This paper considers two different benefit measures of a transportation improvement in a general equilibrium framework. A spatial general equilibrium model with three regions and two primary inputs is used to simulate the impacts of a small as well as large investment in one single link of the network. The first approach is to measure total benefits to the economy as the change in consumer surplus under the general equilibrium demand curve for transports on the improved link. In the second approach the benefits are measured as equivalent variation aggregated across all three regions, which represent the economy-wide welfare improvement. The general equilibrium measure is then compared to the link-oriented measure through a ”true benefit multiplier”, in order to evaluate if the latter correctly captures total benefits to the economy. The simulations are performed both under the assumption of a first-best economy and when increasing returns to scale is present in the production of goods. It is found that total benefits due to the improvement can be fully captured by only considering transport demand on the improved link, given that the economy is a first-best economy. However, if increasing returns to scale is introduced into the model, the link-related measure tends to underestimate total benefits. The diversion between the two measures gradually increases as the scale parameter gets larger. The result shows that given an economy under influence of increasing returns to scale, it is highly relevant to take generated network impacts into consideration. It is also found that the level of flexibility in the economy, reflected by different values in elasticity of substitution in the production process, is a determinant factor for the benefits of an improvement.