The effect of aggregation on nonlinearity
通过蒙特卡洛模拟，研究截面加总、时间加总和系统抽样三种加总方式如何削弱非线性，发现加总程度越大、共同因子越小，非线性简化越明显。
This paper investigates the interaction between aggregation and nonlinearity through a monte carlo study. Various tests for neglected nonlinearity are used to compare the power of the tests for different nonlinear models to different levels of aggregation. Three types of aggregation, namely, cross-sectional aggregation, temporal aggregation and systematic sampling are considered. Aggregation is inclined to simplify nonlinearity. The degree to which nonlinearity is reduced depends on the importance of common factor and extent of the aggregation. The effect is larger when the size of common factor is smaller and when the extent of the aggregation is larger.