Economic Regulatory Systems
An economic regulatory system is an approach to managing or regulating an economy. In this article, we discuss the differences between the two fundamentally different approaches to economic management, that of the centralized linear regulatory model and that of distributed nonlinear regulatory systems. We firstly talk about the key components to a centralized top-down regulatory system and look at some of their competencies and limitations. How they are most effective within simple environments, involving a finite amount of elements with a low level of connectivity and limited change, and how they have a tendency to linearize the system and externalize complexity; limiting their long-term sustainability. Next, we talk about nonlinear forms of regulation, a bottom-up approach that invests in the agent’s capacity to adapt locally, thereby providing them with information, and communication with which to interact, self-organize and better respond in volatile and complex environments.
There are essentially just two different paradigms in system regulation and management, a top-down method whereby we implement a centralized control system that constrains and coordinates the components towards some predefined objective, or we use a bottom-up approach where individual components are enabled to adapt to their environment locally and interact to enable global emergent coordination. A regulatory or control system within system theory and cybernetics is a specialized subsystem that is designed to regulate the operations of the broader system it is a part of in order to maintain homeostasis.1 Homeostasis means maintaining the system within a given set of environmental conditions that are best suited to its preservation and functioning. This form of centralized regulatory system should be very familiar to us as it is the default model to the industrial age management paradigm. It is used in all forms of economic systems in the shape of a hierarchy that is used to regulate most types of economic activity from a business to whole economies.
In order to do this, the regulatory system needs a few key components, firstly, some way of sensing and receiving information about the system it is regulating and the environment it is operating in. Secondly, it needs some model of the system and its environment and the capacity to use that model to generate new decisions based upon a variety of inputted information. Lastly, the regulatory system needs some mechanism for acting out those instructions. It needs what is called an actuator that will alter the state of the system in order that it conforms to the set of instructions. A national government is a classical example of a centralized regulatory system within economics. It gathers a massive amount of data and statistics about the nation’s economy. It employs linear models derived from economic theory, forecasting and policy making to define the desired future state to the system, and then from this generates an output that is acted upon by a number of actuators, namely central interest rates, taxes, tariffs, subsidies etc. An extreme example of this centralized regulatory model would be a communist state like post-war China or Russia, where everything was controlled centrally and regulated through a top-down system. Policymakers made 5 year and 10-year plans, production and consumption quotas were set centrally and everything was expected to be aligned towards that same objective.
Centralized control systems have advantages and disadvantages, but essentially they are a linear model to regulation and they will thus be best suited to regulating simple linear systems within stable environments, meaning there needs to be a finite limited amount of elements in the system that have a low level of interconnectivity and limited capacity for adaptation, and ideally the system will be operating in a stable, predictable and routine environment. Under these conditions, centralized regulatory systems can be very effective due to their centralized nature. They are able to leverage batch processing and their inherent simplicity is also another very desirable feature. With these centralized regulatory systems, we can achieve very powerful coordination between the components that otherwise may be very difficult to achieve. But being designed to regulate simple linear systems, they will, of course, have their limitations. Not all organizations are simple linear systems.2
When we turn up any of the factors that generate complexity, this top-down model will start to become less effective. Centralized models are limited in their scaling, meaning they do not scale infinitely and will work best with a limited amount of elements. As we turn the number of elements up, we will have to develop more and more layers of the hierarchy as it becomes more and more abstract and removed from the actual system that it is designed to regulate. Information and instructions have to be routed farther and farther up and down the system through many layers of command and interpretation. This all places heavy administration costs on the lower levels and reduces the system’s capacity to respond quickly.
If we turn up the interconnectivity between the elements in the system, it becomes more difficult to control them through top-down channels. The hierarchical model is predicated on being able to control the components in the system. This will be most effective when the components are isolated and dependent upon the centralized system for instructions. If the agents are interconnected and able to receive information about each other and their environment, this gives them a much greater capacity to act autonomously. If we also give the agents the capacity to adapt, then this again will make them less dependent upon the centralized regulatory system for processing information, and again make it more difficult to control them.3
Lastly, these large centralized organizations can only deal with a finite amount of volatility within their environment. New information typically has to be routed to some centralized unit before it can be acted upon and their static structure is not designed for change. Although these large centralized organizations may be good at dealing with small shocks, they may also be very vulnerable to systemic shocks or changes within the whole environment. We should note that we do in fact often use this centralized regulatory system within complex organizations and environments, in the same way, that engineers use linear models to describe nonlinear phenomena through what is called linearization. We also do this with respect to economic management. We linearize the system or environment that we are trying to regulate in order for it to fit this model.
As the famous social scientist Herbert Simon once said: “If we want an organism or mechanism to behave effectively in a complex and changing environment, we can design into it adaptive mechanisms that allow it to respond flexibly to the demands the environment places on it. Alternatively, we can try to simplify and stabilize the environment. We can adapt organism to environment or environment to organism.” Many of these economic systems are specifically designed and thus we can try and build them so that they fit in with our linear model even though they may be part of broader nonlinear complex systems. For example, when it comes to forest management we will typically plant all the same trees at the same time in straight rows. We do this in an effort to design a system that is manageable through linear methods.
This also aids in our capacity to model the system. Within a centralized regulatory system, someone needs to understand the whole system. The control is only ever going to be as good as our model, but for many of these complex economic systems like national economies or the global economy, we do not really have any models for them, so we create simplified models and then make the system fit into that model so that we can have some chance of regulating it. But of course, simply linearizing a complex system doesn’t do away with its nonlinear dimension. All we are doing is shifting it outside the system, which means that whereas the system may become more linear, knowable and controllable, its environment may become more nonlinear and uncontrollable. Within these centralize regulatory systems that are driven to optimize, there will inevitably be a drive towards reducing the complexity within the system, that is, to continuously linearize the system in order to be more effective at managing it and directing it towards the perceived desired outcomes.
The result is that everything that is nonlinear and thus does not fit into the model is pushed outside of the system’s immediate context, meaning the system becomes more disconnected from its overall environment. As we continue to try and maintain the system within certain parameters through negative feedback, nonlinear positive feedback phenomena become externalized to the system’s environment. Thus, as the system becomes more linear and controllable, the environment becomes more nonlinear and uncontrollable. The so-called Minsky cycle is an example of this in economics.4 The economic ‘Great Moderation’ that the developed economies enjoyed since the 80’s turned out to be the prelude to one of the largest nonlinear shocks to the system with the Great Depression. By using negative feedback to keep the system within certain parameters – and thus moderate it – we also externalized the nonlinear factors that eventually led to a massive collapse.
Although the linear model to regulation has in many ways proven successful in achieving short-term optimization, it is clearly not designed to deal with complex environments in a sustained fashion. In such circumstances, we need an alternative paradigm. The same factors of complexity that take us beyond the traditional model also enable a new method to economic management.
Effectively managed complex adaptive systems, like the Internet’s routing system, ant colonies or mature ecosystems, are regulated through the distributed interactions between adaptive agents. All of these features need to be present. Firstly, control needs to be distributed out to the local level. This means the components are given autonomy to act according to local information. We then have to invest in the component’s capacity for adaptation and information processing and we have to develop the protocols that facilitate interaction and communications between the agents so that they are able to synchronize their states and we get global patterns of organization. This is before anything about information, communication, and knowledge.5 Agents have the information about their environment. They have the knowledge they need to process that information appropriately and the method of communication so that they can coordinate with others and stay responsive to the environmental context. Interaction is an inherent organization mechanism. The more people interact and communicate, the more opportunity there it for them to differentiate their states with respect to each other, with the result being self-organization. For example, in an ant colony, the queen ant does not tell all the other ants what to do, but instead, each ant responds and adapts its behavior to local information. Ants communicate with their peers through their exchange pheromones, and out of that exchange, we get self-organization, load balancing and the regulation of the whole complex system of differentiated parts.6
This regulatory paradigm will have its capabilities and limitations. On the positive side. Peer-to-peer networked systems like this are scalable to very many or possibly an infinite amount of components. As we do not have many tiers of hierarchy building up or bottlenecks around centralized components, interactions are happening in a distributed fashion and the agents are managing those interactions locally. Thus, we are not getting diminishing returns on scale. It is theoretically infinitely scalable. These networked organizations are normalized for complex nonlinear and volatile environments. Agents can respond immediately to local information. There is no master plan, no projection of what the future will be like. This is beneficial because in complex environments the future is fundamentally uncertain and the whole environment unknowable. The system is inherently nonlinear. There is no inherent drive towards linearization and it may be able to sustain high levels of diversity over time, which will add to its sustainability.
On the other side, nonlinear organizations are not optimized for global coordination and throughput. Global coordination may well be very difficult to optimize, they will typically be far from optimal on the global level. Thus, these systems will typically – but not always – be outperformed within simpler environments. Nonlinear networked guerilla warfare is very effective within the complex environment of a jungle but in a battle that takes place in a very simple environment like a desert, command and control armies will be much more effective. As always, complexity lies somewhere in between the two models of centralized top-down regulation and pure bottom-up distributed regulation. Linear regulatory systems are much better at exploiting known opportunities in simple environments. Nonlinear organizations are much better at exploring and surviving sustainably in volatile changing environments. An efficient and sustainable economic regulatory system would likely involve elements of both and solutions for integrating them.
Of course, this debate surrounding how to manage complex economics systems is not just an academic one. It is of great relevance to our world today. As we increasingly develop an integrated global economy, our traditional centralized linear approach to macroeconomic regulation – the bureaucratic state – is coming under increasing stress and showing its limitation in the face of these global networks, like the financial system, multinational corporations, all of which remain relatively unregulated on the global level. The recent financial crisis has shown how a simple free market approach to this may not be sufficient, and thus a key consideration going forward is how to regulate this global economic network in a more sustainable and secure fashion that limits their current negative externalities on society and the environment.