What would have been
UPDATED: 16 February 2010
The government often claims that they saved Australia from recession, and they base this claim on some Treasury modelling.
The Treasury often does very good work, but I think that some people fall into the trap of assuming that Treasury is some sort of gold-standard of economic forecasting, when the truth is that their track record is mediocre. This is not intended as a criticism of Treasury, but of people who put too much faith in Treasury. Forecasting is a tough game, and economic models only ever provide a rough guide. I should know — I used to do forecasting for the Commonwealth Treasury.
So instead of assuming Treasury is right, I thought I’d make use of the last year’s extra economic data (Treasury’s budget papers and the National Accounts) and try to work out what actually would have happened without the government’s two stimulus packages.
I’ll jump straight to the conclusions.
The stimulus increased GDP in the Dec 2008 (0.7%) and March 2009 (0.7%) quarters, and then contracted GDP in the June 2009 (-0.2%), Sept 2009 (-0.9%) and Dec 2009 (-0.2%) quarters. The net impact over the five quarters was no change in GDP, though the growth was brought forward. This “benefit” needs to be compared with the cost of higher debt and higher future tax.
The reason for the original boost is stronger consumption due to the stimulus handouts. The later contractions are caused by the unwinding of the stimulus as well as the negative impact on net exports due to increased foreign borrowing and relatively higher interest rates.
The GDP growth (with stimulus) for the five quarters starting in September 2008 were: 0.2%, -0.8%, 0.6%, 0.7%, 0.2%. The GDP growth estimates (without stimulus) for the five quarters starting in September 2008 are: 0.2%, -1.5%, -0.1%, 0.9%, 1.1%. I haven’t included the December 2009 quarter because that data is not yet available. Chart 1 shows the actual GDP growth (blue line) with a “no-stimulus” counter-factual (red line)
It is a close call whether the stimulus “saved” us from recession. Two quarters of negative growth are technically a recession, so if my estimates are accurate, then the government is correct by 0.1% (see bolded number above).
However, as I’ve argued elsewhere, it is more important to look at GDP/person. Using a GDP/person basis, the growth rates (with stimulus) were: -0.3%, -1.3%, 0%, 0%, -0.5%. Using a GDP/person basis, the estimated growth rates (without stimulus) are: -0.3%, -2.0%, -0.6%, 0.2%, 0.4%. The interesting thing to note here is that it is the stimulus that caused the Sept 2009 growth to be negative (see bold numbers). Chart 2 shows the actual GDP/person growth (blue line) with a “no-stimulus” counter-factual (red line).
Starting with the national accounts, I took off the benefit from the stimulus package. For private spending, I assumed that 50% of the handouts were saved and 50% was spent (slightly higher than the estimates provided by Andrew Leigh). For direct government investment, I assumed all was spent.
I also assumed that 50% of government debt was financed overseas, leading to international crowding out (ie lower net exports). Of the remaining borrowing, I assumed that domestic crowding out (ie less business investment) was 10%. The results were not sensitive to this last assumption.
For the December handouts I assumed that 40% of the spending was done in the Dec 2008 quarter, 50% in the March 2009 quarter and 10% in the June 2009 quarter. For the March handouts I assumed that 20% of the spending was done in the March 2009 quarter, 70% in the June 2009 quarter and 10% in the Sept 2009 quarter. I assumed that government spending and business incentives were spent roughly evenly. I got the estimates for government spending from the budget documents and from the stimulus progress report.
I assumed that the RBA adjusts monetary policy in order to achieve set montary goals. As effective fiscal policy leads to an increase in broad money (through in a higher credit multiplier) then the RBA should adjust their monetary policy in response to fiscal policy. In other words, without the stimulus, interest rates would have been marginally lower.
However, following Keynesian convention, I assumed that there was a 50% “liquidity trap” where any further monetary policy would have been only 50% effective. That is, I assumed fiscal policy was twice as effective as the monetary alternative. I also assumed that the government spent money just as efficiently as the private sector would have spent the money. I think these assumptions are biased in favour of the stimulus.
If we assume no liquidity trap then the stimulus was about -0.4% less effective over the last five quarters. If we also assume that private spending is 50% more efficient than public spending then the stimulus was about -0.8% less effective over the last five quarters. If we assume that fiscal policy has no impact on monetary policy (or that the “liquidity trap” was 100%) then the stimulus was about 0.5% more effective over the last five quarters.
For international crowding out, domestic crowding out, and impact of interest rates, I assumed the effects were lagged by 3-6 months (50% @ 3 months & 50% @ 6 months).