Existing literature on the link between government expenditure and economic growth shows mixed results. This paper, while analysing the impact of government expenditure on economic growth using the ARDL framework, also investigates whether the difference in data types is one of the reasons for the conflicting findings reported in the literature about Wagner's law and Keynesian hypothesis. We use Sri Lanka as a case study for this analysis and use three different forms of model specifications at the aggregate and disaggregated (sectoral) levels. Results indicate that both Wagner's and Keynesian's propositions are supported by the Sri Lankan data in the long-run for all three model specifications and the results are inconclusive in the short-run. This suggests that Wagner's law and Keynesian hypothesis are sensitive to the data type used and the results could be different depending on whether the analysis is made in the short-run or long-run. Capital and recurrent expenditures are found to have a positive effect on economic growth in the short-run as well as long-run. At the disaggregated level, expenditure on agriculture and health positively impact economic growth in the long-run, while welfare expenditure has a negative effect. These findings provide important insights for policymakers to consider when allocating sectoral level expenditure budget. This study also provides insights on allocating government expenditure towards achieving United Nations’ Sustainable Development Goals which countries across the globe are aiming to achieve by 2030.