Rubbish – far too much rubbish – continues to be written about ‘quantitative easing’. As I have noted before, views on this set of open market operations vary wildly between so-called ‘experts’. On the one hand, Liam Halligan in his Sunday Telegraph column has compared QE to ‘money printing in banana republics’ and said it would lead to hyperinflation. On that basis, QE must be very powerful. On the other hand, Martin Wolf of the Financial Times has on occasion denied that changes in the quantity of money due to QE can matter to anything, since – in Wolf’s view – changes in the quantity of money always do not matter to anything. On that basis, QE is impotent. Two further muddled and misguided contributions have recently appeared in the Financial Times, a piece on 9th July by John Kay (‘Quantitative easing and the curious case of the leaky bucket’) and another by Jonathan Davis on 15th July (‘The art and artifice of Fed-watching’.)
Mr. Davis said that assessing the effects of ending QE ‘is ultimately a matter of subjective judgment, not a simple binary decision that can be derived from objective analysis of data’. As I said in a letter to the Financial Times of 22nd July (see the end of the accompanying note), this is plain wrong. As long as one accepts the standard account of the monetary determination of national income, the relationship between QE and nominal national income is very straightforward in its essentials. (I don’t dispute the complexity of some of the adjustment processes involved, but – as I have been writing about them for over 35 years – I don’t doubt that they exist.) The relationship is discussed in the following note. To summarize, without QE the quantity of money in the UK today would be about 25% lower than it actually is, and so – more or less – would be the equilibrium levels of national income and wealth in nominal terms. QE prevented the Great Recession becoming a second Great Depression. QE was therefore both desirable and necessary.
How QE works, in a nutshell
The initial impact of quantitative easing, as understood in the UK public debate, is to increase the quantity of money. (I am not going to discuss here the Japanese-style version of QE, in which this impact is to increase the quantity of the monetary base.) So QE matters to macroeconomic outcomes in the same way that the quantity of money matters to macroeconomic outcomes. The discussion of QE’s effectiveness is therefore subordinate to the monetary theory of national income determination. (The following two paragraphs borrow from my weekly e-mail of 8th March earlier this year.)
According to standard theory, the equilibrium level of national income in nominal terms is determined by the interaction between the demand to hold money balances and the quantity of money created by the banking system. (This was noted by Keynes at the end of chapter 7 of The General Theory. Neither Keynes nor any other top-rank monetary economist has ever questioned the observation I have just made. Admittedly, the world is full of low-grade, self-styled ‘economists’.)
Non-bank private sector agents have a money demand function, with their demand to hold money depending on the level of income (and/or wealth), the attractiveness of money relative to other assets and other variables. (They have a money demand function, just as they have demand functions for Weetabix, red socks and holidays in Spain, and – as with these demand functions – the variables that determine the quantity demanded are relative price, income and other variables.)
With the quantity of money given, and with the non-income variables in the demand function also set at particular values, the money demand function implies that only one level of nominal national income is consistent with macroeconomic equilibrium. In that sense the quantity of money determines nominal national income. Further, if (quite a big ‘if’ in practice) the non-income variables in the money demand function are stable over time, theory says that changes in the quantity of money and equilibrium national income are equi-proportional. In the real world changes in the quantity of money usually differ from changes in nominal national income in short-run periods of a few quarters or even two or three years. The differences are largely due to agents’ difficulties in matching the demand to hold money with the actual quantity of money in existence. Nevertheless, over the long run and in all countries the differences between the annual rates of change of money and national income are typically – indeed, virtually always – very small compared with the cumulative changes in both money and national income.
What does this mean? Suppose that, over a period of ten years, policy-makers deliberately cause the quantity of money to rise by 100% more than would otherwise have been the case. Then nominal national income will also rise – roughly – by 100% more than would otherwise have been the case. That is the message of standard theory. So policy actions that affect the quantity of money are hugely important to macroeconomic outcomes.
Let us bring the discussion back to QE. The state can always create money by borrowing from the banking system, and using the proceeds either to finance a budget deficit (i.e., the purchase of goods and services from the non-bank private sector) or the purchase of assets from the non-bank private sector. Of course, the payment for the goods and services, or the assets, is in money. So the quantity of money held by the non-bank private sector rises pound for pound, dollar for dollar, euro for euro, or whatever, by the value of the sales to the state. QE is to be understood as the purchase of assets, by the state (either the government or the central bank) from the non-bank private sector, to increase the quantity of money. Suppose that QE amounts to £250b. and the quantity of money at the start of the process to £1,000b. Then QE by itself causes the quantity of money to rise by 25%. Given the equi- proportionality of changes in the quantity of money and nominal national income just discussed, which must be understood to hold only over the long run and ‘in equilibrium’, the increase in the quantity of money causes nominal national income to be 25% higher than it would otherwise be. QE of £250b. has the result of boosting equilibrium nominal national income by 25%. That, in a nutshell, is how QE works and why it matters. As I said, the subject is very straightforward in essentials.
Further details: the relevant aggregate and the key process
The public debate about QE is bedevilled by a large number of misunderstandings, but let me focus here on only two. The first is confusion about the relevant measure of ‘the quantity of money’, and the second relates to a widespread inability to see exactly how changes in this relevant ‘quantity of money’ affect expenditure and hence alter equilibrium national income.
On the first question, I am simply going to say here that the relevant measure is one that
- includes all assets than can be used to make payments, either immediately or with little difficulty and expense after a period of notice, and
- refers to money held by genuine private sector non-bank agent
In other words, the relevant measure is one that is broadly-defined to include time deposits and wholesale deposits, and that does not include money held by the government and its agencies, or by banks and quasi-banks. In the UK context this measure can be equated with M4x, as calculated nowadays on a regular basis by the Bank of England. The M4x aggregate, and in fact all broad money aggregates, are dominated by bank deposits. Indeed, it is not going too far to regard ‘the quantity of money’ and ‘the quantity of bank deposits’ as more or less synonymous in current circumstances.
Some economists think that the key aggregate in monetary economics is the monetary base, i.e., the liabilities of the central bank, including the cash reserves lodged with the central bank by the commercial bank. This is just wrong. In the UK QE has been very large relative to the level of the monetary base in early 2009. Indeed, the monetary base has risen by more than five times since then, providing the rationale – or rather the bogus rationale – for Liam Halligan’s forecasts that QE would lead to hyper-inflation etc. Such forecasts have been invalidated by events.
Secondly, many people are baffled about how a large increase or decrease in the total quantity of bank deposits can affect anything in the economy. (I am not kidding.) I have explained the processes many times in several places, and would refer readers particularly to the opening pages of my 2005 Institute of Economic Affairs’ study on Money and Asset Prices in Boom and Bust and part five (‘How does the economy work?’) of my 2011 collection of essays Money in a Free Society. A paragraph can quickly give the guts of the matter.
I set out above the argument that, with the non-income arguments in the money demand function given, agents would have one and only one desired ratio of their money holdings to income. (With tastes and relative prices given, people have one and only one desired ratio of expenditure on Weetabix, red socks and holidays in Spain to their income.) Suppose that – for whatever reason – the actual money holdings of all agents (i.e., the aggregate quantity of money) are above this level. They therefore all want to reduce the ratio of money to income. Suppose that one agent spends above income in order to get rid of ‘the excess money’. That boosts expenditure, but it doesn’t get rid of the excess money for the economy as a whole, because the money is credited to another agent. This other agent may also try to get rid of the excess money by spending above income, but the money again stays in the economy, this time in yet another bank account. The condition of an excess supply of money is associated therefore with an excess demand for commodities. The desired ratio of money to income is restored when this excess demand has raised the price of commodities by enough to eliminate the excess money holdings. Given the stability of the money demand function, and the assumption of certain values of the non-income arguments in that function, the rise in the price level of commodities has to be equi-proportional with the rise in the quantity of money (i.e., of bank deposits).
The statement I have just given has been made innumerable times by many other economists, beginning with David Hume in the 18th century, and running through Keynes and Friedman in the 20th century. The point I want to emphasize is that the tendency of agents to spend above income arises solely because of the divergence between their actual desired money holdings. That is all, finish, full stop, end of story. (This is the proposal many people find strange. It is basic.) For clarity, the tendency of agents to ‘spend more’ and so get rid of their excess money has nothing to do with any of the following,
– ‘the rate of interest’, whether that means the money market rate or the long bond yield,
– the quantity of new bank credit to the private sector, or
– the spreads charged by banks on loans to the private sector.
(Let me concede that I agree the behaviour of the rate of interest, new bank credit and credit spreads are relevant to macroeconomic outcomes, but such behaviour is emphatically not the point in equilibrating the demand to hold money with the quantity of money created by the banking system. Repeat, none of ‘the rate of interest’, bank lending and credit spreads is pertinent in the present description of the transmission mechanism from money to the economy.)
Much tripe has been written in the last few years by people who think that the purpose of QE is ‘to reduce interest rates’, ‘to boost bank credit’, ‘to narrow credit spreads’ and so on. Tripe. The purpose of QE is to raise the quantity of money and thereby the equilibrium level of nominal national income. The key motivating force in the transmission mechanism is the difference between agents’ actual and desired ratio of money to income. Equilibrium holds only when agents’ actual money holdings (i.e., their holdings of bank deposits, let it be said again) are in line with the quantity of money balances created by the banking system.
Why has QE been necessary since early 2009?
In late 2008, in the weeks following the Lehman crisis, macroeconomic and regulatory officialdom in the leading economies decided that banks must be made safe. The people involved were in such a tizzy that they resolved that ‘the tidying-up of bank balance sheets’ had to be done once and for all, definitely, comprehensively and quickly. (The phrase, ‘the leading economies’, means here the G20, more or less, but we now know that the discussions were very much led intellectually by the USA and the UK, and – in terms of individuals – by Ben Bernanke and Mervyn King.)
Banks across the advanced world were therefore required to shrink their risk assets and to raise more capital. The effect of the official injunctions was
- to make the banks sell off non-core assets, such as securities, to non-banks,
- to oblige the banks to stop new bank lending (which would increases risk assets) and to pull in low-quality loans (even if normally banks are very reluctant to do this, except as a last resort), and
- to issue securities (both new equity capital and bonds) to investors.
All three courses of action led to a fall in the level of bank deposits relative to what would otherwise have happened. They all caused the destruction of money balances. Let me amplify for those new to this subject. When a bank sells anything to a non-bank, the non-bank pays for it by a deduction from his/her bank deposit, which then falls; when a bank loan is repaid, a money balance in a deposit is used to cancel the loan, and both the loan and the deposit disappear from the bank’s balance sheet and the economy; and, when someone subscribes for a new issue of securities by a bank, the investor’s bank deposit falls by the value of the newly-issued securities that are being bought. Over the five years since mid-2008 the banks in all the leading economies have done much to comply with the new and much tighter regulatory standards. The result is that money growth has been negligible or very low almost everywhere ‘in the leading economies’, and so also has been the rate of increase in nominal GDP. That may sound worrying, but it does not go far enough. The truth is that – if not counteracted by an expansionary influence of some sort on the quantity of money – banks’ responses to the new regulatory pressure from late 2008 implied large falls in the quantity of money and, hence, similarly large falls in nominal national income. I know that the statement I have just made – that officialdom caused a collapse in money growth by its regulatory squeeze – may astonish those who regard officialdom as blameless and the bankers as ‘banksters’. Well, let me make my core allegation again: the abrupt, hurried and comprehensive tightening of bank regulation in late 2008 was the main cause of the Great Recession, because it led to a collapse in the growth rate of the quantity of money.
The relative significance of different influences on the growth of the quantity of money can be identified from official data on the so-called ‘credit counterparts’ to broad money growth. The information below relates to the M4 aggregate in the UK. This may seem a bit odd, since I said earlier that M4x is the right aggregate for UK macroeconomic analysis. However, the Bank of England’s resources are finite. It has only a limited amount of funding available for its statistical unit. Numbers are available for the credit counterparts to M4 for several decades back into the past; they are not available for M4x. (M4 is appreciably larger [by about a third in recent years] than M4x, because it includes money balances held by so-called ‘intermediate other financial corporations’, which are not genuine non-banks.) Still the credit counterparts for M4 give us the facts that we want, more or less.
Forces determining UK money growth before and after 2008
The majority of banks exist in order to generate profits for their shareholders. They grow their balance sheets by adding assets on one side of the balance sheet and financing these assets by incurring liabilities on the other side. The profits come of course from charging a higher interest rate on the assets than is paid on the liabilities. The liabilities are dominated by deposits, but they are not exclusively deposits. It follows that
The growth of deposits = the growth of banks’ assets minus the increase in non-deposit liabilities.
Non-deposit liabilities are mostly banks’ capital, notably the equity capital which belongs to the shareholders, but also bond liabilities. Assets can be claims on the domestic public sector (‘the state’), the domestic private sector and the external sector. So the growth of deposits can be seen as reflecting changes in banks’ claims on the three sectors minus the increase in non-deposit liabilities. The bar chart below shows the size of these influences on M4 in the five years to mid-2008 (i.e., the five years before the radical upheaval in bank regulation which followed the Lehman crisis).
The chart below shows the same set of numbers, but for the five years after mid-2008 (i.e., as banks responded to the regulatory onslaught on their businesses). (Please note that I have made an estimate for the very final quarter, i.e., the second quarter of 2013, as UK monetary data are not yet complete for it.)
A comparison of the two charts shows had radically different these two periods were for the UK banking system. In the first five-year period banks were expanding their loan portfolios aggressively, with total new claims on the private sector increasing by almost £1,000b. (But remember that this includes claims on those awkward ‘intermediate other financial corporations’, many of which were in fact bank subsidiaries.) The growth of bank lending exceeded the growth in the quantity of money, which was a bit more than £700b. The main factor explaining this gap was that banks had to increase their capital, evidenced in the £186b. increase in their non-deposit liabilities. (Note that this is a deduction from M4, with a bar in the chart above that is negative and lies beneath the zero line.) Meanwhile the public sector contribution to money growth was small, but negative. Since 1985 official policy has been to ‘fully fund’ the budget deficit, so that until the adoption of QE public sector transactions had little effect on the quantity of money. In this five-year period such transactions reduced M4 by £38b., a very minor influence on the overall picture.
In the second five-year period the M4 quantity of money continued to expand, but at an annual rate of less than a third that in the previous five years. Although the growth of bank balance sheets was therefore much slower than before, the banks had to raise more capital because the regulators’ demands. In this five-year period non-deposit liabilities went up by over £257b. Official disapproval of the amount of risk in bank balance sheets virtually stopped the expansion of banks’ claims on the private sector, which went up by a mere £29b. (Some new lending was made, but it was offset by the shedding of low-quality loans and so-called ‘toxic securities’.) Here we come to the vital conclusion. If the effect of the public sector’s transactions on M4 had been the same in the five years to mid-2013 as in the previous five years, the M4 quantity of money would have fallen. Indeed, it would have fallen substantially, by hundreds of billions of pounds. The sum of the increase in non-deposit liabilities and new bank lending would have been negative by about £200b. and the public sector’s own transactions would have taken the figure down by a further £38b. Sure enough, a monetary squeeze of the sort implied by these numbers would have attracted money balances to come in from the rest of the world and M4 might not in the end have dropped by the full £200b. – £250b. that seems to be indicated by the analysis. Whereas M4 money growth in the five years to mid-2008 was 11% a year, it would have been nil or probably negative in the five years to mid-2013. The Great Recession of late 2008 and 2009 was grim, but it could have been much worse.
Happily, a major policy change occurred in early 2009, with the announcement of QE. The public sector’s transactions were themselves responsible for a large increase in the quantity of money, as the Bank of England bought gilt-edged securities (particularly long-dated gilt-edged securities) from the private sector, including private sector non-banks. According to the statistics, the public sector’s transactions added £463b. to M4 in the five years to mid-2013. The number is somewhat higher than the outstanding QE stock, usually put at £375b., but is ‘in the same ballpark’.
Reasonable conclusions from the data are
- As I said in my letter to the Financial Times on 22nd July (referring to M4x, it has to be said, not M4), the de-risking of bank balance sheets (the shedding of risk assets and the raising of large amounts of capital) from mid-2008 would by itself have cut the quantity of money ‘by hundreds of billions of pounds’, and
- No one knows that the exact size of the fall in the quantity of money that would have occurred without Q But let us take it that QE added somewhat more than £400b. to M4 in the five years to mid-2013. M4 in mid-2008 was about £2,000b. Then, without QE, M4 today would be about £400b. lower than it in fact is. It is in fact at present just under £2,100b. So M4 would be about £1,700b. or so. It would have declined by perhaps 10% – 15 % from its mid-2008 level instead of rising.
Conclusion: QE raised the equilibrium level of UK national income by about 20%, relative to what would otherwise have occurred
The use of data relating to M4 is far from satisfactory, but – as I said – credit counterpart work is possible only with the M4 measure. However, I am confident that – if we could carry out credit counterpart analysis for M4x – it would yield much the same conclusion. The Bank of England does in fact publish figures for bank lending to the non-bank private sector excluding intermediate ‘other financial corporations’ and the data available confirm that such lending has been extremely weak since the regulatory tightening in late 2008. (See the chart on the next page.) In the five years to mid-2008 M4 lending totalled £997.5b. and M4x lending £925.3b.; in the five years from mid-2008 M4 lending was cumulatively £28.6b. and M4x lending was £34.2b. Furthermore, the capital requirements for the IOFCs were very low, since they were intended to bypass official capital rules. As the IOFCs are being closed down gradually, the capital-raising of the main banking system must be similar to that of the main system plus the IOFCs.
It seems plausible that – without some offsetting force – the weakness of bank lending and the capital- raising efforts mandated by officialdom would have implied a drop in M4x in the five years from mid- 2008 of ‘hundreds of billions of pounds’, as I said in my 22nd July letter to the Financial Times. The achievement of QE was to stop that fall taking place. At the start of 2009 M4x stood at a bit more than £1,500b., while the stock of QE today is given as £375b. (The impact of QE operations on broad money may not be exactly the same as the size of the QE operations, but it will be similar. I showed this above, in the analysis of the M4 counterparts.) By implication, in the absence of QE the level of M4x would now be 25% (about £375b.) lower. (But note the UK economy would have been so depressed without QE that a balance-of-payment surplus would have emerged. Money inflows from abroad would have raised UK-held money balances and to a degree have mitigated the slump.)
The deflationary pressures on the UK economy since 2008 would have been ferocious without QE. As I have said before, the QE programme prevented the UK suffering a Great Depression. But it must be said the Great Recession of late 2008 and early 2009 was bad enough, not least because it have been easily avoided if central bankers and regulatory officialdom had had a decent understanding of monetary economics.