It Can and it Does
In light of the upcoming October Fed (non-)decision, we want to briefly revisit a subject that still appears to be causing some confusion. We most recently encountered this confusion again in a quarterly update by the Hoisington Investment Management Company. To be sure, we very often, if not to say almost always, have tended to agree with the economic conclusions of Lacy Hunt and Van Hoisington since we have first come across their work (we may arrive at these conclusions in a somewhat different manner, but the conclusions as such are usually not much different).
Image credit: dreamstime
US true money supply TMS-2: this broad aggregate contains all the items that can be properly defined as money – click to enlarge.
In their third quarter update we have come across one sentence that we believe requires comment, as we have seen similar things asserted elsewhere and we believe it is important to be 100% clear on the topic. In addition to the assertion we want to challenge, which is highlighted below, we also quote the preceding paragraph, because it serves to elucidate a few additional conceptual problems.
“Despite the unprecedented increase in the Federal Reserve’s balance sheet, growth in M2 over the first nine months of this year fell below its average rate of growth over the past 115 years, a time when the growth in the monetary base was stable and quite modest. In addition, velocity of money, which is an equal partner to money in determining nominal GDP, has moved even further outside the Fed’s control. The drop in velocity to a six decade low is consistent with a misallocation of capital and an increase in debt used for either unproductive or counterproductive purposes.
The evidence speaks for itself: the Fed cannot print money. The Fed does not have the authority or the mechanism to print money. They have not, they are not and they will not print money under present laws.”
We actually don’t really need to consult empirical evidence to know whether or not the Fed is able to “print” money (for all intents and purposes, creating its electronic equivalent is the same thing as “printing” it). All we really need to know is in what ways money can be created in the current monetary system. That will enable us to determine what can be done and who can do it. However, even if we were to consult exclusively the empirical evidence, we would still have to conclude that the Fed has been printing money between 2008 and 2014 as the chart below illustrates. If it wasn’t the Fed, it must have been Santa Claus. We don’t believe it was Santa.
The annualized rate of growth of TMS-2 and total bank credit in the US banking system. The most glaring evidence can be seen in the blue rectangle on the left hand side. The growth of fiduciary media (uncovered money substitutes) in the banking system first slowed dramatically and then turned deeply negative between 2008 and 2010 (more bank credit was called in than was extended). The exact opposite happened with the money supply: it soared. How could it do so if not for the Fed? – click to enlarge.
We should first explain why we employ TMS-2 rather than M2 (although this would make little difference with respect to the example shown above). Using M2 to determine the amount of money in the economy is somewhat problematic. Even though it can often be used as a serviceable approximation, there is a specific problem with it that is actually quite relevant to the current juncture.
Rothbard summarizes the guidelines laid down by Ludwig von Mises in the Theory of Money and Credit with respect to the proper definition of money as follows (in Austrian Definitions of the Money Supply, from New Directions in Austrian Economics, 1978, p.143-156):
“Money is the general medium of exchange, the thing that all other goods and services are traded for, the final payment for such goods on the market.”
Obviously, before one can determine what should be included in the money supply, one needs to have a proper concept of money. The concept as laid out above doesn’t provide a final answer to the question of what should be included though, as potential candidates for inclusion have to be considered separately to determine the degree of their “moneyness”.
We won’t go into all the details here, but only focus on the most important components of the “true money supply” and its main difference to M2 (an in-depth discussion of money supply definitions and the concepts behind them can be seen at Michael Pollaro’s “Austrian Money Supply” page at Forbes). Then we will consider who can create this money, and how it is created. Someone’s doing it, after all!
Given the above definition of money, we can state that the money supply in the broadest sense should include all standard money (currency/banknotes), as well as all money substitutes (covered and uncovered ones) that can be transformed into standard money on demand. In the US, this includes deposits in savings accounts, which are de facto available on demand (de iure, banks are able to withhold payment for 30 days – this is in the fine print – but this right is never invoked. If a bank did invoke it, it would start a run on the bank and regulators would be knocking on its doors within hours).
The most important difference between TMS-2 and M2 is money invested in retail money market funds, which is included in M2, but not in TMS-2. Why are retail money funds not money? One could argue that they are considered money from the perspective of the individual holding a check book or a credit card linked with his or her retail money fund. However, retail money fund investments as such can not be used for final payment: the fund manager must first sell the underlying investment (commercial paper or t-bills) before the money becomes available for settlement.
In reality, we are looking at short term credit transactions: the investor buys money fund units, the fund manager lends the investor’s money to companies and/or the government. Thereafter, the money is recorded as a deposit in the accounts of these borrowers. Hence, if one counts both retail money fund investments and demand deposits as part of the money supply, one is in fact double-counting this portion of the money supply.
There is another problem posed by retail money funds that is relevant for all those who want to use money supply growth data as a signal. The amount of money held by individuals in money market funds is often strongly linked with their confidence about the trend in risk asset prices (they move funds between these asset categories). Since risk asset prices tend to decline/increase with a lag relative to money supply growth, retail money fund holdings will often correlate inversely with actual money supply growth. M2 will therefore at times mask the actual rate of money supply growth in the economy. We have highlighted one such period of negative correlation below.
Retail money market funds: not only are they not money according to the proper definition of money, but rather represent investments/credit transactions, they will also tend to distort money supply data at times – usually at times during which it is critical to have precise data at one’s disposal – click to enlarge.
We would love to thoroughly criticize the concept of “money velocity” as well, but that will have to wait for a separate article. We only want to mention here that all declining “velocity” is telling one is that the rate of growth of money printing has exceeded the rate of GDP growth. Well, duh. Velocity is essentially a fudge factor in Fisher’s tautological equation of exchange. It is far more sensible in a conceptual sense to speak of thedemand for money rather than its “velocity” (we certainly agree with Hunt and Hoisington that a lot of capital misallocation has resulted from the Fed’s policies, but we don’t base this conclusion on a chart of money velocity).
The Process of Money Creation
There are essentially two ways in which money is created in the current monetary system. The “normal” way is that banks make loans by simply crediting the accounts of borrowers with money conjured from thin air. In theory, the deposits that are used for lending as well as the deposits that are newly created by it, should be “covered” by a fractional reserve (let us call it 10% for argument’s sake). In practice, these so-called required reserves haven’t mattered in terms of credit expansion since at least 1995, when the Fed greatly boosted inflationary bank lending by allowing the banks to conduct “sweeps”.
In sweeps, banks will simply transfer funds held in the form of demand deposits into so-called money market deposit accounts (or zero interest CDs) overnight, which are not subject to reserve requirements. This is why bank credit and the associated fiduciary media (uncovered money substitutes) were growing like gangbusters between 1995 and 2008, while bank reserves actually declined slightly.
Total bank credit vs. reserves held at the Fed. Prior to the crisis, reserves had ceased to matter in terms of the amount of credit banks could extend – click to enlarge.
Nevertheless, in order to explain the money creation process and how the Fed is backstopping it, let us assume that reserves matter; this will also allow us to explain what reserves can be used for. First of all, the difference between covered and uncovered money substitutes is this: covered money substitutes are “backed” by standard money, either in the form of reserves held at the Fed or vault cash held by the bank. No such backing exists for uncovered money substitutes, which can nevertheless be used for making payments.
These two items, bank reserves and vault cash, represent a bank’s cash assets. Assume by way of example that a bank has $100,000 in vault cash and a customer wants to withdraw $1,000,000. Then the bank will ask the Fed to transform $900,000 of its reserves into cash so it can pay the withdrawal. This is one way in which reserves can be used.
The other use of reserves is for the purpose of interbank lending. Let us assume we are in the pre-crisis period and the required reserve is 10%. Bank A and bank B both hold customer demand deposits in the amount of $100K each and hold reserves of $10K each at the Fed. They could therefore in theory lend out $90K each to become “fully loaned up” (for the sake of simplicity, we are not considering other lending possibilities here, such as the intermediation of time deposits or lending out the bank’s own equity capital).
Let us say that bank A only finds an opportunity to lend out $50K. Assuming that the borrower’s account is with another bank, this will leave it with excess cash of $40K (it has to keep $10K in reserves for its deposit liability). Bank B by contrast wants to lend out $130K. It can now borrow $40K from bank A in the interbank lending market. Note that the amount of required reserves both banks together have to keep with the Federal Reserve remains only $20K – the larger amount of credit extended by bank B is “covered” by the interbank loan it has received from bank A. However, in the aggregate, the two banks have now created $180,000 in uncovered money substitutes from the deposits they hold. System-wide, the both the amount of credit and the amount of money in the economy has increased by this amount.
This process can then be repeated ad nauseam with the newly created fiduciary media (the famed “money multiplier effect”). Assuming that the borrowers hold accounts with bank C, deposits held at bank C will increase by these $180K – of which it can theoretically lend out $162K (as it has to keep 10% in reserve for the new deposit liability). This is in essence how fractionally reserved banks that are subject to a specific reserve requirement create money. As long as more new loans are granted than old loans are paid back, the broad money supply will increase, with most of it consisting of uncovered money substitutes.
How does the Fed fit into this picture? The Fed administers a target rate for interbank loans – the federal funds rate. Let us assume that banks in the aggregate wish to lend so much money at current interest rates that their demand for reserves threatens to push the interbank lending rate above the Fed’s target. In this case, the Fed will make as many reserves available in the form of Federal Reserve credit as are needed to keep the rate at its target (conversely, if the supply of bank reserves offered threatens to lower the federal funds rate below target, it will drain liquidity from the system. However, credit expansion has become a nigh unceasing feature of the system following the adoption of full-fledged fiat money).
The Fed does this in two ways: by means of repo transactions – offering temporary Federal reserve credit by discounting securities held by banks for a limited time period, and by means of coupon passes – outright purchases of eligible securities from banks by the Fed, which will tend to raise the money supply permanently (as long as they are renewed when the underlying securities mature). So this is in essence how the money supply increases in normal times. Most of it is created by means of inflationary bank lending.
Both credit demand and credit supply are as a rule spurred when the Fed lowers its target rate. Demand increases because credit becomes cheaper, supply because firstly, there is a time period during which interest rate margins will be slightly higher than before (with banks usually slow to pass the lower rate on in full), and secondly, given the nature of the lending process, banks will almost always be willing to expand credit as long as there is sufficient demand for it. Anyone able to conjure money into being from thin air would.
However, this process has been briefly interrupted after 2008. Outstanding bank credit actually declined for a little while, an almost unheard of event in the fiat money era. Normally, the money supply would have contracted in view of this decline in total bank credit, specifically the part of it consisting of uncovered money substitutes. Note also that cash withdrawals by worried bank customers rose sharply during the crisis period, which is reflected in the sudden jump in outstanding currency at the time. In a fractionally reserved system, such withdrawals have a “reverse multiplier effect”, as they lead to a decline in bank reserves. Credit will contract even more sharply as a result.
During the mortgage credit crisis and again during the euro area debt crisis, the demand for currency increased sharply – this was mainly due to worried bank depositors withdrawing money – click to enlarge.
And yet, as we noted above, the money supply not only did not contract in the crisis period, it actually increased by leaps and bounds. Between January of 2008 and the end of September of 2015, the amount of money in the US economy has increased by almost 112% (from $5.3 trillion to $11.22 trillion). In light of this, it is highly amusing that the whole world was fretting about “deflation” at the time of the crisis – in reality, the biggest money supply inflation of the entire post WW2 period was set into motion.
This was due to the second way in which money can be created in the modern monetary system: directly by the Fed. In the Hoisington report we mentioned earlier there is an unspoken assumption (others such as Cullen Roche have actually made this assertion explicitly), namely that all that the Fed does when it buys securities in QE operations is to credit the reserve balances of banks in return, and that it then kind of hopes that the banks will increase their lending.
This would be true if the Fed bought securities directly from banks and no-one else. Given that bank reserves are held with the Fed and cannot be spent (they can only be used for interbank lending operations and to pay for customer withdrawals of deposits), they are not considered part of the money supply, as they remain outside of the economy. However, this is not what actually happens. Most of the Fed’s securities purchases in the course of QE are from entities that are legally organized as non-banks – even if many of them are bank subsidiaries, such as the primary dealers (in the course of QE, the Fed has also bought securities from other non-banks, such as Blackrock and Fidelity). Here is a list of the primary dealers with which the Fed does most of its business:
Bank of Nova Scotia, New York Agency
BMO Capital Markets Corp.
BNP Paribas Securities Corp.
Barclays Capital Inc.
Cantor Fitzgerald & Co.
Citigroup Global Markets Inc.
Credit Suisse Securities (USA) LLC
Daiwa Capital Markets America Inc.
Deutsche Bank Securities Inc.
Goldman, Sachs & Co.
HSBC Securities (USA) Inc.
J.P. Morgan Securities LLC
Merrill Lynch, Pierce, Fenner & Smith Incorporated
Mizuho Securities USA Inc.
Morgan Stanley & Co. LLC
Nomura Securities International, Inc.
RBC Capital Markets, LLC
RBS Securities Inc.
SG Americas Securities, LLC
TD Securities (USA) LLC
UBS Securities LLC.
One can already see from the company names that these are not the banks themselves, but rather subsidiaries of banks (at one time, several of the parent companies such as Goldman Sachs were non-banks as well – this changed after the crisis, as they wanted to get easy access to Fed credit). Given that the primary dealers are not deposit-taking institutions, what happens when the Fed purchases securities from them? It will send them a check, which they pay into an account held with their parent bank. The bank will then credit this account with deposit money and present the check to the Fed for settlement – the Fed will in turn credit the bank’s reserve account.
As can be easily seen, in this process both new deposit money and new bank reserves are created. This explains two things at one stroke: 1. how it was possible for the money supply to rise in spite of a contraction in outstanding loans and fiduciary media and 2. that the banks have absolutely no control over the amount of excess reserves piling up at the Fed. The Fed has simply replaced the interbank lending market, and in the process created so much new bank reserves and covered deposit money, that it has easily offset the contraction in uncovered deposit money during the crisis, and then some.
As an aside: the bulk of the Fed’s liabilities consists of the “monetary base”, which in turn consists of currency (part of the money supply) and bank reserves (cash assets of banks and the basis for inflationary lending, but not part of the money supply). Similar to fiduciary media, Fed credit is created from thin air.
In short: the Fed can and does “print” money.
Inconsistencies in the Data
If one looks at the chart showing the growth in bank credit and the money supply further above (chart 2), one can see that during QE1 and QE2, the Fed was the primary source of new money creation in the system, as bank credit was shrinking during most of this time. During QE3 (which started in late 2012), both the Fed and bank credit growth contributed to money supply growth. Since the end of QE3, commercial banks have taken over again and have become the main source of money supply growth – which is why the annual growth rate remains brisk (at currently 8.35% y/y), in spite of QE having ended.
However, isn’t there an inconsistency? Why did money supply growth remain so strong between QE2 and QE3? And why did it weaken during QE3? This can be explained by the fact that the money supply data presented here refer only to the domestic US money supply. However, a great many dollar deposits are held in accounts abroad, the bulk of them in Europe (the euro-dollar market). At the time the euro area debt crisis broke out, there was still an unlimited FDIC deposit guarantee in force in the US, which had been instituted after the 2008 crisis in order to stop electronic bank runs (which had contributed to the demise of several banks and investment banks during the crisis).
As a result, dollars held abroad flowed into the US banking system, keeping money supply growth uncommonly high beyond the end of QE2. By the time QE3 started, the FDIC’s unlimited guarantee had been withdrawn, and the euro area debt crisis was deemed to be under control. Hence, this money flowed back out of the US again, blunting the effect of QE3 on domestic money supply growth (moreover, QE3 started from a much higher base, so the base effect also lowered its percentage growth somewhat). Later, “tapering” contributed to an additional slowdown. The Fed is however eager to keep its contribution to the money supply increase in recent years constant, by continually reinvesting funds it receives for debt securities that mature.
We don’t have precise data on these overseas dollar flows available, but one could probably calculate their approximate size by using the known data with respect to bank lending and QE.
How Money Created by QE Enters the Economy at Large
The next question is, what happens once the Fed’s counterparties have received the newly created money in exchange for securities the Fed buys in the course of QE? They are definitely not leaving it in their accounts, but the process of how it enters the economy is slightly different from a “normal” credit expansion process, in which the banks primarily expand their lending to businesses and consumers.
Instead, the recipients of this money will employ it to buy more securities. Many of the sellers of these securities in the secondary markets will in turn do the same, but not all of them will. Both government and companies will spend the money they receive for newly issued debt (companies will in part pump it back into financial markets again, by buying back their own shares). The major MBS issuers Fannie and Freddie will fund mortgages with it. Thus the new money eventually spreads throughout the economy.
In this case it is however taking an extended detour via financial markets, which probably explains why there has been such an outsized effect on asset prices during the current post-crisis echo boom. This incidentally ensures that the eventual bust should be especially jarring. What the banks have been doing is illustrated by the next chart:
Treasuries and MBS held by commercial banks – they have accelerated their buying during the post-crisis era, mainly because government and the government-backed GSEs were and still are considered the safest debtors, while the Fed’s loose monetary policy encourages “riding the yield curve” – click to enlarge.
The Fed’s Control over Interest Rates
Lastly, we want to briefly comment on how the Fed can possibly raise administered interest rates under current circumstances and why there hasn’t been an explosion in bank lending based as a result of the huge accumulation of excess bank reserves (these are actually related topics).
Consider what we said above about how the Fed controls the federal funds target rate during normal times. It does so by adding to or by draining reserves. However, if it were to drain reserves to a sufficient degree post QE to restore the pre-crisis situation – with a lively interbank lending market in which it can influence the target rate with relatively modest reserve-related open market operations – the money supply would as a corollary shrink rather dramatically. The Fed obviously won’t do that.
This is where interest payments on reserves come in. These have effectively decoupled the amount of bank reserves from the federal funds rate, as the Fed can always keep the banks from offering reserves by keeping the rate it pays on reserves at the upper end of the FF rate target corridor. Similarly, payment of interest on reserves lowers the incentive for banks to lend, as all other borrowers are inherently more risky than the Fed. Moreover, bank lending is not independent of credit demand, which has remained fairly subdued so far (at least on the part of consumers).
Interest paid on reserves and the effective federal funds rate – the Fed simply keeps IOR at the upper end of the corridor. If it wants to raise the FF rate without reducing excess reserves and pressuring the money supply, it will simply have to raise IOR as well – click to enlarge.
In today’s monetary system, both the Fed and commercial banks can create new deposit money. The main difference is that the former will create covered and the latter uncovered money substitutes, but both methods will increase the money supply. The fact that only a small part of money supply growth involves literal printing of banknotes is not relevant to this – what is relevant is that both banknotes and deposit money are accepted as final means of payment. It is therefore perfectly fine to refer to the process as “printing money”.