We explore the normative fiscal assessments of the Finance Commission (FC) of India, and realisation of fiscal policy with regard to Central Finances over the period 1990–2012. We employ the Theil’s inequality coefficient to investigate the magnitude of assessment errors and its partitioning in to bias, slope and random components. Furthermore, this article also evaluates the efficiency, biasedness and persistence of forecast errors. The robustness of the efficiency results are confirmed with the application of maximum entropy bootstrap. The objective of this study is to examine the structural basis on which FCs make their awards rather than examining the predictability of the forecasts. The story of FC’s assessments reflects an interesting political economy theatre of contention between aspirations and outcomes. Our key findings are as follows: First, source of errors for assessments of tax revenue, nontax revenue, interest payments, defence revenue expenditure, plan revenue expenditure and fiscal deficit is principally due to random component. However, the errors in the remaining economic parameters originate due to systematic components i.e. mean and slope errors. Second, the expenditure side predictability is lower than the revenue side predictability.