I have seen a couple of research articles on the effect of the novel covid 19 on macroeconomic variables. I would like to throw an open question that is it econometric ally ideal to map a discrete count variable ( covid 19 cases) on macroeconomic variables such as inflation, GDP, exchange rate etc which are continuous random variable..or perhaps is there are method that supports this.. Knowing fully well that computer programs doesn't recognize the covid number of cases as count unless when specified.
In practice, the macroeconomic variables you mention are less contiuuous than the number of covid19 cases. Take inflation: This is a percentage change of a prrice indix, the format of which is ##.#% for high or #.#% for low inflation. The index usually has the format ###.#. GDP is an estimate for which one can assume a bias of - I think - at least +-1%. That means it is not necessary to have an exactness of more than 3 digigts for statistical purposes. For exchange rates one has, in principle, daily or even hourly data, and the main question is what average to choose for a comparison with other variables. For statistics it seems sufficient to take the 3 first digits of such averages. This means, that, in general, one needs less natural number to "count" these variables than to count covid19 infections (which itself is, of course, be really correct). Finally, a question: Is the percentage change of covid19 cases a continuous or discrete variable? I think, that answering this question shows that this differentiation is senseless.
- Badges
- Science topic