An Oscar nomination earns an extra $2 million. Here's who pays for it.
In the last analysis, I showed that an Oscar Best Picture nomination is worth roughly $2.2 million per nominee in the three weeks after the announcement. Past Lives went from 5 theaters to 188. American Fiction picked up an extra $270K per day. Across the eight nominees with theatrical runs in the 2024 cycle, the total came to about $18 million.
That piece ended with an obvious follow-up question. Where does that money come from?
There are two possibilities. The option is that the Academy brings new people to the movies. Someone who wasn’t planning on seeing anything this weekend hears the nominations, gets curious, buys a ticket. Total spending at the box office goes up. Everyone wins.
The other option is that the Academy just redirects existing moviegoers. Someone who was going to see Aquaman decides to see The Zone of Interest instead, and the total money flowing through theaters stays about the same.
The Three Lines
To answer this, I need to zoom out from individual nominees and look at the entire box office. I compiled every daily box office observation in the United States from 2000 to 2026. That’s 386,866 film-day records across 9,036 films and 27 Oscar announcement dates. For each announcement, I summed every dollar earned by every film playing in American theaters and tracked how that total behaved in the weeks before and after nomination announcements.
Note: The y-axis is an index where each ceremony year’s pre-announcement mean daily gross is set to 100. A value of 110 means revenue is 10% above that year’s baseline. Indexing neutralizes cross-year scale differences (e.g., 2002 vs. 2024) so that percentage deviations from normal are comparable. Lines are 7-day centered rolling averages to smooth weekend/weekday cycles.
After the announcement, the nominee line rises. The non-nominee line falls, and the total barely moves. From a purely descriptive standpoint, it looks like nominations are redistributing revenue and not expanding it.
But Is That Real?
Eyeballing trend lines is how you get fooled, so to test whether total box office actually jumps at the announcement date, I looked for a sharp discontinuity right at day zero. if nominations bring new people into theaters, total daily revenue should visibly jump when the names are read.
It doesn’t. The estimated jump is +6.9%, but the confidence interval runs from -5.4% to +20.7% and the p-value is 0.27. This finding remains true at variable bandwidths as well. I also tested at 14, 28, and 42 days. In none of these windows did box office growth show statistically significant change.
I guess eye-balling it could’ve worked on this one.
Inside the Pie
So the total doesn’t move, but there’s still some work to do on this analysis. Below is a story in three acts. The top panel is total market revenue, which is flat at the cutoff. The middle panel is nominee revenue, which is noisy but drifting upward. The bottom panel is non-nominee revenue, which is flat to declining.
Every dollar at the box office goes to either a nominee or a non-nominee. So if the total isn’t growing but nominees are gaining ground, that money has to be coming from non-nominees.
Both groups are declining in raw terms, that’s just what happens in January and February as the holiday blockbusters wind down. The whole market is shrinking. But nominees are shrinking slower. On average, the total market drops about 24% from pre- to post-announcement. Nominees drop only 11%. Non-nominees drop 26%. The nominees are losing less ground, and in a shrinking market, losing less ground is gaining share.
The Money Shot
Before the announcement, Best Picture nominees collectively earn about 15.7% of total daily box office. After, it jumps to 17.1%. That raw 1.5-point increase understates the true shift, because nominee share was already trending upward from earlier award-season buzz (Golden Globes, SAG, etc.). Once I adjust for that pre-existing trend, the estimated jump caused by the Oscar announcement is 4.1 percentage points (p = 0.009). At the wider 28-day bandwidth: 4.8 percentage points (p = 0.004).
What does 4.1 percentage points mean in dollars? At typical daily market volumes of $20-25 million, roughly 4 cents of every dollar flowing through American theaters shifts toward nominees at the moment of the announcement. That’s about $800K to $1 million per day being redirected. Over a three-week window, you’re looking at something in the neighborhood of $15-20 million per ceremony cycle.
That number should sound familiar. In the last piece, I estimated $18 million in total nominee gains from the film-level data. This market-level analysis lands in the same range, despite using different data, different methods, and 27 years instead of 1. Last time I measured what nominees gained. This time I measured what everyone else lost, and, fortunately, the two numbers agree.
What about the individual non-nominee? The per-film revenue doesn’t significantly change (p = 0.45) since the loss is distributed widely. There’s a few hundred fewer tickets per film per day, across dozens of movies. No single non-nominee gets crushed. It’s a gentle, collective haircut.
What This Means
Oscar nominations redirect about $15-20 million per year from non-nominees to nominees. The 4.1 percentage-point share shift (p=0.009) is the central finding. At typical daily market volumes, that translates to roughly $800K-$1M per day changing hands in the weeks after the announcement. This is consistent with the $18 million I found from the film-level data last time, which is cool.
We find no evidence that nominations grow the overall box office. The estimated total-market effect is +6.9% with a confidence interval that includes zero. I can rule out large expansion (above ~21%), but I can’t distinguish between pure redistribution and modest growth. The honest answer is that the data don’t have the power to detect anything smaller than about an 19% market increase.
The Academy is an information signal, not a demand creator. This is consistent with what I found last time: the bump is largest for small films where the nomination resolves genuine uncertainty about quality. The Academy tells audiences which film to see, not whether to go to the movies.
Non-nominees bear the cost, but gently. No individual non-nominee gets hit hard. The redistribution is spread across dozens of films. At wide bandwidths (42 days), non-nominee aggregate revenue shows a suggestive 9.6% decline (p=0.058), but one marginally significant cell out of twelve isn’t something to build a case on.
The Oscar bump is real, and it comes out of everyone else’s pocket. The $2.2 million per nominee from the last analysis is real revenue. It just appears to come primarily from audience reallocation rather than audience creation. People who were going to see something chose to see nominees instead.
So what is an Oscar nomination worth? About $2.2 million, and the bill gets split across every other movie in theaters.
Appendices
Robustness Checks
Bandwidth sensitivity. The total-market effect stays near zero from 7 to 50 days. There’s no window where the pie-growth story works.
Placebo cutoffs. I shifted the cutoff to fake dates (7 and 14 days before and after the real announcement) and re-ran the analysis. If the method is picking up a real announcement effect rather than seasonal noise, only the true cutoff should show a positive estimate. That’s mostly what happens.
The real cutoff is the only one with a positive point estimate. Two placebos are marginally significant (t=-14 at p=0.066 and t=+7 at p=0.088), likely reflecting seasonal decline or spillover from the Golden Globes and SAG Awards. Not a perfect placebo battery, but the share shift at p=0.009 stands on its own.
Year-by-year variation. The 27 individual estimates are noisy but cluster around zero. The null isn’t driven by any single outlier year.
Excluding COVID years (2020-2021) doesn’t change the story. +6.4% (p=0.32). Splitting by the original 5-nominee format (2000-2009) versus the expanded field (2010-2026) doesn’t either. Neither subgroup is significant.
B: Full Results
Main decomposition (3 outcomes x 4 bandwidths):
Robustness summary:
COVID exclusion (dropping 2020-2021): +6.4% at BW=21 (p=0.316), +7.1% at BW=28 (p=0.205). No change to the story.
Era split: The 5-nominee era (2000-2009) shows +3.5% at BW=21 (p=0.619) and +9.4% at BW=28 (p=0.199). The expanded era (2010-2026) shows +10.9% at BW=21 (p=0.245) and +6.5% at BW=28 (p=0.436). Neither era is individually significant.
I wish Substack rendered .md or LaTeX tables…
C: Methodology
For the curious:
Data: 386,866 film-day observations across 9,036 films, January 2000 to March 2026. Daily box office from The Numbers and Box Office Mojo, merged and standardized.
Nominees: 206 Best Picture nominees across 27 ceremony years, matched to standardized film titles (100% match rate).
Aggregation: For each ceremony year, I summed daily gross across every film in theaters within a 60-day window on either side of the announcement. That gives 3,230 market-day observations.
Method: Regression discontinuity design. The running variable is days from the announcement. The cutoff is day zero. I fit a local linear regression on each side of the cutoff with triangular kernel weights (observations closer to the cutoff get more weight). The jump at the cutoff is the treatment effect.
Controls: Ceremony-year fixed effects (so I’m not comparing 2002 box office to 2024 box office) and day-of-week fixed effects (so I’m not confusing a Tuesday with a Saturday).
Standard errors: Clustered at the ceremony-year level (27 clusters) with a finite-sample correction. P-values use the t(26) distribution rather than the normal approximation.
Bandwidths: Primary: 14, 21, 28, 42 days. Sensitivity: 7 to 50 days in steps of 2.
Outcomes: Log(total daily gross), log(nominee daily gross), log(non-nominee daily gross), nominee market share (level).
D: Limitations
We can’t rule out modest expansion. The 95% confidence interval for the total-market effect spans -5.4% to +20.7%. The data aren’t powered to detect anything smaller than about 18%. It’s possible the Academy creates some new demand, and I just can’t measure it.
Few clusters. With 27 ceremony years as clusters, the standard errors carry some uncertainty. I apply a finite-sample correction and use the t(26) distribution, which is conservative. The null total-market finding isn’t affected (it’s already insignificant), but the significant share-shift result (p=0.009) should be interpreted with this in mind.
Noisy placebos. Two of four placebo cutoffs are marginally significant, likely reflecting seasonal patterns or other award-season events. The true cutoff is the only one with a positive estimate, and the share-shift result provides independent confirmation, but the placebos aren’t as clean as I’d like.
Award-season contamination. Golden Globes, SAG, Critics’ Choice, and BAFTA cluster in the same January-February window. The RD isolates the jump at the exact Oscar date, but pre-announcement trends already reflect earlier awards momentum.
Seasonal decline. January and February are structurally declining for box office. The day-of-week and ceremony-year fixed effects help, but the strong secular downtrend makes it harder to detect a modest upward shift.
COVID years. The 2020 and 2021 ceremonies had depressed baselines and delayed announcements. Results are robust to excluding them.