viewed as something similar to an «income effect». Second, if per capita costs of provision go down, the demand and the optimal level of pro - vision will increase (similar to a «substitution effect»). Note that this last argument does not contradict the definition of a non-rival public good, since it touches upon quantity of provision (e.g., more soldiers, more parks) and not on the quantity of consumers (which of course may be infinite for non-rival public goods without altering costs per defini - tion). In the case of a dominating «substitution effect», the right-hand side of equation (4) should be positive. The more intuitive notion of a domi- nating «income effect» complies with a negative sign on the right-hand side of equation (4). Thus, smaller countries should have relatively higher public expenditure or larger governmental and/or public sectors than larger countries. This is exactly what is meant by diseconomies of scale in the provision of public goods. The empirical examination in Section 3.2 is, moreover, a test whether the right-hand side of (4) is actually 
negative. 3.1.2   Further important determinants and theoretical predictions The simple model in the previous subsection is naturally insufficient to grasp further determinants of government size aside from economies of scale that seem of importance. Specifically, it is only valid when we con- sider economies of scale in the provision of public goods as the sole source for the level and structure of public expenditure. There are many arguments that may question that proposition. Although we have men- tioned a lot of other determinants of government size in Table A.2, the- re are some determinants which are more closely related to country size and population. They will be discussed qualitatively in the following and taken up again in subsequent chapters of this study. Some of them, for which we conjecture a significant effect on public sector size, reappear as control variables in the regressions in Section 3.2: –A lot of publicly provided goods are clearly rival or partially rival. In fact, they display optimal scales of production that are signifi- cantly smaller than the number of inhabitants of even a small coun- 41 
Public sector size and country size in theory