There are those who maintain that central banks will survive so long as people prefer the anonymity of cash to traceable e-money; so long as they need banks to help them distinguish between good and bad credit risks when disposing of their assets; and so long as governments wish to risk taxpayers’ money in trying to control short-term interest rates.136 On the other hand, it has long been recognized that central banks could be dispensed with.137 Indeed, there have been past experiments with ‘free banking’: the United States in the nineteenth century, for example. It is far from self-evident that this did not work. True, the Federal Reserve System was set up after the 1907 financial crisis in the belief that having a lender of last resort would increase the stability of the American financial system. Yet it is worth remembering that, as we have seen, the far worse financial crisis which devastated the American economy in the years after 1929 had a great deal to do with the way the Fed misused its powers. It is at least arguable that if American monetary policy had not been under the Fed’s control, the Great Depression would not have been so severe – and not only in the United States.
To pursue such arguments further, however, it is necessary to turn our attention to a concept that has so far been deliberately left out of account: the rate of interest. The curtain accordingly falls on Goethe’s Faust and rises on Shakespeare’s Shylock.
6
Of Interest
I don’t believe in princerple
But oh, I du in interest.
JAMES RUSSELL LOWELL
In The Merchant of Venice, we never learn at what rate Shylock might have been willing to lend Bassanio three thousand ducats for three months, before the malicious thought occurs to him to lend the money on the security of a pound of Antonio’s flesh. An educated guess would be around 10 per cent.
In the sixteenth century interest rates in Italian commercial centres fell substantially. In the first quarter of the century the interest paid on the forced loans of the city-state of Venice ranged between 6.75 and 9.62 per cent. By the end of the century, when Shakespeare was writing, rates in Genoa (for which we have better records) were as low as 1.88–4.38.1 On the other hand, that was the rate of discount on the declared dividends of the Bank of St George, a semi-public institution with an impeccable reputation; whereas Bassanio wanted to borrow from Shylock on the strength of his merchant friend Antonio’s business. Antonio himself may have been confident that ‘within two months, that’s a month before / This bond expires, I do expect return / Of thrice three times the value of this bond’. But Shylock had every reason to be sceptical:
Yet his means are in supposition: he hath an argosy bound to Tripolis, another to the Indies; I understand, moreover, upon the Rialto, he hath a third at Mexico, a fourth for England, and other ventures he hath, squandered abroad. But ships are but boards, sailors but men: there be land-rats and water-rats, water-thieves and land-thieves, I mean pirates, and then there is the peril of waters, winds and rocks.2
To ask that Antonio pledge a pound of his own flesh – in effect, his life – to guarantee the debt was perhaps to demand an excessive risk premium. Shylock was nevertheless right to recognize that lending on the security of Antonio’s ships was a very different proposition from lending to the Venetian state or the Genoese bank.
YIELDS
This chapter is concerned with rates of interest, and particularly the rates paid by states when they borrow, in the first instance from their own citizens. Largely omitted from the discussion are the rates merchants like Antonio have had to pay for credit through the ages, though it is important to be aware that as early as the sixteenth century a differential had begun to emerge between the rate that a financially well-established state could expect to pay and the rate on commercial bills or bonds. Here the interest rate – usually the yield on a government’s bonds – is of interest because it is the crucial determinant of the cost of government borrowing.
For the sake of uninitiated readers, a few words of explanation may be in order. The ‘yield’ an investor receives from a government bond he has purchased – in effect, the long-term interest rate – is seldom identical with the nominal coupon the bond pays, because bonds generally sell at a price below their face value (‘par’). Thus the 3 per cent coupon on a typical nineteenth-century perpetual bond like a rente in fact represented a yield of 3¾ per cent when the price paid for the bond in question was 80 per cent of par.
But what was it that determined yields? One possibility that has long intrigued economists is that there might be some kind of positive relationship between nominal interest rates and inflation (the ‘Gibson paradox’ or ‘Fisher effect’). The long-run British experience suggest that it was the peculiar fiscal effects of war which produced such an effect.3 As might be expected, there are also statistically significant relationships between the yield on consols (the principal British long-term bond) and measures of monetary growth. One possibility that can apparently be discounted, however, is that of a clear-cut relationship between debt/GDP ratios and yields. Statistical analysis of long-run British data from 1727 until 1997 reveals only negative or very weak relationships between the consol yield and the main indicators of fiscal policy (both the debt/GNP ratio and the deficit/GNP ratio). The only fiscal indicator that comes close to having a statistically significant relationship with consols is the burden of debt service.4 Even when the period is broken up into sub-periods, the results are not much better. One possibility is that it was the increased spending associated with wars, not the increased borrowing, which periodically pushed up interest rates in eighteenth- and nineteenth-century Britain. But it is impossible to separate the effects of increased spending and increased debt as the two moved closely together; and higher yields may partly have reflected changes in the default premium on British bonds and expectations about the future convertibility of the currency into gold.5
One possible explanation for this is that contemporaries simply did not know about debt/GDP ratios. Though the concept of national income or wealth was not unknown,6 estimates were too imprecise and too infrequent for such figures to be calculated on a regular basis. However, even when similar calculations are done for a similar sample of countries over the period 1960–1999, the correlation between the debt/GDP ratio and the long-bond yield is negative instead of positive in five out of seven cases.7 The extreme case is that of Japan, where rapid growth in debt has been accompanied by an almost equally rapid decline in yields. Between 1990 and 1999 Japanese gross government debt rose from 61 per cent of GDP to 108 per cent, and was forecast to reach 130 per cent in 2000. Yet long-term Japanese bond yields fell from above 8 per cent in September 1990 to a nadir of less than 1 per cent in November 1998.8 The reason for this lack of close correspondence between debt burdens and yields is that the current amount of debt outstanding in relation to output is only one of many measures which influence investors’ perceptions; in some cases it may not influence them at all. In the industrialized countries during the 1990s, investors’ expectations of falling inflation – and in the Japanese case of outright deflation – counted for much more than rising debt/GDP ratios.
In economic theory, the yield on a bond is the ‘pure’ or real rate of interest (which is equivalent to the marginal efficiency of capital in the economy) plus a premium for uncertainty that takes into account first the risk of default by the borrower and, secondly, the lender’s expectations of inflation and/or depreciation, with the size of the premium generally being larger the more remote the redemption date. In the simplest possible model, ‘bond rates … reflect the sum of real growth expectations and inflation expectations’.9 In reality yields are also influenced by the liquidity of markets and particularly the availability and relative attractiveness of alternative assets; as well as by legal rules and restrictions (such as those obliging pension funds and life assurance companies to hold government bonds); and by taxation of ‘unearned’ income. But at root yields ought mainly to reflect expected growth and inflation. This is how Keynes put i
t:
The rate of interest … is a measure of the unwillingness of those who possess money to part with their liquid control over it…. It is the ‘price’ which equilibriates the desire to hold wealth in the form of cash with the available quantity of cash … A necessary condition failing which the existence of a liquidity-preference for money as a means of holding wealth could not exist … is the existence of uncertainty as to the future of the rate of interest.10
Expectations about the future course of inflation and the chances of future default are reflected in the ‘yield curve’, which plots the yields of bonds according to their maturity. When (to give the obvious example) inflation is anticipated, the yield curve slopes upwards, meaning that short-term interest rates are lower than longer-term rates.11 Major distributional changes will tend to occur when expectations are badly wrong: to be precise, when there are unanticipated defaults or unforeseen changes in the price level. Problems will also arise when (as happened in the 1980s) expectations of inflation raise the anticipated inflation rate above the actual realized rate.12
The key relationship in debt management is therefore between interest rates, inflation and growth. In particular, when the real interest rate (meaning long-term bond yields less expected inflation) is greater than the real growth rate of the economy, then the debt/GDP ratio is ‘intrinsically explosive’.13 Taking the example of Britain since 1831, Figure 11 shows the difference between real growth and real interest rates (calculated as the difference between the yield on consols in a given year and the average inflation rate for the preceding five years).14 As is clear, there have been relatively few periods when real interest rates have consistently exceeded growth. The worst period in this regard was 1920–1932, and the result was indeed a very rapid increase in the debt burden. (Contrast the French experience between 1921 and 1929, when the real interest rate averaged -2.8 per cent and real growth averaged 6.25 per cent per annum.15) Periods when growth has exceeded the real interest rate – such as the early 1950s and the late 1970s – have of course had the opposite effect.
A complicating factor – which could make debt potentially explosive – is the possibility that high debts may actually drive up real interest rates. For the period 1970–1987, for example, there were significant positive correlations between rising debt/GDP ratios in the world’s main industrial economies and rising real interest rates. Rising debt service burdens have also coincided with falling public sector investment.16 Some recent work suggests a global link between public debts and real interest rates,17 though this is not universally accepted.18
Figure 11. The real growth rate minus the real interest rate in Britain, 1831–1997
Source: Goodhart, ‘Monetary Policy’.
In order to illustrate the interaction of debt, inflation and growth, Table 5 attempts to distinguish the impact on the British national debt of the three key influences: new bond issuance (or amortization); inflation (or deflation); and growth (or recession). The striking point is the distinct periodization which emerges. In the period 1822–1914 there was almost no debt reduction through inflation, but rather a reliance on nominal debt repayments, which reduced the debt by about a quarter in absolute terms, and growth, which reduced it by 90 per cent in real terms over as many years. Between 1915 and 1923 there was an immense sevenfold increase in the nominal debt, which was only offset slightly by inflation, and hardly at all by growth. Between 1924 and 1941, however, the debt was more or less static in both nominal and real terms, but fell by 31 per cent in relative terms thanks to higher growth. Between 1941 and 1946 the debt rose again by a factor of 2.4, an increase that was only slightly mitigated by growth and scarcely at all by inflation. But between 1947 and 1975 inflation and, to a lesser extent, growth wholly negated the effects of a 79 per cent increase in the nominal amount of debt. In real terms the debt fell by 61 per cent, and relative to GNP by 82 per cent. Between 1976 and 1997 there was a more muted interplay between the three factors. Nominal debt increased by a factor of over 7, but inflation reduced this to a factor of just under 2, and growth cut the increase to just over 20 per cent. Similar calculations are possible for the United States, and show a broadly similar trend, though with different peaks and troughs. Between 1969 and 1997, for example, the US federal debt rose in nominal terms by a factor of 13; in real terms by a factor of 3.5; but relative to GNP by a factor of just 1.6.19 These figures reveal the importance of price movements and growth in determining how far large nominal debt burdens persist in real terms.
Table 5. Increase or decrease in the British national debt by sub-periods, 1822–1997
Source: Calculated from figures in Goodhart, ‘Monetary Policy’.
The ease with which real debt burdens have been reduced by inflation in the twentieth century makes it tempting to conclude that such periodic ‘jubilees’ are a recurrent feature of modern political economy. Yet the inflation tax is an effective means of reducing debt burdens only under certain circumstances. When the structure of the debt is tilted towards short-term instruments, lenders may anticipate or swiftly react to inflation by raising the interest rates they demand.20 Even when a government relies mainly on long-term bonds, a rise in inflation will lead to a rise in yields, increasing the cost of any new borrowing. Moreover, inflation is easier to start than to stop under conditions of high public indebtedness. A central bank aiming to halt inflation by raising the short-term interest rate would be likely to fail if the government continued to run high deficits.21 The problem is that the central bank’s rate increase, and expectations of lower inflation, would also tend to raise the real interest rate on government debt, increasing the cost of debt service, widening the budget deficit and thereby undermining the credibility of the bank’s policy. Higher interest rates also tend to reduce seigniorage, as well as reducing revenues and increasing expenditures because of their negative effects on growth.22
Clearly, much depends on the nature of expectations. If these are ‘adaptive’ – if there is only a gradual response to a change in monetary policy because workers and firms base their expectations on an average of current and past inflation – an anti-inflationary policy will inevitably have negative effects on output and employment. If expectations are rational, on the other hand – meaning that economic agents immediately infer lower future inflation from a policy change – then inflation could be brought under control at a lower cost, provided the policy change was ‘once-and-for-all, widely understood and widely agreed upon … and therefore unlikely to be reversed’.23
In the light of the ‘unnecessary randomness’ of ‘partial default via inflation’, some economists have concluded that ‘overall, nominal debt seems to be a bad idea’ and that index-linked (i.e. inflation-proof) bonds are to be preferred.24 However, this course was followed only to a limited extent as inflation fears abated during the 1990s. Instead, many governments have effectively ruled out the possibility of an inflationary default by issuing a high proportion of short-term debt. Table 6 shows that short-term debt counts for relatively little in the total debts of Austria, Germany and the Netherlands, but constitutes more than a third of Italian, French and Spanish debts. In Britain something like a quarter of the total national debt in 1997 had a maturity of five years; more than a fifth has a maturity greater than fifteen years.25 But in the United States around a third of the privately held federal debt has a maturity of less than a year; and 72 per cent – nearly three-quarters – has a maturity of less than five years.26
Such a reliance on short-term bonds stands in marked contrast to the nineteenth century. Quite apart from deterring governments from trying to inflate away their debts, it makes government debt charges a great deal more sensitive to fluctuations in interest rates. This can be advantageous when rates fall, as happened in the 1990s: according to one estimate, the long maturities of British government bonds cost taxpayers in 1999 £3 billion more in debt charges than they would have had to pay on short-term equivalents.27 But short-term debt can quickly lead to trouble w
hen the direction is the other way.
Table 6. The structure of European national debts, circa 1993
* * *
Country
Short-term debt as percentage of domestic debt
* * *
Austria
0.4
Belgium
21.1
Finland
27.9
France
42.4
Germany
3.9
Italy
39.4
Netherlands
4.9
Spain
52.7
Norway
35.4
Sweden
15.3
UK
29.6
* * *
The Cash Nexus: Money and Politics in Modern History, 1700-2000 Page 22