Context: USA economy


Its generally accepted that the growth of e commerce has certain curbing effects on the CPI-inflation. Because search costs are much lower online, people always go for the lowest prices and this makes online retailers fight to price their products lower than everyone else. Plus, there are not much fixed costs in online retail which helps reduce overall prices.

Now with the rise of e- commerce's relevance and ease of use, people are switching to it for grocery shopping and electronics shopping, etc and therefore this is seen to have a curbing effect on inflation CPI, as prices are driven lower.

Consider the two processes:

Let one process be the CPI inflation
Let another process be the growth of e commerce in US(maybe in terms of %of GDP or some other metric)

Is there a way to do a time series analysis(or any other type of analysis if you feel time series is not the way to go) to test if this inflation curbing effect of e commerce exists? Will this test be meaningful? This is for a project so I'd like some input on whether its going to give me interesting results or if this analysis is too general to yield anything useful.

Thank you!

  • $\begingroup$ This is an incredibly broad question, and you haven't given many details. There are many other macroeconomic factors that go into the CPI than just e-commerce. Can CPI be modeled? Yep. Will it be meaningful? No idea, you need to model it. $\endgroup$ – Nathan Calverley Nov 15 '13 at 17:25
  • $\begingroup$ E commerce is a part of CPI calculation ,however small, but that's not whats important here. my hypothesis is that E commerce has a an effect on inflation through another channel, which is that as offline prices(physical stores) go up, consumers tend to switch to lower prices which can be found online. This would result in reduced inflation because then offline stores know they have to reduce prices if they want to attract customers again. I just want some opinions on whether it is possible to test this hypothesis. I hope this comment cleared thing up a bit, @NathanCalverley $\endgroup$ – Siddharth Gopi Nov 15 '13 at 17:34

You can take any two vectors of numbers, and, given that their elements form meaningful pairs (i.e. magnitudes of two variables in the same point in time), estimate their covariance. In your case, it should be negative. This would give you the strength of any linear stochastic dependence.

To try to model CPI as a function of e-commerce alone, say a linear function, and so a linear regression specification, is rather stretched, because so many factors affect CPI. You could include in the specification these other factors, adopting a relevant theoretical model, and add to the regressors the e-commerce variable to test whether it is statistically significant in the presence of the other regressors.

Since this is a time-series framework, you should worry whether your variables are stationary or not, if not what is the nature of non-stationarity (deterministic trend? Stochastic trend? Heteroskedasticity? etc), and whether they are co-integrated or not.

  • $\begingroup$ hey alecos thanks for pointing me towards the right direction, im gonna be reading up in detail on everything youve mentioned in here. $\endgroup$ – Siddharth Gopi Nov 15 '13 at 21:58
  • $\begingroup$ @garciaj ...and come back with more questions!:) $\endgroup$ – Alecos Papadopoulos Nov 15 '13 at 22:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.