REPLY Definition of root cause (SD6840)

[SDMAIL Bill Braun]

Posted by  Bill Braun <bbraun at hlthsys.com>

Jack asks if I have read the Dueling Loops paper (not seeing me use any 
key phrases) and notes that Sterman defines exogenous variables as 
arising "from outside the boundary of the model." (page 95) Thus all the 
variables in the model are endogenous. Finally Jack asks "Perhaps you 
feel that constants or lookup tables are automatically exogenous?"

I am taking the model at the bottom of Dueling_Loops.htm at face value. 
I have not read the paper. My generic question is based on identifying 
exogenous variables as high leverage points.

And perhaps my understanding of exogenous is dead wrong. I draw a 
distinction between constants that represent behavior over which 
decision makers have no control and parameters which are dimensions of 
policy or decision making over which they have control.

I might use a constant to represent the economy if the time frame of the 
model is short enough that changes in the economy are unlikely. The 
economy is an exogenous variable. If I am looking at the time over which 
I adjust inventory, that is a parameter and under my control. According 
to my understanding (perhaps in peril here) all the variables in the 
Dueling_Loops model are not endogenous.

I'm not sure how a lookup table could be exogenous. It either uses 
another model variable or the implicit variable of time as its input 
(the "x" in y=f(x)).

Returning to Jack's specific model, when he notes, "I'm lost here. "The 
policies explicit in the model" are NOT okay. That's the problem!" then 
I would expect to see one of those policies identified as a high 
leverage point, the changing of which would change model behavior. (Or, 
one model that represents current reality and another that shows 
changed, added, or deleted policies.) Stated otherwise, I would expect 
high leverage to include a feedback loop.

I sense the deeper question here is, am I on or off the mark in my 
understanding of exogenous versus endogenous variables. I imagine this 
to be tediously boring to people who do understand the difference, and 
is my last comment on the topic.

Bill Braun
Posted by  Bill Braun <bbraun at hlthsys.com>
posting date  Wed, 26 Mar 2008 07:32:00 -0400