It's a recurring question. How is SD different from other approaches? How is working with SD software different from Excel?

I learned SD at MIT from Forrester and his disciples. I've built many models, both SD and from other points of view, particularly models of consumer markets.

Here are some reflections:

 1. The key difference is the mindset to recognize feedback. Stocks drive the model... Which drives the flows that change the stocks. The marketing models I built in the 70's didn't work this way. SD models do. It can be challenging to step back far enough to find the stocks and feedback. I like the Strategy Dynamics / SYSDEA approach which brings these out early.

 2. It's not about simultaneity. Yes, our SD models technically involve simultaneity, at least theoretically, but we don't calculate them that way. Typically, and as encouraged here, we do a calculation each time period. You can do that in Excel, just as faithfully.  (Forrester's System Dynamics models came closer to simultaneity by breaking time periods into much smaller "DT" with a host of issues for understanding and acceptance. Key question: Is it worth it to shift from period calculations to DT for a given model challenge?)

 3. Key difference between SD software and Excel is where the formulae are kept. In Excel, there's a formula for each cell. Therefore lots of opportunities for errors copying formulae. In SD software, there's one formula that operates on the whole time series. Much easier to keep track of the formulae and watch for errors.

 4. Simultaneity -- The only time I really did simultaneous equations was a complicated model, I wanted to start it in steady state, many stock were not measurable. I solved the simultaneous equations to figure out initial values to start the model in steady state.
Quote 0 0
Working with a financial planner, he sent me an Excel file for 30 year retirement planning.

Buried in the 20th line is a change in formula that's different from all others above in that column. It's change he projects from that year. Drove me nuts trying to check out the calculations.

Make me want to redo the model in SYSDEA... That change would be a step change in a graph that's visible at the top level. Much more understandable.

(No, there's no feedback in this retirement model. Could be... Bigger bank account would encourage us to spend more. But, that's not in the model.)
Quote 0 0
There's a small but important point at the end of this post - that the model does not include feedback ... but the model is still useful !! I think that's because [a] it handles stock-accumulation explicitly and [b] you can see the inter-connected time-charts.
Quote 0 0