A meta-analysis of recession indicators
tldr: recession indicators *suggest* that we aren't heading for a recession
There has been a lot of talk about recession indicators as of late, with the most random of fashion trends being flagged as a signal of an upcoming recession. Blazers, boho skirts, peplums and indie sleaze skinny scarves (à la Timothee Chalamet) aren’t just cyclical trends but instead a sign of an incoming economic slump (according to TikTok). But with fashion trends coming in and out of fashion every few years, and many of these trends being rooted in late y2k nostalgia (hi 2008 recession normcore), it did make me ask if we are simply following along our pre-set trend cycle trajectory or are these actually signals of a shift in our economic situation.
I’m not an economist, but I’m going to do my best here: unlike GDP or inflation, which move more slowly, consumer confidence fluctuates quickly and more directly reflects how people feel about their current financial and economic situation (or at least that is the logic I am following, feel free to correct me in the comments). So I wanted to test: do the fashion and beauty trends being flagged as recession indicators actually fluctuate with consumer confidence in the economy?
A quick overview of recession indicators, if you haven’t been indoctrinated just yet. But this is idea that fashion trends reflect what is going on in the economy isn’t new. Economists, fashion historians, and your favourite TikTok fashion commentator (including me lol) have cited examples like:
The Lipstick Index: When money is tight, small luxuries like lipstick sell better (according to an Estée Lauder executive in 2001). [I tested this one back in Feb 2024]
The Hemline Index: Hemlines supposedly rise during times of economic optimism and fall during downturns, as suggested by economist George Taylor in 1926. [Also tried to test this one in May 2024]
Men’s Underwear Index: Economist Alan Greenspan noted that sales of men’s underwear drop before other sectors do, and rise again when the economy starts to recover (based on data collected over the course of the 2008 recession).
Big Bag Theory: I am not sure if I made this up, it came to me in a dream, or someone else on TikTok made this observation and I picked it up (let me know). But wherever it came from, the theory is that women need big bags during an economic downturn as they are carrying more stuff around, instead of having to buy [lunch, coffee, a Kleenex] when out and about. [Tested this one as well in December 2024]
More recently, TikTok has added some new contenders: peplums, office wear, indie sleaze, and miniskirts (in direct competition with the hemline index). Already, we can see that some recession indicators are more probable than others. For instance: office wear. Speaking as a fellow 9-5 peasant, I absolutely understand the desire to *show up as the best employee you can be* when the economy feels unstable and the #openforwork LinkedIn posts start rolling in.
But not all of these make sense. Take peplums: they flourished during the 1930s because they added a feminine element to a suit set without needing much fabric. But that logic doesn’t map cleanly onto 2008 or the 2020s, when fast fashion made fabric cheap and abundant.
Methods: Search volume and structural equation modelling
To start to answer these questions, I used the best data/stats I had available:
Collected 20 years of Google Trends data across 5 search terms per trend (e.g., "mini skirt," “mini dress”, etc.)
Used structural equation modelling (SEM) to combine those terms into one latent variable per recession indicator.
Ran individual linear regressions against consumer confidence (CCI) as the dependent variable.
(Yes, would love to have 80 years of sales data but that just isn’t possible. All data is US-based, mainly just due to it’s availability.)
I wanted to model it this way because I wasn’t trying to predict fashion trends based on the economy but instead wanted to test whether these trends (like interest in peplums or blazers) reflect how people are feeling. In that sense, fashion becomes a cultural sensor rather than the economy predicting fashion trends (or at least that was the logic in the moment, you can argue with me on this in the comments).
Code and datasets are at the bottom of this post, didn’t want to stick 284 lines of R Code in the middle of a post because I know that it is not for everyone.
Results: Mini skirts and blazers are the two strongest recession indicators
Short version: mini skirts and blazers had the strongest relationships with consumer confidence. Big bags and lipstick were also significant, though weaker. High heels and peplums not so much.
Here, trend interest doesn’t cause economic sentiment (Obviously! I don’t want to see one “correlation doesn’t equal causation” comment! I know! That’s not what I am trying to claim!), but it might reflect it.
For the more stat-y people. I do realise that my R2 values are extremely low, especially if you are coming from an experimental background. However, apparently between 0.1 and 0.4 are acceptable in this type of research when you are not trying to make a prediction and your p-values are low (citation).
Discussion: Recession indicators suggest that we aren’t in a recession
The funniest thing to come out of this research, was that my two strongest indicators of economic sentiment are not trending right now. Mini skirts have been decreasing in volume, while blazers have remained stable (apart from seasonal fluctuations) since about 2022.
In the end (and because I am forever skeptical of my own results) I’m left with an inconclusive outcome. We might be heading into a recession. People are clearly thinking about a recession and feeling economically crushed, given the volume of ‘recession indicators’ being discussed. But the fashion and beauty trends that have aligned with consumer confidence fluctuations over the past 20-years aren’t actually trending right now. As always, we’ll only really be able to make sense of this in hindsight. Sure, I could make a prediction that blazers will spike the moment we actually enter a recession (just for context, the US, where this data is based, isn’t in one yet), but I wouldn’t feel confident doing so.
Ultimately, I think that the recession indicator discussion happening online is less about predicting the economy and more about people trying to make sense of what they’re feeling. Grappling with economic anxiety and instability in a way that’s visible and can be shared with others.
Despite my best efforts, I can’t confidently say: here are the definitive recession indicators. Instead, I take it as one point of evidence and something that should be continued to be tracked and tested. (More data! Sales data! Different economic indicators!) I’ll definitely be keeping an eye on how items that came out significant in this analysis continue to track with different economic indicators going forward (and I’ll keep you updated with interesting results as I go along).
Maybe this will be my life’s work and I’ll have a wikipedia article about the miniskirt index someday, but we are not there just yet.
Obviously these results may be flawed and I recognise that some of the stats and data (e.g., using search volume) is not ideal. I am working with what I have available to me, but if you know where to access better data for free — let me know! I would also love to hear your reccs on how to improve the analysis (if you have any thoughts) in the comments below.
Thanks for reading <3 I appreciate you
(Substack doesn’t support .R or .csv files, so you get a PDF and XLSX instead.)
i live for these analyses
I love this analysis… it feels like economics but more feminine and I’m living for it. Trends and fashion are so interesting imo