How to classify sample by time-variant index?

In finance paper, we have a famous index representing for law is WGI (World Governance Index). I am wondering how we classify countries by this time-variant index as this paper below.

It is easy to classify based on lower or higher than median like developed or developing countries, but I do not know what should I do if the index varies by time such as WGI index

Update: I add a part of one index amongst 6 indices of WGI here for visualization purpose (the full dataset can be retrieved from the link above)

• @"I do not know what should I do if the index varies by time such as WGI index". What you term classifying is basically dimesion reduction - you want to turn a time series into a single value. So it is up to you to decide which features of the time series are important for your question and should be reflected in the summarized value. If you keep this information to yourself we cant help you here Jun 19, 2021 at 9:20

1 Answer

There are various ways of how to achieve that, without knowing the purpose of this classification it is hard to say what is the best way how to do that.

However, a simple and a straightforward way of achieving that would be as follows:

• Calculate mean or median value value of the index - this will give you single time invariant average value for the index.
• Once you assigned each country some time invariant value classify them according to that, so for example country with rule of law above 75th percentile can be classified as countries with "very strong rule of law", in between 50th and 75 percentile as "strong rule of law", countries in between 25th and 50th percentile as "weak rule of law" countries with values between 0th and 25th percentile as countries with very weak rule of law.

In addition, there is no reason why you cannot classify the countries dynamically, if you do not have too many years you can have separate classification for each year, or you can just visualize it as time series. Again unless you provide some details of what this is for it is hard to give a specific recommendation.

• Hi @1muflon1, the purpose is to test whether the laws have different impact on cash equivalent between good governance and bad governance countries. It is similar to test the different of the impact of this law on cash equivalent in developed and developing countries in [this link] (economics.stackexchange.com/questions/45680/…). The problem is developing and developed are invariant while WGI are variant Jul 1, 2021 at 9:27
• I also added one part of the data in the question part, could you please havea look? I am wondering the part "Calculate mean or median value value of the index - this will give you single time invariant average value for the index.", how should we do that can I ask in specific? Jul 1, 2021 at 9:31