With many factors and risks to consider, identifying the impact of policy change can be a challenge.
Learn why simulation is used to make evidence-based policy decisions, improve program outcomes and deliver services more efficiently to the public.
Using real-life examples from healthcare to smart cities, Tom Stephenson shows the benefits of using simulation for evaluating policy changes.
Can simulations be useful for organizations to help understand their places as components of the bigger system and so find compromises that are good for all rather than brilliant for a few and dire for the rest?
Yes, absolutely. I think that is a good question and it is a scenario where we see simulation used a lot, as it enables you to look at the whole system and understand the links between the services.
If I take healthcare as an example, imagine if a hospital emergency department is really full. You are going to have people come into the ED, they are going to see long queues and as a result, they are going to look for alternative care. It can work the other way around too, like if the GP is closed early then there is no alternative except to go to the emergency department. You can look at all those links between the different services and you can see how they are likely to work together. In the simulation, you can isolate those results of the areas where people have had to use an alternative service in an area because a separate one was not maybe appropriate. Finding that balance is often quite tricky and it is certainly an area where simulation can be useful.
Have you dealt with difficult policy colleagues in the past and how do you convince them of the usefulness of simulation?
Anywhere you can have difficult colleagues but I think it is often more people who will have their set ideas. You are looking at people who have worked in the service a long time. They will have their way of working or they might not really be very open to change or are just very certain that their methodology is likely to work. That is probably where you can have the most struggle with people and that is where you have to make sure that the simulation is very easy to understand. You have to make sure that it is very visual and that the results are very clear, so that it helps communicate those ideas in a way that is going to allow people to understand policy changes or ideas. Computer-based simulation draws out these views and offers the impartial insight needed to facilitate and foster collaboration.
The other thing that I always make sure I do whenever I feel like it might be difficult for people to maybe accept the assumptions of a simulation is always involve them all the way through the simulation build. You do not want to build a simulation and show the people who you are working with the results at the end. You want to be including them the whole way through, so collecting input and data from them, getting them to test the simulation at various points of development and run workshops as the simulation is being built. The best way to mitigate against those difficult colleagues is to have them involved throughout the process so they can understand the thinking that has gone in to the policy changes and simulation.
Senior stakeholders I work with as an internal consultant often get very excited about simulation when they first encounter it, but can have very unrealistic expectations. Do you have any tips on how to manage those expectations without undermining their faith in simulation as an approach?
Yes, it is a tricky one because I will see it a lot of the time where you will have a simulation and it is exciting for people to see that simulation and it sparks new ideas that they want to build into the simulation. This is great because it shows interest and it shows that people are becoming engaged with the simulation, but often it might not be so feasible.
What I always do as a consultant is make sure I have a really clear specification of the initial simulation. There is a three step methodology that I go through. What is the problem? What problem I am trying to solve with this simulation? That is the main one because if you are saying what problem you are trying to solve that prevents a lot of scope creep. Then you say, ‘If this is my problem how can I possibly solve it?’ What are the possible techniques that I might use to solve it? What changes could I make? For example, in a prison system could I look at having increased tags on people? Could I look at reducing young offending and having schemes early on? You want to have those ideas for change ready before you start building the simulation.
Then to make those changes you want to know what strings can I pull to make those changes and what is possible to do. Once you have that sorted at the start of the project you want to remind people as you are doing the presentation, ‘This is my problem. These are my ideas for trying to solve the problem and these are the results’. There is nothing wrong with that moving on to a phase two, but you just have to keep very clear about what problem you are trying to solve.
In your opinion what approach seems to fit better for long-term policy testing, especially in a social care setting – discrete event simulation or system dynamics?
As a specialist in discrete event simulation I apologize for any bias that may come up in this answer, but I think both of them have a place. System dynamics is great for looking at that big picture and getting a really quick result, because you can look at things like population change really quickly. If you need to know any kind of detail beyond that, that is where discrete event is more useful. If I am making a policy change and I want to know how many people are going to use services, then it is good to use system dynamics. If you need to know what is that going to mean for staffing or how are waiting times going to change as a result, this is where discrete event is useful.
Of course you can use both methods together, I am lucky enough to have collaborated on a project where we used both discrete event and system dynamics. I think it depends a lot on the problem you are trying to solve.
I tend to manage both high level and detail just using discrete event simulation, so even if I am looking at the whole system and I am just looking at high level changes I would often do that with discrete event simulation, so that it can feed into the lower level detail seamlessly.
One of the benefits of simulations are that they can be a way of exploring changes that may be too expensive or lethal or both to deal with in real life. Would you agree?
Yes, absolutely, I completely agree with that statement. Especially with policy, we are dealing with people, so there is a lot of risk if things go wrong that those people are going to be impacted in a negative way. I think that test bed to give more confidence that changes are going to be positive is a really powerful thing and it is certainly something that simulation is useful for. I always try to build simulations in a way that enables very quick testing of different scenarios, because that is the main point. It is not about building the model, it is about being able to answer that question and being able to answer that question quite quickly.