2.4. From Events to Dynamics and Feedback – Drug-related Crime
A shift of mind (from event-oriented thinking to feedback systems thinking) is not easy to achieve. The best way to make progress is through examples of feedback systems thinking applied to real-world situations. Instead of hot water showers we now consider something entirely different – drug-related crime. A typical description of the problem, by the victims of crime, might be as follows.
Drugs are a big worry for me, not least because of the crimes that addicts commit to fund their dependency. We want the police to bust these rings and destroy the drugs. They say they're doing it and they keep showing us sacks of cocaine that they've seized, but the crime problem seems to be getting worse.
Expressed this way drug-related crime appears as a series of disturbing events. There is a concern about crime among the members of the community affected by it. They want action backed-up with evidence of police attempts to fix the problem by busting rings and seizing drugs. But, despite these efforts, more crimes are happening. The feedback systems thinker re-interprets the description and draws out those aspects concerned with performance through time (dynamics) that suggest an underlying feedback structure, one or more interacting feedback loops, capable of generating the dynamics of interest. Of particular significance are puzzling dynamics, performance through time that people experience but do not want or intend. Some of the most interesting and intractable problems in society and business appear this way.
Figure 2.10. Unintended dynamics of drug-related crime - a rough sketch
Figure 2.10 shows the unintended dynamics of drug-related crime that might be inferred from the brief verbal description above. This is just a rough sketch to provide a focus for structuring the problem. On the horizontal axis is time in years. On the vertical axis is drug-related crime defined in terms of 'incidents per month'. There are two trajectories. The upper line is a sketch of crime reported by the community. We assume a growth trajectory because 'the crime problem seems to be getting worse'. The lower line is a sketch of tolerable crime, a kind of benchmark against which to compare the actual level of crime. We assume a downward sloping trajectory because the community wants less crime and fewer drugs, and the police are taking action to achieve this end by seizing drugs and arresting dealers.[]
[] You may be thinking this method of creating time charts is rather loose and in a sense you are right because we have very little data about the problem. But even in practice, with real clients, the information sources for modelling are always a pragmatic blend of informed opinion, anecdote, objective facts and clear reasoning. For a good example of this balanced approach in the area of drug policy see Homer 1993 and Levin et al. 1975.
The divergence between reported and tolerable crime is of particular interest to the feedback systems thinker. What feedback structure could explain this phenomenon? Reported crime is growing and we know that growth arises from reinforcing feedback. So where could such a malignant feedback process come from and why would it exist at all if those involved want less crime, not more? The persistence of unwanted growth in crime suggests a feedback loop that weaves its way around society (crossing the boundaries between police, the community and drug users) and by doing so it goes unnoticed.
Figure 2.11. Causal loop diagram for drug related crime
2.4.1. A Feedback View
Figure 2.11 is a causal loop diagram for drug-related crime. First, consider the words and phrases alone. They provide the basic vocabulary of the causal model, the factors that drive up crime, or at least are hypothesised to do so. They also give clues to the boundary of the model, which parts of society are included. Of course there is drug-related crime itself, the variable of central interest and concern to the community. There is a 'call for police action' and drug seizures that take us inside the police department. Then there is supply, demand and price that belong in the world of drug users who commit crime.[]
[] Notice that all the terms in the diagram are nouns or so-called 'noun-phrases'. This is an important diagramming convention because you want concepts to denote things, attributes or qualities that can, in imagination, be unambiguously increased or decreased. Then, and only then, is it possible to assign polarity cleanly to causal links and thereby deduce the loop types – balancing or reinforcing. Take, for example, price and drug-related crime. It is easy to imagine the price of drugs going up or down and separately to imagine drug-related crime increasing or decreasing. Therefore, when a causal link is drawn between these two concepts, it is meaningful to ask whether an increase in one leads to an increase or decrease in the other. This thought experiment would make no sense if one or other concept were labelled as an activity, say pricing instead of price.
These factors join up to make a closed loop of cause and effect. The loop brings together disparate parts of society to reveal a surprise. Hidden in the connections is a reinforcing feedback process responsible for (or at least contributing to) escalating crime. To confirm, let's trace the effect, around the loop, of an imagined increase in drug-related crime. In this kind of analysis the reason for the initial increase does not matter, it is the feedback effect that is of central interest. The story begins at the top of the diagram. An increase of drug-related crime leads to a call for more police action. More police action (raids and arrests) leads to more drug seizures. So far so good. But the paradox lies in what happens next as available drugs are traded on the streets. An increase in drug seizures causes the supply of drugs to decrease. This supply cut then causes the price of drugs to increase, just like any traded goods subject to market forces, assuming of course that higher price does not depress demand. And crucially for illegal drugs, price has little effect on demand because most users are addicts, dependent on their daily fix. So an increase in price merely boosts crime as desperate drug users steal even more to fund their addiction. The reinforcing loop is plain to see. There is a 'crime spiral' in which any increase of drug-related crime tends to amplify itself through the inadvertent actions of police, drug dealers and addicts.
2.4.2. Scope and Boundary of Factors in Drug-related Crime
There could be more, much more, to the problem situation than the six concepts shown. I am not saying these six factors and this single reinforcing loop is a perfect representation of escalating crime in a community plagued with drug addicts. Rather it is a useful way of thinking about the problem that raises the perspective above the narrow confines of a single stakeholder. In fact, three stakeholders are united in this particular view and, just as we noted in the shower case, there is a lot going on behind the scenes of the stark causal links; detail that would need to be fleshed out in thinking more carefully about the problem and in building a simulation model to test alternative intervention policies. There is the community suffering from crime and calling for police action. There is the police department, concerned with all sorts of law enforcement, allocating police officers to priority tasks, among which is drug busting. Then there is the shady world of drug dealers sourcing drugs and covertly selling them to addicts who must consume, no matter what the cost. In the next chapter, we see how this qualitative feedback diagram is transformed into a full-blown simulator, but for now I want to end the discussion of drug-related crime by inviting you to think about what else might be included in a conceptual model of the problem.
One area of the diagram to expand is demand and supply. (Another good idea in practice is to gather more time series data to help refine the dynamic hypothesis, but we will bypass that step in this small illustrative example.) What if there is growth in demand because addicts and dealers themselves recruit new users? This possibility adds a whole new dimension to escalating crime not dealt with in our current picture, a new theory if you like. What if, as is surely the case, the available supply of drugs increases as the price rises? Does that mean drug seizures perversely expand the whole illegal drug industry (in the long run) by artificially boosting prices? Such industry growth could exacerbate the crime problem, particularly if the relevant time frame is a decade or more rather than just a few years. These questions, and others like them, are worth probing and may usefully expand the scope and boundary of our thinking. The point, however, in any such conceptualisation task, is to avoid unnecessary complexity and focus on finding plausible loops, often unnoticed in the pressure of day-to-day operations, that not only challenge conventional event-oriented thinking but also produce dynamics consistent with the observed problem.
2.4.3. An Aside – More Practice with Link Polarity and Loop Types
I have explained the origin of the reinforcing loop in Figure 2.11 by tracing an imagined change in crime all the way around the loop and showing it leads to even more crime. As mentioned earlier, another way to find the loop type is to use the counting rule. Count the negative links around the loop. If the number of links is odd then the loop is balancing and if the number is even the loop is reinforcing. Let's do this exercise now. First we need to assign link polarities using the standard test. Any individual link connects two concepts A and B where A is the cause and B is the effect. For each link imagine an increase in the cause A and then work out the effect on B. In this thought experiment all other influences on B are assumed to remain unchanged, the ceteris paribus assumption. The link is positive if, when A increases, B increases above what it would otherwise have been. The link is negative if, when A increases, B decreases below what it would have been. Note that the mirror image test works too. So when A decreases and B also decreases the link is positive, but when A decreases and B increases the link is negative. What matters for polarity is whether or not there is a reversal.
We start at the top. All else equal, if drug-related crime increases then the call for police action (complaints from the community) increases above what it would otherwise have been, a positive link. When the call for police action increases then drug seizures increase, another positive link. Note there is a large leap of causality here that relies on all else remaining equal, ceteris paribus. We implicitly assume that a call for action really leads to action (in this case more police allocated to drug busting), rather than being ignored. Moreover, we assume that more police leads to more seizures. In the next link, an increase in seizures leads to a decrease in supply, below what it would otherwise have been, a negative link. Then a decrease in supply leads to an increase in price, another negative link coming this time from a mirror image test. Here there is a particularly clear instance of ceteris paribus reasoning because price depends both on supply and demand. The assumption behind the polarity test is that demand remains constant. An equivalent test on the demand-to-price link shows it is positive: an increase in demand leads to an increase in price, assuming supply is held constant. Finally, an increase in price leads to an increase in drug-related crime, a positive link that completes the loop. Counting up there are two negative links around the loop, an even number, so the loop type is reinforcing.