I am planning to take GDP (current), Real Inflation rates, Real interest rates and exchange rate fluctuations for four countries USA,UK, India and Japan. Would I be able to analyze the data using time series analysis, as I'm taking values for the variables for the years from 1993-2013 (on a yearly interval)?
Yes ,correct . You can . You can ask any questions to this team : www.Stats4Edu.com
I believe it is panel data and time series analysis may not produce reliable (BLUE) results. Moreover, panel data models are easier compared to time series.
It is not clear the type of data you are referring to, but bear in mind that Multivariate time series data and panel data are two different types of data.
Multivariate time series data are multiple variables that are observed over time, and the variables may be correlated with each other.
while, panel data is longitudinal, and can be unbalanced or balanced.
Chuck A Arize This is what the data looks like. Isn't this Panel data? Or am I wrong in my understanding?
The data in the Excel file you provided is in time series format, as there is only one entity (i.e., country/state/district/industry/firm, etc.) and multiple variables gathered over consistent time intervals. In contrast, panel data emphasises several entities.
Therefore, you can use time series econometric techniques by examining each of the four countries separately.
However, if you want to perform panel data analysis, you must first prepare your data set in a panel format. First create a column entitled Countries after or next to Year in your shared Excel file to account for this. Following the data for India for all the variables for the sample period you selected, arrange the data for the other three nations in the same manner.
For your panel dataset preparation, can refer to this: https://www.automateexcel.com/charts/panel-template/
Can I ask you to set out the model that you are trying to estimate? You have four variables GDP (current), Real Inflation rates, Real interest rates, and exchange rate fluctuations. Say you wish to estimate a model
y = b0 + b1 x1 +b2x2 + b3 x3 +u
I don't know which of your variables correspond to y, x1, x2, and x3 Looking at your countries I would think that the b-coefficients would be different for different countries. This would rule out panel estimation as it would require these to be the same in each country. There is also the problem that there would be contemporaneous correlations between the shocks in different countries. The technique to deal with this would be Seemingly Unrelated Regressions (SUR). I can not confirm that this technique is appropriate in your case.
If you plan 4 separate time series analyses and can overcome any stationarity problems 21 observations are unlikely to give you satisfactory results.
You might also have a look at the Penn World Tables
https://www.rug.nl/ggdc/productivity/pwt/?lang=en
There is a lot more data there that might enable you to fit a more elaborate model to your data. You might also consider the IMF IFS database. This and several other databases have data that someone has attempted to make consistent between countries. You should take note of any metadata on the database files.
The data would be panel data because you are analysing multiple countries over multiple time periods
Aberkane mouna Mouna The web page for Stats4Edu claims that
98% of Ph.D students Get Better Grades with Stats4Edu's Assistance.
No person with any understanding of statistics could make such a claim. I would advise anyone seeking advice to be very careful if they are thinking of seeking help from that site.
For example, if 5% of Ph.D. students use Stats4Edu then it is possible that 4.9% of Ph.D. students get better grades with Stats4Edu's Assistance. As I have never met a PhD student who used Stats4Edu's Assistance, I think 5% is an overestimate and the actual numbers are much smaller. It would be interesting to know what the statistical theory, if any, underlies their estimate.
Your data consists of different cross-sectional units (countries) over a specified time (1993-2013) which means it is panel data. However, you can treat the data as time series if you intend to analyse these units seperately (comparative study). This means that you will be analysing the USA, UK, Japan and India separately. I dont know you study objective but analysing the data as panel is recommended.
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication