When are we just speaking for a model

And then catching flak for it

A wizened and grumpy looking turkey at the zoo

For the past 6 months or so, something has been nagging at me and while I don't think I fully have a grasp on what is bothering me just yet, I'm going to use this space this week to poke around at the idea in my head.

The thing that is bothering me originates from the swirling debate around inflation in the US and how the US Federal Reserve is attempting to bring inflation down. More specifically, as interest rates were being cranked up at a relatively breakneck pace towards the end of 2022, there was a lot of accusations that the Fed was, and still is, deliberately causing unemployment in order to bring down inflation. There’s something of an implication that there are potentially alternative actions that the central bank, or the Federal government can take, that would be more effective at lowering inflation.

The accusations aren't completely unfounded because in multiple public statements by the chairman, variations of the sentiment that the unemployment rate was too low and the employment market is too tight. The explanation goes that the resulting pressure of rising wages is one component (though not the biggest) of inflationary pressure. But of all the known factors at work that do contribute to inflation, economic activity and by extension unemployment are the only factors within control of the central bank, so it is a lever that they are going to pull in order to follow their mandate to bring price stability back.

My simple understanding of the situation is that a key linchpin of this reasoning and current economic models is originally grounded in the Philips curve — an observation that inflation and unemployment are inversely related. It implies that an economy can trade inflation for more or less overall employment levels to some extent — meaning slowing an economy (and thus hiring) with economic policy has a lowering effect on inflation. It is currently a hot debate in economics as to whether the Philips Curve has any relevance in the past 30 years because it has flattened significantly. Despite the current debate, for the moment, modified variations of the concept are still important to economic models and economic policymakers. Since I'm not an economist, nor particularly savvy about economic models and alternative proposals and mechanisms for inflation that get floated, I'm not here to discuss the models themselves.

But I couldn't help but feel that there is something fascinating about the situation. Here, the leader of the banking system for the US, heading a large organization of people who are definitely experts at the complexities of measuring, understanding, and predicting how economies function and are modeled, has been backed into a pretty undesirable situation of saying that in order to bring inflation down, unemployment will also have to go up. It’s effectively a consequence of how the current models and theoretical understanding of how things say things will turn out. I even remember a couple of instances in press conferences where the chairman had to reiterate that they don’t think increasing unemployment is good, but that there’s room for it compared to historic levels.

For people distrustful of central banking and sensitive to the economic inequalities of today, that has been interpreted in some circles as the Fed is deliberately engineering for people to lose their jobs in order to “control” inflation while the root cause is likely somewhere else like corporate greed, ongoing war, or COVID. I’ve even seen comments floating around on social media couching it in class warfare rhetoric of the rich sticking it to the poor.

Even a more charitable reading still implies that the Fed will keep increasing interest rates and attempting to slow economic growth and will stomach any subsequent unemployment rate increases as a necessary evil. The Fed is still going to pull the one monetary policy lever that they have, and the only material difference is what intent you ascribe to that lever pull.

Because of how the current economic models are set up, once it has been determined that inflation must come down, the fate of the unemployment number is effectively sealed as a consequence of the modeling machinery. The rhetoric about grand conspiracies, greed, or unfortunate correlations are just projections of intent based on where people believe the many complex, interacting arrows of causality within the model are pointing.

If you accept that the set of equations and assumptions that create the economic models are the best-known approximations of how the economy behaves, you’ve effectively committed to the narrative that moving one parameter implies a correlated shift of other parameters. It just becomes a giant public flashpoint when said models and actions will affect millions of people and everyone has an opinion. Even more so when there isn’t a single generally accepted theory and model for economics.

I find the whole situation fascinating. Here’s this massive apparatus of statistics collection, modeling, and heated academic debate, and the essential argument is “we find inflation unacceptable and so the only way we can do something about it without lawmakers making new policies is to do this painful attempt at slowing the economy without crashing it.” The models dictate thusly and we haven’t found any alternatives that say otherwise.

I don’t particularly believe that the economists behind the wheel are completely blind to the fact that changes in unemployment statistics or inflation represent real suffering by millions of people. Nor are they plotting some grand scheme in fancy board rooms like cartoon villains. But when they get put on press conferences and TV and that’s the gist of what they can present to convince the skeptical public to go along with the plan… (a-la the sentiment in Benn’s recent post about persuading people to go along with things)… while also attempting to set people’s expectations for inflation (which some models are theorizing are an important factor)… and it’s a mess I very much enjoy not having to deal with.

At what point in our job do we effectively become mere mouthpieces for a model that we’ve built and accepted as the guide for our course of action? When has our agency been subsumed by the model as we continue to believe in the grinding gears until we convince ourselves to stop?

At work, the most I’d probably have to deal with with any frequency are trivial examples of this like how how raising prices will affect customer retention and satisfaction — such a model of reality isn’t particularly controversial. Everyone knows what the rough cause and effects are, so we can very quickly move on to discussions about what levels of discontent are acceptable to decision-makers, what method of executions will lead to the best result possible given the situation. While even such scenarios can have ethical implications (imagine you’re setting the prices for critical safety equipment), they very often merely inconvenience people.

So, despite writing a thousand words in bemusement about the whole situation, I’m not sure if I’ve wound up anywhere particularly different from where I started. Us data scientists have to live in a world of models and measurement to do our work. Sometimes, when things go off the rails and we must use our models to help guide people into taking a certain path, it can get very difficult. Especially when people are going to be unhappy no matter what.

It does raise some questions in my mind as to what kind of models do I ever work with that would leave with with enough believe in their inherent truth that I would be willing to use them as a “plan” that needed to be followed despite opposition. There’s probably one or two out there, but I can’t think of them off the top of my head. Most of the models in my head are flexible and up for for debate since we haven’t taken the time to establish the foundation with rigor. We in industry don’t deal with hard durable truths — we move too quickly for that to happen except across the span of maybe a whole career of experience. It’s vibes backed by circumstantial data for the most part, stuff that’s fairly easy to cast doubt on when headwinds arise.

So perhaps a part of the reason why I’m not likely to go up and defend a model to the point of becoming a public punching bag for one is because… I don’t believe in them nearly enough to warrant that level of conviction.

If only I can tell whether that’s a good thing or not…

---

If you’re looking to (re)connect with Data Twitter

Please reference these crowdsourced spreadsheets and feel free to contribute to them.

• A list of data hangouts - Mostly Slack and Discord servers where data folk hang out

• A crowdsourced list of Mastodon accounts of Data Twitter folk - it’s a big list of accounts that people have contributed to of data folk who are now on Mastodon that you can import and auto-follow to reboot your timeline

---

Standing offer: If you created something and would like me to review or share it w/ the data community — my mailbox and Twitter DMs are open.

Guest posts: If you’re interested in writing something a data-related post to either show off work, share an experience, or need help coming up with a topic, please contact me. You don’t need any special credentials or credibility to do so.

---

About this newsletter

I’m Randy Au, Quantitative UX researcher, former data analyst, and general-purpose data and tech nerd. Counting Stuff is a weekly newsletter about the less-than-sexy aspects of data science, UX research and tech. With some excursions into other fun topics.

All photos/drawings used are taken/created by Randy unless otherwise credited.

randyau.com — Curated archive of evergreen posts.
Approaching Significance Discord —where data folk hang out and can talk a bit about data, and a bit about everything else. Randy moderates the discord.

Support the newsletter:

This newsletter is free and will continue to stay that way every Tuesday, share it with your friends without guilt! But if you like the content and want to send some love, here’s some options:

• Share posts with other people

• Consider a paid Substack subscription or a small one-time Ko-fi donation

• Tweet me with comments and questions

• Get merch! If shirts and stickers are more your style — There’s a survivorship bias shirt!