A fiscal tsunami is soon to wash across the globe with barely a ripple of dissent. That is in spite of oceans of conflicting economic research on whether fiscal packages actually raise output during downturns. Politicians and academics, however, need not worry about being proved wrong because the other feature of fiscal boosts is that it is impossible to say conclusively what effect they have.

The confusion arises from problems common to all economic analysis. For a start, no two fiscal packages are ever launched amid the same circumstances. That makes it wrong to extrapolate from the past. The era of Roosevelt’s New Deal followed by the US entering the second world war, for example, could not be more different from today. Another issue is choosing the dates between which to crunch the numbers. Does fiscal policy begin when governments press the button or when individuals and companies first hear of the plans? Finally, there is the impossibility of measuring the counterfactual – what would have happened if there had been no stimulus.

But the biggest headache to overcome, according to the International Monetary Fund, is that fiscal policy and economic fluctuations interact perpetually with each other. Government deficits always increase in a downturn as tax revenues fall and unemployment payments kick in. Measuring the separate effect of any one-off fiscal boost from this “automatic” extra spending is therefore hard. The IMF concludes that large fiscal packages can help during a downturn, albeit only modestly.

It is not surprising that measuring complex economic systems is tricky. After all, academics cannot even agree on how fiscal policy influences individuals. Do most people spend tax cuts or save the rebate in the expectation that it will have to be repaid eventually through higher future taxes? Naturally, such arguments are absent from speeches such as Barack Obama’s yesterday. Perhaps his ambitious fiscal plan will work. But even if it does not, it will be almost impossible for anyone to prove it one way or the other.