Research Showcase: Memory and Financial Decisions

Psychologists have long shown that people imagine the future by drawing on the past. In joint research with Zhengyang Jiang, Hongqi Liu and Hongjun Yan, Dr Cameron Peng of the LSE Department of Finance brings this idea to financial markets. Surveying more than 17,000 Chinese retail investors and linking their answers to their trading records, the team shows that what investors remember depends on what the market is doing, and that these memories shape both their beliefs and their trades.

People imagine the future by drawing on the past

The broad question I study is how people form beliefs about the future. Psychology offers a clear answer: we imagine the future by drawing on the past. When we try to picture what will happen next, we retrieve relevant experiences from memory and use them to simulate what might happen.

Memory is not a fixed archive that we consult in full. It is associative: what comes to mind depends on the cues around us. A particular room, a sound or a smell can bring back an episode we had not thought about in years. The French call this a madeleine de Proust, after the cake that brings back the narrator's childhood in Proust's novel. In this project we asked whether financial markets have their own madeleines, and whether they shape how investors see the future.

Recalled experiences explain investors' beliefs better than their actual experiences

Working with a large financial institution in China, we surveyed a nationally representative sample of retail investors across the branch offices of the country's 60 largest brokers. We asked investors to recall past market episodes and their own past investment returns, relying only on memory. We then asked them to forecast the market and their own portfolios. For more than 5,000 of them, we could link their answers to their actual trading records. This let us compare what investors remember with what actually happened.

We rolled the survey out over six weeks. Because the market moves every day, different investors took the survey under different market conditions. This let us test whether the market’s movement at that moment acts as a memory cue.

It does. When the market was up on the day an investor took the survey, they recalled their own past returns more positively. This held even after we controlled for what their returns had actually been. Memory, in other words, is not a static record of the past. It shifts with the current environment.

These recalled experiences matter. Investors who remembered better past returns expected higher returns in future and saw a market crash as less likely. When we compared recalled experiences with actual experiences, recalled experiences did a better job of explaining investors' beliefs. It is how investors remember their past, not the past itself, that drives what they expect.

Memory-driven expectations show up in what investors actually buy

Those beliefs also show up in behaviour. Investors who expected higher returns on their own portfolios bought more stock, both on the day of the survey and over the following month. The part of their optimism driven by memory had an especially strong link to their trading.

For individual investors and households, the lesson is to be careful about trusting your memory. If your investing track record feels good, that feeling may reflect today's market as much as your actual results. Before an important financial decision, check the numbers.

Cued recall offers a memory-based explanation for extrapolation

For markets as a whole, the findings offer a memory-based account of why investors extrapolate past returns. That tendency is widely thought to contribute to booms and busts. Rising prices bring good memories to mind, which makes investors more optimistic and more willing to buy. Falling prices work the other way. For policymakers, understanding what investors remember, and what sticks in their minds, is part of understanding market sentiment and how financial risks are perceived.

Cameron Peng is Assistant Professor of Finance at the London School of Economics and Political Science.

Read more about his work at https://sites.google.com/site/cameronpengresearch/

Read the Research: Investor Memory and Biased Beliefs: Evidence from the Field, Quarterly Journal of Economics 140(4), 2749–2804 (2025)

Open-access version: NBER Working Paper No. 33226