Paul Romer's 1987 paper "Growth Based on Increasing Returns Due to Specialization" examines dynamic models with specialized inputs and "variety in production." It demonstrates that monopolistically competitive equilibria with unceasing growth can be explicitly calculated. The paper explores how specialization can lead to increasing returns, a concept that has been recognized in economics for a long time. Romer builds on earlier work by Dixit and Stiglitz, who used a utility function to capture preferences for variety. Romer reinterprets this function as a production function, showing that output increases with the number of specialized intermediate inputs. The paper introduces a production function that incorporates both labor and intermediate inputs, with a focus on the role of specialization. It also considers the implications of fixed costs and the effects of spillovers of knowledge. Romer argues that while models of increasing returns from knowledge accumulation are well-established, models based on specialization are less developed. The paper presents a model where the range of intermediate inputs (M) and the amount of each input (N) determine output. It shows that output increases with M, the number of different inputs, when labor and total quantity of intermediate inputs are held constant. The paper also discusses the implications of a decentralized equilibrium where firms are price takers. It shows that the equilibrium price of a resource (Z) is determined by the requirement that profits for intermediate goods producers must be zero. The paper highlights the suboptimality of the decentralized equilibrium compared to the first best social optimum, where the marginal value of an additional unit of Z is higher than the market price. This suggests that the decentralized equilibrium leads to less accumulation of Z than would be socially optimal. The paper then extends the static model to a dynamic model, allowing for the accumulation of the primary resource Z, interpreted as a durable, general-purpose capital good. It shows that the dynamic model can support unbounded growth, with consumption and the stock of Z growing at a constant rate. The paper also discusses the implications of this growth model, showing that it behaves similarly to a model with exogenous technological change. The paper concludes that the model captures the idea of external economies associated with specialization, similar to the intuition behind Marshall's use of the term. The analysis shows that the model is not one with a true positive externality but behaves as if one were present. The paper also discusses the implications of different functional forms for the production function and the cost function, showing that the results are robust to these changes.Paul Romer's 1987 paper "Growth Based on Increasing Returns Due to Specialization" examines dynamic models with specialized inputs and "variety in production." It demonstrates that monopolistically competitive equilibria with unceasing growth can be explicitly calculated. The paper explores how specialization can lead to increasing returns, a concept that has been recognized in economics for a long time. Romer builds on earlier work by Dixit and Stiglitz, who used a utility function to capture preferences for variety. Romer reinterprets this function as a production function, showing that output increases with the number of specialized intermediate inputs. The paper introduces a production function that incorporates both labor and intermediate inputs, with a focus on the role of specialization. It also considers the implications of fixed costs and the effects of spillovers of knowledge. Romer argues that while models of increasing returns from knowledge accumulation are well-established, models based on specialization are less developed. The paper presents a model where the range of intermediate inputs (M) and the amount of each input (N) determine output. It shows that output increases with M, the number of different inputs, when labor and total quantity of intermediate inputs are held constant. The paper also discusses the implications of a decentralized equilibrium where firms are price takers. It shows that the equilibrium price of a resource (Z) is determined by the requirement that profits for intermediate goods producers must be zero. The paper highlights the suboptimality of the decentralized equilibrium compared to the first best social optimum, where the marginal value of an additional unit of Z is higher than the market price. This suggests that the decentralized equilibrium leads to less accumulation of Z than would be socially optimal. The paper then extends the static model to a dynamic model, allowing for the accumulation of the primary resource Z, interpreted as a durable, general-purpose capital good. It shows that the dynamic model can support unbounded growth, with consumption and the stock of Z growing at a constant rate. The paper also discusses the implications of this growth model, showing that it behaves similarly to a model with exogenous technological change. The paper concludes that the model captures the idea of external economies associated with specialization, similar to the intuition behind Marshall's use of the term. The analysis shows that the model is not one with a true positive externality but behaves as if one were present. The paper also discusses the implications of different functional forms for the production function and the cost function, showing that the results are robust to these changes.