So I am writing an econ paper and will be using regressions. I have written my intro / lit review / collected relevant data. I have Swiss quarterly and yearly export data by industry from 2009-2018. In 2015, the Swiss National Bank removed its fixed exchange rate (was pegged to the euro)… I am interested in measuring the effect of large sudden currency appreciation which occurred after the depeg. I have all my data, and ultimately just want to estimate the impact using pre and post depeg data. I am not entirely sure how to build my model (I am not a great statistician or econometrician). I know that export levels will be my dependent variable, and exchange rate will be in my explanatory. Would simple linear be sufficient? Would RDD be appropriate? How would I incorporate a depeg dummy variable into my model if my only explanatory is exchange rate – would this present multicollinearity problem? Any advice would be greatly appreciated.
A simple linear regression would not be appropriate as each observation (i.e. a quarter in a given year) is not independent of the other observations and they are also ordered.
A regression discontinuity design would also struggle due to the time series nature of your data... What you are probably after would be a time-series model as this accounts for the autocorrelation between observations and would allow you to account for cyclic variation in the data (See this link). There is approaches which are an amalgamation of both time series and RDD (RDiT / ES) which you could also explore, although I am not sure how easy they are to implement.