Epistemic humility

Forecasting is especially inappropriate when there's no relevant history to draw upon

“Have you ever noticed how when something’s banned, there’s less of it?”

This week’s hullabaloo about measures of economic activity amidst shutdown felt like an unintentional stand-up bit. Why -24 instead of -19? How countries compare? Quarter-wise forecasts for rest of the year. It might have been helpful to split the quarter gone by into two divergent halves, but what’s useful isn’t feasible. Since this sort of macro thingy offers entertainment rather than utility even in normal times, extra humour seemed par for the course in abnormal times (technically, it’s pre-new-normal times).

But comedy, however unintentional, carries deeper truths. When I study something in the messy world, I usually pose a few precise questions, on the following lines:

a)       What am I dealing with here?

b)      Have I seen something like this before?

c)       Can (b) help me make sense of (a)?

I know this sounds very L-board, but bear with me. The first question gives me rudimentary yet lifesaving context such as: stock is part ownership in a business; risk is bad things happening; cereal is dessert, not breakfast. Sophisticated academics often skip the first question only to get lost in XL sheets, Greek letters, meaningless acronyms and diabetes.

The second question acknowledges that I cannot reason my way through a messy social world, using non-existent axioms, made-up causality and deductive reasoning. My starting point has to be empirical, by seeking out some relevant history. How have diverse businesses fared over time? How have they been valued? How did different risks pan out? If I have a historical frame of reference that’s relevant enough and robust enough, I have a chance of addressing my third question. Else, no matter how confident I feel about my story, the situation at hand is likely ‘too tough’ to assess, let alone predict. For buggy humans in a messy world, historical patterns and odds go hand-in-hand. Without patterns, there’s no basis for odds.

Back to locked down GDP and our ongoing ‘recession’. What are we dealing with here? As I see it, something unique and unprecedented. It’s abrupt. It’s voluntary. By-diktat, if you prefer. It’s global and synchronized. Its cause is exogenous. It involves unspecified behavioural scarring. Governments, with popular backing, chose public health over economic activity and shut down economies that were generally doing fine till then. While I’ve seen correct references to the exogenous part, I haven’t seen much attention paid to other unique aspects of this thingy we’re all suffering from (thanks to an illegal immigrant from our unfriendly neighbour). Sure, economists study recessions, but these tend to be endogenous and involuntary, due to accumulated excess of many years causing Minsky moments within economic systems. As a digression, how’s that for pretending to know something I don’t? So, back to my second question. Has anyone seen anything like this before? Nope, not even close. Then, what’s the basis to predict what happens next, that too quarter-wise with decimal point accuracy?

In an unprecedented situation, there’s no frame of reference to guide us on how things will pan out from here. This doesn’t preclude overconfident folks from making predictions. There are a variety of alphabets floating around including an unlikely K, with a few special symbols thrown in for variety. It just feels like the current round of macro predictions are even less reliable than customary ones (which is like referring to someone as taller than Danny DeVito).

There’s another danger to theorizing amidst extraordinary times: recency bias on steroids. This behavioural quirk is strong even in normal times. When we have just lived through vivid, dramatic events, they have a disproportionate effect on our buggy minds. It’s collective PTSD. Analysts claimed, plausibly, that there will be heightened reluctance to allowing strangers in to paint homes. Counter-intuitively, this segment is now seeing higher activity than a year ago. While it’s reasonable to expect the guy shouting “Jarugandi, jarugandi” to have an easy 2020, I wouldn’t bet money on it.

I am averse to forecasts, macro or otherwise. I eschew situations where there’s no frame of reference. I am wary of overweighting stark, over-publicized, recent information. I find data sets from an unrepresentative phase more dangerous than useful. As we’re amidst a lollapalooza of all of these and more, I felt like rehashing my views on epistemic humility.

Messy world is unknowable, even when we seem to know what we’re dealing with. In unprecedented times, we don’t stand a chance. While I understand people in real jobs modelling a range of scenarios to ensure preparedness, I don’t understand people without real jobs making a pointless point forecast. It’s a good time to replace ‘forecast’ with ‘guess’.