Small Business Property Tax Reductions and Firm Productivity

We investigate the productivity spillovers from the UK government’s decision to use extensive property tax reductions as a key instrument to stimulate innovation in smaller businesses and drive local growth. To capture the complex interaction and clustering of hierarchical effects, we apply non-parametric Random Effects Expectation Maximisation algorithm that complements more standard econometric estimators, namely matching to control for endogeneity and control functions to estimate total factor productivity. These approaches enabled us to incorporate various contextual configurations in comparing the recipients of these reductions to non-recipients with regard to productivity, in which the UK has experienced a considerably worse performance than its peers since the great recession. Contrary to policy assumptions and business community expectations, we show that generic tax reductions, when significant, are mostly associated with lower productivity and thus have been unsuitably chosen as a policy mechanism to stimulate productivity growth. We further show how instruments that are not built for causality could be beneficial for policy evaluation. Reductions in small business property tax will not solve UK productivity problem. We investigate the productivity spillovers from the UK government’s decision to use extensive property tax reductions as a key instrument to stimulate innovation in smaller businesses and drive local growth. We incorporate various contextual interactions in comparing the recipients of these reductions to non-recipients with regard to productivity, in which the UK has experienced a considerably worse performance than its peers since the great recession. Contrary to policy assumptions and business community expectations, we show that generic tax reductions, when significant, are mostly associated with lower productivity and thus have been unsuitably chosen as a policy mechanism to stimulate productivity growth. Thus, the principal implication of this study is that instead of generic tax incentives based on building and land value, policy should be more nuanced and targeted, enabling the inclusion of finer-grained characteristics. We also contribute to methodology development by showing how configuration approaches, centring on non-linearity and equifinality, could be enabled through the application of machine learning to study organisations.