We provide one of the first estimates of elasticities of substitution across inputs supplied by suppliers within the same industry. This elasticity is particularly relevant for the transmission and amplification of supply shocks across the production network. We obtain new real-time administrative tax data on product-level prices and quantities with firm-to-firm transactions. We leverage geographic and temporal variation from the Covid-19 lockdowns in India to estimate these firm-level elasticities of substitution and quantify the fall in trade. If suppliers are complements rather than substitutes in production, this shock can amplify by further transmitting downstream and upstream through the supply chain. We find that even at this very granular supplier level, inputs are highly complementary, with an estimated elasticity of 0.55. Causally estimating these micro-level elasticities of substitution at the firm level allows us to understand how shocks propagate through supply chains, affecting aggregate GDP. We use our elasticities and simulate the impact of the Covid-19 lockdowns to find that under our estimated elasticities, the overall fall in output is substantial and widespread.