To accomplish the first task, the Data Envelopment Analysis (DEA) model will be applied to estimate technical efficiency of government spending (ΔG) in raising GDP (ΔGDP) with government spending multipliers being calculated first. The inefficiency scores of each country for each year will be recorded for the second task.
The Extreme Bounds Analysis (EBA) approach in association with the truncated Tobit regression will be adopted to carry out the second task. As to testing for factors that could cause inefficiency of government spending, several hypotheses will be examined in the ΔG -ΔGDP nexus. The first is whether private sector’s activities (such as consumption, investment, and foreign trade) foster government performance. The second is the government corruption hypothesis. The third hypothesis is about the relationship between monetary expansion and government spending in promoting growth. The fourth hypothesis is about the effect of government size. The fifth hypothesis to be tested is about the effect of government debt. In order to perform the robustness test using EBA, several macro variables will be chosen as exogenous (Z) variables.
The results of DEA show that the United States, New Zealand and Germany are the countries having the highest efficiency scores in the OECD sample. As the OECD sample is concerned, EBA method in association with Tobit regression indicates that private sector activities exhibit a robust negative relationship with government inefficiency, which means that increasing the share of private activities in the economy helps reduce the inefficiency of public spending. The Corruption Perception Index (CPI) indicator reveals a not robust effect on government inefficiency in OECD group. The reason might be due to the fact that OECD countries under study are mostly of less degree of corruption and with less variation among their indices. The EBA results indicate that M3 expansion is a robust positive indicator that remains significant and positive within the range. Government size indicated by the Revenue/GDP ratio is not robust. Although the debt/GDP ratio carries a negatively significant coefficient in the basic models discussed above, it does not come out as a robust variable in the EBA robustness tests.