Independence of distribution of costs across countries and over time

[irexyu]

Thank you for presenting. It is an interesting topic.

I notice that in the model assumption, you assumed the distribution of costs is i.i.d across commodities and over time, and costs are independent across cities. However, we know that the distribution cannot be independent across cities and over time, e.g., the shortage of one kind of raw material in one city could result in an increase in cost for surrounding cities for a period of time. So I am wondering if there is any possibility to introduce a Hidden Markov Model to obtain a higher precision. If so, what would be the difficulty when implementing HMM?

[tamos]

I noticed this assumption too. It's interesting that even making such an assertion, the model performs quite well. I wonder if its a case of balancing precision vs. accuracy?

[zundaxu]

I don't think it's a problem to make that assumption, since the parameter 'theta' of the Weibull distribution represents the inverse measure of price dispersion, which means the variation of cost across cities of the same commodity, thus from my point of view, we can make this assumption that the costs are independent across cities, and the results also shows that the model performs quite well.