Learning from Experience Flashcards
(14 cards)
Standard Assumptions about belief formation in Economics and Finance
- Preferences and beliefs are exogenously given
- Balis are formed rationally using all publicly available information
Learning from experience hypothesis
- Preferences are endogenous what we experience over our lives
- Beliefs overweight recent information and things that happened during our own lifetime
“Depression Babies” Study
- Studies how people experience average stock return depending on their previous experience
- They calculate weights using time lags since return realization
- The shape of the weighting function is set by lambda
The form of the weighting function dependent on lambda
- lambda < 0: increasing and convex as k -> age
- lambda = 0: equal weights
- lambda > 0: weights decreasing in k
- 0 < lambda < 1: concave
- lambda = 1: linear
- lambda > 1: convex
How is the study “Depression Babies” set up?
- The authors set up a formula to calculate average experienced stock market returns, with individual weights depending on time since realization and the age of the individual.
- In the first step, both the shape of the weighting function (lambda) and the effect of experiences on participation are estimated a larger sample. In the second step stock market expectations are estimated from individual experiences using fixed values for lambda from the first step.
What are the findings in “Depression Babies”?
- Negative experiences predict non-participation
- Mostly based on beliefs
- Estimated recency bias, lambda > 1
- Experience + fixed costs yield some explanation why wealthy individuals do not participate
Setup: Effect of Inflation experience on inflation expectations
- Respondents are split into age cohorts
- Agents are assumed to estimate an AR(1) process of inflation only using realizations during own lifetime.
- Allows for different weights
- Expected inflation significantly decreases with age of the cohort.
- Recent observations matter more for younger people with less experience stock
Local Experiences might matter because
- More available to agents
- Explains cross sectional heterogeneity in beliefs and decisions
- At odds with “full-information rational expectations” if they have predictive power
Empirical evidence of local experiences
- Local house price growth predicts nationwide house price growth
- Job loss makes households pessimistic about aggregate unemployment
More empirical findings of local effects
-Stock investment behavior and local firm bankruptcies
- Household inflation expectations and local grocery prices
- Firm managers inflation expectations and industry level shocks with no aggregate effect on economy
Channels for transmission of experiences:
- Vertical transmission from parents to children
- Horizontal transmission through collective memory
Market level implications of learning from experience when incorporating into an asset pricing model:
- Cross cohort variation in experiences generates heterogeneity in beliefs, portfolio choice and trade
- Replicates several known pattern in asset prices, excess volatility and return predictability
Learning through experience cannot explain:
Why agents make different decisions in different situations
Can Learning through experience explain excessive pessimism?
YES, as beliefs and preferences are endogenous in this model
