model thinking

1.4 analysing data

1. see patterns (horizontal, rising, falling, cyclical...)

2. predict points (for ex., price of the house basing on metrage)

3. predict boundaries ( infl 0..3%)

4. retrodict - test models by running from past

5. predict sth else (planet with big mass)

6. inform data collection - which data do we need?

7. estimate hidden parameters (number of sick people and sickness progress)

8. calibrate - make the model closer to real world - forest fires.

1.5 using models to analyse

1. rral time desicion aid (monty hall paradox)

2. comparative statics (as-ad curves)

3. conterfactuals (with or without recovery plan - what would happen?)

4. identifying leverages (if greece fails, all europe fails)

5. experimental design (models instead of real experiments)

6. institutional design

7. help choose among institutions (democratics or communism?)

1. sorting and peer effects

sorting effect: people tend to communicate with similars (detroit - black and white or smokers)

peer effect: peers start acting like surrounding

2. schelling segregation

1. financial - by income (model of NY)

2. racial - "ghettos"

3. checkers model

how many people around you are rich

3/7 good; 2/7 bad

3. measuring segregation

1. rich/poor

24 blocks: 12 rich, 6 50-50, 6 poor

b=rich in block; B - rich total; y/Y - poor

| b/B - y/Y | - if equals 0, totally not segregated - index of similarity

for blue - 1/15; yellow - 1/9; green - 1/45

total index: 6*1/45 + 6*1/9 + 12* 1/15= 72/45 (total index)

for perfectly segregaged - 2; for perfectly mixed 0

4. peer effects

model

N individuals

each one has a threshold (how many people you need to join them)

avg value - how likely that everyone will do it

Subtopic