Vladimír Gazda

• prof since 2013 in Finance

• currently working in:

  1. Internet of Things (5G Telecom. Networks)
  2. Complexity of Economics - applications in various fields
  3. Modern Teaching Methods

• teaching

  1. Microeconomics, Econometrics, Game Theory, …

• programming in

  1. R – moderate level
  2. NetLogo – pre-intermediate level

Mainstream Economics (Shortages and Limitations)

• static equilibrium analysis
• the existence of the representative agent
• agents are rational
• assumption of the perfect competition
• models based on a limited number of variables (max. 5-6)

Is this equilibrium????

Black Friday 1929

Is the evolution of DJI in equilibrium???

## [1] "DJI"

Law of Diminishing- vs. Increasing Returns

The neoclassical approach assumes:

1. scarcity of the resources implies the strictly concave increasing shape of the production function.

2. (1.) => marginal products/returns (in other words, average costs are increasing).

Current economic experience demonstrates:

1. production factors are of high disposal and are often implemented in an optimal amount.
2. (1.) => the deficit production factor demonstrates increasing returns (i.e., average costs are decreasing (see 1st stage of the production function). Therefore, it implies the explosive character of economic growth.

Negative- vs. positive feedback loop.

Economic dynamics in mainstream economics is based on the negative feedback loop. From the mathematical point of view, it corresponds to fixed points in the role of attractors. The best example is a Keynesian multiplicator of economic growth.

Positive feedback loop causes explosive economic growth and destroys stability. The best example is the investment in mobile phones marketing support. If one buys a phone, all of his/her friends are (probably) buying it. If marketing reaches a (wo)man in the central position of the social network of friends, it triggers a massive increase in mobile sales. (See similar situation with the graph Giant component provided later)

Representative agent???

Representative agent !!!

🙂 🙂 🙂 🙂 🙂

🙂 🙂 🙂 🙂 🙂

🙂 🙂 🙂 🙂 🙂

🙂 🙂 🙂 🙂 🙂

🙂 🙂 🙂 🙂 🙂

🙂 🙂 🙂 🙂 🙂

Heterogeneous Agents and Statistics

Linear Regression Results

Mean of the Sushi restaurant visits is 26, and mean Income is 3611.

Linear regression result: \[SushiRestVisits_i = 5.843+0.0055 Income_i\]

## integer(0)

Regression Interpretations

The regression results mean that Income growth by 1000 USD raises visits to the Sushi restaurant by 5.5; But is it definitely TRUE ??? Did we take into consideration heterogeneity of the population?

Answer: Proper application of the OLS regression needs applying the so-called Mixed linear models (running scarcelly by the economists).

Perfect Competition assumptions and consequence

• Infinite number of producers and consumers
• No transport and transaction costs
• Perfect information
• Free entry to any business sector
• Homogeneous goods

Under the above-given assumptions, the competitiveness prevails, and the prices that maximize producers’ profits and consumers’ utilities lead to supply-demand balance.

Is Perfect Competition still working?

• BlackRock and Vanguard control the overwhelming majority of the EU and US stock exchange companies. As a result, companies have small incentives to compete with each other.
• information is quite flexible but is provided by a few information monopolies (see bad examples of Amazon, Google, …) misusing the information asymmetry.
• gasoline prices in Komráno and Komárom are far away from being equal

Perfect Rationality vs. Bounded Rationality

Decisions of the agents are assumed to be optimal in the traditional approach. On the other side, there is an experience that

• Cognitive Limitation
• Information Imperfection
• Time Scarcity

Prevent optimal decisions from the Math point of view.

Herbert Simon, later also Vernon Smith, Daniel Kahneman, Richard Thaler show that the decisions are sub-optimal.

Complexity economics contribution:

• Frequent applications of the bio-inspired models of competition and cooperation.

• Stressing the evolutionary aspects of the agents’ behavior (Darwinian / Mendelian principles).

Agent-Based Modelling (ABM) I.

The general methodology of ABM consists of:

1. Agents Definition:

• Properties of the agents (income, propensity to consume, etc.)
• Objective function definition (utility function, profit function) that each agent is to optimize (max/min).

2. Definition of the binary agents’ interaction - for example buy/sell, cooperate/non-cooperate,…
3. Macro-parameters definition: Interest rates, tax rate, legal rules, etc…
4. Definition of the agents’ environment: 2D space, social network, cellular automata, All-in-One Club, etc…

Agent-Based Modelling (ABM) II.

5. Programme coding and debugging. Most popular developing tool becomes NetLogo.

6. Calibration of the model (calibration consists of the parameters’ optimal setting). Evaluation of the pivot simulations.

7. Extensive computer simulations run by specialized computer servers.

8. Analysis of the parameters’ sensitivity - i.e., research of the impacts of various parameters settings (including initial conditions).Simulations run hand in hand with R software to enable deep statistical analysis.

9. Model interpretations and publication of the results.

Conclusion

Thanks for attention,

Vladimír Gazda

gazdav@ujs.sk