Guest Blog: Collective Impact and SROI – Why Scale Matters
Collective Impact and SROI- Why Scale Matters
GUEST BLOG BY GARRY HAYWOOD, 6TH DECEMBER 2011
In his opening blog, Jeremy uses the anecdote of his son Jack playing under the desk as a device to explore if an ‘integrated development strategy’ should be simple or not. People like simplicity, it’s easy to understand and in situations like Granby/Toxteth it also fits linear expectations, induced from the deficit model, which suggests a causal link between good policy, right levels funding, a good dose of hard work leading to the creation of the desired impact/value: doing A leads to B. If it was really that simple we’d have solved the problems associated with economic and social deprivation a long time ago. The relations between various agents in the two systems we call society and economy are complex, and this operates at all scales from the human-scale to the global-scale.
However, complexity doesn’t have to be complicated. Sending an astronaut into outer space is a very complicated affair, getting vehicles to proceed around a traffic island is a complex affair. However, accepting complexity does mean a change in paradigm, a loss of pseudo-control in exchange for a better understanding of reality. Complexity is the outcome from dynamic interactions of simple processes such as getting traffic to take turns in crossing the junction from contributory roads and becoming part of the rotary flow around the island until they exit. Complexity increases with the level of variation (distinct factors), dependency (connecting factors) and dimension. A traffic island with 3 contributory roads is less complex than one with five entry points.
As complexity increases, the system becomes more non-linear; it becomes increasingly unlikely that A will lead to B. Making small changes to the operation of the system, can effect the whole system or just parts in a multiplicity of ways, or even not at all. This is why, in policy terms, we lose some pseudo-control. Why do I use ‘pseudo’ here? Because quite often the effect of our intervention is out of our control.
For example, all road users will have experienced a busy traffic island when one car comes out from a junction too early/sharp/etc. We are each in control of our car, but we are not in control of the traffic. The eager driver might cause a local effect so only cars at that contributory junction are effected. It may just as easily cause the whole traffic island to become momentarily clogged-up. And it is equally plausible that this single event causes a tail back of one block or even several. This might even cascade into other events a long way from the initiating event. A simple event can have localised or wider system implications and to know how extended the effect is we may have to consider the system at several scales. If we only look at one junction we might not see the effect to the whole traffic island or the extended road network. If we are looking at a system, the less complexity we have the less we might see any effect, yet equally if we have too much complexity the effect might be buried. This suggests there might be an optimum level of scale(s) for certain types of interventions.
It is important to stress this when we’re talking about systems such as a local economy; we need to think of them as complex systems in which there is a dynamic (flow) from all the actual interactions within the local system and without. We will need to consider different scales of the system, to understand different complexities, different levels of variation and connectivity. We might not be able to predict the outcome because the simple interventions we can make, such as helping to start a firm, have to interact in a system with other firms in a complex way. We certainly can’t measure all the interactions in the system. Yet we can still test if our interventions had any impact, desired or unintended. To do this we must understand at what scale(s) we should consider measuring any changes to various systems that we are trying to affect.
The objective of the strategy in Granby/Toxteth was to intervene in the local system of firms so that we could have more firms offering more employment opportunities, while at the same time improving education and training provision for residents so that they might take up those opportunities. We might summarise that we had four simple groups of interventions to achieve our objective: adding more new firms; helping existing firms to grow to take on more employees; improving human capital; improving the match between job opportunities and human capital development. So far, so simple.
Where this becomes complex is the firms that are supported must interact with other firms in the market place. A new firm would add some employment to the area, but it might destroy employment in a competitor firm. Imagine supporting a community shop that wins customers from an existing shop which must now close because it cannot sustain itself on less custom. The intervention has negated itself. Likewise supporting an existing firm to become stronger may make it impossible for new firms to gain traction and become a sustainable enterprise. The same events might just as well work. Starting a new business in an area may make the existing business look for customers from a different market. On a programme wide basis we might expect all these to be true. Consequently, matching the training needs of residents to firms future needs is reliant on getting the firm interventions right, otherwise the mismatch may continue or increase.
When we start out we just don’t know what will happen, only what we hope will happen. That’s what happened with the GTDT strategy. Lots of careful preparation. Lots partnership working. A dash of optimism. Yet, as Jeremy wrote, there was little overall effect from the interventions. Without a full evaluation it is difficult to know why this was, but on reflection I suspect that we didn’t sufficiently consider scale, for both the intervention or the measurement of its outcome.
There is lots to think about when considering scale. Foremost in my mind is the area we were working on. It had an arbitrary border, which was largely a political construct hewn out of administrative boundaries. What if our area was not actually a system but only part of system? For example, consider a circle on a piece of paper. If we zoom out at a certain scale the circle will become a dot and then disappear. If we zoom in we will lose the edge and we will only see space. If the spatial scale of our intervention or measurement is wrong then we may not create any impact or, worse still, we may have some impact but not be able to see it, or even worse, we may see some impact that was generated by other influences and claim it as our own leading us to bias our own methodology.
We certainly didn’t consider the scale of dependency between the SRB Partnership and other areas. The GTDT area is an inner-city area that borders on to the CIty Center. It is possible that any effect our interventions had were limited by the proximity of this more potent force.
So what to do about scale? I’m not entirely sure although I have some thoughts about statistical distributions, particularly those of a power-law nature and also about interventions that are specified to have wider-system effects but the only measurable value is at the human scale. I’ll save developing these thoughts for a future contribution to this discussion. But I’d like to end this blog with a brief return to our traffic island metaphor. If we wanted to regulate the traffic flow onto the island, would we do it with lights at the junctions to the island itself or further back on the approach road? I don’t want to propose an answer here, but let you think about it while highlighting that it is a question of scale.