Integrated Random Walk with Kalman Filters

[Shouro Dasgupta]

I have 20 years of data for damages suffered from cyclones and would like to estimate an Integrated Random Walk (IRW) with Kalman Filters along the lines of "On the relation between weather-related disaster impacts, vulnerability and climate change". I have attached the paper.

I would like to replicate what the authors do in Section 3.1 "normalized disaster burdens...along with IRW trends". I have also attached the resulting plot. The authors mention that they used S-PLUS 8.1 software for these analyses, I have Stata 13. I was wondering if these results can be replicated in Stata. I hope I have provided enough information. Thank you!

Sincerely,

Shouro

[Sean Lambert]

Shouro,

I'm somewhat surprised that no one has responded to your post, as it's been nearly a month. I only came across it because it was tagged under time series. I'm interested in whether or not you have made progress on this project. I study economics so my background isn't exactly lined up with what you are doing, but reading the source that you posted leads me to believe that you would need a lot more information on what they actually did to generate the results of the paper. It seems that in following the format of the journal that they submitted to, they omitted important details for replication.

Looking at some of the cited literature in the paper, I found that Normalizing economic loss from natural disaster: a global analysis by Neumayer and Barthel provides a better starting point if you were to replicate the results of the Visser paper.

In particular, the definition of normalized loss is very important, especially if your data does not already have it normalized.

As for Stata, I think you could easily replicate those charts. Personally, I would use a VAR approach to do it as VAR's postestimation commands are very good in Stata 10 and beyond. For instance, -fcast graph- would be able to make exactly the results you are looking for.

For reference material on VAR, and time series in general, I highly recommend that you pick up a copy of Walter Ender's Applied Econometric Time Series. It is a veritable trove of tricks for doing the kind of work that you are interested in. Additionally, Sean Becketti's Introduction to Time Series Using Stata is also a good resource, especially for examples on how to make good looking time series graphics.

Good luck with your research!