Anchoring

Anchoring operates through a two-stage mental shortcut: the mind seizes on a salient starting value, then adjusts outward from it toward what the evidence implies. The failure lies in the second stage. Adjustment terminates as soon as the revised figure feels plausible rather than when the evidence is fully incorporated, so the final estimate remains tethered to its origin.

What makes the bias notable is that the anchor need carry no informational content. Experimental work shows that arbitrary numbers presented immediately before an unrelated numerical judgment still shift responses in their direction. In markets, anchors are rarely arbitrary but are frequently stale: a cost basis reflects conditions at purchase, a 52-week high reflects a sentiment regime that may have ended, and a prior price target reflects assumptions since invalidated.

Anchoring surfaces wherever a prominent number precedes a judgment. Equity analysts revise price targets in small increments from their own prior published figures rather than rebuilding a valuation from current fundamentals, producing target-price series that trail the market. Forecast revisions cluster near consensus, since consensus itself becomes the anchor and departing from it carries career risk.

The purchase price is the most damaging personal anchor. It governs the disposition effect, where investors refuse to realise losses until a position returns to cost, and it distorts position sizing when average-cost thinking substitutes for a forward-looking assessment. Round numbers on indices, IPO indicative price ranges, prior takeover bids in M&A negotiations, and listed asking prices in real estate all serve the same function: they frame the range within which subsequent judgment moves.

Because anchoring is a cognitive error rather than an emotional one, it responds to process discipline, information and structured reasoning. The standard corrective is to force the estimate to be constructed rather than adjusted: build a valuation from cash flows, growth and discount rates without consulting the previous target, then compare. Where an anchor is unavoidable, generating independent estimates from separate starting points and reconciling the difference limits the pull of any single reference value.

Two practical tests are widely used. The first asks whether an investor would establish the position today at the prevailing market price with no legacy holding; if not, the cost basis is doing the work. The second requires that valuation ranges and rebalancing triggers be documented before new information arrives, so subsequent revisions must be justified against a pre-committed framework rather than nudged from a comfortable number.

Level I questions typically present a short vignette and ask for correct identification. The discriminating feature of anchoring is the presence of an explicit reference number that the subject will not move far from. If the vignette instead emphasises reluctance to revise beliefs after new evidence, conservatism is the better fit; if it emphasises judgment by superficial resemblance to a category or a stereotype, representativeness applies; if it emphasises weighting recent or vivid information, availability applies.

Candidates should also be able to state the consequences and the corresponding remedies. Anchoring produces underreaction to news, sluggish forecast revisions, holding periods driven by cost basis rather than expected return, and portfolios that drift from target weights because the last known allocation feels correct. Each is addressed through documented process, objective valuation inputs and periodic review against the investment policy statement.