Conservatism bias in investing: under-reacting to new information that contradicts existing beliefs
Conservatism bias is the tendency to update beliefs too slowly when new information arrives. Investors who hold a prior view of a company, an asset class, or the broader economy tend to anchor on that view and revise it less than the new information warrants. The bias is the structural opposite of recency bias (which over-weights recent information) and is documented across both retail and professional investor populations.
What conservatism bias is
The bias was identified by Edwards (1968) in early behavioural-economics experiments. Subjects updating beliefs in response to evidence updated less than Bayes' rule prescribed; the under-updating was systematic and persistent across multiple experimental designs. The mechanism appears to be a combination of effort cost (updating is cognitively demanding) and identity cost (changing one's mind feels like admitting prior error).
In financial markets, conservatism is one of two ingredients in the Barberis, Shleifer & Vishny (1998) model of investor sentiment. The model proposes that investors alternate between conservative regimes (under-reacting to news, producing momentum-like under-reaction) and representativeness-driven regimes (over-reacting to patterns, producing reversal-like over-reaction). The combination explains both the empirical persistence of momentum and the eventual mean-reversion that follows.
How it manifests in investing
The clearest manifestation is the slow incorporation of information after corporate news events. Earnings surprises produce price movements that continue in the direction of the surprise for several weeks (post-earnings announcement drift), consistent with conservatism—investors update their views slowly even though the news is publicly disclosed and unambiguous. Analyst forecast revisions produce a similar pattern: a single revision is incorporated quickly into prices, but the residual information leaks into prices over subsequent weeks.
At the broader portfolio level, conservatism shows up as anchoring on a prior allocation that no longer matches the investor's circumstances or the market environment. An investor who built a portfolio for one set of objectives often retains that allocation past the point at which the objectives have changed, because the cognitive effort of revising the allocation is high and the dissonance from acknowledging the old allocation as outdated is uncomfortable.
The cost
The financial cost of conservatism is the slow recognition of broken theses. A company whose competitive position has materially deteriorated continues to be held by investors who anchor on the position's original case. A market regime shift—from low-rate to high-rate, from disinflation to inflation—is acknowledged only with a lag, and portfolios optimised for the old regime are slow to be repositioned.
The aggregate cost is the absence of timely repositioning. Empirical studies of professional investor portfolios (Coval & Moskowitz, 1999; Wermers, 2000) document that even institutional investors who in principle should update faster than retail are still slower to incorporate new information than the strict Bayesian benchmark. The bias is universal even when the cognitive cost should be minimised by professional infrastructure.
What helps
The structural defence is to externalise the updating process. A rules-based system that incorporates new information on a defined schedule—quarterly rebalancing based on recent factor exposures, for instance—performs the updating that conservative human discretion would otherwise delay. The rule does not feel any emotional discomfort about revising prior conclusions.
Periodic, deliberate review at the portfolio level achieves part of the same effect. An investor who schedules an explicit annual review of the portfolio's strategic assumptions creates a forced opportunity to acknowledge information that has accumulated during the year. The review does not eliminate conservatism within the year, but it bounds the cumulative cost.
Related articles
Disclaimer
Get started now
