Hi everyone,
I only have a small doubt, I am calibrating my model by the method of moments, adjusting certain parameters to the great ratios, these are: internal-public-debt-to-GDP, external-public-debt-to-GDP, consumption-to -GDP, Trade-balance-to-GDP, Government-Investment-to-GDP, etc.
When calculating these proportions from the data in quarterly frequency, (after adjusting for seasonality), I find that their behavior is not stationary, in many cases there appear to be trends. My questions are:
-
Is it still valid to take the average of these series as steady-state values?
-
I was thinking that since they are steady-state values it is better to obtain the trend component of each of the variables and divide it by the trend component of GDP, (using one-sided HP filter). Could this be a better option?
Thanks in advance.
There is no good answer here. It is well known that in open economy models the average trade balance does typically not match the annuity required to maintain the average net foreign asset position. So you need to take a stand on what you consider to be the steady state. Detrending would not help you to pin down the great ratio as you cannot compute a ratio of mean 0 variables.
1 Like
Thanks for your quick response.
Ok, so I think I’ll use the investment to GDP ratio and leave the trade balance as residual, I guess.
I was referring to taking the trend component of consumption and dividing it by the trend component of GDP (for example). Since that way I would be obtaining the long-term ratio, I think.
Though on second thought, I don’t think the steady state value will improve much, since I doubt that the cyclical component is causing the trend behavior. I think what I would do is eliminate short-term movements, which in the end are zero mean.
I think I answered myself. I’m just looking for your confirmation.
Exactly. By their very definition, cyclical fluctuations cannot explain a trend.