The 5 categories you spend most on when you're stressed

Cost Me Research Desk · May 27, 2026

TL;DR. Stress, sadness, and loneliness reliably increase spending in five categories: food delivery, alcohol, online shopping, gambling, and comfort products. The pattern is consistent across decades of research and probably accounts for more impulse spending than any other single driver.

Almost everyone has noticed it about themselves at some point. A hard day at work, a fight with a partner, a layoff, an anniversary of a loss — and the bank statement shows a strange cluster of charges that don't look like the rest of the month. Different category, different time of day, different size.

Affective spending — spending driven by emotional state — is one of the most consistently studied phenomena in consumer behaviour. Here are the five categories where the research is most clear, and what to do about each.

1. Food delivery and convenience food

This is the most common stress-spending category by transaction count in essentially every consumer-finance dataset that's been examined. The reasons are straightforward: stress depletes the executive function needed to plan and cook a meal, food is a reliable short-term mood lifter, and delivery apps have removed every possible point of friction between craving and arrival.

Cryder, Lerner, Gross, and Dahl (2008) ran a now-classic experiment showing that sad subjects were willing to pay roughly four times as much for an identical commodity (a water bottle) as control subjects — without being aware of any difference in their valuation (Cryder, Lerner, Gross & Dahl, 2008).¹ The mechanism, they argued, was self-focus: sadness narrows attention to the self, and a purchase reads as self-soothing in a way that overrides normal price-checking.

The food-delivery version of this is the same effect on a different stimulus. The cost differential between a $35 DoorDash order and a $7 meal cooked at home maps almost exactly onto Cryder's “sad surcharge.”

2. Alcohol and recreational substances

Self-medication is the textbook explanation here, and the data is consistent. Stressful life events — job loss, divorce, bereavement — correlate with measurable upticks in alcohol purchases at scanner-data resolution. The effect is largest in the first weeks after the stressor and decays over months.

From a spending standpoint, alcohol has two compounding properties. The unit cost is higher than most categories when bought at bars and restaurants. And it impairs judgment in a way that increases other stress spending — late-night online orders, taxi rides, food delivery — that wouldn't have happened sober.

3. Online shopping (retail therapy)

Atalay and Meloy (2011) ran the most-cited modern study of the “retail therapy” phenomenon. They surveyed and experimentally manipulated mood in consumers, then tracked whether unplanned purchases improved subjective well-being (Atalay & Meloy, 2011).²

Their finding, which surprises people: retail therapy actually works in the short term. Subjects who made an unplanned purchase to lift their mood reported meaningfully better affect afterward than control subjects. The lift was real and not driven by buyer's remorse later.

This is awkward to acknowledge in personal-finance writing, because it means the behaviour isn't irrational in the moment. The cost is real, the intervention works, and the only honest argument against it is opportunity cost over time.

Retail therapy works. That's the problem. A cheaper intervention that worked half as well would be a much better deal.

4. Gambling (lottery, sports betting, online casinos)

Gambling deserves its own category because the emotional dynamics are different. The appeal isn't soothing — it's the introduction of variance into an emotional state that feels static and bad. A depressed or anxious mood is, in some sense, a flat line. Gambling produces sharp highs and lows that interrupt the flatness, even when the expected value is negative.

The app-based versions of gambling — sports betting, online slots, crypto trading framed as gambling — have intensified this pattern dramatically in the last decade. Their UX is designed around variable-ratio reinforcement, the most powerful and most addictive reward schedule known. Stress increases the appeal of variable reinforcement because it's a faster emotional state-change than the slow grind of mood recovery through normal life.

5. “Comfort” categories — candles, blankets, skincare, snacks

Pieters (2013) ran a longitudinal study showing that loneliness and materialism reinforce each other over time — lonelier people buy more material goods in search of comfort, and acquiring more material goods predicts increased loneliness six months later (Pieters, 2013).³ The cycle is bidirectional, which means breaking it requires intervening on whichever side is easier to change.

The specific subcategory that shows up most often in the data is what you might call cozy consumption: candles, soft blankets, premium skincare, comfort snacks, hot drinks. These purchases are explicitly framed by their marketers as self-care, and the framing matters — they convert what would be ordinary spending into something that doubles as emotional labor.

None of this means a candle is bad. It means a $40 candle bought on a hard Tuesday evening, every other Tuesday for a year, is $1,040 — and probably wouldn't have been bought without the underlying emotional state.

What the five have in common

Pattern across all five: they reduce friction at the moment of the emotional state. Delivery apps, one-tap purchases, in-state lottery apps, scrollable shopping feeds, Amazon. None of them require leaving the couch, leaving the moment, or pausing long enough for the affect to shift.

Stress-spending isn't a personality trait. It's the interaction between a known psychological pattern and an environment that's been specifically engineered to capitalize on it.

What this means for you

Three reframes that the research supports. First, notice the categories. If your stress spending is concentrated in two or three of the five above, that's the inventory worth looking at — and the most actionable place to intervene.

Second, find cheaper interventions that work. Atalay and Meloy's data is a useful starting point: the mechanism of retail therapy is the brief sense of agency and self-care, not the goods themselves. A walk, a conversation with a friend, a hot shower, a nap — these produce a similar lift at meaningfully lower cost. They are not as instantly available as a cart full of Amazon, which is exactly why pre-committing to them matters.

Third, add friction where it matters most. Remove saved cards from delivery apps. Uninstall gambling apps. Move your shopping bookmarks somewhere two taps deeper. The goal isn't to eliminate the spending; it's to make sure the spending that happens is the spending you'd still endorse twelve hours later.

The honest limitation

These studies measure short-term affect and immediate purchase behaviour. They don't establish that any individual stress purchase is bad for any individual person. Some genuinely earn their cost — and some people are dealing with situations where the alternative interventions aren't available. The argument is not that stress spending is a moral failure. It's that recognizing the pattern is the cheapest, easiest first step toward changing whatever portion of it you want to change.

References

1. Cryder, C. E., Lerner, J. S., Gross, J. J., & Dahl, R. E. (2008). Misery is not miserly: Sad and self-focused individuals spend more. Psychological Science, 19(6), 525–530. https://doi.org/10.1111/j.1467-9280.2008.02118.x
2. Atalay, A. S., & Meloy, M. G. (2011). Retail therapy: A strategic effort to improve mood. Psychology & Marketing, 28(6), 638–660. https://doi.org/10.1002/mar.20404
3. Pieters, R. (2013). Bidirectional dynamics of materialism and loneliness: Not just a vicious cycle. Journal of Consumer Research, 40(4), 615–631. https://doi.org/10.1086/671564

Want more like this? Decision fatigue and late-night spending or the “I deserve it” reframe. Or head back to costme.io.