The study http://www3.eeg.uminho.pt/economia/n..._WP_6_2009.pdf is the most comprehensive to date,covering data for 145 countries from 1960 to 2007. Its principal conclusion:
"...an increase of government consumption by 1% of real GDP immediately reduces consumption (investment) by approximately 1.2% (0.6%).with the decline continuing for about four years when the cumulative decrease in consumption has reached approximately 1.9% (1.8%)."
This effect does not depend on the phase of the business cycle.This finding contradicts the Keynesian model that promotes government spending on infrastucture and social welfare in recessions in order to exploit economic slack capacity and unemployment and to support consumer spending. If the study is correct,maybe Keynesian spending can only be guaranteed to work in economic depressions and ruinous financial panics as demonstrated in the 1930s Great Depression in the US,the UK and Germany. Hitler's economic programme in the 1930s has been described as military Keynesianism,while the US didn't fully emerge from the Depression until WWII's surge in military spending.
An interesting question is suggested by the study: if increases in government consumption depress the private sector,doesn't that strongly imply that cuts in government spending would have the opposite effect and grow the private sector? Some research suggests that the effect of cuts could be moderately negative in the short run and beneficial after a few years. In my opinion,the higher the level of government spending relative to the GNP,the more beneficial the effect of the cuts: Economies with government spending around 60% of GNP such as Sweden in the 1980s and Ireland at present would benefit more from cuts than third world economies with parsimonious spending of 12 to 15% of GNP.