The following note considers whether the aggressive monetary easing now apparently under way in Japan will prove effective in boosting the economy.
The Bank of Japan has indicated that it will double the monetary base by the end of 2014. But the important monetary aggregate for the macroeconomic prospect is the quantity of money, broadly-defined. The analysis in the note shows that earlier surges in the monetary base and M1 over the last 18 years have not led to lasting improvements in demand and output, and they have not been enough to deliver meaningfully positive inflation.
I. Money and the economy
By common consent, the passage of events in the Great Recession shows that what happens in the banking industry matters to macroeconomic outcomes. But what aspect of banking is most relevant, and how does it affect expenditure and wealth? Before discussing current Japanese monetary policy, a few brief remarks are needed to answer this question. They are valid in any modern industrial economy, but the Japanese monetary authorities – like those in other countries – seem not to be altogether sure of them.
Total expenditure consists of expenditure by the private sector and the state. The state’s expenditure is not affected by its money holdings, because uniquely it has fiscal powers, i.e., the power to tax and the power to print money. The situation is very different for private sector agents, which can be divided into banks and non-banks. Most private sector expenditure on goods and services is carried out by non-bank agents, that is, by individuals and companies. Only a trivial proportion of expenditure is carried out by banks themselves for their own purposes. So, to the extent that changes to banks’ balance sheets affect the wider economic situation, they do so via changes in non-banks’ claims on or liabilities to banks.
Non-banks interact with banks in two main ways. First, they borrow from banks, either by taking out bank loans or issuing securities which are bought by banks. Such borrowing is almost invariably to acquire a capital asset of some sort, with the asset acquired usually pledged to the bank as security for the loan. Asset purchases financed by bank credit are a small fraction of all asset transactions, although they can be important in some parts of the economy, particularly in the commercial property market and housing finance. Further, when bank credit is used to facilitate asset purchases, it has no direct effect on the income-expenditure circular flow which is so prominent in the undergraduate teaching of macroeconomics. Banks do not lend to so-called ‘liquidity-constrained households’, that is, to households that have no wealth and want to spend above income, because liquidity-constrained households cannot pay them back. It follows that bank credit is by itself of little importance in the determination of national income and wealth. (When banks extend new loans, they create new money, and the money balances are important to national income determination. But that is another subject, to which I will now turn.)
Secondly, non-banks hold claims on banks in the form of deposits. The deposits, which in principle were originally of legal-tender cash, can be converted back into legal-tender cash. (Note that, as recent events in Cyprus have shown, this may not always be so.) The depositing of cash with banks reduces non-banks’ transactions costs, because banks are specialists in money transmission and the settlement of debts. (Banks have clearing houses and associated settlement arrangements, plus domestic branch systems and networks of international correspondents.) The banks are so efficient in making payments that they charge next to nothing for complying with customers’ payment instructions. Instead they make a profit (or try to make a profit) by earning a higher rate of return on their assets than on their deposit liabilities.
Partly because of the efficiency of bank settlement in a modern economy, the value of payments by non-banks is nowadays a very high multiple of national income and expenditure. Nevertheless, neither the value of payments nor the level of national income can be infinite. The nominal value of both payments and income/expenditure is limited by agent’s inability to make money circulate an indefinitely large number of times in any given period. Can the limit be pinned down more precisely?
The monetary theory of national income determination proposes that agents (let it be emphasized again, private-sector non-bank agents) have a demand to hold money balances, just as they have a demand (say) for the services provided by the housing stock. Further, a demand function for money can be described in which the quantity of money an individual wants to hold depends partly on his or her income, and partly on other variables that affect the relative attractiveness of money and other things (such as, for example, the rate of return on money compared with alternative non-money assets). Individuals’ money demand functions can be combined in an ‘aggregate money demand function’ for all private-sector non-bank agents together. For certain given values of the variables that affect the attractiveness of money relative to other things, there is one – and only one – desired ratio of money to income and payments. It follows that the equilibrium level of national income is determined by two variables, the non-bank private sector’s desired ratio of money to income and the quantity of money.
However, there is an ambiguity here. ‘Money’ can be defined in more than one way. Some economists focus on a so-called ‘narrow’ aggregate (often called ‘M1’), which includes only current accounts or sight deposits, and believe that such narrow aggregates can be interpreted as a multiple of the monetary base (i.e., the legal-tender cash, into which deposits are supposed to be wholly convertible). Others – including myself – judge that an aggregate that includes all money (or money- like) assets is the correct one to use in macroeconomic analysis. One reason for concentrating on an all-inclusive or broadly-defined money measure is that narrow money can be changed by transfers between different kinds of money balances, while such transfers have no effect on aggregate demand. Large changes in the relative attractiveness of different types of money balance can lead to widely divergent rates of change between narrow money and broad money, and between different concepts of narrow money, where such divergences have no macroeconomic significance whatsoever.
To summarize, developments in the banking industry have huge relevance for the macroeconomic situation generally. The critical relationship is between, on the one hand, a broadly-defined measure of money – in practice dominated by bank deposits – and, on the other, the nominal values of national income and wealth. Roughly speaking, a particular % change in bank deposits is likely to be associated with a very similar % change in nominal national income and wealth over the medium and long runs. The key variable in macro analysis is the quantity of bank deposits (i.e., money) held by non-bank private sector agents. A large body of evidence – from virtually all countries over all long runs for which meaningful data exist – confirms the validity of this theory. The theory, which has clear microeconomic foundations, is valid despite many perplexities of interpretation and analysis, particularly in short runs (of, say, two or three years).
(Let me repeat myself here for the umpteenth time, that this relationship – the relationship between money and the equilibrium levels of income and wealth – turns on the stability of the money demand function. It has nothing to do with ‘credit’, in the multiple usages of that word. ‘Credit availability’, ‘credit spreads’, ‘companies’ ability to fund themselves from the markets’ et hoc genus omne are irrelevant to the relationship between money on the one hand and income and wealth on the other. The relationship between money and income would hold if banks’ assets consisted entirely of cash and government securities, and banks extended no credit whatsoever to the private sector.)
II. The Japanese approach to monetary policy
The Japanese authorities have recently announced a radical shift to expansionary monetary policy or, at any rate, to ‘expansionary monetary policy’, as they define it. In early April the newly-appointed governor of the Bank of Japan, Haruhiko Kuroda, announced that large-scale asset purchases would be intended to double the monetary base within less than two years. To quote from the Financial Times of 4th April, the Bank of Japan said
it would double Japan’s monetary base from Y135tn ($1.43tn) to Y270tn by December 2014, mainly by buying more long-term government bonds. That will raise the average remaining maturity of its holdings from about three years to seven years, keeping downward pressure on yields all along the curve.
Plainly, the Bank of Japan makes no reference here to the level of bank deposits or to ‘the quantity of money’ as conventionally understood. The objective is to avert deflation and indeed to have a positive (if very low) rate of inflation of about 2% a year. This is to be secured by stimulating the economy (meaning, giving a boost to private sector expenditure) by two means,
- The doubling of the monetary base, and
- A reduction in ‘the rate of interest’ (i.e., bond yields) all the way down the yen yield curve, with even seven-year yields to be lowered by central bank bond purchase
But I have just argued that it is the level of bank deposits which is the key variable linking the banking system and macroeconomic outcomes, not ‘the monetary base’ or bond yields. I am not going to spend any time on bond yields, other than to remark that in the 1970s and 1980s Japanese nominal GDP grew rapidly when bond yields were often well over 5%. Of course, a fall in bond yields relative to what they previously were has a stimulatory effect on the growth of aggregate demand relative to what would otherwise have occurred. But bonds are only one asset in private sector wealth portfolios, and – given the obstinate weakness of spending in Japan for many years despite very low interest rates – I am sceptical that much is to be gained by further downward pressure on bond yields. (In fact, I would expect bond yields to rise as and when the stimulus plan began to work, simply because investors would need to be compensated for the return to inflation.)
The heart of the Kuroda plan is therefore to double the monetary base within less than two years. Will this necessarily have the desired effect on the economy?
III. The meaning of the term ‘the monetary base’
The monetary base has two components,
– Notes and coin in the hands of the general public, and
– Bank’s cash holdings, which consist of
- so-called ‘vault cash’ or ‘till cash’ (i., notes and coin held ‘behind the counter’ and available to meet deposit withdrawals at the branch network) and
- cash reserves at the central bank (just a balance-sheet entry in principle, although I am told that in the UK the Bank of England’s vaults contain huge stockpiles of notes labelled according to the ownership of the settlement bank, which makes some sense, but is still rather astonishing).
Let us think about how the two components of the monetary base might be relevant to the macroeconomic scene.
First, notes and coin are used in retail transactions, often when the cost of using a cheque or a credit card is disproportionate because of the small consideration involved. The smallness of the average cash transaction needs to be highlighted. In general, notes and coin are unsuitable in large business or asset transactions, and such transactions are the overwhelming majority (over 99%) by value of those that take place in a modern post-industrial economy. The cost of counting and bundling up notes is substantial, while moving significant amounts of cash around increases the risk of theft and special security measures must sometimes be taken. By its intrinsic nature, the note circulation can never account for a high proportion of the payments in a modern economy. This is true in Japan, as in any other rich economy. Further, people adjust their note holdings to their spending plans by occasional cash withdrawals (a typical money-into-money transaction, with no direct effect on the income- expenditure circular flow). They do not adjust their spending plans to their note holdings. (In jargon, the note circulation is ‘endogenous’.)
In summary, we can ignore the general public’s note holdings in a macroeconomic analysis, on the grounds that they are de minimis relative to the big and important variables. In any event the size of the public’s holding is determined by other variables rather than having the ability to determine something else.
Second, what is to be said about banks’ cash reserves? It cannot be disputed that, in one sense, they are basic to the whole discussion. If banks cannot repay depositors with cash and meet obligations in inter-bank settlement, they have ceased to be ‘banks’ in the usually understood sense of the term. Of course, if they cannot repay cash because they are insolvent, a major crisis may erupt. During the Great Recession events in such countries as Iceland, Ireland and Cyprus demonstrated the risks all too clearly. However, in the normal course of events depositors are confident that banks can repay with cash and banks do meet their obligations, and the great bulk of banks’ customers do not bother themselves to check banks’ cash reserves and liquidity. (Apart from anything else, they assume that bank managements, central banks and bank regulators are doing their jobs properly.)
Let us then notice that banks’ cash reserves belong to banks. It was argued in the opening section that private sector expenditure is predominantly by non-banks. So banks’ cash reserves are not directly relevant to expenditure by non-banks. Indeed, monetary economists have understood this point for a very long time, so that banks’ cash reserves are excluded from the quantity of money, as usually defined. (Irving Fisher was clear about the matter in his 1913 classic, The Purchasing Power of Money.)
To summarize, of the two components of the monetary base, the note and coin circulation (i.e., the circulation outside the banks) accounts for such a tiny fraction of expenditure that it cannot have any significant bearing on macro outcomes, and banks’ cash reserves are not in fact normally deemed to be ‘money’ at all when they are held by banks rather than non-banks. More generally, the monetary base by itself is not of any great importance in the determination of national income and wealth, given modern monetary arrangements in which deposits and cash are regarded as interchangeable. (I am not disputing that – when deposits have become unsafe because of widespread bank insolvencies [as, for example, in Cyprus at the moment] – the size of agents’ holdings of monetary base becomes vital to the macro prospect. But such circumstances are very exceptional.)
IV. Japan and the monetary base
Let us now look at some evidence, with an examination of the Japanese data in the period since the bursting of the bubble of the late 1980s, a period in which the Japanese economy has been continuously in the doldrums with low output growth and occasional deflation. The chart above shows the growth rates of Japanese nominal GDP, and the monetary base and M3 in Japan, in the 18 years from 1994 to 2012. Two points are clear.
First, the year-by-year correlation between nominal national income and the money aggregates is poor. If an analyst were searching for a tight annual relationship, the temptation might be to chuck out the money numbers altogether.
Secondly, the year-by-year correlation is much better with the M3 aggregate than with the monetary base. Wild swings in the monetary base between 2001 and 2008 had no effect on the remarkably steady and sedate sideways movement of nominal national income. The swings in the monetary base were due to the Bank of Japan’s first experiment in ‘quantitative easing’, in the five years from March The Bank of Japan bought large quantities of securities from the banks, but it did not make similarly large purchases from non-banks or target a higher rate of growth of broad money. The purchases of securities from the banks caused massive surges in banks’ cash reserves, which – for much of the five-year period – were more than double their end-2000 level. But we have just explained that
- banks’ cash reserves are of course not in non-banks’ ownership and
- banks themselves account for only a small fraction of private expenditure, as the vast majority of such expenditure is by individuals and non-bank companies.
Not surprisingly, the leap in the monetary base was not reflected in nominal national income. From March 2006 the Bank of Japan reversed the purchases and indeed sold securities on a large scale to banks and others, causing the monetary base to contract. In 2006 the monetary base actually plunged by 20% because of these operations. Ironically, 2006 was in fact quite a good year for the Japanese economy. Evidently, over the 18 years surveyed here the base was useless in understanding the behaviour of Japanese demand and output.
In April 2013 the monetary base was just under 1.5 trillion yen (about $15b.), over a quarter of GDP. In early 2001 – ahead of the first QE experiment – it was about 0.8 trillion yen, less than a sixth of GDP. Kuroda’s planned doubling of the monetary base by the end of 2014 – Japan’s QE2 – is therefore an appreciably more dramatic effort than the previous one. However, QE1 turned out to be unsuccessful, arguably because it had no effect on the quantity of money, as usually understood. (By that I mean of course M3.) I am concerned that the Bank of Japan continues to express its intermediate target in terms of the monetary base rather than the quantity of money. In fact, I do wonder if the key officials in the Bank of Japan have any organized rationale for their approach.
V. Will Japanese banks respond to excess cash reserves by growing their assets and creating money?
I hope it is clear from what I have said so far that the monetary base is by itself of little direct importance to the spending behaviour of private sector non-banks. However, a common argument tends to be deployed at this point and, if it were always true, the monetary base would indirectly be of immense macroeconomic significance. The crux of the argument is that, just as non-banks have a well-specified and relatively stable demand function for money (i.e., bank deposits, mostly), banks have a well-specified and relatively stable demand function for monetary base assets (i.e., for their cash reserves). An extension of this claim is that banks respond to injections of cash reserves by expanding their assets, and actions to expand their assets are also actions to expand the deposit liabilities which constitute the bulk of the quantity of money. Indeed, in the simplest versions of the argument, an increase in the monetary base of x % is accompanied quickly and almost automatically by expansion of the quantity of money also by x %.
The story usually runs in terms of bank lending. Suppose a bank has cash reserves of 50b. yen and total assets of 1,000b. yen, and that it is comfortable with the implied cash reserve ratio of 5%. Suppose that the central bank wades into the money markets, buys securities from the banks in general and 50b. yen from this bank in particular, so that its total cash is 100b. yen and it has excess cash of 50b. yen. Its cash ratio is 10%, instead of the 5% with which it feels comfortable. The bank therefore lends a company 50b. yen, which results in an extra 50b. of loan assets, an extra 50b. of bank deposits and an asset total of 1,050b. yen. Obviously, the cash ratio is reduced from 10% (i.e., 100b. yen divided by 1,000b. yen) to just above 9.5% (i.e., 100b yen divided by 1,050b. yen). In general, the bank will keep on expanding its loan portfolio until total assets are 2,000b. yen, taking the cash ratio down once again to 5% (i.e., to 100b. yen divided by 2,000b. yen).
Some economists regard this expansion of bank credit as the most important element in the transmission mechanism of monetary policy. In other words, they are ‘creditists’, viewing bank credit to the private sector in its own right as a key variable in macroeconomic analysis. I have already debunked this notion in the opening section (as well as much other work) and am not going to discuss it further here. However, even for ‘monetarists’ (i..e., those who believe that ‘the quantity of money’ is the variable that matters) the process described in the last paragraph needs to be highlighted. The point is that, when banks extend new loans, they create new deposits. Until the Great Recession a large body of evidence could be cited for many countries that the ratio of banks’ cash reserves to M1 balances was stable over many medium- or long-run periods. (The data almost invariably failed to find relationship of similar quality between banks’ cash reserves and broad money.)
This fact led many economists to conclude that measures taken to control or manage the monetary base were tantamount to measures to control or manage the quantity of money. Further, because of a reasonably reliable relationship between ‘the quantity of money’ and nominal national income, operations to control the base ought to be sufficient to deliver control the growth of nominal GDP. By ‘the quantity of money’ these economists meant M1, since regulators typically only require cash reserves against sight deposit/current accounts. I hope the chart above will persuade the M1 enthusiasts to think again. In Japan M1 grew quite strongly in the 1990s, as non-interest-bearing sight deposits took a rising share of all deposits in a very low-interest-rate economy. M1 also surged during the QE1 experiment, with banks stepping up their expansion of some assets perhaps, but with the continuing shift to non-interest-bearing accounts in a virtually zero interest rate economy being more important. When the monetary base collapsed by 20% in 2006, M1 fell by only 1½%. At any rate, no relationship holds between changes in M1 and changes in nominal GDP.
VI. Conclusion: broad money is the monetary variable which plays a strategic role in macroeconomic analysis, but the Japanese authorities have not mentioned it
Over the 18 years to 2012 Japan had very low or zero short-term interest rates almost continuously, so that the Bank of Japan could not stimulate demand by reductions in interest rates. These were very unusual conditions for macroeconomic policy-makers. Many observers claimed that ‘monetary policy was ineffective’, alleging a breakdown in traditional relationships. It is certainly true that measures to expand the monetary base and, at a further remove, the M1 money aggregate did from time to time lead to rapid growth in these two monetary variables. However, the strong advances in the base and M1 had little bearing on the wider macroeconomic picture.
The table below shows the average annual growth rates of national income and the main monetary variables, and their standard deviations, in the 18 years to 2012. It is obvious that the average annual growth and standard deviations of the growth rates of nominal national income and M3 are similar, while the monetary base and M1 must be grouped separately. There is no room for rigorous substantiation of the argument, but the suggestion must be that the traditional relationship between money and national income did obtain, when ‘the quantity of money’ is understood to have been broadly-defined.
My further conclusion is that – to the extent that Japan’s QE2 does boost M3 – it will be effective in raising Japan’s equilibrium national income and wealth. But that – if QE2 does not impact on M3 – the recent surges in the stock market, and the related fall in the yen, will prove transient. I am underwhelmed, to say the least, by the Bank of Japan’s decision to express its target in terms of the monetary base instead of in terms of broad money. But that does not mean QE2 will have no effect on broad money. In particular, if the Bank of Japan finances asset purchases (such as long-dated government securities) from non-banks by means of borrowing from the commercial banking system, as is presumably intended, the annual rate of broad money growth could well move into the 5% – 10% area. That would lead – at least for a year or two – to a rip-roaring boom.
Japan’s QE2 is very new, and both the exchange rate and the stock market seem to have jumped the gun. In the month of April, when the expansionary operations were fully under way, M3 did not rise particularly quickly. But the numbers need to be watched month by month before dismissing QE2 as ineffective as Japan’s QE1.