A major component of the policy response to the economic crisis in the United States, Japan and especially Britain has been ‘quantitative easing’. The policy is having very little effect, which should not be surprising given the weakness of its theoretical underpinnings.
Quantitative easing involves the central bank buying longer-term financial assets and paying for them by crediting banks’ reserve accounts. This action leaves net private financial assets unchanged, but alters their composition, which affects yields and returns.
One intention of the policy is to create excess bank reserves. This may reflect a traditional belief among policymakers and orthodox (neoclassical) economists that bank lending is constrained by insufficient reserves, and that increasing these will encourage banks to lend to households and firms who will then use the funds to consume or productively invest.
Another intention of the policy is to reduce longer term interest rates, which the orthodoxy believes will encourage private investment. By buying longer-term financial assets, the central bank hopes to cause the prices of these assets to rise relative to shorter-term asset prices, and bring down longer-term interest rates.
So neoclassical supporters of quantitative easing – including Ben Bernanke on the monetarist right and Paul Krugman on the liberal left – believe the policy will boost lending and private investment. Neoclassical critics of the policy fear that it will be inflationary, a concern shared by economists in the economic-libertarian Austrian School.
Quantitative easing was actually first introduced in Japan almost a decade ago as the economy struggled to recover from the Asian crisis of the late 1990s, so the policy approach has some history prior to the current crisis. The experience of Japan strongly suggests that the neoclassical supporters and critics of quantitative easing are both wrong. The policy does not boost lending and private investment; nor is it inflationary.
Bill Mitchell provides a good discussion of these points:
In the late 1990s, Paul Krugman joined a number of academic economists in urging the Bank of Japan to introduce large-scale quantitative easing to kick start the economy. The Bank, reluctantly, heeded their advice and in 2001 they increased bank reserves from ¥5 trillion to ¥30 trillion. This action had very little impact – real economic activity and asset prices continued their downward spiral and inflation headed below the zero line.
Many economists had also claimed that the huge increase in bank reserves would be inflationary. They were also wrong. …
In this 1998 article on Japan’s trap, Krugman claimed that Japan was “in the dreaded ‘liquidity trap’, in which monetary policy becomes ineffective because you can’t push interest rates below zero”. …
He said that when the nominal interest rate is at zero and therefore stimulatory interest rate adjustments can no longer be made, the real rate of interest that is required to “match saving and investment may well be negative”. …
Krugman [said] that the nation [needed] a dose of expected inflation so that the real interest rate [would become] negative (and flexible).
[H]e was completely wrong in this diagnosis. The only thing that got Japan moving again in the early part of this Century was a dramatic expansion of fiscal policy.
Mitchell’s complete argument can be found here.
To many heterodox economists, the ineffectiveness of quantitative easing is not at all surprising. Orthodox expectations to the contrary appear to be based on flawed ‘money multiplier’ reasoning as well as a discredited theory of the determinants of private investment.
According to the traditional money-multiplier model, if the central bank increases bank reserves, this will lead to more lending and a multiplied increase in deposits. If banks are not lending, the reasoning goes, it must be because banks have insufficient reserves.
The main problem with this theory is that banks do not require reserves prior to lending. In a credit economy, if there is demand for loans from good credit risks, the banks will extend loans (credit the bank accounts of borrowers). Loans create deposits. The borrowers might then use the loans to consume or productively invest, adding to production and income. If necessary, banks can always make up any shortfall in reserves at the end of the process by borrowing from other banks or from the Fed.
The reason banks are unwilling to lend during a crisis is that there is insufficient demand for loans from credit-worthy borrowers (in the assessment of the banks). Increasing bank reserves doesn’t do anything to alter this problem. Lending is demand, risk and capital constrained, not reserve constrained.
The orthodox claim that lower interest rates must induce higher private investment is also incorrect. For a start, the claim relies on the notion of a well-behaved ‘demand for capital’ function, which was shown to be invalid in the capital debates (see Nobel-nomics). Besides that, in a dynamic setting, investment will always depend on the comparison of shifting revenues and costs. If revenue prospects are poor because of weak demand, lower interest rates may not make the cost of investment more attractive for firms relative to expected revenues.
While neoclassical arguments in favor of quantitative easing are dubious, the inflationary fears of the neoclassical and Austrian critics arise from an equally flawed theoretical conception: the ‘quantity theory of money’. This theory is based on a particular interpretation of an accounting identity known as the ‘quantity equation’:
MV = PY
In this identity, M is the quantity of money (defined in this context as currency plus demand deposits), V is the income velocity of money (the number of times a unit of money, on average, circulates in the accounting period), P is the general price level (price index) for the economy as a whole and Y is real output. The identity simply says that the amount of nominal output sold (PY) must equal the quantity of money times the average number of times a unit of money is used in transactions over the period (MV). For example, if the general price level is 1, real output sold is 100 and a unit of money is used in an average of 5 transactions each period, the quantity of money must be 20.
The relationship is true by definition, but in itself says nothing about how the different variables affect each other or what determines each variable. The quantity theory of money relies on particular assumptions: (1) the quantity of money is assumed to be exogenously controlled by the central bank; (2) the velocity of money is assumed to be constant; (3) real output Y is assumed to be ‘fixed’ at the full-employment level in the long run; (4) causation is assumed to run from left to right (i.e. changes in MV are assumed to cause changes in PY).
Given these four assumptions, the theory suggests that if the central bank increases the quantity of money M, the only effect in the long run will be to increase the general price level P. In other words, a monetary expansion causes inflation in the long run, with no effect on real output, which will supposedly gravitate to the full-employment level irrespective of monetary policy.
The assumptions are unfounded. First, the central bank is incapable of controlling the broader money supply. This is because the broader money supply (currency plus deposits) depends partly on private bank lending, which has an effect on the amount of deposits that is largely independent of the central bank’s policy actions. All the central bank can do is control the short-term interest rate (and other interest rates if it wishes) and let the quantity of money vary. This was made very clear in the early 1980s when central banks temporarily tried to put this monetarist theory into practice by controlling the money supply. The attempts failed abysmally and the approach was quickly abandoned in favor of interest-rate targeting.
Second, the velocity of money is not constant. It varies over the economic cycle and can move inversely to the quantity of money. An increase in M may coincide with a reduction in V, leaving MV unchanged. At the onset of the crisis, the velocity of money collapsed. It has remained low, though variable, due to the weakness in economic activity.
Third, the notion that market economies automatically tend to full employment is false, on both theoretical and empirical grounds. The neoclassical argument is that unemployment will be eliminated in the long run through adjustments in wages and interest rates. The steps of the argument are that: (a) unemployment puts downward pressure on real wages; (b) this results in higher demand for labor and an increase in aggregate employment; (c) the increase in employment results in higher output; and (d) demand adjusts to this higher level of output through price and interest-rate adjustments.
Step (c) of this argument is reasonable. Step (a) is also plausible, although to hold it requires prices not to be falling faster (or rising more slowly) than money wages under conditions of weak demand. However, steps (b) and (d) are based on fallacies of composition (aggregation problems). It is not valid to suppose an inverse relationship between real wages and aggregate employment, or between real interest rates and total private investment. Another problem for the neoclassical argument is that employment and private investment are dependent on conditions in product markets. Lower wages, for instance, have effects on product demands, which then react back on employment. These feedback effects make the neoclassical ‘demand for labor’ and ‘demand for capital’ functions unstable. In general, the effects of wage and interest-rate changes at the aggregate level are indeterminate.
Since there is no automatic tendency to full employment, it is illegitimate to assume that real output is ‘fixed’ at the full employment level in the long run. Therefore, if MV increases on the left-hand side of the quantity equation, the corresponding increase in PY on the right-hand side will not necessarily be due to an increase in the price level P. It could be due to an increase in real output Y. In fact, whenever there is unemployment and underemployment, higher spending is likely to result in an expansion of output, not a mere increase in prices. There is competitive pressure on firms to respond to increased demand in this way to protect market share. For this purpose, they tend to operate with a planned degree of excess capacity, which enables them to respond to fluctuations in demand without necessarily altering prices.
Fourth, causation does not run from money to prices (i.e. from MV to PY). It is the reverse. Nominal output PY reflects the level of economic activity, which influences the demand for loans from credit-worthy borrowers. Banks then extend loans accordingly, altering the amount of deposits and the quantity of money M.
To sum up, in the identity MV = PY, quantity theorists suppose M to be directly controllable by the central bank (which it isn’t), Y to be at the full-employment level (which it isn’t), V to be constant (which it isn’t), and causality to run from money to prices (which it doesn’t).
The claims that quantitative easing are likely to be inflationary therefore rest on a false theory. The policy does not result in an expansion of loans, and so cannot cause spending that tests the supply limits of the economy. Even if it could succeed in boosting lending and expenditure, the primary effect under conditions of high unemployment and underutilization of productive capacity would be an increase in real output to meet the greater demand. Price effects, if any, would be secondary.
In spite of the weakness of the case for quantitative easing both in theory and practice, mainstream supporters continued to advocate it. With the onset of the current crisis, Paul Krugman wrote in support of the approach for the US in two NYT articles, “It’s the Stupidity Economy” and “Bernanke’s Unfinished Mission”. In them, he argued that the Fed should buy a further $2 trillion of longer-term assets, which he suggested “could do a lot to promote faster growth, while having hardly any downside”. But this policy action would only have a positive effect on real economic activity if it actually encouraged more loans to be extended to productive investors and consumers. Since an injection of liquidity does not have this effect, the impact on the real economy will be minimal.
Krugman has also argued that the Fed should “credibly commit to higher inflation, so as to reduce real interest rates”. He thinks this is necessary because neoclassical theory suggests that the downturn in economic activity is caused by a failure of real interest rates to drop low enough to encourage private investment. For given nominal rates of interest, higher expected inflation would deliver lower real interest rates. But the argument is weak for reasons already discussed. Quantitative easing is not inflationary, and even if it was, lower real interest rates cannot be relied upon to induce higher private investment. The hope is a forlorn one, and quantitative easing is an ineffective policy.
The way to stimulate the economy during a period in which the private sector is deleveraging is for the government to undertake expansionary fiscal policy. The deficit expenditure directly adds to demand and helps to sustain output and employment levels. The higher incomes help private firms and households pay off debts and get their balance sheets in better shape. Once firms feel ready to invest productively and consumers are willing to lift spending levels, the demand for loans from credit-worthy borrowers will strengthen, and banks will be willing to lend again.
Krugman is of course fully aware that fiscal policy is an alternative means of stimulus, but considers it politically difficult, which may well be true. However, even when Krugman discusses fiscal policy, he is hampered, like other neoclassical economists, by the false notion of a ‘government budget constraint’ that is supposedly analogous to a household budget constraint. The analogy simply doesn’t hold for a sovereign government that is the issuer of its own fiat currency under a flexible exchange-rate regime. A private household needs income or a loan before it can spend. The government does not.
The faulty notion of a government budget constraint leads neoclassical economists to suppose budget deficits result in: (i) higher interest rates that crowd out private investment; and (ii) unsustainable public debt that imposes a burden on future generations.
The idea that budget deficits put upward pressure on interest rates is based on the ‘loanable funds doctrine’. According to this doctrine, there is supposedly a pool of savings available to fund investment (whether private or public). If the government draws on this pool of saving, there will be less savings left over to fund private investment. The higher demand for funds, given a fixed supply of existing funds, will drive up interest rates.
But this is a faulty depiction of the relationship between saving and investment. First, a sovereign government that is the issuer of its own fiat currency does not need to draw on existing savings. It spends without requiring prior funds. The expenditure adds to aggregate demand and income, and saving and tax revenue rise as some fraction of income. Government expenditure adds to savings rather than depleting preexisting savings. Second, the loanable funds doctrine does not even apply to private investment. As has already been noted, in a modern credit economy, banks can extend loans to firms wishing to undertake productive investment without possessing prior reserves. The private investment adds to aggregate demand and income, and therefore to saving and tax revenue.
The neoclassical idea that budget deficits put upward pressure on interest rates also contradicts the way in which the monetary system actually operates. When the government net spends, this involves it crediting private bank accounts (spending) more than it debits private bank accounts (taxing). The immediate result is an increase in private bank deposits and, as a consequence, bank reserves. This means that if the central bank did nothing in response to a budget deficit, there would be strong downward pressure on interest rates. The excess bank reserves would push the cash rate down to zero (or to a support rate in countries where there is one). So budget deficits in themselves cause interest rates to fall, not rise as neoclassical theory claims. If the central bank wants the short-term interest rate to be above zero, it is forced to step in to prevent the budget deficit from pushing the cash rate below its target rate. It does this either by draining the excess reserves (selling bonds) or, as has occurred in some countries post-crisis, paying interest on reserves.
The neoclassical claim that budget deficits impose a debt burden on future generations is also without basis. There is no such debt burden. The government does not actually need to issue debt at all to net spend. It does so for (expendable) operational and political reasons that have nothing to do with funding the expenditure. The real burden society places on future generations is its failure to maintain full utilization of resources, because this means potential investment in technology, education, infrastructure and other activity that increases future productive capacity is forgone.