Skip to main content

Table 1 Key features of analyses of intervening in complex systems

From: Policy experimentation and innovation as a response to complexity in China’s management of health reforms

Requirements/approaches Core principles
1. Scope system and dynamics; Identify stakeholders and critical change points for intervention - Attempt to identify key stakeholders, networks and ‘critical points for change’ in advance of reforms [5], but be ready to learn rapidly about system dynamics and adjust as reforms are implemented
- Make use of tools such as scenario planning and agent-based models to identify CAS phenomena, attempt to anticipate changes and guide interventions [5], but be prepared to adapt
- Engage multiple stakeholder groups in conceptualising and planning interventions to help to capture dispersed understandings of system complexity [68]
2. Experiment and encourage innovation and policy pluralism - ‘Plan for unpredictability’: create conditions for change, rather than trying to directly engineer change; allow flexible implementation by implementing units (sub-national governments, departments and others) over rigid adherence to initially-conceived plans [5]
- Use experimental policy frameworks to stimulate systemic innovation; allow and encourage trial and error by implementing units and tolerate implementation failure within reasonable bounds [23]
- Use incentive systems that incentivise achievement of agreed goals (however these are achieved) over carrying out of specific processes as a way to encourage pragmatic problem solving and foster emergence of multiple policy solutions [23]
- Directly experiment and contrast multiple approaches where needed and/or where fostering dispersed innovation is failing to find workable policy solutions [69]
3. Foster methodological pragmatism and appropriate policy solutions - Encourage methodological pragmatism and pluralism in problem solving by implementing units over preconceived approaches (fit solutions to problems; not problems to solutions) [70]; recognise that there may be multiple different solutions to a given problem or multiple practices; accept diversity in policy solutions adopted by implementing units
- Look for policy solutions that are contextually appropriate and show ‘fit’ with the environment, rather than getting hung up on ‘best practice’ solutions; recognise that apparently technically second best approaches may function as ‘stepping-stone’ policies, may promote system adaptation, and can be improved on in subsequent rounds of reform [71]
- Avoid ‘isomorphic mimicry’, or practices that mimic supposed best practice, but have little substance [70]
4. Screen for, and learn from, policy innovations and useful practices - Accept that in a CAS, solutions to problems must be discovered, and that there are limits to ex ante design [72]; decrease reliance on planning and ex-ante analysis, and increase monitoring, learning and adaptation during experimentation-implementation [73]
- Incorporate multiple types of evaluation to capitalise on emerging understanding of intervention processes and effects during roll out [68]
- Screen for ‘positive deviance’, or naturally occurring useful practices and emerging innovations, and encourage dissemination of these through appropriate ‘fitness functions’ that can promote adaptive learning while discouraging inappropriate practices
- Promote appropriate and replicable (though not necessarily optimally efficient) policy innovations over optimally efficient policy solutions that will be hard to replicate [74, 75]
5. Promote rapid learning and diffusion of useful practices, but allow flexibility during scale up - Carry out iterative programming / reforms; promote rapid learning, and ‘fail fast’ – directly terminate failing or inappropriate practices or allow implementing units a degree of discretion over this [23]
- Make use of diagnostic tools to analyse staging of reforms and think through the process of removal successive binding constraints to reform [21]; use such tools to help sequence reforms, but recognise that variation in starting points and conditions will lead to implementing jurisdictions or units simultaneously occupying different stages of reform trajectories
- Proactively propagate and support useful policy innovations, but be wary of imposing one-size-fits-all policy solutions during scale up [10]