Systems Thinking and the Long Term View of Costs

One of the key ideas that surfaces again and again when working with complex systems is that time matters. A lot. You can't just look at a complex system at the current moment in time and gain any understanding of the over-all behavior. For example, if you look at the stock market on three consecutive days you will not get an accurate idea at all of how the the stock market really behaves--for that you need to examine what the stock market does over the course of, say, 30 consecutive years. "The system" could be anything from weather to oil production to the social services for autistic people.

It can be hard to understand what the long-term behavior of a complex system will be like without a demonstration. And how does one demonstrate what something would be like over the course of 30 years in anything less than 30 years time? One answer to this is to model the complex system on a computer and simulate its behavior over time. Computer simulation is both a way to quickly demonstrate what long term outcomes will be like, and a way to try out a bunch of different ideas to see which have a better chance of succeeding before making costly real world commitments.

Off Reuters comes the headline Government should spend to save on autism. Sounds like an oxymoron? Perhaps not,

A computer model showed that increasing the rate of people identified with the less severe forms of the condition from 1 percent to 8 percent in local communities could lead to savings of nearly 70 million pounds per year.

The economic benefit would come from getting sufferers into specialist care units that help them remain in full-time employment, so reducing welfare payments. It could also lighten the economic burden on families caring for autistic relatives.

For background and source information on the Reuters article, see the National Audit Office (UK) Supporting people with autism through adulthood page (thank you to the reader who emailed me that link!).

I'm reminded immediately of an example in Sterman's Business Dynamics wherein a company found that spending a large amount of money in the short term to improve their equipment saved them a much, much, much larger amount of money in the long term due to less downtime, less maintenance fees, and less parts costs.

This is one of the many reasons why the argument that services--particularly non-crisis services--"cost too much" has never washed well with me. As a systems scientist, it's hard for me avoid seeing the much greater long term costs of these short term gains.

Hopefully the work reported on from NAO will be taken seriously, and will enable the people with the power to make change to see the long term view. I'd love to see a U.S. organization undertake a similar analysis.