Evidence from the years of the Great Recession justify renewed attention to a broadly defined concept of the quantity of money in central bank research
Controversy over the use of monetary aggregates undermined the impact of the monetarist counter-revolution of the 1970s and early 1980s. Top central bankers accepted Milton Friedman‘s dictum “money matters” was valid in some sense, but they were unsure exactly how and why it mattered. Practical application was elusive when their research departments produced data on half-a-dozen monetary aggregates. Which concept of money was of greatest importance – or at any rate of some relevance – to the determination of macroeconomic outcomes? Anthony Harris, one of the Financial Times’ leading commentators, compared the quarrel to that between the ‘Big-endians’ and ‘Little-endians’ about the best way to open a boiled egg in Jonathan Swift‘s Gulliver‘s Travels.
Some economists favoured ‘broad money’, which included all bank deposits and occasionally even included liquid assets that had arisen outside the banking system. Others supported ‘narrow money’ – generally taken to mean notes and coins in circulation plus sight deposits. The majority of monetary economists did not regard the monetary base (the liabilities of the central bank) as equivalent to ‘the quantity of money’ based on any definition. Instead, they believed the change in the base influenced the change in narrow money and hence affected expenditure at a further remove. However, some participants in the debate thought that the monetary base by itself, regardless of its role in banks’ creation of money, had a significant bearing on spending and the economy.
The purpose of this article is to propose that the monetary experiences of the Great Recession have settled the debate about the aggregates. At least, they ought to have settled it to every reasonable observer. In all of the prominent nations, only broad money has continued to hold a clear, roughly proportionate relationship with national income and output since 2008. Of course, this relationship is medium-term and rather imprecise in nature, but Friedman and his associates always contended that it worked best with large movements in the relevant variables. They did not claim exact short-term correspondence between money and income.
Two sets of facts, set out in two tables, are the basis for the proposition being made. Table 1 gives key information on broad money and national income for economies that traditionally have accounted for well over half of world output. In the US, the eurozone and the UK, the message is clear. Not only does a relationship hold in the long run between the rates of growth of broad money and nominal national output, but also this relationship has survived the turmoil and uncertainties of the Great Recession and its aftermath. Broad money has had a tendency to increase slightly faster in recent decades than income and output, with the gap typically being 1 % or 2% a year.
The explanation is probably to be sought in the increased efficiency and competitiveness of banking systems, which have led to banks offering interest on a wider range of their deposit liabilities. Bank deposits, the dominant constituent in broad money, have become higher-yielding on average and therefore more attractive to hold. After allowing for this pattern, the link between broad money and national income is robust and persistent.
A vital feature is that, over the five years to 2015, in all of the US, the eurozone member states and the UK, the rates of money growth have been the lowest since the 1930s, as have the rates of increase in nominal GDP. The point is easy to establish for the US and the UK, as fairly consistent sources of pertinent statistics are available. To prove the claim more precisely for the eurozone is more problematic. Strictly speaking, data needs to be prepared for each of the eurozone member states before the introduction of the single currency in 1999 and that data should then be compared with the post-1999 statistics.
However, there can be little doubt that in the immediate post-war decades, bank balance sheets – and hence broad money – were growing rapidly in such nations as Germany, France and Italy, as they enjoyed economic recoveries from wartime devastation. In the countries of single-currency Europe, too, it must be true that the last five years have seen the lowest increases since the 1930s in both the quantity of money, broadly defined, and nominal GpP. The association since the Great Recession between, on the one hand, low money growth and, on the other, low rises in nominal GDP, low inflation and even occasional deflation is evident in the major nations of North America and Europe.
Japan does not fit so simply into the jigsaw. The increase in nominal GDP in the five years to 2015, at an average annual rate of0.4%, is the most meagre in Table 1, as is the increase in broad money. But these are not the lowest numbers since the 1930s. In the five-year periods to all of 2001, 2002, 2003, 2004 and 2005 Japan’s nominal GDP actually fell, according to the International Financial Statistics database maintained by the International Monetary Fund.
Connecting the dots
Fortunately for the thesis of this article, these declines again had an obvious connection with the behaviour of money, since in the same period the rate of increase in broad money was under 0.5% a year. As far as Japan is concerned, it was the first five-year period in the new millennium – not the latest five-year period – that saw the lowest increases in both money and nominal GDP since the 1930s.
Table 2 and the subsequent discussion are related only to the US. This could be justified by the liveliness of the US debate on the money aggregates, but space constraints prohibit a survey of the data in every nation of interest. The table shows the increases in two measures of broad money (M3 and M2), narrow money (Ml), and the monetary base in the seven years to 2015 and in the preceding 49 years, and compares them with the increases in nominal GDP. One result is obvious. Whereas the growth of nominal GDP and broad money (on the M3 measure) collapsed in the period of the Great Recession and after, the growth rates of both narrow money and the monetary base were much higher in this period than they had been in preceding decades. In the Great Recession and afterwards, the rise in nominal GDP has decelerated, while that in narrow money has accelerated.
Indeed, in the seven years to 2015, the growth of narrow money and the base far exceeded the contemporaneous growth of nominal GDP. While the recital of figures does not itself prove anything, the sharp divergence between nominal GDP and the narrow money concepts (Ml and the base) is question-begging. On the face of it, something has gone wrong with narrow money. In the 1980s, when officialdom’s interest in monetarist ideas was at its height, narrow money tended to become more popular in central banks’ research exercises relative to broad money. A common argument was that narrow money ought to have a good connection with household transactions, particularly retail sales. But retail sales in the US have certainly not advanced at double-digit annual rates since the Great Recession hit in early 2008
Narrow money downside
One drawback of narrow money concepts has been compelling in recent years. Although sight deposits have the advantage of being available for spending without notice, they suffer from not paying interest. The general level of interest rates was at moderate levels until the Great Recession, with the Fed funds rate averaging 4% in the two years to mid-2008. The consequent loss of interest deterred people and companies from holding too much of their overall money balances in sight deposits. That kept down Ml relative to where it would have been with a zero Fed funds rate. But interest rates were slashed in late 2008, as the Fed responded to the onset of the Great Recession.
An effective zero Fed funds rate became a major financial landmark as oflate October 2008. With time deposits and wholesale deposits no longer providing a meaningful return, non-interest-bearing sight deposits increased in relative attraction. A big shift from time deposits into sight deposits then began, causing narrow money to expand far more quickly than broad money. Ml shot up by 18.1 % in the year to June 2009 and by 20.8% in the year to August 2011. But the explosions in Ml did not mean that spending and output would revive strongly and over the medium term, they had no relationship with nominal GDP.
Advocates of broad money deny that sight deposits are of unique importance to transactions. They insist that people and companies regard all money balances, including time deposits and even wholesale money, as available for any transaction with little inconvenience. Their view has surely become more plausible in the light of recent experience. After all, banks must help customers or risk losing business. On this basis, the split between sight and time deposits, and transfers between the two categories of deposit, have little importance for the economy. The argument justifies an emphasis on as comprehensive a measure of broad money as possible, since – by definition – a measure of money that includes every money balance cannot be altered by transfers between different balances.
A more long-term point emerges here. As noted earlier, since the 1950s, improvements in financial technology and the extra competition resulting from deregulation have encouraged banks to pay interest on a higher proportion of their liabilities. The lack of interest on sight deposits ought therefore to have caused Ml to decline in size relative to broader measures of money. That is exactly what has happened. The Fed’s money data series begin in 1959, when Ml was about $140 billion and nominal GDP was little more than $500 billion. With M2 and M3 similar in size at just under $300 billion, Ml accounted for almost half of broad money. In 2008, before the recent surge in Ml began, Ml was about $1,400 billion, barely 10% ofM3, which had almost reached $14,000 billion and modest compared with nominal GDP, which was approaching $15,000 billion
In short, Ml had dropped as a share of GDP from almost 30% in 1959 to less than 10% just before the Great Recession. In their 2009 book Animal spirits: how human psychology drives the economy, and why it matters for global capitalism (Princeton University Press), Nobel-prize-winning economists George Akerlof and Robert Shiller observed that demand and cheque-able deposits in Ml amounted to only $600 billion in 2008, which was “about 1 % ofnational wealth”. They warned: “How can it be that by managing the quantity of demand deposits, the Fed can fix all the problems? … It can’t.” The Akerlof-Shiller comment on Ml was intended as a rebuke for those monetarist economists who persisted in recommending the analysis of Ml in research work and the targeting of Ml in Fed policy-making.
What about the monetary base? It has two components, notes and coins in circulation with the public, and banks’ cash reserves, which are mostly in central bank accounts. Banks’ expenditure (on staff, premises and so on) is a tiny part of aggregate demand, and is in any case not related to their cash reserves. It has therefore been a long-standing international convention to exclude these cash reserves from officially recognised ‘quantity of money’ concepts.
In the Great Recession, it was the cash reserve element in the base that climbed dramatically in the US, as a by-product of the Fed’s expansionary open• market operations. The irrelevance of cash reserves by themselves to aggregate demand has been amply confirmed, since no connection can be observed between the base – which more than quadrupled between mid-2008 and late 2014 – and sluggish nominal GDP. The international convention has been vindicated. Some authorities have worried about the inflationary risks supposedly inherent in the Fed’s ‘money printing’, appealing to the textbook multiplier that links – or is claimed to link – the base to the overall size of bank balance sheets. But the increase in their base holdings has not prompted US banks to seek faster growth in their assets. As economist Charles Goodhart has demonstrated in influential papers, the base multiplier model of money supply determination has been discredited.
An important problem with both the monetary base (dominated by cash) and Ml is that nowadays few large US businesses or financial institutions keep sums in cash or sight deposits that are meaningful relative to their operations. Without doubt, much of the instability in the Great Recession reflected decisions taken by ‘big business’ and in the financial sector, but the virtual absence of sight deposits from their assets means that narrow money cannot have been an explanatory variable behind those decisions. In 2009, companies complained of shortfalls in revenue relative to plans, and squeezes on their cashflow and money balances. The money aggregate at work must have been broad money.
However, in the US context, the idea of broad money is ambiguous in two respects. First, in the past, two measures of broad money – M2 and M3 – were published by the Fed. M2 was Friedman’s favourite aggregate for most of his career, and data for it are still prepared and published by the Fed. But it now has a major weakness. The statisticians who put the numbers together have a tendency – arguably a very unfortunate tendency – to think that money exerts its influence on the economy because of its role in transactions, where ‘transactions’ and ‘retail spending’ are virtually synonymous. In 1959 – the first year of the Fed’s money series – deposits with a value of more than $100,000 were rare, and a sum of that size could not be readily spent ‘in the shops’. Deposits with a value of over $100,000 were therefore excluded from M2.
Of course, inflation and the growth of incomes since 1959 mean that deposits over $100,000 are now common. In fact, the bulk of bank accounts in the ownership of large companies and financial institutions must, individually, be worth more than $100,000. It follows that M2 no longer captures the money• holding behaviour of the corporate US. This failing ought to have caused the Fed’s statisticians to rethink their definitions and procedures, and to shift their focus towards M3. But that has not been their response. On the contrary, in a bizarre, move the Fed stopped publishing M3 in February 2006.
A case can be made that the old Fed M3 – composed of all banks’ deposit liabilities and also the money market funds that have long been non-bank financial institutions’ main form of liquidity – was the comprehensive money measure that supporters of broad money wanted, and that they still want. Several critics of the Fed’s 2006 decision have said that M3 served a vital analytical purpose and should be restored. In the event, a private sector research company, Shadow Government Statistics, has taken on the job of guesstimating M3 from publicly available information on the constituents of M3 that are not in M2. Shadow Government Statistics’ post-2006 figures are used in tables 1 and 2
The second ambiguity with broad money arises from a difference between the Fed’s practice and the international norm. Since the early years, International Monetary Fund member states have been obliged to send it data on both sides of their consolidated banking system’s balance sheet. (The consolidated system combines the central bank and the commercial banking system.) The thinking was that the data would help to identify the monetary origins of external payments imbalances, in line with an approach pioneered in the late 1940s by Edward Bernstein, the IMF’ s first director of research, and more particularly by his successor, Jacques Polak. The IMF’s data collection has over the decades created the world’s most important body of monetary information, embodied in the International Financial Statistics (IFS) database.
The IFS database has long-run series on broad money and its credit counterparts for scores of countries, including the US. Whether the Fed likes it or not, and regardless of its 2006 decision, an official US broad money number is available. Further, the Paris-based Organisation for Economic Co-operation and Development also includes broad money figures for its member states in its own publicly available economics database. For the US, it calls them ‘M3’! But there is a problem for analysts. The M3/broad money concept endorsed by the IMF and the OECD is not the same as the old Fed M3. The internationally favoured M3 relates specifically to the banking system, and therefore excludes money market mutual funds. Just before the Great Recession, these funds were central to the financial sector’s liquidity, and they are still important today, but their growth has stopped because of new regulations.
To conclude, the Great Recession provoked central banks into an abrupt cutting of interest rates in 2008, causing money-holders to radically increase the amounts in their non-interest-bearing accounts. That change, plus the multiplication of the monetary base because of quantitative easing and other stimulatory operations, shattered the links between narrow money concepts and nominal GDP that were thought to hold by some monetarist economists. By contrast, broad money and nominal GDP continued to move together over the medium term. In North America and Europe, the Great Recession and its aftermath saw the lowest rates of increases since the 1930s in both broad money and nominal GDP.
That parallelism of movement is striking. It surely goes a long way to revalidate a monetary approach to national income determination. Can a case be made that in the US, in particular, the Fed’s economists and statisticians need to align themselves more closely with international practice and other central bank research departments? If so, Fed economists need to reintroduce the measurement of M3 and analyse it in their macroeconomic prognoses. Let it immediately be conceded that the facts and figures in this note are far from being a complete treatment of vexed and problematic issues.
Nevertheless, the article states a position and marshals a worthwhile body of evidence to back it up. If money matters at all, then all money balances must matter in some way to the economy’s equilibrium. The events of the Great Recession justify renewed attention to a broadly defined concept of the quantity of money in central bank research