Introduction: Srivastava and Sen (1997) have advanced a framework for the study of government subsidies in India. Their approach estimates subsidy as the un-recovered costs in the provision of goods and services by the government (see the next section for various definitions of subsidy). Specifically, for the state of Maharashtra, the subsidy was estimated at Rupees 9607.41 Crores (Annexure 15, pg 151, NIPFP report) for the year 1993-94, while the Gross Fiscal Deficit (GFD) for the same year stood at Rupees 2265.3 Crores. As a proportion of Gross State Domestic Product (GSDP), these magnitudes were 8.5 and 2 per cent, respectively. The estimated subsidy constituted about 65 (73) per cent of the total (revenue) expenditure.5 Contrast this with the budgetary subsidies estimated between 2 and 2.35 per cent of GSDP by the Finance Department of the Government of Maharashtra (GoM) as shown in Table 1. Excluding the grants-in-aid, the estimate of subsidy varies between 0.8 to 1.05 percent of GSDP. Thus, estimates of subsidies vary widely between the official report (of GoM) and the NIPFP report. The present paper is not an attempt at comparing or reconciling these two estimates. Instead it focuses on the fundamental question of whether the governments budgetary subsidies, estimated as un-recovered costs, can exceed the GFD. The query came up specifically in the context of deciphering the sources of financing of the subsidies (both explicit and implicit) and thus ascertaining who bears the costs of the subsidies. Certain costs are borne by the society at large in terms of loss of productivity and efficiency. These maybe estimated as the social dead weight losses but they may or may not impinge upon the government budget. Subsidies that impact upon the budget must be a part of the expenditures (on goods and services provision) of the government. So long as public expenditure (on goods and services) is financed by tax and non-tax revenues, subsidies represent inter-personal transfers and redistribution, with the government acting as facilitator. The inter-personal transfers are generally achieved through price discrimination across different sections of consumers. So long as such price variations are revenue neutral they have an impact on the resource allocation mechanism but do not influence the sustainability of the government expenditure program. Often government expenditure exceeds the sum of tax and non-tax revenue. The revenues then constitute the recovered costs of government expenditures, while the un-recovered costs have to be financed by borrowings. The total borrowing requirements of the government from all sources are known as the GFD. The GFD is, therefore, a measure of the extent to which the economy is living beyond its present means (income). In reality a substantial component of the GFD may actually represent investment with only a part of it subsidizing the present consumption plan. Even if all the borrowings were assumed to be financing the present consumption plan, this measure of subsidy should not exceed the GFD. A relevant objective here would be to minimize this component of GFD. In this paper we explore the reasons for the wide gaps in the measure of fiscal deficit and the estimate of aggregate subsidy and suggest an improvement in the methodology to estimate the latter. The plan of this paper is as follows. Section II discusses the meaning, scope and definition of subsidy to dispel some of the myths associated with the term. In section III a simple algebraic structure is presented to provide a theoretical ceiling on aggregate subsidy. Section IV elucidates the economic rationale for subsidies and the need to study their impact / incidence as a significant policy tool. Section V critically analyses the methodology followed by NIPFP and outlines the reasons for the errors in the estimates. An alternative formulation to estimate the un-recovered costs (net aggregate subsidy) is then advanced. Finally, section VI concludes by emphasizing the need for reconciliation between the fiscal deficit and aggregate subsidy estimation and the consequent need for broader macroeconomic consistency.