This is a difficult question which has been often studied in many different ways. Now, Sebastian Gechert of the German Institut für Makroökonomie und Konjunktur has published a working paper, in which he performs a statistical meta-study of 104 international studies on this topic.
His conclusion should be familiar to anyone with even the slightest Keynesian bent: public investment has the best value-for money to kick-start/speed up economic growth, or in other words, it has the biggest multiplier.
Public investment seems to be the most effective fiscal impulse […] (p.35)
and in particular this:
Ninth, setting up a crisis scenario with a fixed nominal interest rate (ZLB), lowered ability of consumption smoothing (increased share of non-Ricardian agents) and a concerted action of the country under investigation and its trading partners (closed economy assumption) for the average of the whole set of models in our simulation-based sample
yields multipliers close to two, which implies a higher effectiveness of fiscal policy in times of the recent crisis years and stronger negative impact of the concerted austerity measures in the Euro Area.
Obviously, the assumptions are not entirely realistic but:
As an overall conclusion, reported multipliers very much depend on the setting and method chosen. Thus, economic policy consulting based on a certain multiplier study should signal, how strongly results are influenced by specification. Our meta analysis
may provide guidance concerning such influential specifications and their direction and
In other words, public investment is a good idea but its estimation of effect depends very much on the model used. This is why it is very important to be critical: the estimates produced by official actors can easily be more or less robust depending on the model chosen to be used.