Beyond Friedman; Adding Asset Inflation and 52 Extra Years
Updated: May 1, 2020
This paper shows that in modern USA history asset (share price) inflation mostly responds to changes in monetary policy in the same direction as goods inflation. However, in certain important episodes, asset inflation booms while monetary policy is well controlled and goods inflation is low
. Recognising these ‘aberrant episodes’ has important implications for monetary policy, and explaining them satisfactorily provides a challenge for economic theory and macroeconomic analysis.
We update and extend the analysis of Friedman and Schwartz, A Monetary History of the United States, 1867 – 1960. We add share prices to the measures of inflation used by Friedman and Schwartz, then we extend the period of analysis to 2012, using a version of their historical analysis and simple statistical tests.
Readers familiar with the London underground will recall, perhaps fondly, the recorded message: 'Mind the gap'. We have long suspected that there is a gap in monetary history and analysis, a gap that is far more important than the gap between train and platform in the London underground. The gap in question concerns the role of asset prices in the analysis of the effects of monetary policy.
Milton Friedman famously said: 'Inflation is always and everywhere a monetary phenomenon' (Friedman, 1963, p. 17). Inflation in this context is goods inflation, or goods and services inflation. This was based primarily on the masterful historical analysis of Friedman and Schwartz (1963), but has been confirmed with statistical analysis of different levels of complexity and is now part of the canon of macroeconomics.
There has, however, been a long and we judge inconclusive debate about the interaction of asset prices and monetary policy. There has been resort to complex econometric techniques of different sorts and simulation analysis of simple models. But this work has not resolved the issue. We approach this question by repeating in summary form the analysis of Friedman and Schwartz including (as they did not) asset prices as represented by US share prices, specifically the S&P 500 index, with the advantage of five decades of additional data.
We must stress that this paper is fundamentally empirical, as we believe were Friedman and Schwartz. Our aims are to determine: what are the facts; and what does the data suggest about how to include asset inflation in existing models.
US share prices are more volatile than the monetary, price and real income variables used by Friedman and Schwartz. But all the major episodes, and most of the sub-periods, analysed by Friedman and Schwartz, show trends in both share price inflation and goods and services inflation directionally consistent with changes in monetary policy as represented by growth of the money stock.
But there are also cases when share prices are suppressed, such as in the first world war and the first part of the second world war, when goods and services prices rise much faster than suggested by their normal relationships with money growth. Conversely, when goods and services prices are suppressed, share prices rise especially rapidly, as they did in the second part of World Was II. The Roaring Twenties and the share booms of 1948 to 1960 and 1990 to 2000 are far more important examples in which money growth and wholesale inflation were both low but there were large booms in share prices. These 'aberrant episodes' (as we call them) are of especial interest and when appropriately allowed for will improve both theory and empirical analysis of how economies work.
In our penultimate section we suggest a theoretical reason for the aberrant episodes. In our concluding section we suggest how central banks should work with the 'alternative inflations hypothesis' given the current state of economic theory and the particular results of this paper.
II Monetary policy and Asset inflation : the Literature
Mayer (1998) explains why the USA experienced the Great Inflation from the middle 1960s. 'Inflation' in his account is solely goods and services inflation. This is surely still the common usage among economists, despite the recent upsurge of interest in asset inflation.
Akerloff and Shiller in their fine book Animal Spirits (2009) and Shiller in his earlier book Irrational Exuberence (2005) effectively say there is no systematic way to explain asset inflation. For example, the former book's chapter 11 opens as follows: 'No one has ever made rational sense of the wild gyrations in financial prices, such as stock prices'. (p 131).
A conference that is reported by Hunter, Kaufman and Pomerleano (2003) assembled considerable material on 'Asset Price Bubbles'. Key points include the following (quotes are from the editors' introduction) : it is difficult to identify asset price bubbles at the time; monetary policy is a 'blunt instrument' for responding to a narrow class of asset markets; however, asset markets contain 'at least some information that might be useful for policymakers'.
The conference included a paper by Charles Goodhart, who bases his contribution on several earlier papers, including by Goodhart and Delargy (1998). Goodhart (2003) summarises thus: 'business cycles in more normal periods, as occurred before 1939 and after 1985, have been driven mainly by fluctuations in private sector expenditures, notably in private sector investment and exports, and in associated fluctuations, frequently even more extreme, in asset prices, notably in equity and property prices' (p 470).
Charles Goodhart has also contributed to the question of whether and how (if necessary) to include asset prices in the measurement of inflation, building on the pioneering theoretical work of Alchian and Klein (1973). There is a strong theoretical and common sense case to do this, and thus potentially taking asset prices into account in setting monetary policy. But Goodhart (2001) is unable to find any asset price data worth including, although he concludes that housing prices are the best candidate. We observe that a central bank can (and should) allow for asset prices as at least subsidiary guides to monetary policy, despite the volatility of asset prices and incomplete or unreliable data. And, to anticipate the results, our analysis indeed makes a strong case for their use as a subsidiary indicator rather than incorporation as part of a single measure of 'inflation'.
The relationship between monetary policy and asset prices is investigated by Bordo and Olivier (2002a and 2002b) using a general boom-bust framework (ie not just discussing 'bubbles') and asking 'Does "benign neglect" make sense?' They conclude it does not, because whether or not intervention using monetary policy makes sense depends on 'economic conditions in a complex non-linear way.'(p 139). We agree with the spirit of this conclusion but for a simpler reason that asset inflation is related systematically to goods and services inflation and money growth but in a non-linear way, as we will later explain.
Bordo and Wheelock (2004) examine asset booms in USA history. They conclude that 'booms do not occur in the absence of increases in real economic growth and perhaps productivity growth. We find little indication that booms were caused by excessive growth of money or credit, though nineteenth century booms tended to occur during periods of monetary expansion. ... These questions leave unsettled the question of how monetary policy should respond to an asset price boom' (p 41).
It is hard to imagine a really rollicking asset boom occurring without financial accommodation, even if this is allowing buyers to do so on small deposits, or with buying on margin; which is a wonderful way to get rich while the boom goes on and to be ruined when the bubble bursts. Borio and Lowe (2002) argue that 'financial imbalances' - rapid growth of credit accompanied by rapid asset inflation - can and often does create financial instability and often crisis. This is a common theme in the literature on financial crisis, as emphasised by Minsky (1986) and his followers.
An important recent paper, by Schularick and Taylor (2010), makes a major contribution in several respects. Using data from fourteen countries, including the USA, over the period 1870 to 2008 they establish several important relationships. They establish that leverage in the financial sector increased greatly in the second half of the twentieth century, due to a 'decoupling' of money and credit aggregates and implying a decline in the quality of banks' balance sheets. Financial crises have remained important despite monetary policy responses being 'more aggressive' post-1945. Credit growth 'is a powerful predictor of financial crises', and 'policy-makers ignore credit at their peril'. (These points, including words in quotation marks, are all from this article's abstract.)
A different approach is to address the issue of whether asset prices should be taken into account in formulating monetary policy comes from simulation analysis of simple theoretical models, for example Gilchrist and Leahy (2002). These authors survey the relevant literature, which leads them to conclude that: 'whereas there are good theoretical grounds for including asset prices in monetary rules, in practice this adds little to stabilising output and inflation' (p 84). Gilchrist and Leahy then consider the impact of 'shocks whose main impact comes in the future' in three simple theoretical models. They conclude: 'We do not find a strong case for including asset prices in monetary policy rules'. (p 95). This supports the conclusions of Bernanke and Gertler (2001) which however differs from that of Cecchetti et al (2000). In our view simulation studies of simple theoretic models are incapable of resolving this issue, at least until essential stylized facts are allowed for.
White (2004) surveys most of the literature, including highly technical attempts to differentiate asset booms from bubbles and to explain the big runs (and subsequent busts) in share prices. 'The conclusion I draw from this comparison of the booms and busts of the 1920s and 1990s [and the related literature survey] is that the central bank should not respond to stock market booms'. (p 29).
There is a literature on the so-called 'Fed model', which seeks to explain 'the strong negative relationship between stock returns and inflation' as being due to 'stock returns anticipating future economic activity and inflation acting as a proxy for expected real activity'; with experience of stagflation important in establishing these relationships. (Fama (1981), cited by Bekaert and Angstrom (2010)). The latter authors claim to confirm these conjectures, (especially the importance of stagflation), with their VAR econometric model, and that this model holds for 'the 2008-2009 period'. We agree that the stagflation of the seventies was indeed important in supporting Friedman's 'single inflation' hypothesis, but note that there are many episodes in which goods and services and asset inflation respond similarly to the strength or weakness of money growth.
A different insight comes from looking at 'Bubbles and Crashes' from the perspective of Hayek and Harberler, in a paper by Bernard and Bisignano at a conference whose proceedings were edited by Kaufman (2001). These writers ask one central question: 'is financial instability inherent in an environment of low and stable inflation and liberalised domestic and international capital markets' (p 4). The facts discussed in this paper support a positive answer to this question.
An overview of the inter-war literature is provided by Laidler (2003). This paper includes the following prescient remark: 'And yet, just as things went wrong in the US in 1929, even though the [goods and services] price level seemed to be behaving well, so they went wrong in Japan at the beginning of the 1990s under similar circumstances, and again in the United States, and to a lesser degree in Europe too, in 2001. In each case an asset market bubble seems to have developed and collapsed without this event being preceded by any general upsurge of [goods and services] inflation'.
'With the passage of time, it has become tempting to treat the boom of the late 1920s that ended with the bursting of the stock market bubble in 1929 as an outlier, but the recent occurrence of two apparently similar episodes suggests that there are forces at work in the economy that are not properly integrated into those explanations of the cycle and doctrines about macroeconomic stabilization that derive from the quantity theory of money' (p 8).
The profession has so far failed to integrate these 'aberrant episodes' into the standard framework. In our analysis such episodes include, as well the cases noted by Laidler, US experience in the second world war and the share boom of the 1950s, which continued into the 1960s.
Integrating these 'aberrant episodes', plus the findings of Schularick and Taylor (2010), we believe will produce a major advance in macroeconomic modelling and in the analysis of monetary policy.
III The analysis
We have used the method of Friedman and Schwartz (1963), utilising the historical episodes, (or 'natural experiments' as Miron (1964) calls them) provided by US monetary experience from 1867 to 1960. We find that adding US share prices to the analysis and viewing all episodes or 'natural experiments' with fresh eyes yields new and rich hypotheses. These hypotheses survive the test of adding data from 1960 to 2012.
Table 1 (p 8 below) is the prime source of the data from twenty-four 'natural experiments'.
We begin, as do Friedman and Schwartz, with the famous 'Quantity Theory' identity. (1) Mv = Py Where M is the money stock, v is its velocity of circulation, P is an index of goods or goods and services prices and y is an index of real goods or real goods and services output.
In any complete theory of how the economy works, the equality of demand for and supply of money should be seen as a constraint to be obeyed in the long term and the interactions between all four of these variables would be spelt out, with allowance for lags, expectations, (both realised and unrealised), buffer stocks, different classes of goods and assets, and even policy reactions including policy interest rates, growth of the money stock and growth of credit. This has been the program of those who build econometric models (including one of the current authors) and it must be described as a program that generally has not produced widely agreed-upon results.
Friedman's famous statement quoted in the introduction is based on the view that, in any reasonable medium term, y and v are essentially constant, or growing at a constant rate (y) or changing in a predictable way (v). However, Friedman and Schwartz demonstrate impact effects of a change in monetary policy on real output, and there is plenty of evidence of direct money supply or interest rate effects on aggregate demand, and of demand effects on the course of inflation. Current macroeconomics is based on such responses.
Assuming in equilibrium the stock of money must equal the demand for money, a typical 'real' (inflation adjusted) demand function for money is often written as follows. (2) md = moyebr = M/P
Or, in growth terms: (3) Dlog M = Dlog P+ Dlog y + b.Dr + c
From an historical perspective, there are nineteen episodes analysed by Friedman and Schwartz (1963) (1867 - 1960), and when later data (1960 - 2012) is added twenty-four separate episodes, or 'natural experiments', far too few for standard statistical analysis. (1) Both P and y could be thought of as vectors, but for this paper we are going to resist trying to define functional forms that impose particular structures.
The data from 1867 to 2012, is divided into sub-periods using Friedman and Schwartz's timing for the period to 1960 and their preferred measures of money (M2) and one of their measures of goods inflation (the wholesale price index) (P1). Both of these series are updated to 2012 and with the S&P 500 Index of share prices (P2) added, we calculate averages for the entire period and for each 1867-1960 and 1960-2012. Theses averages are reported at the bottom of Table 1 below.
For the full period, the sum of both types of inflation is not far from equal to growth of the money stock. This allows in principle room for some positive GDP and real asset growth and some room for interest rate effects.
If the data is divided into the period analysed by Friedman and Schwartz and the inflationary years since then, the sum of inflations is less than the growth of money for the Friedman and Swartz period (allowing in a standard demand for money equation for positive real growth and interest rate effects) and more than the growth of money in the second period. The averages for each variable in the inflationary period are hard to reconcile with a standard money demand function with positive real effects and negative interest rate effects. We note however, that the twelve years from 2000 to 2012 contains a lot of volatility and that this unusual experience may be distorting the overall picture.
Table 1 also shows average annual rates of growth (strictly, averages of D log for each variable during segments within 1867 - 2012. We use a band around the average money growth for the full period to define 'tight', 'moderate' and 'easy' monetary policy. Specifically, 'tight' monetary policy we define as DlogM below 4; 'moderate' between 4 and 8 (inclusive); and 'easy' above 8. Note the exception in classification for 2000-12, which is explained below.
Our analysis is in two parts. The first part covers the same period as Friedman and Schwartz - that is 1867 to 1960. Data from this period we use to construct hypotheses, and we reach different conclusions to Friedman and Schwartz when share price inflation is included. Then we test the hypotheses on data from 1960 to 2012.
Despite generally higher inflation on average since 1960, the hypotheses from the earlier period are supported when data from the latter period are examined. (2)
We bequeath to others the task of comparing all hypotheses reported or developed here with later data, perhaps in 2050, in what will be a purer scientific test.
We also ignore two areas of analysis that Friedman and Schwartz spent a lot of time working on - the behaviour of 'velocity of circulation' and the sources of monetary growth. Velocity is ignored because results using it throw little light on the main subject and because if this concept is to be used in future work it is (in our view) desirable that it be analysed in a model such as that in equation 2 with both goods and non-monetary assets. Friedman and Schwartz's work on the sources of money is very sound and leaves the money stock as sufficiently exogenous (despite feedbacks that can be important, especially in major booms and busts) to allow us to proceed in the same way as they did.
IV. Historical analysis, extending Friedman and Schwartz.
Friedman and Schwartz divided US monetary history into seven major episodes, each consisting of at least two but often more sub-periods, with the exception of the time from 1948 to 1960. The divisions, based on differences in the state of monetary policy, are represented by growth in the money stock.
The ninety-three years covered by Friedman and Schwartz is broken down into nineteen sub-periods, and results for average annual growth of the stock of money, the wholesale price index and the share price index are summarised in Table 1, along with comparable results for the major sub-periods of the period from 1960 to 2012.
First we summarise the results of the nineteen sub-periods of the original analysis, then we cover the post-Friedman and Schwartz period in greater detail.
Easy money. There are five periods out of the nineteen analysed by Friedman and Schwartz when monetary policy was clearly easy, with double-digit growth of money in each case. Two are in times of peace, and in those episodes both goods inflation and share inflation are strong. The remaining three episodes are in times of war.
From 1914-19, money growth averaged 12.5 %, commodity inflation 13 % while share prices rose only slightly, presumably because of the deep uncertainties and horror of what was a brutal and slow-moving war.
From 1939 until 1945 money growth was also strong, averaging around 16 % taking the two segments together. But there was a vast difference in the behaviour of goods and share inflation between the segments. In the period to 1943, goods prices boomed and share prices fell. Then the US government imposed controls on various prices of goods and services and goods inflation rose at a relatively slow annual rate of 1.86 %, while share inflation averaged almost 20 % per annum. It is also relevant that in the second segment of World War II it became increasingly clear that the allies would win, restoring buoyancy to equity markets previously missing.
A lazy analyst might find it easy to dismiss the aberrant wartime experience as 'obviously irrelevant', but we beg to differ. In fact it suggests an amended hypothesis: In peacetime, easy money creates both goods inflation and asset inflation. But if extraneous forces (such as the effects of World War I and the first part of World War II) restrain share prices while money growth is strong, goods and services prices will rise faster than they would otherwise have done.
Conversely, in the second part of World War II when controls held down goods and services prices at a time of strong monetary growth, share price inflation became very strong once an allied victory became likely.
Tight money. Of the nineteen sub-periods analysed by Friedman and Schwartz, seven can be categorised as times of 'tight money' - with annual money growth below 4 %. In six of the seven sub-periods wholesale inflation is negative, and in five of these cases share prices either fell or rose only very slightly. The two-year period from 1867 saw rising share prices despite the general deflation, plausibly a result of post Civil War recovery. Just as the periods of easy money produced both strong goods inflation and share inflation except in times of war, tight money usually produced low or negative outcomes for both types of inflation. The case of the severe post-war deflation in 1920-21 is an extreme example, and we find it impossible to imagine any equilibrium money demand function represented by the numbers for that period.
The final period of tight money, from 1948 to 1960 is a period of post war reconstruction, extension of trade and capital flows, virtually a golden age of economic development. Money growth is low (below the 4 % rate), commodity inflation is even lower and share prices rose by an annual rate of over 10 %. In the historical period covered by Friedman and Schwartz it is most similar to that of the Roaring Twenties, when money growth was low at 4.5 %, commodity inflation virtually zero and share prices rose at an annual rate in excess of 13 %. We regard both of these episodes, plus Japan's experience in the 1980s already alluded to, as calling into question Friedman's 'single inflation hypothesis'.
We suggest that this effect might work in reverse to that of wartime suppression of one type of inflation: if extraneous effects (such as rapid technical innovation and generally strong growth, along with powerful 'animal spirits'), when restrained money growth inhibits goods and services inflation, share prices will rise far faster than would otherwise be expected.
This we call the 'alternative inflation' hypothesis, which should be contrasted with Friedman's 'single inflation' hypothesis.
Moderate money. There are also seven episodes of 'moderate¡' money in the analysis of Friedman and Schwartz. As it happens, money growth in each case is in the range of 4.5 % to just over 6 %. Given this similarity in money growth, the central tendency is somewhat puzzling - initially negative goods (wholesale) inflation (1868-73, 1885-91), rising to low positive goods (wholesale) inflation (1902-08, 1908-14), moderate (1933-39) and high (1945-48). The corresponding rates of asset inflation are all low positive except 1933-39 when share inflation averaged 8.5 %.
Putting aside the wartime goods inflations, the tendency of rising goods inflation, from deflation in 1868-73 to inflation from 1902-14, is presumably related to the global swing from deflation to inflation over the same period, related to the major new gold discoveries in the 1890s. Friedman and Schwartz were of course well aware of the effects of global inflation when the USA was on the gold standard, as it was from 1879. (See Friedman and Schwartz, Chapter three, especially pp 89 - 97.)
The interwar period contained the severe deflation of 1920-21, followed by the so-called Roaring Twenties, when goods (wholesale) inflation was virtually zero but average share inflation was over 13 %. With money growth 4.54 %, near the low end of the 'moderate' range, this episode is most similar to the period 1948 - 60, which also experienced low goods (wholesale) inflation and high share inflation.
Summing up for 1867 - 1960
Chapter 13 of A Monetary History of the United States 1867 to 1960 sums up. We agree with almost all of its conclusions except for the gap represented by omission of any systematic discussion of fluctuations in stock prices. When this data is added, rich new conclusions can be drawn.
We amend the authors' most general statement about their study as follows, with the amended words in bold italics.
'Of relationships revealed by our evidence, the closest are between, on the one hand, secular and cyclical movements in the stock of money and, on the other, corresponding movements in money income, in stock prices and in goods (wholesale) prices' (Friedman & Schwartz, (1963), p. 678, as amended.)
We further suggest, on the basis of discussion in Galbraith (1955) that monetary policy, while best represented by growth in the money stock in the long run, in some episodes must be buttressed by other measures. The clearest example of this concerns the market bubble at the end of the long boom in share prices from 1921 to 1929 when banks were borrowing from the Fed at low rates in order to lend at high rates of interest to buyers of shares on low margins.
But a more general conclusion concerns the influence of credit growth. Taylor and Shulerick (2010)show that in the post WWII world, bank credit grew faster than bank deposits (the major component of 'money') and financial deregulation and a tendency to lower prudential standards by bankers increased leverage. While this is outside the scope of the research reported here, it is clearly relevant to the overall conclusions about how economies work.
A more specific conclusion from the research summarised here concerns the relative effects of monetary growth on goods and services prices and asset prices, specifically share prices.
The onset of World War I initially reduced share prices despite the onset of rapid money growth. The unusually sharp rise in the stock of money was reflected in strong goods (wholesale) inflation, and growth of both money and goods inflation are well above the average growth of each variable for the whole period from 1867 to 1960. (Table 1). This was repeated at the start of World War II, then reversed, with share prices leaping when goods prices were constrained by policy dictat.
The so-called 'Roaring Twenties' shows another type of disconnect, one that more dramatically refutes Friedman's 'single inflation' theory. The stock of money was rising (at an average rate over 4 per cent per annum from 1922 to 1929) while commodity prices were on average almost steady. Share prices were rising, fitting Friedman's theory better than goods and services inflation. The fact is that share prices from 1923 to 1929 were rising far faster than in any previous episode analysed here. This is the most important example of our 'alternative inflation' hypothesis.
The post war period from 1948 to 1960 is another candidate for the 'alternative inflation' theory. During this extended period, money growth was even lower than that from 1922 - 1929, and share prices again rose much faster than goods (wholesale) prices. This episode ended without the great crash that began in late 1929, although the late 1960s and all of the 1970s saw adverse trends in all three variables.
The following chart shows the scatter diagram of the 24 points in Table 1.
Chart 1: Scatter diagram of money growth and twin inflations 1870 -2012
There is a moderate degree of correlation between of the two measures of inflation and a generally upward trend when (as in Chart 1) of both inflation measures in relation to money growth. The regression coefficients between the two measures of inflation and each of them in relation to Dlog M are shown in Chart 2.
When similar scatter diagrams are prepared for the different states of monetary policy (easy, tight or moderate) results are far less regular.
V A New Age of Inflation - beyond 1960.
Experience beyond the time considered by Friedman and Schwartz has included faster money growth, higher average goods (wholesale) inflation and higher average share inflation. We have chosen to break the five decades into five separate periods, though as we shall see there is a case to split the final sub-period, from 2000-12 into two or even three segments. Certainly with two episodes of near zero cash rates and, in the second case 'quantitative easing' and bailouts of major financial institutions, monetary policy became highly innovative, and in our view far more expansionary than indicated by average money growth.
The data in Table one shows only one post-1960 episode of 'easy money', 1970-80, as defined for this study. This was a clearly inflationary decade, somewhat analogous to World War I and the first part of World War II, with goods (wholesale) inflation close to money growth and share prices subdued. This was the decade of oil shocks with a deep recession in 1974-75. This was also the decade at the end of which the US Fed, led by Paul Volcker, called 'enough' to inflation and introduced a new operating policy for monetary policy that saw cash rates highly volatile and at times of the order of 20 %.
In terms of goods (wholesale) inflation, the inflationary sixties was a resumption of business as usual after the decidedly non-inflationary fifties. Economists generally ascribe the gradual build up of US goods inflation in the sixties to fiscal overreach, with both the Great Society and Vietnam war expenditure straining budgets and monetary policy generally accommodating.
Excess spending and war in the sixties, sharply rising goods (wholesale) inflation and successive oil price crises all plausibly restrained share price inflation by hammering investor confidence in the 1970s.
Following Paul Volcker's dramatic breaking of the general inflationary regime of the sixties and seventies, the next twenty years included the so-called 'Great Moderation'. Moderate money growth, low wholesale inflation, especially in the decade of the nineties went with powerful share price increases, averaging over 12 % in the 1990s and even higher in the last five years of that decade. The decade of the 1990s in the USA was, as in the 1920s and the 1950s, a time of innovation and high confidence, moderate money growth, low goods (wholesale) inflation and dramatic share inflation. We classify the 1920s, the 1950s and the 1990s in the USA 'aberrant' episodes' due to the conjunction of low goods (wholesale) inflation and high share inflation.
The 2000s were a nervous decade which included the American housing crisis that turned into the Global Financial Crisis, with its aftermath sometimes now labelled as the Great Recession. Students of history were entitled to worry that the surge in share prices in the second half of the 1990s, comparable to the rate of share price inflation in the nineteen-twenties, might again be followed by a great crash. But a cut in the Fed Funds rate to 1 % and a reasonably stable banking sector helped prevent such an outcome, as did the operation of fiscal 'automatic stabilisers' and discretionary fiscal expansion considerably stronger than tiny efforts in the late 1920s and 1930s.
For the twelve years as a whole, money stock grew at an annual rate of 7.7 % (above average), goods (wholesale) prices by 3.5% (about average) while share prices were virtually flat. However, these summary statistics are totally inadequate to convey the size of the shocks and of their impact at the time, or even as seen in retrospect.
The recession of 2001 was relatively mild, although the sharp drop in share prices was followed by a series of cuts to the Fed funds rate that was to reach a near-record low of 1 % in mid-2003, shortly after the share price index reached its bottom after a fall at the rate of 23% per annum. Fed Funds rate at 1 % was a policy innovation that in our view, with hindsight, should have worried people more than it did. The then Fed chief, Alan Greenspan, is widely credited with inventing the 'Greenspan put'. This involves not seeking to modify a share price boom but cleaning up (with large rate cuts) after the boom has broken. Critics, including the current authors, say the general belief that this was the Fed's operational rule encouraged people to speculate heavily in shares (and other assets) knowing that the downside would be protected, at least to some extent, by rate cuts.
Industrial production and wholesale prices had fallen slightly during the recession, and then began to grow moderately during the general economic recovery, recovery turned into boom, and share prices peaked in late 2007 just before the general downturn in the overall economy. Having risen at an annual rate of 13 % from early 2003 until the peak in late 2007, share prices, however, were not the full story of USA asset inflation. US house prices had boomed, and the boom was based on general optimism and lending by banks on the poor foundations of 'sub-prime' loans, outcomes copied in many other nations at that time. Having exhausted the appetite of middle class borrowers, mortgage lenders began to lend to people called Ninjas - with no income, no jobs and no assets (Skidelsky, 2009, p 5.).
But this almost unbelievable (to many) development was by no means the full story. Starting with British mortgage lender Northern Rock in September 2007, the bad loans began to cause real problems. The really adventurous banks, especially banks with investment banking arms (legal again in the USA after the repeal in 1999 of the 1933 Glass-Steagall Act) also got into trouble with complex derivative positions that no one understood, or which were based on excessively optimistic stress-testing, based on parameters from the years of 'moderation'. Bail-outs of major banks were deemed necessary but, in the case of Lehman Brothers, lack of a bailout caused the global interbank credit markets to freeze up. Share prices fell at an annual rate of 50 % from October 2007 to March 2009.
Mr Greenspan's put became Mr Bernanke's put when the Fed again began to cut the Fed funds rate, which was effectively zero by the end of 2008. Bank bailouts, 'quantitative easing' and commitments to continue very easy money (until conditions improved) further underpinned the Bernanke put.
The key point, in our view, about the 2000s is that it was another time of monetary innovation. Rather than Paul Volcker's dramatic monetary tightening, Alan Greenspan responded to the end of the share boom of the 1990s by reducing Fed funds rate to 1 %, a depth not previously plumbed. This restarted the share boom which ran hard until the onset of the Global Financial Crisis saw Ben Bernanke slash Fed Funds rate almost to zero, as well as introducing 'Quantitative easing' and bailing out a number of major financial institutions.
Monetary policy in this troubled decade - with growth of the money stock about average for the period since 1960 - was far easier than suggested by growth of money. Indeed, the operational focus of monetary policy was on 'Taylor Rules' that assigned weights to goods and services inflation and the output gap or the rate of unemployment. But in responding to the share market falls and other indications of serious trouble in both the early 2000s and in 2007 and 2008, the Fed Funds rate was slashed well below the rates indicated by Taylor Rules devised in calmer times. This is why we classify in Table 1 the 2000-2012 period as one of 'easy' money, but also 'aberrant'.
We observe that a simple diagram (drawn in the air or on a whiteboard) with demand for and supply of money, combined with a near zero 'Chairman's discretion' cash rate illustrates the disequilibrium in monetary policy in much of the period from 2000 to 2012 and into 2013. A recent paper on 'Credit crises, money and contractions' by Bordo and Haubrich (2010) concludes as follows: 'The current episode combines elements of a credit crunch, asset price bust and banking crisis. A large drop in share prices, and large rise in relative yields of commercial loans co-existed with strong growth of the money supply'. Clearly, for this episode, as in the 1920s, money growth alone is not a fully descriptive measure of monetary policy.
It is far too soon to evaluate these monetary policy innovations, except to note one further point. Simply inspecting the history of major booms and busts from the time of the Tulip Mania of 1636 provides a strong suggestion of shorter times between episodes. In addition, the swings in share prices from the early 2000s provides an impression of larger and more frequent swings than in all of the previous history examined in this paper. The volatility of global financial markets as the US Fed begins the dialogue of how to exit from current very easy monetary policy shows the dangers faced by all participants in the modern, globalised economy.
One of the current authors wrote in late 2010: 'The biggest threat to modern capitalism is the possibility of instability caused by policy swings: expansion; recovery; asset inflation; goods inflation; policy tightens; economy falls back; recession starting the whole process anew'. (Jonson (2011), p 285.)
'So far, so bad' is the progress report.
VI Implications for economic analysis.
There are three analytic reasons to take seriously the conclusions of this paper.
The first is that strong money growth usually means both strong commodity inflation and strong asset inflation, and the converse occurs when money growth is very weak or negative. These extreme cases show inflation or deflation mostly affects both asset and commodity price inflation in the same direction. Indeed, the regression coefficient between money growth and share inflation (Chart 2) is slightly lower than that between money growth and goods (wholesale) inflation, as Chart two illustrates.
Chart 2: Simple regression results: sub-period averages 1870 - 2012
The second point qualifies the single inflation hypothesis in a different way. During World War I, share prices were subdued throughout (despite ebbs and flows loosely correlated with the success of the war from an Allied perspective). Strong money growth went with strong commodity inflation, with wholesale inflation eight times its long-term average and money growth only two-and-a-half times its long-term average.
In World War II the initial experience (1939 to 1943) was very strong money growth, strong goods and services inflation and slightly negative share price performance, the latter result presumably influenced by negative confidence due to the state of the war. Then from 1943 until 1945 money growth increased to almost three times its long-term average, and share price inflation was almost 20 per cent - almost five times its average growth - presumably because confidence about the outcome of the war was high. In this case there was also the mechanical effect of strong money growth while goods and services inflation was suppressed by wartime price controls.
The third and most important point is that our 'alternative inflation' theory fits the facts better than Friedman's single inflation hypothesis because of three puzzling but vitally important 'aberrant episodes' previously ignored by most economists. In US history there are three key examples - the share boom of the 1920s, the period from 1948 to the middle 196os, especially 1949 to 1956, and the two decades from 1980 to 2000, and especially 1994 to 2000. In the first and third of these cases, monetary policy, as represented by growth of the stock of money, was moderate but not tight. In the middle case, money growth was below the bottom of the range we call 'moderate', and in all three cases commodity inflation was subdued while share prices rose powerfully.
The lesson of these three episodes is that messages from asset markets are vitally important, and should engage economists and central bankers far more systematically than they have.
Clearly some plausible theory needs to be brought to bear if these episodes are to be incorporated into macroeconomic theory and policy modelling. We suggest that allowing for the state of confidence, Keynes' 'Animal Spirits', would be a good way to start. Powerful confidence seems to have been an important part of the story of the USA in the 1920s, the 1950s and the 1990s, in each case associated with powerful innovation and strong GDP growth.
Shiller (2005), whose famous book is called Irrational Exuberance, recognises the same three great American bull markets that we do, but his most explicit discussion of their causes points out that 'These examples show that earnings growth and [share] price growth do not correspond well at all'. (Pp 184 and 185) His whole book is designed to show there is no systematic connection between any recognised economic measure with the long swings in share prices. As already noted, his 2005 book with George Akerlof begins its Chapter 11 by asserting: 'No one has ever made rational sense of the wild gyrations in financial prices, such as stock prices'. (P 131). With appropriate modesty, we claim to have at least clarified key facts.
Powerful confidence might be expected to boost demands for both money and shares. Increased demand for money would presumably mean lower goods inflation for any given rate of money growth, and contained money growth would itself boost confidence. Powerful confidence would presumably make bankers more inclined to advance credit, and as Schularick and Taylor (2010) show so convincingly, strong credit growth is a vital part of the generation of asset booms.
We are aware of no systematic measure of the confidence of investors that stretches back to post-Civil War America. When questioned about this matter, share brokers usually comment that the best measure of confidence is the behaviour of share prices, which is of course circular if used to explain strong share booms.
The Australian historian, Geoffrey Blainey (1988) provides a systematic and persuasive analysis of the effects of the 'Great Seesaw' of swings in confidence in the history of the Western world from 1750. Kindelberger and Aliber (2005) say: 'Asset price bubbles - at least the large ones - are almost always associated with economic euphoria'. (p 115). Galbraith (1955) documents in telling detail the progress of euphoria during the boom of 1922 - 29, and especially in 1928 and 1929.
Daniel Yergin's wonderful 1991 book, The Prize, summarises in just a few pages the dramatic post-WWII development of America. 'The inexorable flow of [cheap] oil transformed everything in its path'. (p 532). Suburbanization gathered pace requiring widespread ownership of cars. Great highways were built, drive in restaurants and movie theatres were built, cars added fins and other symbols of innovation and success. (Pp 530 - 536).
Ben Bernanke (2013) says 'low and stable inflation over a long period supports healthy growth and productivity and economic activity'. (p 40) Then in discussing the so-called Great Moderation, he remarks: 'With twenty years of relatively calm economic and financial conditions, people became more confident, willing to take on more debt'.
These contributions, and others like them, especially Akerlof & Shiller (2009), provide useful hints for modelling investor confidence, and Bernanke's general statement suggests a way to relate low and stable goods inflation with the genesis of asset inflation.
Measurement or modelling of 'investor confidence' in a convincing way is the next step in overcoming the skepticism of the economics profession about the causes of long swings of share prices and its relation to monetary policy.
The aberrant experiences deserve emphasis because all three of the relevant episodes analysed in this paper ended badly, in the latter case with continued uncertainty more than five years after the start of the crisis itself. As previously noted, Japan's experience in the 1980s, with a massive asset bubble while monetary policy seemed to be restraining goods and services inflation, (also followed by a massive asset bust), fits this 'aberrant' scenario well, and was followed by two decades of poor economic performance. The build-up to the Global Financial Crisis also involved great confidence, especially the almost manic confidence of the bankers of Wall Street and the City of London.
Apart from further efforts to quantify 'confidence', or 'Animal Spirits', there are several obvious extensions of this work - to include other assets, especially real estate - and to investigate more systematically alternative or subsidiary measures of the impact of monetary policy.
Application to smaller economies is not straight-forward because, even if they have flexible exchange rates, small country share prices are influenced directly by swings in Wall Street, as well as by in-country monetary policy, real growth, goods inflation and other domestic factors.
Despite the difficulties of wars and depression, the regularity of the relationships exhibited here should provide fresh impetus to modelling of interactions between share prices and other key economic variables.
VII Implications for monetary policy.
'Inflation is always and everywhere a monetary phenomenon' is a good guide to monetary (and macroprudential) policy so long as asset inflation is included. Goods and services inflation is the single best indication of 'inflation', measuring as it does the broadest set of economic prices, and should therefore remain as the principal guide to monetary policy action by central bankers. But there are two other indicators that must be closely monitored, and one absolutely vital implication.
When central banks make their decisions about monetary policy, the secondary indications should include asset inflation, growth of the money stock and relevant measures of credit. Indeed, the conjunction of low but positive money growth (normally indicating firm monetary policy), low goods inflation and strong asset inflation is a vital warning sign. Beyond this vital point is the need to monitor carefully all of the massive amount of financial and non-financial data, and liaison information that central banks collect in order to create a consistent picture of the overall economic situation and prospects. The bottom line for analysis of monetary policy is a clear economic understanding and narrative, not any single indicator such a growth of some measure of goods or goods and services inflation. Some, indeed most, central bankers regard these views as self-evident, but in our opinion measures of (goods and services) inflation often get overemphasised relative to constructing and presenting the broader narrative.
Tightening monetary policy, as in 1928 and 1929, on top of a general expectation that markets are over-valued, will bring any asset boom to an end. But, as Friedman and Schwartz say, there are 'great difficulties in seeking to make [monetary] policy serve two masters' ... and 'the Board [in the 1920s] should not have made itself an "arbiter of security speculation or values" ' (p. 291). In our view, policy-makers need to have a clear way to modify asset booms when the evidence suggests the strong risk of such a boom getting out of control. Ironically, the evidence suggests that firm monetary policy is likely to promote asset inflation, which is a powerful reason not to rely on monetary policy to control asset inflation. The literature generally has missed this point, which has resulted in many wasted and unproductive articles on the subject of this paper.
We hasten to assert that the aim should not be to strangle a boom for its own sake, since much that is constructive occurs in the boom times. Rather the aim should be to control the asset boom so that it is not based on 'irrational exuberance', or excessive credit growth, especially not credit based on lending with narrow asset backing or highly risky credit standards.
Successive Fed Chairmen have said that knowing when an asset boom is likely to cause trouble is very hard or even impossible in practice. In the light of the evidence discussed in this paper, and by Borio and Lowe (2002), Richards and Robinson, (Eds), 2003), Bordo and Wheelock (2004), Reinhart and Rogoff (2009) and especially Schularick and Taylor (2010), we beg to differ. More effort devoted to asset markets, especially share markets, with systematic attempts to relate their behaviour to monetary policy, general economic conditions and the state of confidence ('Animal Spirits'), will be repaid many times over.
This effort is best undertaken by central bankers, investors and economists interested in market dynamics. The challenge of responding more systematically to the messages from asset markets will best be handled by central bankers, either alone or working with specialised financial regulators where this function has been removed from central banks under the general rubric of 'macro-prudential policy'. We note in particular the important insight of Borio and Lowe, that: 'Currently, the primary focus [of prudential authorities] is on preventing the failure of individual financial institutions, ... and there is a tendency to treat macroeconomic risks as exogenous with respect to the institutions' behaviour'. (2002, p 25.)
Ben Bernanke, in his recently published book The Federal Reserve and the Financial Crisis (2013) seems to us to be moving toward the view that what he calls 'Regulatory Policy' is the appropriate tool, independent of monetary policy, to control asset inflation and the overall stability of the financial system.
Monetary policy should aim to keep goods and services inflation low and predictable. Normally that will occur with low and stable money growth and official cash rates modestly positive, but we emphasise that at present there are unresolved issues in defining the stance of monetary policy.
The great amount of discussion of Regulatory policy in Mr Bernanke's final chapter shows there has been a lot of debate and various changes, in the USA as elsewhere. In our view, however, what Regulatory policy needs is some rule equivalent to the 'inflation targets' designated as a key indicator of monetary policy. Rules that increase prudential ratios when growth of asset inflation exceeds some predetermined rate based on historical experience might be worth considering. Asset inflation in excess of, say, 10 % per annum for more than two or three years might be such an trigger and trigger for action by regulators, but we leave this to experts.
And, perhaps most difficult of all, is to analyse the causes of the excessive greed and insufficient prudence amply demonstrated by financiers and their clients during the early twenty-first century; and to devise remedies. Removing the one way bets with improved regulatory methods will be one important step, necessary but far from sufficient.
1. For spirited discussions of heavy duty ‘single equation’ econometrics verses historical analysis and simpler regression analysis plus ‘natural experiments’, see Friedman and Schwartz (1991) and Miron (1994). Jonson (1975) provides a discussion of attempts to use ‘full system’ model building with money as a buffer stock with the assumption of economies in continuous disequilibrium but adjusting to a long-term equilibrium.
2. Sources of data and the data itself will be provided on request once this article is published. For M2 and wholesale prices (hereafter goods inflation) we use data that is the same as that of Friedman and Schwartz, updated to 2012. Share price data is the S&P 500 Composite Index. We use the National Bureau of Economic Research (NBER) timing on periods of general expansion or contraction. Note that the NBER timing on the start and end of some business ‘cycles’ appears slightly different to those dates used by Friedman and Schwartz due to revisions in business cycle dating decisions, but we judge the differences have no material influence. A set of charts similar to those of Friedman and Schwartz is available on request.
Akerlof, George A. & Shiller, Robert J., Animal Spirits. How Human Psychology Drives the Economy and Why it Matters for Global Capitalism, Princeton University Press, Princeton and Oxford, 2009.
Alchian, A. A. And Klein, B., 'On a correct measure of inflation', Journal of Money, Credit and Banking, February 1971, 5(1), pt. 1, 173-91.
Bank for International Settlements, (2003), 'Monetary stability, financial stability and the business cycle: five views', BIS Papers, 18, September 2003.
Bean, C. (2003), 'Asset prices, financial imbalances and monetary policy: are inflation targets enough?', BIS Working Papers, 140, September 2003.
Bean, C. (2010), 'Joseph Schumpeter Lecture: The Great Moderation, The Great Panic, and The Great Contraction'. Journal of the European Economic Association, 8(2©\3), 289-325.
Bekaert, G. and Engstrom, E. (2010), 'Inflation and the stock market: Understanding the "Fed Model" ',Journal of Monetary Economics, 57(3), 278-294.
Bernanke, B. S. (1983),'Non-monetary Effects of the Financial Crisis in the Propagation of the Great Depression', NBER Working Paper,1054, January 1983.
Bernanke, B. S. (2004), 'The Great Moderation, Remarks at the meetings of the Eastern Economic Association', Washington, DC, February 20, 2004.
Bernanke, B. S. (2013), The Federal Reserve and the Financial Crisis. Lectures by Ben Bernanke, Princeton University Press, 2013
Bernanke, B. S. and Gertler, M. (2001), 'Should central banks respond to Movements in Asset Prices?',American Economic Review, May 2001
Bernard, H. and Bisignano, J. (2001), 'Bubbles and Crashes: Hayek and Harberler Revisited', in Kaufman, George G., Asset Price Bubbles: Implications for Monetary and Regulatory Policies, Elsevier Science.
Blainey, Geoffrey, The Great Seesaw. A New View of the Western World, 1750 ¨C 2000, MacMillan, Australia, 1988.
Board of Governors of the Federal Reserve System (2013), Currency Component of M1 Plus Demand Deposits, 1960-2012, Series ID CURRDD, Board of Governors of the Federal Reserve System. Accessed 01 February 2013<http://research.stlouisfed.org/fred2/series/CURRDD?cid=25>.
Board of Governors of the Federal Reserve System (2013), Large Time Deposits at Commercial Banks, 1960-2012, Series ID LGTDCBSL, Board of Governors of the Federal Reserve System. Accessed 01 February 2013 <http://research.stlouisfed.org/fred2/series/LGTDCBSL?cid=28>.
Board of Governors of the Federal Reserve System (2013),Savings and Small Time Deposits at Commercial Banks, 1960-2012, Series ID SVSTCBSL, Board of Governors of the Federal Reserve System. Accessed 01 February 2013 <http://research.stlouisfed.org/fred2/series/SVSTCBSL>.
Bordo, M. D. and Haubrich, J. G.(2010), 'Credit crises, money and contractions: An historical view', Journal of Monetary Economics, 57, 1-18.
Bordo, M. D. and Olivier, J. (2002a), 'Boom-busts in Asset prices, Economic Instability and Monetary Policy', NBER Working Paper no. 9866, May 2002
Bordo, M. D. and Olivier, J. (2002b), 'Monetary Policy and Asset Prices: Does "Benign Neglect" Make Sense', International Finance, 5(2), 139-164
Bordo, M.D. and Wheelock, D. C. (2004), 'Monetary policy and asset prices: a look back at past US stock market booms', NBER Working Paper, 10704.
Borio, C. and Lowe, P. (2002), 'Asset prices, financial and monetary stability: exploring the nexus', BIS Working Papers,114, July 2002.
Bureau of Labor Statistics (2012), Producer Price Index- Commodities, 1914-2012, Series ID WPU00000000, Bureau of Labor Statistics. Accessed 01 February 2013<http://bls.gov/ppi/#data>
Cecchetti, S. B., Genberg, H., Lipsky, J., and Wadhwami, S. (2000), 'Asset Prices and Central Bank Policy', Geneva Reports on the World Economy 2, International Center for Monetary and Banking Studies, Geneva.
Clarida, R., Gali, J., and Gertler, M. (2000), 'Monetary Policy Rules and Macroeconomic Stability: Evidence and some Theory', Quarterly Journal of Economics, 115(1), 147-180.
Fama, E. F. (1981), 'Stock returns, real activity, inflation and money', The American Economic Review, 1981, 71, 545 ¨C 565.
Federal Reserve Bank of St. Louis (2013), NBER based Recession Indicators for the United States from the Period following the Peak through the Trough, 1867-2012, Series ID USREC, Federal Reserve Bank of St. Louis. Accessed 01 February 2013<http://research.stlouisfed.org/fred2/series/USREC>.
Friedman, M. (1963), 'Inflation: Causes and Consequences'. Asia Publishing House.
Friedman, M. (1968), 'The Role of Monetary Policy', American Economic Review, March 1968.
Friedman, M. and Schwartz, A. J. (1991), 'Alternative approaches to analyzing economic data'. American Economic Review, 81, 39-49.
Friedman, M.and Schwartz, A. J. (1993), A Monetary History of the United States, 1867-1960. Princeton University Press. (Original work published 1963)
Galbraith, J. K. (2009), The Great Crash 1929, Marriner Books. (Originally published 1955.)
Gilchrist, S. and Leahy, J. V. (2002), 'Monetary policy and asset prices', Journal of Monetary Economics, 2002, 49, 75-97.
Goodhart, C. (2001), 'What weight should be given to Asset Prices and the measurement of Inflation?' The Economic Journal, June 2001, 335-355.
Goodhart, C. and Hofmann, B. (2000), 'Do Asset Prices Help Predict Consumer Price Inflation?' ,Manchester School, 68 (supplements), 122-40.
Hendry, D. and Ericsson, N. R. (1991), 'An Econometric Analysis of U. K. Money Demand in Monetary Trends in the United States and the United Kingdom, by Milton Friedman and Anna J. Schwartz', American Economic Review, 81, 3-38.
Hunter, W., Kaufman, G. And Pomerleano, M., (Eds), Asset Price Bubbles: The Implications for Monetary, Regulatory and International Policies, Cambridge, MIT Press, 2002.
Jonson, P. D. (1975), 'Money, Prices and Output: an Integrative Essay', Kredit und Kapital, 1975, Reprinted in D. E. W. Laidler(ed), Foundations of Monetary Economics, Volume II, 1999.
Jonson, P. D. (2011), Great Crises of Capitalism, Connor Court, 2013.
Kaufman, G. G. (Ed.) (2001), Asset Price Bubbles: Implications for Monetary and Regulatory Policies, Elsevier Science, 2001.
Laidler, D. (2003), 'The price level, relative prices, and economic stability: aspects of the inter-war debate', BIS Working Papers, 114, September 2003.
Mayer, T. (1982), 'Monetary Trends in the United States and the United Kingdom: A Review Article', Journal of Economic Literature, 20(4), 1528-1539.
Mayer, T. (1998), Monetary Policy and the Great Inflation in the United States: The Federal Reserve and the Failure of Macroeconomic Policy, 1965 - 79, Edward Elgar.
Minsky, H. P. (2008). Stabilizing an unstable economy. New York: McGraw-Hill. (Original work published 1986).
Miron, J. A. (1994), 'Empirical methodology in macroeconomics. Explaining the success of Friedman and Schwartz's "A Monetary History of the United States, 1867 - 1960", Journal of Monetary Economics, 34, 17-25.
National Bureau of Economic Research (2012), Index of Wholesale Prices for United States, 1890-1914, Series ID M0448BUSM336NNBR, National Bureau of Economic Research. Accessed 01 February 2013 <http://www.nber.org/databases/macrohistory/contents/chapter04.html.>.
Nelson, E. (2003), 'The future of monetary aggregates in monetary policy analysis', Journal of Monetary Economics, 50, 1029 - 1059.
Reinhart, C. M. and Rogoff, K. S. (2008), 'Is the US subprime financial crisis so different? An international historical comparison', American Economic Review, 98(2), 339-344.
Reinhart, C. M. and Rogoff, K. S. (2009), 'The aftermath of financial crises', NBER Working Paper,No. w14656,National Bureau of Economic Research.
Reinhart, C. M. and Rogoff, K. S. (2009), This Time is Different. Eight Centuries of Financial Folly, Princeton University Press.
Richards, A. and Robinson, T. (2003), Asset Prices and Monetary Policy, Proceedings of a Conference, Reserve Bank of Australia.
Schularick, Moritz and Taylor, Alan M., 'Credit Booms Gone Bust: Monetary Policy, Leverage Cycles and Financial Crises, 1870 - 2008', NBER Working Paper 15512, February 2012.
Schwartz, A. J. (2002), 'Asset Price Inflation and Monetary Policy', NBER Working Paper, No. 9321, National Bureau of Economic Research.
Shiller, Robert J. (2005), Irrational Exurberence, Second Edition, Broadway books.
Shlaes, A. (2007), The Forgotten Man. A New History of the Great Depression, Harper Perennial.
Skidelsky, R. (2009), Keynes: The Return of the Master, Allen Lane.
White, E. N. (2004), 'Bubbles and Busts: The 1990s in the Mirror of the 1920s', Paper presented at Duke University and the University of North Carolina conference, Understanding the 1990s: The Long-run Perspective, Durham, North Carolina, March 26-27, 2004.