2/4/26: Lecture 2

How do economists think about human behavior?

Goal-driven individual behavior: constrained optimization

1. Utility function/preferences: what makes people happy?
   • at a moment in time (instantaneous utility function)?
   • when time, risk, or others are involved (time, risk, and social preferences)?
2. Beliefs: what do people believe about their environment?
   • Physical environment and others' behavior
   • Use of information to update their beliefs
3. Choice/Decision-making: How do people use the above to make decisions?
   • Some influences on behavior aren't about utility or beliefs
   • Frames, defaults, and nudges; heuristics

Utility and Social Preferences

Much of classical economics assume people are selfish (most true).

People care about others in other ways than pure altruism, like: warm glow, inequity aversion, reciprocity, social/self image.

Dictator game: one player (the "dictator") is given a sum of money and decides how much, if any, to share with a second, passive recipient. We analyze behavior between your decision being made public vs not.

[TODO]: Does behavior change based on how much money?

Betting on losing weight: This is an odd bet because this is an outcome over which one has complete control. However, bettors often lose! People do this for self motivation to make it higher stakes.

Often a conflict between short-term desires and long-term pleasures.

present bias: tendency to overvalue immediate rewards and instant gratification over larger, delayed, or long-term benefits. People tend to be relatively patient for far-off decisions, but less patient for immediate relevant decisions.

Beliefs

heuristic and biases: mental shortcuts our brains use for quick judgment, but also leads to systematic biases in judgement. This is what behavior economists study.

Biases in Subjective Probability Judgement: Survey on probability of yourself landing a high paying job vs percentage of class. Men tend to overoptimistic about their prospects abilities relative to others.

Choice

Reference dependence: value is computed relative to a salient baseline

Anchoring: bias where irrelevant numbers pull your judgment toward them, even when you know they shouldn’t matter.

Could the survey above just be result of anchoring?

Loss Aversion: same situation is perceived as worse if it is framed as a loss, rather than a gain

Applications

• There is a much less willingness of quitting social media alone vs advocating everyone quitting social media
• Adequate sleep significantly enhances physical and mental performance despite less time during the day.
• Despite so many people feeling lonely, why don't people interact with one another?

2/6/26: Recitation 1

Utility is the satisfaction from consuming a bundle of goods or services.$u: X \rightarrow \mathbb{R}$.

Marginal utility $MU_x$ is the additional satisfaction from consuming one more unit of a good or service. Formally, the partial derivative with respect to $x$.

Increasing in its arguments (more is always better) is called non-satiation.

Diminishing marginal utility states that the more you consume, the less utility you get from an additional unit.

With uncertainty, we aim to maximize expected utility.

Over multiple time periods, exponential discounting of future utility relative to current utility.

Note that agents are self-interested and utility is over final outcomes.

Utility Maximization over Two Goods

Problem usually becomes

If $(x^*,y^*)$ is an interior solution, then

Lagrangian Approach

Getting first order conditions and setting them equal give the same result.

Utility Maximization over Two Time Periods

Utility becomes

2/9/26 - 2/11/26: Time Preferences (Theory)

Setup: utility over time

Many decisions trade off costs/benefits across dates.

Model with:

• per-period utility $u_t$ (often from consumption, effort, etc.)
• a discount function that weights future utility

Exponential discounting (Samuelson)

At time $t$, preferences are:

$\delta$ is the discount factor. Higher $\delta$ $\Rightarrow$ more patient.

• Discount function $D(\tau)$ is the weight given to utility at time $\tau$.
• Discount rate $\rho$ is the rate at which the discount function decreases, or $\rho = -\frac{dD(\tau)/d\tau}{D(\tau)} = -\frac{d(\delta^{\tau}/d\tau)}{\delta^{\tau}}=-\frac{\delta^\tau\log(\delta)}{\delta^{\tau}}=-\log(\delta) \approx 1-\delta$.

Inferring $\delta$ from simple choices

If an option has an immediate cost $1$ and future benefit $B$ arriving $k$ periods later, then choose it iff:

Observed accept/reject decisions give inequalities that bound $\delta$.

Money choices: implied annual discounting

Indifference between $\$Y$ now and $\$X$ in $t$ years (assume linear utility in money) implies:

Annualized (continuous) discount rate:

Empirical pattern from such questions: implied impatience is much larger for short delays than long delays.

Key implication: exponential discounting is time-consistent

If at time 0 you prefer $A$ to $B$ based on future streams, then at time 1 you will still prefer $A$ to $B$ (no preference reversal) because exponential weights preserve rankings when you move forward one period.

Evidence: present bias, reversals, commitment demand

Behavior often shows:

• present bias: disproportionately strong preference for immediate gratification
• preference reversals: plans for the future differ from choices when the moment arrives
• commitment demand: people restrict their own future choices

Quasi-hyperbolic ($\beta$--$\delta$) discounting

A simple way to model present bias:

Interpretation:

• $\beta<1$ creates a discrete drop between “now” and “later” (present bias)
• $\delta$ controls standard long-run discounting across future periods

Two canonical implications (with $\delta=1$ for intuition)

Leisure good (benefit now $B$, cost later $C$):

If $B-\beta C>0$ but $B-C<0$, you eat today but plan not to later $\Rightarrow$ over-consumption vs. plans.

Investment good (cost now $C$, benefit later $B$):

If $-C+\beta B<0$ but $-C+B>0$, you procrastinate $\Rightarrow$ under-investment vs. plans.