The effectiveness of decision making
Group decisions are critical to investment firm organizational effectiveness. Teams, committees, informal groups, and boards rely on the socialization and governance attached to decisions, but biases limit the effectiveness of decision outcomes. With new methods and technologies incorporated, biases can be reduced.
KEY POINTS
- The issue for investment firms is the effectiveness of their decision processes for both investment portfolios and business issues. Subconscious biases exist at a personal level that are often reinforced at the group level. This situation is problematic, not least because one of our subconscious biases is overconfidence. The highest-level decisions are the most challenging, involving complex applications of values and beliefs in situations where such issues are not fully settled or even understood.
- The first line of response is to increase the measurement and benchmarking of results to decisions. Investment portfolios are regularly assessed with tightly specified benchmarks, though interpreting the causes of performance variances relative to benchmarks generally proves difficult. In short, attribution errors—not in terms of measurement but in judgment and inference—are endemic.
- Decisions taken on business issues—from hiring and performance management to longer-term strategic decisions—have less data attached and have looser benchmarks. The absence of data creates most of the problems with interpretation. So, the quality of these decisions can be difficult to assess, largely because their measurement is limited.
- A rational response is to develop better foundational thinking and processes. An example of foundational thinking is ensuring the organization draws on considered and well-socialized values and beliefs. Beliefs have substantial relevance to all investment firm challenges—from portfolios to products, client service, and the business model.
- Given greater awareness of biases, firms can adapt to improve decision accuracy while still incorporating consensus.
- In this narrative, there is a role for change. This role is split into two parts: First, better measurement and big data help inform the quality of actions (an example here is the use of group polling; see Dalio 2017); second, technology itself becomes an integral part of the process (for example, AI methods making assessments of job candidates, which is happening already).
- We should qualify the outcomes from AI and machine learning methods, recognizing there are biases with data sets themselves and with the humans that program them. In short, even with impressive leaps in technology we are vulnerable to “garbage in, garbage out” errors.