Modeling The Economy

NERD ALERT:

WHY MODEL THE ECONOMY???

  1. In this case it is not to figure out what equity to short…it is so that we, as a people, as a scoiety, don’t have to live through objectively governmental policy choices.
  2. We can peer into the looking glass and say - yeah that does not play out well.
  3. Of course if we model the economy “pretty good” - aka create a decent crystal ball - then of course industrius individuals easily profit with sufficiently good crystal balls.

Take the Example of:

  • Argentina
    • People
    • Resources
    • Rules
  • Australia
    • People
    • Resources
    • Rules

But with very different macro economic outcomes…

Economic Primatives:

  • People
    • Purpose
    • Belief
    • Constraints
  • Governments
    • (Collection of People)
    • with Power to:
      • tax
      • spend
      • borrow
      • redistribute
  • Technologies
    • produce goods
    • produce services
    • produce physical capital
    • produce human capital
  • Processes
    • Describes Information Flow
    • Describes Economic Shocks
  • Equilbs
    • Describe Markets, Rules, Regs, reconcile people diverse purpose and possibilities

Principal Agent Problems - but the economy is a multi-agent problem

The goverment has strategies and the people have counter-strategies - Chinese Proverb 1000 BC

decision(t, forecast(t, beliefs))

To model the economiy
You are modeling things - that themselves have models.

What is a model?
A PDF over a seqence - A Stochastic Process that is indexed by some params

Communism of Models - everyone has the same model - conditioned with different information

Can’t fool all the people all the time. - Abe Lincoln

Thoughts about the future influence the actions today

Given past, the present and future are simultaneous determined

Two Main Polar-Opposed Models:

  • model of “bank runs”
  • model of “moral hazard” (aka bailing out banks)

Joint Density

  • aka stochastic process
  • observe data - infer params

Models beat model

Our Goal Should be: No Camps (Chicago vs Cambridge vs etc)

  • Direct Problem: theta ==> Joint Density
  • Inverse Problem: One slice of joint densities ==> theta (Max Likeli or Bayesian)

Markov Process - finding the state is an art
Sequential Inference:
- Abrahams Wald’s sequential probabilty ratio
- Dynamic Programming - Richard Bellman
Filtering and Smoothing:
- Kalman fitlering
- Nonlinear Filtering

Bellman Equation

  • Maximizer Eq
  • let v(x)

The Economy is a bellman equation with a government (with policies) + a vector of agents. (aka a multi agent problem) + tech + process.

Each agent is a bellman equation where they are influenced by the government.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Simulation
	Timeline
	Technologies,
	Process,
	Government(
		[Policies
			Tax,
			Spend,
			Borrow,
			Redistribute
		]
	[Agents
		Constraints,
		Purpose,
		[Beliefs
			Past Experience
			Values
				Risk
				Greed
				
		], N=lots
	]
MonteSim
	[Simulation(IC)]
	Halt Conditions: iterations, fulfilled condition, 
GenSim
	Simulation t+n  = Simulation((t,n), IC)
MonteGenSim
	[Simulation t+n  = Simulation((t,n), IC)]

What is the processing model for the sim?

How to scale horizontal?
Process per simulation?
Thread per simulation?

What if our simuluations moved towards:

  • 1000 sims

  • 5 yr long

  • popN = 330Mppl

  • Can we fit it into am AWS Lambda func execution model?

    • λ MetaSimulate = ƒ(s3://config.json)
    • λ MetaSimulate = ƒ(vecSims)
    • λ Simulate = ƒ( right-sized coro on agent size)
    • λ Simulate = ƒ(s3://config.json)
    • λ SizedSimulate = ƒ(s3://config.json)
    • λ SizedSimulate = ƒ(runfor, environmentIC₀, vecAgents₀, govt₀, tech₀, process₀)

How Will I interogate the results?

  • Time series:
    • events happening within a sim
    • fs:// //.jsonl
    • s3:// //.jsonl
    • animated gif from plots of each metrics as time cycles through

Need I simulate:

  • Timeline Resolution in:
    • days
    • Suppose it depeneds on resolution of predictions.
    • 2, 4, 8, 10, 20, 50, 100 yr out predictions. (max 3650 step run)

Various Distributions:

  • Democrats vs Republicans
  • Types of Work (small cal, mid cap, large cap)

For People:

  • Purpose
    • is this the maximizer function?
      • max dollars?
      • max happiness?
  • Age Spread as of 2017
  • Family and Reproduction
    • aka: Population growth over time
  • Communal Cohesion Coeffcient
    • Spread of Ideas
  • Acts of Crime and Justice (population inactive and cost)
    • or call it ethics - the continuous vs the discrete
  • Personal Traits
    • Goal Orientation
    • Listening
    • Self-Aware
    • Confidence
    • Discipline
    • Adaptive
    • Integrity
    • Communication
      • Articulate (Convicingness)
    • Integrity
      • Trustworthy
      • Trusting
    • Productive
    • Courage
    • Articulate
    • Balanced
    • Collaborative
    • Committed
    • Courageous
    • Decisive
    • Detail-Oriented
    • Emotionally Competent
    • Friendly
    • Hard Working
    • Humorous
    • Intelligent
    • Organized
    • Prepared
    • Productive
    • Relationship-Oriented
    • Responsible
    • Sincere
    • Self-Sacrificing
    • Trustworthy
    • =================
    • Convincingness
    • Openness
    • Conscientiousness
    • Agreeableness
    • Neuroticism
    • MBTI x16 combinations
      • Extraversion / Introversion
      • Judging / Perceiving
      • Sensing / Intuition
      • Thinking / Feeling
  • Cultural Beliefs
    • would just be the aggregate feelings of the population
business
analytics economics models