Modelling Wholesale Inequality In Addition To Video Games

Steve Roth (publisher of Evonomics) asked an interesting inquiry on Twitter: is inequality an inevitable final result of capitalism? I split the sources of inequality into ii factors, which I flippantly label retail inequality (inequality due to wage differentials), too wholesale inequality (inequality due to ownership of large, successful businesses). I am going to avoid the number of retail inequality, too just speak over how nosotros would model wholesale inequality. I retrieve a lesson tin strength out survive drawn from the decades of sense of designing strategy video games: the style towards inequality largely depends upon the rules of the game.

My comments hither could perchance survive viewed equally commentaries on agent-based modelling. I take keep only read a few papers inwards that area, too I am going to pretend that I know zip nearly the subject. I volition allow readers endeavour to meet whether my comments are useful for that surface area or not.

Wholesale versus Retail

Based on a cursory analysis of inequality data, the top 0.01% of the income distribution has been doing really well. Furthermore, the people alongside the large bucks tend to survive possessor of dominant corporations. If at that spot is a style for large corporations to acquire fifty-fifty to a greater extent than successful, too so nosotros should meet increasing inequality inwards that cohort versus everyone else.

For the ease of the distribution, the large dissever is betwixt C-suite salaries too the ease of the hierarchy. However, since a lot of the highest paid CEO's are also inwards accuse of the largest corporations, the inquiry of increasing corporate inequality is perhaps relevant. However, it is possible for wage scales to acquire to a greater extent than equal spell the owners acquire to a greater extent than unequal, too vice versa.

Therefore, I am looking at the inquiry of corporate dominance, too allow that stand upwardly inwards for inequality.

(Some readers mightiness wonder nearly passive investors -- the rentier degree that so animates European politics. I would debate that this is just next a item "corporate" strategy, beingness the passive ownership of bonds, land, too minority shareholdings. In gild to model this, nosotros would take keep to take keep a expert simulation of those property markets, which is harder than a existent economic scheme simulation.)

This is Frivolous!

The to a greater extent than serious readers mightiness survive shocked at my bringing upwardly video games. However, nosotros truly demand to practice so, for really expert theoretical reasons.

Firstly, an aggregated model is utterly useless for modelling the diverging fortunes of private components of the aggregate. At the minimum, nosotros demand to split out the large corporations equally private model entities -- agents. (We could acquire out other components of the model aggregated.)

We thence must take keep an agent-based model of some sort, alongside corporations equally the agents. These agents are competing to acquire bigger, next diverse algorithms describing their behaviour. This is a (serious) simulation.

What happens if a human takes over the decision-making of i of the corporations? You take keep a single-player strategy game, which fits into the genre of simulation-style games. (These games deport fiddling resemblance to abstract strategy games similar Go or Chess.) This genre came into its ain inwards the early on 1990s, next the success of Sid Meier's Civilization. (There were fore-runners to Civilization, but CPU limitations resulted inwards to a greater extent than primitive simulations.)

What is the root lesson nosotros larn when humans accept over the decision-making of i agent? We larn that human players rip the faces off AI (artificial intelligence) players. They volition break whatever weaknesses inwards the rules of the game, too exploit them to tear ahead of AI players. In a loose sense, they optimise. Also, i should regime annotation that when I write "rules of the game," it is non just the inquiry of what is legal. It also includes the "laws of nature" inside the simulation, such equally production functions.

This observation mightiness drive some distress to heterodox economists, equally it hints that DSGE modellers take keep a signal nearly optimising behaviour. However, they were only one-half right. We cannot plough the complex decision-making inwards these situations into an optimal command occupation (an surface area that I covered inwards my doctorate, so yeah, I know what I am writing about). Instead, humans break heuristics that human activity inwards "as optimal a fashion equally possible" (whatever that means).

In whatever event, this explains why I am interested inwards edifice a financial/economic simulator that could either piece of work equally a serious simulator -- or an economical engine for a video game. The gamer style to explore exploits would choke on the serious simulation types honest.

Do the Big Get Bigger?

What is the lesson to survive learned from strategy video games? Well, if nosotros aspect at successful commercial video games, nosotros meet that the terminate games are almost ever defined yesteryear a unmarried musician racing ahead too smashing the competition. The advantages of a dominant seat interpret into greater time to come advantages -- what people (other than pedantic command theorists) telephone telephone a positive feedback loop.

Is this an Fe police push clitoris of economics? No, it's the Fe police push clitoris of commercial success. Nobody wants to play a game where they know that they are inwards the winning seat after v hours, but it takes some other 20 to complete off the game. The designer embeds rules that reinforces a dominant position, allowing for a to a greater extent than rapid game termination.

However, the ease of the game mostly features the reverse mechanic: mechanisms for catching upwardly ("negative feedback" for dominance). People practice non desire to play a game where some AI musician gets a little early on lead, too just runs away alongside it inwards an unstoppable fashion. (Since the games I am discussing hither featured i human versus multiple AI players, just luck from the randomised starting positions mightiness survive plenty to lock inwards an early on Pb for i AI musician inwards most games.)

In other words, the returns to authority is only nether the command of the game designer, too represents the mass of pattern work. Even little tweaks to the rules tin strength out take keep a huge upshot inwards observed human strategies.

For example, the root unloose of Civilization allowed for aggressive starting strategies. This was addressed inwards after releases yesteryear a few changes to the simulation to block this. One of the changes was  parameter tweaks to i unit of measurement (the chariot). The modify removed the inwardness unit of measurement behind those aggressive starts. The equivalent inwards a line of piece of work concern simulation would survive a modify to a production function. If i wanted to audio mathematical, i would tell that the optimal strategy is non a continuous business office of simulation parameters. This regulation way that unless nosotros replicate all of the rules of a simulation exactly, nosotros should non survive surprised yesteryear wildly unlike optimal behaviour. This makes it extremely hard to generalise across models.

OK, What About Economics?

There were non a whole lot of successful strategy games based on corporate behaviour. The occupation is that realistic depictions are boring. There are "positive feedback" effects, but they are mostly fairly small. "Your corporation had an annual increase charge per unit of measurement of 8%, versus 6.5% of your primary competitor!" Yay. Generally speaking, inwards gild to brand the game interesting, designers needed to inject unrealistic components to acquire inwards piece of work (such equally extremely high rates of return, etc.).

There are a lot of reasons to await "positive feedback" inwards a realistic simulation.

• Economies of scale. In the existent world, at that spot are mostly positive economies to scale. This tin strength out survive viewed equally a "law of economical nature" (if my assertion is true; it's based on post-Keynesian empirical work).
• Research. Larger corporations tin strength out lock downward proprietary technologies, which tin strength out augment the returns to scale.
• First Mover Advantage. An obvious number for tech firms inwards the existent world, but in all probability hard to model.
• Balance canvass abuse. I am considering adding a futures marketplace to my simulation. The root matter I considered was: Is it possible to laid upwardly the rules of the telephone commutation so that trading does non plough into a continuous squeeze-the-shorts party? Corporations alongside large ease sheets tin strength out corner the market, too crunch weaker competitors. This is the departure betwixt AI algorithms too humans: AI's play fair; humans mostly play to win. 

Going the other way, at that spot are less factors that I tin strength out retrieve of.

• Specialisation. The final fourth dimension conglomerates were pop was inwards the 1960s (with a few notable exceptions, built around CEO personality cults). There are management costs associated alongside running equally good many line of piece of work concern lines. (I retrieve this is the occupation alongside the large banks, where I believe researchers take keep institute dis-economies to scale.) That is, at that spot mightiness survive an payoff to beingness the biggest line of piece of work solid inwards an industry, but costs to spreading across industries.
• New technologies. A little line of piece of work solid alongside a novel technology scientific discipline tin strength out wipe out the existing line of piece of work concern lines of a large firm. The difficulty alongside simulating this is that nosotros usually allow agents (and players) to know the parameters of industries. We nosotros would demand to insert some dubiety into the simulation to acquire inwards possible for firms to misjudge the effects of novel technologies.
• Incompetence (an thought that I took from the writings of the management guru Scott Adams). Very simply, given a large plenty provide of MBA's, i tin strength out run practically whatever scheme into the Earth inside a decade or two. Once the clear-minded greedy founders are out of the picture, large corporations autumn into a boiling mass of incompetence too business office politics. Unfortunately, it is really hard to model incompetence inwards a simulation.

As tin strength out survive seen, it is slow to come upwardly up alongside simulations where authority snowballs, but it would non survive that much harder to come upwardly up alongside catch-up mechanisms. Furthermore, rules laid upwardly yesteryear (simulated) regulators would tend to bird the playing field.

However, it seems clear that a strategy of wiping out your competitors is going to survive a dominant one. In gild to choke on marketplace shares stable inwards a simulation, the simulation rules (and/or algorithmic behaviour) in all probability has to survive stacked inwards a way to forestall firms from knocking out weaker competitors.

Can things similar taxes brand a difference? If you lot are busy driving your competitors out of business, it is unclear to me how tweaking corporate revenue enhancement rates volition brand a difference. The generic strategy of losing coin inwards brusque term yesteryear engaging inwards predatory pricing (to allow time to come super-normal profits) is going to piece of work inwards a lot of unlike simulation structures. Wealth taxes mightiness choke after the corporate founders, breaking the link betwixt corporate success too private wealth. Another angle of laid upwardly on is anti-trust law; I practice non know plenty nearly anti-trust police push clitoris to know how to implement it inside a simulation. (My excuse is that such laws take keep non been enforced since I was a fiddling kid.)

Concluding Remarks

Saying that wholesale inequality depends upon the rules of the game is in all probability non going to satisfy anyone. That said, whatever attempts to copy the dynamics of wholesale inequality volition survive extremely sensitive to the parameters of the simulation, too the only way to bear witness this is to experiment alongside the model. Anyone who is just stuck alongside a enquiry newspaper is non going to survive able to practice such experiments.

Appendix: Multi-Player

Single musician strategy games are no longer of much commercial interest. However, nosotros tin strength out pull less conclusions from multiplayer games.

There is non a whole lot of departure betwixt a 1-player too 2-player game; both players are trying to musical rhythm out each other too the AI players.

For 3-8 players, the dynamics of human politics dominates the play ease decisions of the designer. People volition tend to gang upwardly on the leader, since they lose fifty-fifty if they are inwards 2nd place. This is non typical demeanor inwards real-world corporations.

One mightiness promise that at that spot are lessons from games alongside massive numbers of players. However, the economical systems inwards those games are extremely fine-tuned for play balance. The designers demand to payoff the to a greater extent than enthusiastic players, but they cannot survive allowed to survive equally good dominant. As a result, they characteristic interactions that are far to a greater extent than egalitarian than the existent world. Nobody wants to log inwards too take keep to follow the orders of a pointy-haired boss; they practice that all twenty-four lx minutes menstruum inwards the existent world.

(c) Brian Romanchuk 2017