Features: June 7th, 2013

How to reduce debt is an issue facing many governments. Spending to promote growth is an argument that is growing in popularity.  In this article Ryan Bourne and Tim Knox of the Centre for Policy Studies look at both sides of the argument.

The debate over fiscal policy has been getting personal of late. The revelation that the work of Ken Rogoff and Carmen Reinhart (R&R) contained mistakes within the spreadsheet coding has re-opened the debate about fiscal policy which has been rumbling for the past four years. If you remember, their work used a threshold debt-to-GDP ratio of 90% to show how real GDP growth slows as the burden of public debt increases. Many economists have seen the subsequent revisions to their work as a huge game-changer in the policy debate about the risks of a high-debt burden. But as my joint article with Vuk Vukovic for Pieria recently outlined (and Carmen Reinhart’s recent open letter to Paul Krugman explained), the key results of their debt paper remain unchanged, and there is good reason to think high debts will feed back into lower growth over time, even if the initial increase in debt arises as a result of financial crises.

The forthcoming pressures of an ageing UK population and the subsequent commitments to higher entitlement spending in the future multiply the risks. Furthermore, the important conclusion developed by R&R that growth tends to be retarded for longer periods following financial crises has proved prescient.

The hard facts

Forget all the economic models we hear so much about for a minute. The simple arithmetic is that debts have to be paid back, defaulted on or inflated away. The debate then is really about how we go about maximising our chances of healthy economic growth to get our debts under control in order to avoid the negative consequences which default and inflation would entail. To many of us who advocate the need for fiscal consolidation, this means permanent spending cuts and reforms to entitlements to lower the long-run debt path whilst undertaking robust supply-side reforms to raise the sustainable growth rate of the UK economy by giving productivity growth a chance. This view entails looking beyond the very short-term, and thinking about the medium-term growth rate of the economy. We have criticised the government for front-loading tax rises (which affect incentives), and some infrastructure spending (which if properly targeted, can reduce supply constraints), but have been supportive of the need to bring the deficit down overall. Current public expenditure, which had increased so quickly under the previous government, should have taken the brunt of the axe.

The self-financing spending?

Many of those on the opposite side of the debate have instead argued that, in current conditions, fiscal policy has a big effect on economic growth and that we should be willing to accept more debt today in order to try to reduce unemployment and boost real GDP. This was largely played out last year in a well-documented debate over the size of so-called “fiscal multipliers”. But more recently the case put forward by those in favour of higher deficit spending has subtlety changed. Not only is fiscal consolidation bad for growth in the short-term, they say, and fiscal stimulus required to boost GDP and ‘confidence’, but doing fiscal consolidation now is even self-defeating in terms of lowering the path of debt-to-GDP in the medium term. In other words, more spending now would lead to a lower debt-to-GDP ratio than we would otherwise face in the future!

The intellectual case for this was put forward by Brad DeLong and Larry Summers in their non-empirical paper ‘Fiscal Policy in a depressed economy’. The academics developed a simple model which made a raft of assumptions about the size of fiscal multipliers, hysteresis effects, interest rates, and an inability or unwillingness for Central Banks to undertake even looser monetary policy. The model showed that, under these conditions, a fiscal stimulus today would be self-financing in the longer term. This line, that consolidating public finances today can actually undermine growth prospects and thus the debt-to-GDP path in the future, has since been adopted by many UK thinkers, including Lord Skidelsky and Marcus Miller, and is beginning to be repeated by others.

It’s worth noting that the assumptions behind the DeLong-Summers model are not insignificant:

– The Central Bank being “unable or unwilling to provide additional stimulus through quantitative easing or other means”
– The fact that fiscal policy works as Keynesians say it does, and the fiscal multiplier is much higher in today’s circumstances
– That temporary spending can permanently alleviate potential hysteresis effects
– That interest rates are likely to remain low

Thus, if all of these types of conditions are met, we are left with a sort of neo-Keynesian inverted Laffer curve or inverted Say’s Law – “demand creates its own supply”. We can borrow more today to borrow less in the future. This view suggests Milton Friedman was wrong. There is such a thing as a free lunch, and its name is “fiscal stimulus”.

Do the assumptions stack up?

But do these assumptions stack up in reality? For as DeLong and Summer make clear, the result “is dependent on multiplier and hysteresis effects, the assumption about government borrowing costs, and the assumption that government spending once increased can again be reduced”.


Perhaps the most important assumption is that of the “hysteresis” effect. Hysteresis is the idea that a protracted period where output is below potential leads to an erosion of the future potential of the economy. This could be because the skills of workers waste away if left idle, or because a dearth of investment reduces the capital stock. Thus, so the argument goes, keeping ‘demand’ higher through more government spending prevents this erosion of skills and underinvestment from occurring, and so prevents this hysteresis effect from taking hold and means the economy has greater productive potential in the future than would otherwise be the case.

Now, it’s worth noting here that the hysteresis phenomenon is a well-known economic concept and there is lots of evidence that it exists. In fact, it’s a concept that both left and right embrace. The left object to fiscal retrenchment on the basis that the lowering of unemployment owing to sticky wages will put people out of work, and over time lead to a deterioration of skills such that it might be difficult in future for individuals to obtain jobs. Many on the right make similar points with regards to incentives and welfare. For example, many believe that if welfare is too generous, or minimum wages are set too high, this will trap unskilled workers out of the labour market, leading to a loss of skills and eroding the possibility of the individual getting a future job.

Much of the early literature looked at this in the context of labour unions, and the lock-out of workers. Yet, in these cases, there were self-evident solutions: reform the supply-sides of the economy in order to eliminate this “lock-out” of workers.

The case advocated by DeLong and Summers is much more contentious. As the economist Valerie Ramey says in her discussion of the paper, their calculations assume both that a short-term spending boost today will reverse the hysteresis effects that would have occurred and that hysteresis itself is a near-permanent phenomenon. She tests this by examining whether temporary increases in government spending have permanently increased output in the past (which we would expect if hysteresis was a permanent phenomenon). She finds that government spending does not lead to permanent increases in output, and that actually increased government spending is associated with lower investment. She concludes that the evidence for the hysteresis effect that the model assumes, particularly through the investment channel, is therefore thin.

The hysteresis story therefore requires a huge leap of faith. If hysteresis exists, it assumes a temporary increase in government spending can prevent it from being realised. This assumes that the government can effectively bridge the gap in the short-term and generate the private sector skills or investment to replace it when the government spending is withdrawn. The importance of the hysteresis parameter is also influenced in turn by the assumption of a high government spending multiplier in order to have a big effect on output today. If we think that either of these micro assumptions is in doubt, then it might be better to think of hysteresis, where it exists, as a supply-side phenomenon.

In the UK, the hysteresis explanation seems even less likely still. Given the depth of the recession, and through the poor recovery, there has been surprisingly low unemployment compared to what we might have expected (in both numbers and duration), and very low levels of corporate bankruptcies. Demonstrating the existence of large hysteresis effects therefore requires that the types of jobs that would result from public sector borrowing are either more productive than, or prevent hysteresis to a greater degree than, the very many private sector jobs that are being created already.

The costs of debt

Another key assumption of the DeLong and Summers model is that it measures the costs of extra debt simply as the cost of making the extra interest payments and the distortionary effect of taxes to pay these extra payments. But, as Martin Feldstein outlines in his critique of their paper, they ignore numerous other potential and, in some cases, likely costs. First, the potential for crowding out of private investment (due to expectations of future taxes). Second, the fact that a large proportion of debt is held by foreign investors, meaning that financing it requires an increase in net exports (which might have to be achieved by a fall in the exchange rate relative to trading partners, and thus a fall in domestic real incomes). Third, that higher debts make countries more vulnerable to upward movement in interest rates. Fourth, that higher debts give governments less room to manoeuvre if other unexpected events occur.

Other assumptions

As mentioned above, the DeLong-Summers model also makes assumptions that government spending multipliers are much higher at the moment and that interest rates are likely to stay low for a long period. It’s fair to say that on both of these fronts there has been vigorous debate over whether these assumptions are justified, so we won’t go over familiar territory here.

Yet, it’s worth noting a final key feature of the model. In it, the government doesn’t ever act to reverse increases in debt, or try to slow the down the current upward trend we are seeing. Instead the model simply examines a single jump in debt. But we know that eventually debt has to be dealt with in some way, and this model fails to explain how. We also know that the debt-to-GDP ratio is currently on an upward path and that an ageing population will bring an upward pressure on the debt path after the next decade or so. That’s why DeLong and Summers rightly outline the need for a long-term plan to get debts under control. But their model doesn’t include the costs of doing this. The choice facing the government therefore isn’t really a choice between spending or not spending from a standing start, but really about whether they should borrow significantly more when they are already running a huge deficit, and debt is already at very high levels.


The debate over fiscal policy has raged since the start of the economic crisis. Most economists suggest that a medium term plan is required to close the deficit and to thus get the debt burden back on a downward path. They believe this will require a combination of both cuts to government spending and policies to induce economic growth over the medium term. Recently, however, the idea that there might be a ‘fiscal free lunch’ and that borrowing more today might help to lower the debt-to-GDP ratio in future has been touted as justification for looser fiscal policy today. In our view, the assumptions for this are hopeful rather than realistic.

Last week, the FT’s Chris Giles said: “you have to have a rather weird faith in the notion that raising capital spending a bit today provides a much more powerful boost than the hit that inevitably comes next year when that temporary support is removed.” Yet the intellectual model developed by DeLong and Summers, whilst an interesting way of presenting fiscal choices, does just that. To our mind, it has provided intellectual cover for an idea which requires unrealistic assumptions about hysteresis effects, while ignoring risks of other costs associated with high debts. The paper does an important public service, however. It lays out in detail the scale of the assumptions required to believe that spending has a long-term beneficial debt impact.

Must Be Read: