Thread by @GorwayGlobal

On the eve of the #2ndAnniversary of the #DeMonetisation that we know was a #DemonetisationDisaster I present a Tweet thread on my analysis of just one of the many claims made for it - that it would boost #cashless #digital transactions. It did not do that. Here's my evidence:

The analysis and paper (Draft 1) is here:
on issuu: issuu.com/home/published…
and
on Scribd: scribd.com/document/39252…
But here in this thread I present the main results. Previous analysts used a simplistic approach of doing a before-after comparison. But this is a spurious method

It leads to the Post hoc ergo propter hoc fallacy. See rationalwiki.org/wiki/Post_hoc,…
My analytical approach was simple. I used the figures from RBI and took the monthly data upto Oct 2016 on each idicator and treated it as a regular time series data. (see robjhyndman.com/hyndsight/fpp/)

I used ARIMA (Auto Regression an Integrated Moving Average) to do a statistical prediction (with 80% and 95% Confidence bands) of the following 22 months to Aug 2018. I then overlaid the actual data for these 22 months to see if there was an shift that was not 'predictable'

In each of the following charts: The black line is the trend before #DeMonetisation, the white line is the mean prediction of the model, the brown bands are the 80%/95% confidence intervals, and the red line is the actual observed numbers post #Demon. A series of charts now:

Note how mobile Banking values were lower than could have been predicted with the data we had before #DeMonetisation, but the volume of mBanking transactions were above it. True enough the average value of each transaction fell dramatically. See

And finally the total digital payments. This is a composite sum of RTGS, CCIL, mBanking, Prepaid instruments the lot.

And finally here is what the data looks like:
You can download it here: go to dbie.rbi.org.in/DBIE/dbie.rbi?… and click on 'Payment Systems Indicators'
The R program code for the full analysis is on my git hub repository here: github.com/JammiNRao/DeMo…
Comments welcome

Acknowledgements: #rstats

Analysis in R : cran.r-project.org

Charts in ggplot2: ggplot2.org by @hadleywickham

ARIMA from package 'forecast': cran.r-project.org/web/packages/f… by

[e-mail]