The optimal speed of fiscal adjustment
March 30, 2012 12:28 pm
by Gavyn Davies
.
How rapidly should governments correct their fiscal deficits, which in the long run are unsustainable in the US, UK, Japan and many countries in the eurozone?
That is a question which continues to dominate the policy debate among economists. Rapid correction undoubtedly damages near term economic growth, but is intended to reduce the risk of a sovereign debt crisis coming suddenly out of the blue. Slow correction does the opposite. There is no theoretically “correct” policy on this. The result depends on how the near term loss of output should be weighed against the risk and consequences of a fiscal crisis, which is an empirical matter. (See this earlier blog: Assessing the risk of a financial crisis, which attempts to measure the risk of fiscal crisis.)
.
It is possible for reasonable economists to disagree about this, and for the “right” policy to be different in different countries. However, occasionally a piece of research comes along which changes the “dial” on the debate, and I believe that applies to the important Brookings Paper published last week by Brad DeLong and Larry Summers. This paper, which is well summarised here and here, essentially implies that the trade-off between near-term GDP growth and the probability of fiscal crisis can be irrelevant, because temporary fiscal expansions, at a time when interest rates are at the zero bound, are eventually self-financing.
The intuition behind this result is reasonably straightforward. It depends first on a belief that the “multiplier”, which measures the relationship between a fiscal injection and the resulting rise in GDP, is unusually high at present, so that fiscal easing would have a large beneficial effect of GDP in the near term.
And it also depends on a belief that the current recession will have permanent effects on the long run future path for output in the economy. This results in a “hysteresis” coefficient, which is defined as the relationship between a near term shortfall of GDP relative to trend, and the permanent loss in trend GDP which this causes.
If either the multiplier or the hysteresis coefficient is equal to zero, then there is no possibility that temporary fiscal expansions will be self-financing, and the DeLong/Summers result becomes irrelevant. However, if both of the coefficients are positive, then it is possible that fiscal expansion can be self-financing in the long term, and that fiscal contraction can actually make public debt ratios worse.
Basically, a fiscal expansion today raises GDP today; higher GDP today permanently raises the path for GDP; and a permanently higher path for GDP results in permanently higher tax receipts. Eventually, these higher tax receipts might fully finance the original cost of today’s budgetary injection.
All of this has been in the public debate for quite a while (see Paul Krugman for example), but the important additional contribution of DeLong and Summers is that we can now quantify the parameters involved. Their methodology, which was apparently unchallenged at the Brookings meeting last week, enables us to plug in estimates for the multiplier and the hysteresis coefficient, and then calculate whether a fiscal injection will be self- financing.
The authors, for example, suggest a case in which the multiplier is 1, and the hysteresis parameter is 0.05. Their equation then suggests that a temporary fiscal expansion will be self-financing unless the real yield on treasuries exceeds the long-term growth of GDP by more than 2.5 per cent per annum, which has hardly ever happened in recent US history.
.
How robust should we take this result to be?
A fiscal multiplier of around 1 in present circumstances seems reasonable. In normal times, the Fed might well tighten monetary policy in the face of fiscal expansion, eliminating most or all of the boost to GDP, in order to keep inflation on their pre-determined target. If so, the multiplier would be zero, and a budget stimulus would not be self-financing.
But these are not normal times for the Fed. To judge from what Ben Bernanke has been saying recently, the Fed would actually welcome some easing in fiscal policy in the immediate future, provided that this is accompanied by credible measures to reduce public debt in the long term. In that event, the Fed would be unlikely to raise short rates (though it might do less QE), and the resulting fiscal multiplier could be unusually high. Many independent studies suggest it might be in the range of 0.8-1.5.
.
Much less is known about the hysteresis coefficient. In the US, GDP has in the past shown an encouraging tendency to return eventually to its long-term trend, even after very major shocks like the 1930s depression. In that case, the hysteresis coefficient would be zero, and the DeLong/Summers result would collapse. But that was certainly not the case in Europe in the 1980s or Japan in the 1990s.
.
It seems optimistic to suggest that the current deep recession will leave no mark whatsoever on either the capital stock or the active labour force in the US, and the suggestion that the coefficient might be around 0.05 or more seems somewhat conservative.
.
So the DeLong/Summers results deserve to be taken seriously in present economic conditions. But I would pose one more question to Brad and Larry, concerning policy credibility. They take this seriously, but make the assumption that a temporary easing in fiscal policy would have no effect on the risk premium which the market expects on US treasuries, either now or in the future. They no doubt justify this by pointing to the large demand for low risk investments in the present climate of excess savings. And they would argue that it is not rational for investors to expect a higher risk premium if a temporary fiscal injection is actually likely to be self-financing.
.
These are valid points, but investors are not always rational, and they would be sceptical about any government which takes the “make me chaste, but not yet” approach to policy. Easier fiscal policy today might well be taken as a signal that the political will to tighten policy tomorrow simply will never exist. Rational or not, that remains a serious problem.
.
It is possible for reasonable economists to disagree about this, and for the “right” policy to be different in different countries. However, occasionally a piece of research comes along which changes the “dial” on the debate, and I believe that applies to the important Brookings Paper published last week by Brad DeLong and Larry Summers. This paper, which is well summarised here and here, essentially implies that the trade-off between near-term GDP growth and the probability of fiscal crisis can be irrelevant, because temporary fiscal expansions, at a time when interest rates are at the zero bound, are eventually self-financing.
The intuition behind this result is reasonably straightforward. It depends first on a belief that the “multiplier”, which measures the relationship between a fiscal injection and the resulting rise in GDP, is unusually high at present, so that fiscal easing would have a large beneficial effect of GDP in the near term.
And it also depends on a belief that the current recession will have permanent effects on the long run future path for output in the economy. This results in a “hysteresis” coefficient, which is defined as the relationship between a near term shortfall of GDP relative to trend, and the permanent loss in trend GDP which this causes.
If either the multiplier or the hysteresis coefficient is equal to zero, then there is no possibility that temporary fiscal expansions will be self-financing, and the DeLong/Summers result becomes irrelevant. However, if both of the coefficients are positive, then it is possible that fiscal expansion can be self-financing in the long term, and that fiscal contraction can actually make public debt ratios worse.
Basically, a fiscal expansion today raises GDP today; higher GDP today permanently raises the path for GDP; and a permanently higher path for GDP results in permanently higher tax receipts. Eventually, these higher tax receipts might fully finance the original cost of today’s budgetary injection.
All of this has been in the public debate for quite a while (see Paul Krugman for example), but the important additional contribution of DeLong and Summers is that we can now quantify the parameters involved. Their methodology, which was apparently unchallenged at the Brookings meeting last week, enables us to plug in estimates for the multiplier and the hysteresis coefficient, and then calculate whether a fiscal injection will be self- financing.
The authors, for example, suggest a case in which the multiplier is 1, and the hysteresis parameter is 0.05. Their equation then suggests that a temporary fiscal expansion will be self-financing unless the real yield on treasuries exceeds the long-term growth of GDP by more than 2.5 per cent per annum, which has hardly ever happened in recent US history.
.
How robust should we take this result to be?
A fiscal multiplier of around 1 in present circumstances seems reasonable. In normal times, the Fed might well tighten monetary policy in the face of fiscal expansion, eliminating most or all of the boost to GDP, in order to keep inflation on their pre-determined target. If so, the multiplier would be zero, and a budget stimulus would not be self-financing.
But these are not normal times for the Fed. To judge from what Ben Bernanke has been saying recently, the Fed would actually welcome some easing in fiscal policy in the immediate future, provided that this is accompanied by credible measures to reduce public debt in the long term. In that event, the Fed would be unlikely to raise short rates (though it might do less QE), and the resulting fiscal multiplier could be unusually high. Many independent studies suggest it might be in the range of 0.8-1.5.
.
Much less is known about the hysteresis coefficient. In the US, GDP has in the past shown an encouraging tendency to return eventually to its long-term trend, even after very major shocks like the 1930s depression. In that case, the hysteresis coefficient would be zero, and the DeLong/Summers result would collapse. But that was certainly not the case in Europe in the 1980s or Japan in the 1990s.
.
It seems optimistic to suggest that the current deep recession will leave no mark whatsoever on either the capital stock or the active labour force in the US, and the suggestion that the coefficient might be around 0.05 or more seems somewhat conservative.
.
So the DeLong/Summers results deserve to be taken seriously in present economic conditions. But I would pose one more question to Brad and Larry, concerning policy credibility. They take this seriously, but make the assumption that a temporary easing in fiscal policy would have no effect on the risk premium which the market expects on US treasuries, either now or in the future. They no doubt justify this by pointing to the large demand for low risk investments in the present climate of excess savings. And they would argue that it is not rational for investors to expect a higher risk premium if a temporary fiscal injection is actually likely to be self-financing.
.
These are valid points, but investors are not always rational, and they would be sceptical about any government which takes the “make me chaste, but not yet” approach to policy. Easier fiscal policy today might well be taken as a signal that the political will to tighten policy tomorrow simply will never exist. Rational or not, that remains a serious problem.
0 comments:
Publicar un comentario