Problems with estimating beta

Suppose that the beta estimate and $R^2$ for a model vary greatly depending on whether the NASDQ or S&P $500$ is used as a market portfolio proxy.

The beta estimates most likely suffer from which of the following forms of bias?

Time interval bias

Nonstationary beta problem

Market index bias