Looking for the Holy Grail of asset pricing

Could you briefly explain what your thesis is about?

"My thesis deals with understanding and refining statistical tools that are commonly used for analyzing financial data. I focus on one of the largest financial markets, the stock market, with the general aim to identify relevant factors that determine changes in stock prices. Building on well-established methods in this context, I point out that the structure of the data has an important effect on how the different tools perform. Furthermore, I suggest corrections for already existing methods to account for the specific features, such as noisiness and persistence, of the data."

Why did you choose to study this subject?

"I find this topic fascinating because we are surrounded everyday by a vast amount of financial data, yet we have relatively limited understanding of the patterns that generate them. It seems almost impossible to say something about the price change of a specific stock on a specific day. However, there is a lot to learn about the bigger picture: how stock markets overall respond to changes in the economic environment. And to study this, we need the rigor of mathematics to help us process and structure the available information."

So what are the key findings?

"The general finding of the thesis is that the performance of the statistical models that we use highly depends on the structure of the data that we feed into them. I analyse the predictive regression, a simple yet commonly used tool, and I look at estimation in two specific cases. Firstly when the predictor is a noisy proxy of return expectations, and secondly when the explanatory variables are persistent. In both cases, the explanatory power of the regressions improves, once these dynamic properties of the variables are taken into account. Furthermore, I analyse a test on return predictability that is based on a simple accounting identity between returns, prices and dividends. I show that even this economically well-grounded test leads to statistical distortions if the time-series dynamics of the parameters of the model are not treated properly."

What are the implications of these results?

"An important implication is that empirical work with financial data should always consider the dynamic structure of the variables, and, if necessary, make adjustments to the statistical tools that are used in the analysis. In my thesis, I provide these adjustment tools for the two specific cases above, which can easily be implemented in practice. I think the results are important, because they extend our understanding of how stock returns behave and they mean a few steps forward in the quest for finding the true empirical model of asset returns, the Holy Grail of asset pricing."

Who could benefit from your study and how?

"The thesis can benefit mainly two distinct groups. First, the topic is relevant for individual investors, who want to make informed asset allocation decisions. For them, understanding patterns in the asset markets can directly translate into material gains, in terms of higher risk-adjusted returns. Second, my study is relevant for financial supervisory authorities, because understanding how financial markets actually work helps devise a regulatory system that protects well against the malfunctioning of the financial system."

Now that you completed the doctoral programme, what are your future plans?

"Apart from continued research on the statistical properties of stock return predictions, I would also like to look at the interaction between financial markets and the real economy more closely. Specifically, I want to study what kind of information from the real economy is incorporated in asset prices. Related to that, I have two potential projects in the pipeline. The first one is on macroeconomic announcements and asset returns, where I plan to look at the heterogeneous effect of the arrival of macroeconomic news on beliefs about firms performance and hence on individual stock prices. The other idea is about using textual analysis to connect textual information in firms’ annual reports or earnings announcements to stock returns."

Tamás Kiss defends his thesis "Predictability in Equity Markets: Estimations and Inference" at the School of Business, Economics and Law 29 May 2019.