Attention Deficits and Asset Prices: A Theory of Information Backlogs
A study uses 500,000 YouTube finance videos to show news can outpace investor attention, creating backlog. Authors define NPAG and a backlog state persisting 58 to 97 per day. When NPAG is positive, index returns jump 24 to 29 bps, then fall 6.3 bps per day with a one to three day half life. VIX futures drop 66 bps with ample attention and 38 bps when constrained. Effects appear in indices, not Fama French factors.
What it examines
The paper develops a theory of information backlogs that form when news flow exceeds investor attention. Using ~500,000 YouTube news videos, it measures supply (video length) and demand (views) from the same source and estimates a dynamic backlog model to test predictions for market returns, volatility, and state dependence.
What it concludes
Attention deficits lift market returns immediately, but accumulated backlogs predict lower returns for 1--3 days and weaken news’ volatility reduction, proving processing capacity—not availability—limits efficiency. Applications include trading, risk management, and disclosure timing. Limits: YouTube view timing, partial equilibrium, homogenous agents. Future work: other platforms, heterogeneity, algorithmic interactions.
Evidence objects
Analyzing 500,000 YouTube finance videos, the study jointly measures news supply (duration) and attention demand (views) to define NPAG and a backlog state $B_t$ persisting $58%$--$97%$ per day after shocks.
key_findings bullet 1 · key_findings · validation V0
When $NPAG>0$, aggregate index returns jump immediately by 24--29 bps; the ensuing backlog depresses returns about 6.3 bps per day, with a 1--3 day half-life. Punchy, reversible pricing pressure emerges.
key_findings bullet 2 · key_findings · validation V0
Volatility state-dependent: news lowers VIX futures 66 bps when attention ample, 38 bps when constrained; indices-only, not Fama--Frenchsignaling cognitive bottleneck. Methods: fixed-effects, winsorization, Newey--West, profile-likelihood; limits: timestamps, scope, partial-equilibrium, proxies.
key_findings bullet 3 · key_findings · validation V0
Offers a testable, state-dependent mechanism: attention-driven information backlogs link YouTube news pressure to subsequent returns and volatility. Innovatively measures supply (video duration) and demand (views) from identical channels, resolving identification. Introduces geometric-decay backlogs with a 1--23-day predictability window, yielding aggregate-only effects, informing asset pricing and mid- to low-frequency trading practice.
key_findings bullet 4 · key_findings · validation V0
Raw abstract and provenance
- … We document a new source of return predictability: information backlogs that form … persists each day, creating return predictability at a previously undocumented frequency …
Source row: 260 · abstract type: snippet