Overcoming Mental Inertia
I'm back after a very long hiatus. I've been doing a lot of analytics lately that ended up going against expected results, and it got me thinking about how people form ideas and hold onto them, and how that can be overcome.
A few weeks ago I came across this great substack that discusses how little the average American knows about how their government works. Lest you might think this is another typical “punching down” article that mocks Americans for not knowing how many senators there are or how long they serve, it is quite different. This is really about the fact that Americans, all of us, yes you included (non-Americans too, probably), do not fully understand the responsibilities and functions of the various different agencies that make up our government, both local and national. This is absolutely true! Even the people closest to the government do not know its functions. One of my favorite examples is Rick Perry, who somewhat infamously did not understand the responsibilities of the Department of Energy, which he eventually ran during the Trump administration. The author, Hamilton Nolan, goes on to argue that journalists should not take polling on whether or not an agency should be abolished, or similar questions, as representative of national attitudes, because the subjects of the poll can not be expected to understand what they are talking about.
I came away from that writeup with a completely different takeaway. It made me think about how difficult it is to overcome what I guess I’d call mental inertia. I'm sure there’s a psychological term for this. Once an idea has taken root, it is incredibly difficult, even in the face of overwhelming factual evidence, to overcome the initial idea. Often if you try to convince a person that their deeply held belief is wrong, they will simply dig their heels in even further. It’s a difficult problem. Sometimes it’s obvious that certain beliefs are deeply tied to an emotional response, and that’s what makes them difficult to overcome. People that believe that the Department of Education needs to be abolished are not actually mad about the functions they serve, they are mad about some aspect of public education that they have an emotional response to, there’s not much logic to the position. I think those cases are easier to understand. What’s more difficult to understand are more benign beliefs, untied to emotion, that are difficult to overcome with evidence. This happens all the time in business.
Early in my career, I had first hand experience with government officials that had a bad hypothesis about a very important but very in-the-weeds bit of government business. I worked for a consultancy that specialized in deeply understanding federal lending programs. Most of our work was for agencies who were trying to accurately calculate the current net present value of their loan portfolios for budget reasons, and wanted to ensure they were appropriately accounting for the various account transactions that happened during budget season. At the time, Paul Ryan was the Speaker of the House, and he (really his Chief of Staff) had a pet project where they wanted to change the accounting for these loan programs. Importantly, this was not really a partisan project (though there were partisan elements), it was mostly pure wonkiness. This is not an issue that the average American cares about, and they’d fall asleep if you tried to explain it to them, but that’s not going to stop me here!
The federal government has dozens of federal lending programs, and there is a statute that governs how the accounting is done for these programs. To simplify, each agency evaluates the credit risks in their program in a manner similar to a combination of how a bank would evaluate credit risk and how rating agencies like Moody’s and Fitch evaluate credit risks. They calculate a net present value for their portfolio accounting for the timing and severity (read $$ amount) of expected loan losses (delinquencies and defaults), and if it’s a negative result, that program is deemed to cost the federal government money, if it’s a positive result, the program saves money against the deficit. The Speaker’s team thought that programs, on net, were understating expected losses by not accounting for certain market risks, which would inflate the value of these programs. Our team was brought in by a trade group to see if 1) their hypothesis was true and 2) if so how big was the error. It’s a pretty straightforward evaluation. We had, at the time, over 20 years of history with this accounting process, and had gone through at least four different extreme business cycles as a country. So the team simply evaluated whether expected losses were different from actual losses. They were not, and that lack of a systemic bias meant there was no systemic risk being missed in these evaluations. The response to this huge analysis, and all of this evidence, was “that can’t be true”.
I find this happens all the time in business, and as a data professional I’m often caught right in the middle of these problems. Often leaders, particularly very senior leaders, have a strong hypothesis about what is going on in a particular business area. In my current work, leaders have strong beliefs about different headwinds that may be facing certain business areas and impacting customer behavior. Data people are brought in to evaluate what is actually happening, but the trap often is that the data gets shaped to the original hypothesis, and we sometimes fail to test for what’s actually going on in the business. Furthermore, if the data do not support the original hypothesis, often the data is dismissed as anomalous rather than the hypothesis itself. It becomes a string of hypotheses about data anomalies that all circle back to say “our original belief is true”. And this is a direct result of mental inertia. People really do not like letting go of their beliefs. So what to do?
To go back somewhat to the original article at the top of the post, the best weapon is a good story. Voters do not have a deep understanding of budgetary processes and which agencies are underfunded and which agencies have too much money. They don’t know whether they spend money efficiently, whether the consultants they hire are good, or whether that job function should even require consulting support. What they do understand are stories that they are told by people they trust about how all of this should work. The same goes for the relatively unemotional hypothesis your leader has about some change in the company’s operations. It is a story, it is a logical story they have concocted based on their wealth of previous experience. If you want to convince them it’s wrong, you too must have a compelling story. “This number is X% not Y% and thus this is not what’s happening” is not a compelling story.
A compelling story does not guarantee you will win the argument, but it gives you a better shot. My previous story about government accounting was not compelling. I said that expected loss = actual loss and thus there’s no bias. Case closed. What I could say is something like this:
In public markets, market risks are associated with the broader movements of the entire economy or certain sectors of the economy. These movements are, in general, uniform and measurable. If a company was failing to take into account these risks, there would be a “bias” between their forecasted performance and their actual performance that was exactly equal to the market risks they failed to account for. We can evaluate the government’s loan portfolio the same way. On a global portfolio level and within each sector, we can look at the forecasted losses for these programs and compare them to the actual losses. If there is a consistent difference in forecasted vs actual losses over time, then we can conclude that the government failed to account for some risks, perhaps market risks.
In fact we did say something like that, and went on to spell out our conclusion, and obviously we were still unconvincing. But the point is there are no numbers there, there is only a narrative. If you are not using narrative to shape your argument, and you’re only using numbers and counting on your audience to reach the same conclusion, you’re going to fail. At least a story gives us a fighting chance.
By the way, all of that government lending data is publicly available. You can check the work!