Dear all!

My research is about the price behavior of investment grade wines. Of each wine, critics have estimated an ideal drinking date (e.g. 2050-2070). The Midpoint Drink Date Range ("MDDR") is 2060 in this case. Lets say that it is currently 2020, then the CurrentTimeTo_MDDR (CTT_MDDR) is 40 years (CTT_MDDR = MDDR - current year; 2060-2020=40)
  • From the year 2020 on, the CTT_MDDR decreases until the year 2060. The rationale behind this variable is that the price increases if CTT_MDDR decreases because the wine becomes more mature (and therefore 'better')
  • The tippingpoint is in the year 2060; After 2060 the CTT_MDDR becomes negative. Then the wine reaches full maturity after which the price can go two ways
    • 1) for high quality wines: further increase despite the drink-ability is lower (due to the fact that it becomes a collectors item)
    • 2) for low quality wines: decrease since the drink-ability is lower and there is no more demand for the wine.
I suspect that including the CTT_MDDR as a linear variable is not suitable (right?). I have now included the CTT_MDDR as a third degree polynomial variable. Is this correct? Or are there other ways to implement this relation?


Some insights in my model:
- panel regression (109,000 monthly price observations for ~1250 wines over 23 years)
- proposed regression formula: LnPrice = fixed chateau effects + fixed time effects + CTT_MDDR (3rd degree polynomial) + Quality measure [ + CTT_MDDR (3rd degree polynomial) * Quality measure]
- the last term serves as interaction variable


Thanks in advance for your help!

Best,

Raphael