June 23, 2017
Complexity Theory Offers Insights (Part 1)
by Ron Ginn
The conceptual framework best suited to understanding our networked world is complexity science. It shows how insurance must evolve.
In the first of this series of four segments, we will look at the current state of the risk markets and the insurance industry; the emerging peer-to-peer (P2P) segment of the risk markets; how blockchain technology is enabling a new taxonomy in the risk markets; and what changes may occur as a result of these new technologies and methods.
The purpose of this series hails from the open source movement in the software industry. Key to the open source philosophy is the transparent and voluntary collaboration of all interested parties. While this work has been kept fairly close to the vest for the past few years, I have taken meetings with two Fortune 500 insurance companies’ strategy and venture teams, both of which asked for a proof of concept — as well as with a handful of other large international insurance companies and one of the big four accounting firms.
At the other end of the spectrum, I have also spoken with other founders of P2P insurance startups around the world, and I have participated in the communities surrounding blockchain technology. I feel that these handful of folks have already enjoyed early access to these concepts, and my motivation with this series is to achieve a more level playing field for all parties interested in the future of the risk markets.
There are links at the bottom of this article to join the conversation via a LinkedIn group and get access to the whole series.
To begin, let’s take a look at the current state of risk markets. It is important to distinguish between drivers of economic systems and the impact they have on business models in the industrial age vs. in the information age.
See also: Should We Take This Risk?
Hardware and technology was a key driver throughout the industrial age, which saw a growing batch of new technologies — from cars and planes, to computers and smart phones, to industrial robots, etc.
Industrial age business models were almost always “extractionary” in their nature. The business model engages with some market, and it profits by keeping some portion of the market’s value.
Extracting value from the market
The strategies of the industrial age were:
- Standardization — interchangeable parts
- Centralization — big factories, vertical integration, economies of scale
- Consolidation —an indication that an industry is about to experience a phase change
In the information age, business models almost always embody some creation of “network effect.” When the business model engages with a market, the individual actors all benefit as more actors engage with the business model. The value creation is usually tied to a network’s graph, and the value creation will grow exponentially as the network’s density grows.
Creating value for the market, not extracting value from the market
The strategies and efficiency-drivers in the information age are:
- Cheap connections — enabling multiple paths through the network’s graph
- Low transaction cost — in terms of time, effort and money
- Lateral scaling — not vertical structures, which will be flattened out (“top down” increases network fragility)
- Increase in node diversity — and in the ways each node can connect
All of these drivers lead to increasing network density and flow. Things are moving away from large, brittle centralized organizational structures and toward “distributed,” P2P, “crowd” or “sharing economy” types of organizational structures.
Moving away from command-and-control organizational structures is almost impossible for organizations that profit from efficiency gains derived from a centralized effort. It is this attribute of their business model that necessitates startups and new business models coming in and bringing improvements to the market — challenging incumbent economic and business models.
The information age is all about networks (not technology), and building graphs that create positive network effects.
The conceptual framework best suited to understanding networks and the networked world we now live in is complexity science. The study of complex adaptive systems has grown out of its roots in the 1940s and has proliferated since the 1990s and the explosion of computer networks and social networks. Here is an introduction:
When looking at complex systems, we start by looking at the system’s graph. To get an idea of what a graph is, let’s look at a few examples of “graph companies.”
- Facebook built the “social graph” of acquaintances; it did not create acquaintances.
- Linkedin built the “professional graph” of coworkers and colleagues; it did not create coworkers and colleagues.
- Google built the “link graph” for topics searched; it did not create back links for the topics searched.
Notice that, in each of these cases, the company built and documented the connections between the things or nodes in the network and did not create the things or nodes themselves. Those already existed.
To start looking at the risk markets, we must first understand what is being connected or transferred between the nodes (a.k.a. the users). It should be of little surprise that, in the risk markets, it is risk that is being transferred between nodes, like a user transferring risk to an insurance company. In terms of risk graphing, there are currently two dominant graphs. A third is emerging.
Let’s take a look at the graphs that make up the risk markets and the insurance industry.
- Insurance — is the “hub and spoke” graph.
- Reinsurance — is the decentralized graph connecting risk hubs.
- P2P Coverage — will be formalized in a distributed graph. (This is the one that does obviously not exist formally, but, informally, you see people calling parents/friends and using GoFundMe/their church/their office/other community organizations to spread risk out laterally.)
In today’s risk markets, insurance companies act as centralized hubs where risk is transferred to and carried through time.
The reinsurance industry graph is enabling second-degree connections between insurance companies, creating a decentralized graph. In the current industry’s combined graph structure or stack, only these two graphs formally exist.
While an insurance company’s ledgers remain a hub where risk is transferred to and carried through time, reinsurance enables those risk hubs to network together, achieving a higher degree of overall system resilience.
See also: Are Portfolios Taking Too Much Risk?
The P2P distributed graph currently exists via informal social methods.
Stack all three graphs, and you can observe how total risk is addressed across all three graph types. Each has its strengths and weaknesses, which leads to its existing in its proper place within the risk markets.
The fact that insurance as a financial service gets more expensive per $1,000 of coverage as coverage approaches the first dollar of loss means that, as a financial service, there is a boundary where insurance’s weaknesses will outweigh its strengths.
My expectation is that much of the risk currently being carried on the hub-and-spoke insurance graph will accrue to the P2P distributed graph because of improved capital efficiency on small losses via a trend of increasing deductibles. This may lead to some of the risk currently carried on the reinsurance decentralized graph being challenged by centralized insurance.
The proportion of total risk — or “market share” — that each graph carries will shift in this phase change.
When people say insurance is dropping the ball, they are expressing that there is a misunderstanding or poor expectation-setting about how much of total risk the first two graphs should be absorbing. Users are unhappy that they end up resorting to informal P2P methods to fully cover risk.
To increase the resilience of society’s risk management systems and fill the gaps left by the insurance and reinsurance graphs, we need the third risk distribution graph: a distributed P2P system.
Society needs a distributed system that enables the transfer of risk laterally from individual to individual via formalized methods. This P2P service must be able to carry un-insurable risk exposures, such as deductibles, or niche risk exposures that insurance is not well-suited to cover.
Much of this activity already occurs today and, in fact, has been occurring since the dawn of civilization. KarmaCoverage.com is designed to formalize these informal methods and enable end users to benefit from financial leverage created by the system’s network effect on their savings.
When observing a system through the complexity paradigm, another key measure to observe is a system’s level of resilience vs. efficiency. Resilience and efficiency sit on opposite sides of a spectrum. A system that is 100% resilient will exhibit an excess of redundancy and wasted resources, while a system that is 100% efficient will exhibit an extreme brittleness that lends itself to a system collapse.
When we look at the real world and natural ecosystems as an example, we find that systems tend to self-organize toward a balance of roughly 67% resilient and 33% efficient. Here is a video for more on this optimum balance.
Industrial-age ideas have driven economics as a field of study to over-optimize for efficiency, but economics has, in recent years, begun to challenge this notion as the field expands into behavioral economics, game theory and complexity economics — all of which shift the focus away from solely optimizing for efficiency and toward optimizing for more sustainable and resilient systems. In the risk markets, optimizing for resilience should have obvious benefits.
Now, let’s take a look at how this applies practically to the risk markets, by looking at those three industry graphs.
Centralized network structures are highly efficient. This is why a user can pay only $1,000 per year for home insurance and when her home burns down get several hundred thousand dollars to rebuild. From the user’s point of view, the amount of leverage she was able to achieve via the insurance policy was highly efficient. However, like yin and yang, centralized systems have an inherent weakness — if a single node in the network (the insurance company) is removed, the entire system will collapse. It is this high risk of system collapse that necessitates so much regulation.
In the risk markets, we can observe two continuing efforts to reduce the risk of an insurance system collapse. We observe a high degree of regulation, and we see the existence of reinsurance markets. The reinsurance markets function as a decentralized graph in the risk markets, and their core purpose is to connect the centralized insurance companies in a manner to ensure that their inherent brittleness does not materialize a “too big to fail” type of event.
Reinsurance achieves this increase in resilience by insuring insurance companies on a global scale. If a hurricane or tsunami hits a few regional carriers of risk, those carriers can turn to their reinsurance for coverage on the catastrophic loss. Reinsurance companies are functionally transferring the risk of that region’s catastrophic loss event to insurance carriers in other regions of the globe. By stacking the two system’s graphs (insurance and reinsurance), the risk markets’ ability to successfully transfer risk across society has improved overall system resilience while still retaining a desired amount of efficiency.
Observations of nature reveal what appears to be a natural progression of networks that grow in density of connections. Therefore, it makes sense that the reinsurance industry came into existence after the insurance industry, boosting the risk markets’ overall density of connections. Along the same line of thought, we would expect to see the risk markets continue to increase in the density of connections from centralized to decentralized and further toward distributed. A distributed network in the risk markets will materialize as some form of financial P2P, “crowd” or “sharing economy” coverage service.
A network’s density is defined by the number of connections between the nodes. More connections between nodes mean the network has a higher density. For example, a distributed network has a higher density of connections than a centralized network. However, a higher density of connections requires more intense management efforts. There is a limit to how much complexity a centralized management team can successfully organize and control.
See also: 5 Steps to Profitable Risk Taking
When a network’s connections outgrow centralized management’s capacity to control, the network will begin to self-organize or exhibit distributed managerial methods. Through this self-organization, a new graph structure of the network’s connections will begin to emerge. As this process unfolds, an entirely new macro system structure will emerge that shows little resemblance to the system’s prior state, much like a new species through evolution.
What emerges is a macro phase change (aka “disruption”) that does not necessitate any new resource inputs, only a reorganization of the resources. For example, the macro state of water can go through a phase change and become ice. The micro parts that make up water and ice are the same. The macro state, however, has undergone a phase change, and the nature of the connections between the micro parts will have been reorganized.
In his book “Why Information Grows: The Evolution of Order from Atoms to Economies,” MIT’s Cesar Hidalgo explains that, as time marches forward, the amount of information we carry with us increases. That information ultimately requires a higher density of connections as it grows. This can be understood at the level of an individual who grows wiser with experiences over time. However, as the saying goes, “The more you know, the more you know you don’t know.”
In the history of human systems, we have observed the need for families to create a tribe, tribes to create a society and society-organizing-firms to achieve cross-society economic work. We are now at the point of needing these firms to create a network of firms that can handle increased complexity and coordination.
It is this network of firms that will be achieved via distributed methods because no individual firm will ever agree to let another single firm be the centralized controller of the whole network — nor could a single firm do so.
In the next segment of this series, we will look more closely at the distributed graph that will become formalized, creating a P2P system in the risk markets.
I have started a LinkedIn group for discussion on blockchain, complexity and P2P insurance. Feel free to join here: https://www.linkedin.com/groups/8478617
If you are interesting exploring working with KarmaCoverge please feel free to reach out to me.