Saturday, August 6, 2022

No, a Planned Economy Can’t Actually Work

By Richard Fulmer of EconLog.

"In 2019, Leigh Phillips and Michal Rozworski published an article in Jacobin arguing that, “Yes, a Planned Economy Can Actually Work.” Their article, based on their book, The People’s Republic of Walmart: How the World’s Biggest Corporations are Laying the Foundation for Socialism, argues that large companies like Walmart have solved the problem of central planning:

In The People’s Republic of Walmart, we show how contrary to the historic argument of the likes of free market economists Ludwig Von Mises and Friedrich Hayek, economic planning of millions of products and services involving infinitudes of variables in supply chains and lots of non-price information is not just feasible, but works incredibly well…

Walmart, the largest company in the world, employs more workers than any other private firm; it is the world’s third-largest employer after the US Department of Defense and China’s People’s Liberation Army. If it were a country, its economy would be roughly the size of Switzerland.

Walmart, of course, sells goods on the market. Under capitalism prices are still inputs into the planning process for corporations and states alike. In addition to prices, however, firms today have at their disposal exponentially increasing amounts of information that is directly about people’s preferences or the use of resources…

Walmart engages in large-scale planning without the direct intermediation of markets at scales to make Hayek bristle. Internally, like nearly all firms large and small, it is a dictatorial planned economy: managers tell workers what to do, departments realize goals from on high, and goods flow by fiat.

 

Sam Gindin, also writing in Jacobin, throws a wrench into their claims:

Aside from the fact that the scale of organizing a total society in a nonmarket way is of a different order of magnitude than addressing a single, even vast, corporation, internal corporate calculations under capitalism have an advantage that centralized socialist planning would not have: they have external market prices and market-driven standards by which to measure themselves.

 

Moreover, though they admit that “under capitalism prices are still inputs into the planning process for corporations and states alike,” the authors never grapple with the problem of how to compare the relative values of disparate goods and services other than to wave it away:

There is a hard question about how we relate things to one another — cotton to steel or mind-numbing drudgery to art — but it is a poverty of imagination to think only markets can determine these multidimensional comparison questions rather than we ourselves, democratically.

 

While someone unencumbered by an impoverished imagination could conceivably devise a way in which to democratically answer “these multidimensional comparison questions,” no one, including Karl Marx, has succeeded in over a century and a half.

In addition, the question of how to compare different production processes and capital goods without market prices is not mentioned at all.

The authors’ solution to the Mises-Hayek socialist calculation problem is to get a bigger calculator:

Jack Ma, the founder of China’s Alibaba Group — one of the largest and most valuable companies in the world — argues that previous state planners in the Soviet Union and the early People’s Republic of China failed due to insufficient information. He has predicted that over the next three decades thanks to artificial intelligence and the sheer volume of data to which we now have access, we will finally be able to achieve a planned economy.

Before conceding that an Artificial Intelligence program running on a supercomputer can direct an economy, however, let’s consider solving something vastly simpler: the game of chess. Compared to an economy, chess is child’s play. First, there are only thirty-two “individuals” on a chessboard, and none of them has free will.  Their movements are severely constrained: only one piece may move at a time, each side moving a single piece in alternating turns; each piece’s position is restricted to no more than sixty-four squares arranged in two dimensions; and each piece has a set way in which it can be moved.

But even in this simple world, there are over 318 billion ways of playing just the first four moves on each side.  It has been estimated that in a 40-move game there are more possible moves than there are atoms in the universe. As a result, no computer programs have yet been able to solve the game, even though they can now routinely beat even the best human players.

Let’s make the game a bit more like real life. Suppose that each piece is magically given free will and the ability to move itself. Now the pieces may all choose to move at once or to never move at all. A pawn could decide to act like a Queen or a Knight. A Rook might turn traitor and attack pieces of its own color, or the Kings could negotiate a peace settlement and leave the board entirely. The number of possible board positions after only the first move just became infinite.

Yet even this revised game of chess is immeasurably simpler than an economy consisting of millions of people – all acting and interacting at the same time – each with his or her own motivations, incentives, responsibilities, cares, and circumstances; and each thinking and acting in four-dimensional space and time.

But it’s even worse than that. A supercomputer trying to solve the game of chess needs no data; it’s just a matter of pure number crunching and access to an impossibly large storage array. By contrast, “solving” an economy requires the ability to acquire vast amounts of knowledge, some of which can’t even be articulated, and some of which is generated by myriad market transactions that occur second-by-second. Worse still, this data must be collected, transmitted, formatted, and “crunched” in real time.

In 2007, computer games expert Jonathan Schaeffer reportedly solved the  game of checkers by performing 100,000,000,000,000 (1014) calculations on an array of as many as 200 desktop computers over a period of 18 years. Checkers is much simpler than chess (with a mere 1020 possible board positions vs 10120) and is vastly simpler than a nation’s economy.

So, assuming a lower bound of 18 years to determine the relative prices of billions of products and services, even if we could accomplish the impossible tasks of:

  1. identifying all the world’s products and services,
  2. creating a supercomputer, or computer array, big enough to handle the calculations,
  3. creating a storage array big enough to handle the data, and
  4. instantaneously acquiring and uploading all the needed data to our computer

the uploaded information would be out of date and useless years before the computer finished its work.

Like many a politician, Phillips and Rozworski claim that the country can be “run like a business.” Such a presumption is based in unfathomable ignorance and arrogance – a deadly combination."

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.