolio has increased is not really good for them since they want to keep accumulating more USD denominated assets and high yields are better for them. Also they just fall into the neoliberal myth of inflation caused by increase in US bank reserves. At a more general level since there are all sorts of misunderstandings around, they will use any misunderstanding which works their way to get preferential political treatment on other matters.

  38. Tschäff says:


    Accumulating dollar assets, I imagine isn’t the goal of the PBoC, but the consequence of trying to undercut the competition by keeping the the dollar value of their exports low. So China has this constant flow of dollars coming to exporters. Those exporters want to convert it to yuan (to pay their taxes, and workers who pay taxes), but doing so would increase the demand for yuan raising the exchange value of it. The currency board doesn’t want this happening so they do some mixture of increasing the supply of yuan (they are the monopoly supplier) and taking some of those dollars away from the exporters holding them in interest bearing US treasuries. I could be wrong, I’m still learning.

    By undercutting the global competition they are able to tap into global demand and buy all the imports they can’t produce domestically to accelerate growth. Yes, they’d be able to buy more imports if their exchange rate was allowed to float, BUT demand for their exports might well be lower, with a stronger yuan resulting in less imports and less $ savings. Conversely, they could let their exchange rate float, and keep cutting the exporter’s yuan wages to remain competitive, but imports will be cheaper in yuan too so maybe it’s a wash?

    What is scary to me about China having such a large pool of dollars is that they’ll be able to profit forever off compound interest and the US might not be able to tax them the same way we can tax Americans. Even more tragic is that the US government doesn’t need to issue treasuries to them in order to fund US government spending.

  39. Tschäff says:

    regarding the wash, I guess it would depend if the Marshall-Learner condition occurs or not.

  40. Ramanan says:

    Hi Tschaff,

    China actually does not run a currency board and its currency is sovereign. Hong Kong on the other hand has a currency board. Wikipedia has a list of countries with a currency board arrangement. The PBoC can peg its currency well because its holding of the US dollar is so high. So in their logic, the holding of USDs serves to peg the exchange rate and also serves as a “reserve” for imports. So they are still using the gold standard logic. They can easily float their currency (gradually or sharply ??) and have domestic growth. Currency sovereignty offers a lot more than what 99.99…% of economists think – one does not have to be at the mercy of the US consumers to grow.

Leave a Reply

Your email address will not be published. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

To prove you're a person (not a spam script), type the answer to the math equation shown in the picture. Click on the picture to hear an audio file of the equation.
Click to hear an audio file of the anti-spam equation