Empirical analysis of bitcoin price

This paper, merging the stories of monetary theory, management analysis, computer science, and finance, comprehensively studies different forces that affect the bitcoin price. After conducting various stationarity tests and cointegration test, I choose the VEC model as the baseline to estimate the bitcoin price empirically. I also include competing methodologies, which have been used in previous studies, on my data set. These methodologies are VAR and ADRL models. With the daily data of 2009–2019, my baseline model shows that in the short run the bitcoin price is mainly affected by the medium of exchange and financial expectation forces, while blockchain technology factors only show a small impact on the bitcoin price. Moreover, using different econometric models yields different results in the short run. To investigate whether some conflicts among previous research can be explained by different specifications of the data, I also conduct two kinds of robustness checks. One is to estimate the same variables between bear and bull states. The states are found using Markov switching model. The other check is to focus on supply and demand forces, particularly during high volatility periods. Both checks find that in different states the effects can be different both in the size or in the significance. Lastly, I also find that effects are different between the short run and the long run.