Behavioral Economics GuideSAFE

🔬 A curated collection of 23,000+ agent skills for empirical research across 8 social science disciplines. | 精选 23,000+ AI Agent 技能库,覆盖8大社会科学学科的实证研究。CoPaper.AI 20分钟完成一篇可复现的规范实证论文,并支持用户上传 Skills。-- Maintained by CoPaper.AI from Stanford REAP.

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: behavioral-economics-guide
description: "Behavioral economics research methods and key frameworks"
metadata:
  openclaw:
    emoji: "🧠"
    category: "domains"
    subcategory: "economics"
    keywords: ["behavioral economics", "microeconomics", "development economics"]
    source: "wentor-research-plugins"
---

# Behavioral Economics Guide

Conduct behavioral economics research using experimental methods, prospect theory, nudge frameworks, and key empirical tools for studying decision-making under bounded rationality.

## Core Theoretical Frameworks

### Prospect Theory (Kahneman & Tversky, 1979)

People evaluate outcomes relative to a reference point, with losses looming larger than equivalent gains:

```
Key features:
1. Reference dependence: Utility is defined over gains and losses, not absolute wealth
2. Loss aversion: lambda ≈ 2.25 (losses hurt ~2.25x more than equivalent gains)
3. Diminishing sensitivity: Marginal impact decreases as you move away from reference
4. Probability weighting: Overweight small probabilities, underweight large ones

Value function:
v(x) = x^alpha            if x >= 0  (alpha ≈ 0.88)
v(x) = -lambda * (-x)^beta if x < 0  (beta ≈ 0.88, lambda ≈ 2.25)

Probability weighting function (Prelec, 1998):
w(p) = exp(-(-ln(p))^alpha)   (alpha ≈ 0.65 for gains, 0.69 for losses)
```

### Dual Process Theory (Kahneman, 2011)

| System 1 (Fast) | System 2 (Slow) |
|-----------------|-----------------|
| Automatic, effortless | Deliberate, effortful |
| Intuitive, heuristic-based | Analytical, rule-based |
| Parallel processing | Serial processing |
| Emotional | Logical |
| Prone to biases | Can override biases |
| Default mode | Activated when needed |

### Nudge Theory (Thaler & Sunstein, 2008)

Nudges alter choice architecture to influence decisions without restricting options:

| Nudge Type | Example | Mechanism |
|-----------|---------|-----------|
| Default setting | Opt-out organ donation | Status quo bias |
| Salience | Calorie labels at point of sale | Attention focus |
| Social norms | "9 out of 10 neighbors recycle" | Conformity |
| Commitment device | Pre-commitment to savings plans | Present bias correction |
| Simplification | Pre-filled tax forms | Reduce cognitive load |
| Feedback | Real-time energy usage display | Information salience |
| Framing | "90% survival" vs "10% mortality" | Reference frame |

## Key Behavioral Biases and Experimental Tests

| Bias | Definition | Classic Experiment |
|------|-----------|-------------------|
| **Anchoring** | Over-reliance on first piece of information | Wheel of fortune + estimation task |
| **Endowment effect** | Overvaluing what you own | Mug trading experiment (Kahneman et al., 1990) |
| **Status quo bias** | Preference for current state | Default choice experiments |
| **Present bias** | Overweighting immediate outcomes | Discount rate elicitation |
| **Sunk cost fallacy** | Continuing due to past investment | Theater ticket scenario |
| **Overconfidence** | Overestimating own knowledge/ability | Calibration tasks |
| **Availability heuristic** | Judging probability by ease of recall | Frequency estimation tasks |
| **Representativeness** | Judging probability by similarity | Linda problem |
| **Framing effect** | Choices depend on how options are presented | Asian disease problem |

## Experimental Methods

### Lab Experiments

```python
# Example: Dictator Game implementation with oTree
# oTree is the standard platform for behavioral economics experiments

# models.py
class Player(BasePlayer):
    dictator_give = models.CurrencyField(
        min=0, max=100,
        label="How much do you want to give to the other participant?"
    )

# pages.py
class Decision(Page):
    form_model = 'player'
    form_fields = ['dictator_give']

    def vars_for_template(self):
        return {'endowment': 100}

class Results(Page):
    def vars_for_template(self):
        return {
            'kept': 100 - self.player.dictator_give,
            'given': self.player.dictator_give
        }
```

### Field Experiments and RCTs

```
Design checklist for a behavioral field experiment:

1. RESEARCH QUESTION
   "Does changing the default retirement contribution rate from 3% to 6%
   increase average savings?"

2. TREATMENT ARMS
   - Control: Default contribution = 3% (status quo)
   - Treatment 1: Default contribution = 6% (higher default)
   - Treatment 2: Default contribution = 6% + active choice prompt

3. RANDOMIZATION
   - Unit: Individual employees
   - Method: Stratified randomization by age, salary, tenure
   - Balance checks: t-tests on observables across treatment arms

4. SAMPLE SIZE
   - Power calculation: N = 1,200 per arm (power=0.80, MDE=2pp,
     alpha=0.05, ICC adjusted for clustering by department)

5. OUTCOME MEASURES
   - Primary: Contribution rate at 6 months
   - Secondary: Total savings at 12 months, opt-out rate
   - Administrative data (no survey needed)

6. PRE-REGISTRATION
   - Register on AEA RCT Registry before treatment assignment
```

### Survey Experiments

```python
# Example: Willingness-to-Pay (WTP) elicitation using BDM mechanism
# Becker-DeGroot-Marschak procedure

import numpy as np

def bdm_auction(stated_wtp, item_cost_range=(0, 20)):
    """
    Becker-DeGroot-Marschak incentive-compatible mechanism.
    Random price drawn; participant buys if WTP >= price.
    """
    random_price = np.random.uniform(*item_cost_range)
    buys = stated_wtp >= random_price
    payment = random_price if buys else 0
    return {
        "stated_wtp": stated_wtp,
        "random_price": round(random_price, 2),
        "purchased": buys,
        "payment": round(payment, 2)
    }

# This is incentive-compatible: truthfully reporting WTP is optimal
# because the price is determined independently of the stated WTP
```

## Time Preferences and Discounting

```python
# Estimating discount factors from multiple price list (MPL) choices

def estimate_discount_factor(choices, amounts, delays):
    """
  

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

What the source does

Filesystem

none-observed

Network

none-observed

Shell

none-observed

Dependencies

pinned

Secrets in source

none-found

Findings (0)

No findings outside the package's declared scope.

Gates applied: no_behavioural_pass.