The Supply-Use input-output model of the SNA and Eurostat is examined. For the product-by-product IO tables, two hypothesis are possible: "product technology", largely adopted (Eurostat A) and examined here, and "industry technology" (Eurostat B). One examines the calculability of the model. Negatives are an issue; (i) they are systematical in the inverse supply matrix: negative probabilities are impossible, Markov chains become impossible; (ii) negative flows are nonsense in symmetric matrix of coefficients or in inverse matrices. Matrices must be square what removes much of their interest. Same conclusions can be transposed to industry-by-industry IO tables under fixed-industry-sales-structure assumption (Eurostat C).