As the economy begins to show signs of strength, people naturally want to know, how long will the damage from the financial crisis linger? A new paper from the Bank of England suggests it may be much longer than we would like.

If that's right, then official economic projections remain too optimistic, even though they are more subdued than they were before the crisis.

The Bank of England analysis, by Nicholas Oulton and Maria Sebastia-Barriel, examines the well-known Reinhart-Rogoff episodes of financial crisis globally (named for the economists Carmen Reinhart and Kenneth Rogoff). The BOE analysis finds that crises reduce short-run productivity growth by 0.6 to 0.7 percentage point per year, on average - a sizeable effect. Over the past five years, U.S. annual productivity growth has averaged only about 1.5 percent, so a drop of 0.6 to 0.7 percentage point is relatively big.

Perhaps more disturbingly, Oulton and Sebastia-Barriel find a significant long-run effect: For each year of a financial crisis, the level of gross domestic product per capita is reduced in the long term by 1.5 percentage points. In other words, a crisis lasting five years permanently lowers GDP per capita by a whopping 7.5 percentage points.

The authors include the appropriate caveats about their analysis, including that such estimates are based on averages and that each episode is unique. Nonetheless, they conclude that "banking crises as defined by Reinhart and Rogoff have on average a substantial and statistically significant effect on both the short-run growth rate and the long-run level of labour productivity."

In contrast, the Congressional Budget Office projects a much more modest effect. Its latest projections (which will be updated next week) assume that "potential output will be about 1.5 percent lower in 2022 than it would have been without the recession and the ensuing economic weakness." The difference between the long-term effects of the crisis in these two forecasts is enormous, amounting to perhaps $1 trillion a year.

Why would financial crises have long-lasting effects? One reason is that workers' skills often atrophy during the downturn that follows such a crisis, and some people never get jobs again. Indeed, because the share of population with a job falls permanently, Oulton and Sebastia-Barriel find, the effect on GDP per capita is about twice the effect on GDP per worker. (In the United States, look no further than the rapid increase in enrollment in the disability-insurance program. The number of beneficiaries is now 8.8 million compared with 6.8 million six years ago. The extra 2 million are people who are not going back to work.)

A second reason for the lasting hangover is that after a financial crisis capital spending (for example, on new factories and equipment) tends to decline, and the lower level of capital may then be perpetuated, reducing output permanently.

Finally, it's possible that financial crises have lasting effects on how well capital and labor can be combined to produce income - what is called total-factor productivity. By this theory, innovation declines during a crisis, and since innovation has a cumulative aspect, the impact lingers.

Even without long-lasting effects from the crisis, the official projections may be on the rosy side. An analysis by John Fernald, a senior research adviser at the Federal Reserve Bank of San Francisco, is largely agnostic about whether the impact from the financial crisis will last. But Fernald also raises another troubling point: that productivity growth had been slowing even before the crisis hit in 2007.

Fernald estimates that, as a result of the slowdown in productivity growth (along with an aging population), in the long run we should expect annual growth to average only 2.1 percent. Yet government forecasts are higher. The White House is projecting long-term growth of 2.5 percent per year, and the CBO puts it at 2.4 percent.

These differences may seem small, but don't forget the power of compound interest (which has large consequences for income) or the budget deficit's sensitivity to growth (which is surprisingly acute). Over the next decade, if Fernald is right, the deficit will be about $1 trillion larger than official projections suggest. Keep in mind, his analysis takes a benign view on whether the crisis itself will further diminish income in the long run.

Fernald's thesis, by the way, may partially solve a puzzle - why, over the past two years, annual economic growth has averaged about 2 percent even though the unemployment rate has declined by about 1.5 percentage points. The traditional relationship between growth and unemployment suggests that at about 2 percent growth rates, unemployment should remain little changed. If Fernald is right, the break-even growth rate to reduce joblessness would be lower than the traditional view suggests. So there would still be a puzzle, but less of one.

The crucial question about longer-than-expected effects from the financial crisis is, what should we do about them? Passive despair, after all, isn't an ideal strategy.

Which brings me back, once again, to the idea of a barbell fiscal policy. Policymakers in Washington should couple substantial upfront stimulus spending with even bigger, but delayed, deficit reduction. Both ends of this barbell are crucial: The stimulus can help to reduce the lasting effects of the crisis, and, given that the official deficit projections may be too sunny, the austerity will help prevent a future fiscal crisis. Furthermore, the combination is more politically feasible than either component alone.

Unfortunately, it seems likely that neither side of the barbell will be enacted. In failing to act aggressively, we are increasing the risks for both slower growth and future trouble.

-----

Peter Orszag is vice chairman of corporate and investment banking at Citigroup and a former director of the Office of Management and Budget in the Obama administration.