Wednesday, July 20, 2011

econometrics of life

Making big decisions in life is sometimes require courage, particularly if it will affect our life in long term (e.g. Choosing education, career). It may be different than short term, daily decisions, which outcome will cointegrate in the long term. If errors exist, we can simply hope that it will back to normal tomorrow, and offset today's bad experience. If dynamic forecasting is too risky, we can simply use static projections, by updating the data and reestimate the model frequently.
In econometrics, short term forecasting is often fit well by simple autoregressive models, that only require historical data of variable projected. But for long term, advanced theory and more complex methods are often required. we also have to use many exogenous and instrumental variables, together with assumptions and scenarios. Therefore, it require much more information.
So what it has to be in the real life? To choose a career, we have to collect large scale of historical information about story of others, including exogenous variables such as particular individual characteristics, and control variables such as social acceptance and support. We have to make assumptions or scenarios to set values of exogenous shocks. Indeed, once assumptions are flawed or scenario we choose is misleading, we will face unexpected future. Nevertheless, information between sets of choices are often unbalanced.
Suppose there are two long term choices, let's consider them equivalent to long maturity assets that have very high transaction costs. One has sufficient number of observations and known exogenous data, while the other has not, but shows higher return in that small number of observations. Which asset should we choose?


Gaffari said...

Small number of observations will probably lead you to a wrong conclusion. Empirically you dont have a plausible judgement that can guide you to take the right decision.
But as long as you consider to put an error correcting variable in your asset model, all you need is to make sure that you already have suitable speed of adjustments in correcting your short-term errors and immediately can get them back to meet the long term equlibrium.
Note: assume that errors in your short term model are stationer. :)

Anonymous said...

Hey guys,

Would you like to watch Final Destination 5? It is not released yet but you can watch it already!

Click here to [url=]watch Final Destination 5 online[/url]