Jonathan
Introduction#
I like the structure of the introduction: you introduce QE in a general way, and show anecdotally that QE might have a differential impact on fin. markets and ‘real’ economic outcomes such as unemployment.
- If you want, you can say something about the method / results / structure of the thesis at the end of the introduction.
Literature Review#
I also agree with this structure: QE $\Rightarrow$ Financial Markets, QE $\Rightarrow$ Economic Growth, FM $\Rightarrow$ Economig Growth,
Methodology#
I find it convincing that you use a VAR-framework to estimate your models: it is a system of equations after all (as we talked about last time), and it allows for both contemporaneous and lagged impact of Fin. Markets and QE on Economic Growth.
With a VAR regression each independent variable will be a function of its own lag.
It can also be a function of a contemporaneous (and lagged) value of other variables, which I think is crucial for your analysis (you want to explain Ec. Growth as a function of lagged versions of itself, but also as a function of contemporaneous and lagged QE. VAR looks at the data as two interrelated stationary processes which are dependent on each other via a number of lags, and the dependence of models as being linear.
-
Later, I noticed that this was exactly what you had in mind (page 16)
-
You can probably do a Hausman-like test to comapre the coefficients of QE on Fin. Market Growth and Ec. Growth. (A Hausman test generally tests whether sets of coefficients from different regressions are equal).
-
Page 16/17: The pairwise analysis can just be mentioned in one sentence, I guess you don’t need to spell it out in full!
Analysis#
Very nice that you have done some analysis already! Make sure to not use raw Stata output. You can use the package outreg to export to latex or Word: https://www.princeton.edu/~otorres/Outreg2.pdf
As these variables are time series, show them graphically! Construct a few graphs with Stata’s twoway or R’s ggplot!
You can also think of “forest plots” to plot contemporaneous and lagged coefficients to visualize the different between various models.
You can also create Impulse Response Functions from the VAR estimates and visualize what will happen to the three interrelated series if there is a stochastic shock in one of the variables.