Theory-based Empirical Asset Pricing Research (TBEAR) Network

The TBEAR Network

How do risk premia for financial assets vary over time and across states of the world? How does this variation affect the behavior of asset prices over time and differences across assets?

While these questions are not new in asset pricing research, the aim of the TBEAR network is to promote theory-based empirical research in asset pricing to approach these questions. For that purpose, we organize annual workshops that bring together leading researchers working at the nexus of empirical and theoretical asset pricing. By presenting and discussing recent research, we expect to gain insights into the sources of risk premia that go beyond their mere statistical description, which has often been the focus of past and current asset pricing research.

The initiative is funded by the German Research Foundation (DFG). Type of funding: Scientific network. Project number: 454790565.


Since the seminal article by Fama and French (1992), the empirical asset pricing literature has been dominated by studies that statistically demonstrate alleged relations between stock or firm characteristics and unconditional expected returns in the cross-section of stocks. This approach is problematic for two reasons. First, empirical tests should always start with a clear hypothesis derived from economic theory, which allows a clear interpretation of the test results. However, many empirical asset pricing papers lack this theoretical foundation. Instead, papers often define statistical factors and argue that these factors must be related to the marginal utility of investors, without giving a theoretical justification. Second, due to the lack of theoretical discipline in the derivation of testable hypotheses, research has largely focused on screening the data for regularities and correlations between stock characteristics and subsequent returns. Cochrane (2011) and Harvey, Liu, and Zhu (2016) have established the term “factor zoo” and conclude that the usual criteria for statistical inference no longer applied, since empirical asset pricing research had put itself in a multiple testing environment.

The approach of our network is to eschew the roots of this statistical problem. The aim is to search specifically for implications from asset pricing theory regarding the cross-sectional and time series properties of risk premia in financial markets and to translate these into clear-cut testable hypotheses. Since large parts of the empirical literature test unconditional factor models, but asset pricing models often imply conditional factor models (i.e., models with time-varying risk premia), sophisticated econometric methods will be necessary to bring the testable hypotheses to the data. In order to advance asset pricing research in this direction, it is necessary to gain an integral view of the research question discussed above. To this end, the network is supposed to bring together researchers in the field of asset pricing with special expertise in theory, empirics, and econometric methods. It serves as a forum to stimulate collaboration between researchers from these different areas.