Thomas Jefferson, no fan of deficit spending, has a clearer view of the US fiscal stimulus than any living politician. From his memorial by the Tidal Basin in Washington he can watch a $12.4m stimulus project to fix a seawall and stop him sinking into the water.
The site is cluttered with machinery and rolls of steel mesh. It employs about 20 workers. That makes it hard to deny that the stimulus is doing something – but Republican critics point to a 9.5 per cent unemployment rate to argue that it has failed. Democratic prospects in elections this autumn and any push for more spending rest on proving otherwise.
That is difficult. At one Congressional hearing, Mark Zandi, chief economist at Moody’s Analytics, said each dollar spent on infrastructure was creating $1.57 in gross domestic product within a year, and that without the stimulus there would be 2m fewer jobs today.
He was followed by John Taylor, an economist at Stanford University, who said government spending “had little to do with the turnaround in economic activity”. Mr Taylor estimated the effect of the stimulus at about 70 cents of GDP per dollar spent.
The trouble is that economists cannot try the recession again – this time without government-built seawalls – to see what would happen. Instead, they must rely on models and historic data to judge how bad the recession would have been without a fiscal boost. In some models, stimulus works well, while in others it cannot work at all, which is why economists get such different results.
The highest estimates of stimulus effects – as much as $2 of GDP per dollar spent – come from large macroeconomic models run by the Federal Reserve and firms such as Moody’s Analytics. These models describe every sector of the economy and are where the Obama administration gets its stimulus numbers.
Some academic economists criticise large macro models for relying on ad hoc number-crunching rather than theory. Robert Barro of Harvard University argues that they find big effects for stimulus because they assume that higher government spending causes higher output, whereas it might be the other way around. Gus Faucher, director of macroeconomics for Economy.com, responds that the assumption is reasonable if the spending comes before the output rise.
Mr Barro’s own work finds that an extra dollar of defence spending increases GDP by only 40-50 cents in the short term. But critics say that estimate is not relevant because it relies on numbers from the second world war, when rationing kept private spending down.
Mr Taylor says large macro models have “less emphasis on expectations” than the “New Keynesian” models he uses. Many of these models, which are common in academia, assume that people anticipate future taxes to pay for the stimulus. That limits any rise in their spending.
New Keynesian models have also come in for criticism because most of them do not include a financial sector and so struggle to explain the recession at all.
“Until folks like Taylor can explain how we got into the recession then it is hard to be convinced by their conclusions on stimulus,” says Mr Faucher.
Many New Keynesian modellers estimate the effect of stimulus at between 50 cents and $1 per dollar spent in normal times. But the same researchers find numbers as high as $3.90 when interest rates are zero, as in 2009 and today, and the central bank cannot cut them to stimulate the economy.
“In an economy with an output multiplier for government purchases of just under 1 in normal times, the multiplier rises to 1.7 when monetary policy becomes passive with a zero nominal interest rate,” wrote Robert Hall, president of the American Economic Association, in a paper.
That is something close to a consensus among economists – albeit with high-profile dissenters such as Mr Barro and Mr Taylor – and suggests that the stimulus did have a large influence. Without more direct evidence, however, proponents of more spending will struggle to overcome fears about adding to the national debt.