This note studies graphical identification in the canonical competitive demand-supply model. Using Pearl's (2009) latent projection with explicit demand and supply shocks, it derives an acyclic directed mixed graph (ADMG) representation that accommodates equilibrium feedback while preserving standard econometric assumptions. Within this framework, the familiar instrumental-variables estimand arises on observables, but graphical identification of a structural slope requires an explicit normalization of the structural equations. When price is normalized as the dependent variable in the demand equation, Pearl's instrumental-variables identification theorem applies directly to recover the demand slope; an alternative normalization yields identification of the supply slope. An extension to instruments with side effects shows how graphical separation criteria make the required conditioning set and resulting estimand transparent when conventional IV arguments become cumbersome. The results clarify the relationship between reduced-form equilibrium representations and structural causal identification in simultaneous-equation models.