Seeing like a CEO

What really happened to Boeing?

A cliché has emerged to explain Boeing's decline. Following the merger of the two firms, McDonnell Douglas' "financial", "Wall-Street-oriented", "bean-counter" culture displaced an engineering-centered culture that had hitherto prevailed at Boeing.

I find this explanation unsatisfying, because I do not know what it means. "Culture" is a kind of catch-all in social affairs that can be deployed to explain almost everything everywhere, and so it usefully explains nothing at all. If the advice your theory yields is "change the culture", then you have no meaningful advice.

Boeing's turning point was arguably when Harry Stonecipher, formerly of McDonnell Douglas, more formerly of Jack Welch's General Electric, ascended to the firm's leadership. Stonecipher's successor, James McNerny, also hailed from GE.

Jack Welch managed GE as a certain kind of rationalist. The business of a business is to maximize revenue and minimize costs. Plant, labor, and raw materials are costs. GE was already a conglomerate, not a unified firm. Welch sought to rotate its holdings towards lines of business that would bring in greater revenue with fewer of those costs, which meant he increasingly favored finance over industry. He gutted activities like basic research, whose connection to revenue was too loose to reliably value. He imposed very blunt, clear, incentives — ranking all employees, firing supposed underperformers and giving bonuses to outperformers in fixed percentages.

A thing to notice about this style of management is that you don't need to know anything about GE's particular businesses or technical processes to understand it. Welch's prescriptions reflected a form of economic rationality that is universally accessible, that could be taught in business school and tried anywhere. In my own little lingo, they reflect unsituated virtues.

But if you think the one weird trick to managing highly technical cutting-edge business processes at scale is a checklist of B-school bromides, you might just be mistaken. Basic research is very difficult to value. It remains essential to many firms' long-term viability regardless. Performance in highly collaborative processes may be hard to measure and easy to game. Canning supposed underperformers and rewarding "your best" might in practice lead you to shed your best and reward the cunning and disloyal. The mere exercise of ranking performance might undercut trust between employees, hobbling productive collaboration.

So a simple, not wrong, story about what happened at Boeing is that Jack Welch acolytes were elevated to the firm's leadership, and began to impose Welch's caricature, one-size-fits-all, business-school-friendly management style on what had already been an excellent enterprise, and in doing so they destroyed it.

I want to tell a slightly more sympathetic story. What, after all, was Harry Stonecipher supposed to do?

A modern aircraft is a marvel of technology, built upon an edifice of hard, empirical, science. Every element of an aircraft can be precisely described, characterized, and tested. But each choice about how any of those elements should be designed and all of those elements should be combined introduces tradeoffs, benefits and costs that are not straightforwardly commensurable.

You might think aeronautical engineering would be the ultimate "hard-information" kind of enterprise, but you would be wrong. The design space of an aircraft, the alternative combinations of plausible choices at every level from the tiniest part to the shape of the fuselage, is effectively infinite. There is no one true optimal design. Beneath the broadest specifications — how big, how far, how fast, burning how much fuel — an endless array of choices may have been made well or poorly, and expensively or cheaply. There is no objective measure by which to judge the quality of the process from outside of it. Designing and building — "engineering" — an aircraft is a soft information exercise. You can only have any idea if it is being done well if you are situated within the process and interacting with participants.

The key fact to note about Harry Stonecipher is not that he was influenced by Jack Welch. It's that he was, for the first time in the firm's history, an outside hire. Stonecipher had been President of McDonnell Douglas, which had under his stewardship been merged into Boeing. But he had no access to, no visibility into, no way to monitor, judge, understand, or discipline the je ne sais quoi that made Boeing work.

In order to do his job, Stonecipher undertook to rationalize and simplify and render legible the firm. Lacking access to "soft information" that could only come from developing relationships within Boeing over years of collaboration, he sought to manage the firm on the basis of "objective" so-called hard information. Among the most salient hard information in any business enterprise is cash flows, so the sway of "bean counters" was elevated automatically. Among the least legible information in a firm, to an outsider, is the quality of employee judgment calls, so internal expertise was demoted.

The Boeing Harry Stonecipher inherited was an integrated powerhouse. Every step, from the conception of new aircraft through their design, manufacture, and sale, was controlled and performed in-house. The new CEO infamously sought to transform the firm into a narrow specialist that would outsource all but a few aspects of the process. Again, you can blame Welch's simplistic B-school dogmatism for that. Specialization and trade are better than bureaucratic central planning. Focus on your core technology and let the market provide the rest. In retrospect, those ideas were obviously stupid for a domain where modification of any one element can cascade into requirements that myriad other elements change, requiring intensive coordination and costly renegotiation of contracts if those other elements are outsourced. Ronald Coase is also taught in business schools.

But the terms of contracts with outside vendors are objective and tangible in a way that continually evolving plans by managers and employees are not, especially when you don't know those managers and employees well enough to know whom to trust. Outsourcing was a business school fad by the early 2000s, but it also solved an information project for Boeing's new CEO, who, fêted but flying blind, was trying to manage a company that ran on processes he could not evaluate or even perceive.

Contracts, however, he could read. And understand, and negotiate. Organizational competences are soft information. The terms of legal documents are hard information.

In Seeing Like A State, James Scott famously described the perverse results that can occur when attempts to rationalize a system necessarily simplify it, eliminating as extraneous factors elements that in fact are essential, but whose role is not included in the reformer's simple model.

How was Harry Stonecipher supposed to run a firm whose excellences were embedded in talents and habits that could not be tabulated, systematized or measured? Was he just supposed to trust the judgment of "old Boeing hands"? Which ones? With whom would the buck stop?

That was his job. He had to reorganize firm into something he could see, understand, and manage. So he did. Of course it did not help that Stonecipher and his successor James McNerny came out of Jack Welch's orbit. They had Welch's playbook too close at hand. But if you think the Boeing they inherited was a national treasure, the root of the rot was to bring in outside hires.

Contrary I think to many of our perceptions, the vast majority of firms, especially larger firms promote their CEOs internally. The idea of the superstar CEO turnaround artist, endowed with a managerial g factor enabling excellent leadership at any firm, is bullshit. Complex human institutions are their own realities, whose most important characteristics subsist in habits and relationships and invisibly distributed skills.

Effective leaders certainly do strive to surface and measure objective, quantitative data about their institutions, and incorporate it into their decision making. But those kinds of measures cannot be remotely sufficient to guide and govern the living organism that we call a firm. Track your EKGs and blood oxygen and body mass and lung capacity all you want. You'll never think out how to get peristalsis right. A brain detached from, incapable of regulating, the most of what we are that is submerged and "autonomous" is the brain of a creature not long for this world.

Firms like Boeing, that are not mere holding companies, but that develop and sustain and expand unique and complex competences, are each their own "deep state". Making outsiders brain and dictator of such an organism is an incredible disruption, an incredible sort of gamble. Perhaps when we have decided that the pathologies of our deep state so outweigh its capabilities, a YOLO on effectively killing the organism and reconstituting something else from its assets might make sense. But usually all you end up with is a corpse.

Which is what Boeing is, or would be, if its customers and the government that regulates it didn't require it so much. It's a firm on a ventilator. Perhaps it can be restored to rude good health, but that will take a great deal of time and dedication and care, and I think recruiting a lot of help from people likely now retired who knew it, and made it, when.

What it does not need is further bloodletting by shareholders. Even the most productive workers demand subsidy when they find themselves in intensive care.


Indirection and the character of capitalism (Part II)

When last we met, we saw that extending the basic capitalist circuit from M-C-M' to something much more indirect, like M-(M-(M-(M-C-M')-M')-M')-M', changes the character of capitalist activity. The controlling party in the more indirect case has little information about or ability to improve the core business process — the combination of commodity C and labor into a product salable for M'.

Our indirect capitalist has basically one meaningful lever that she can exert down the chain of control, which we've called the squeeze. She can try to encourage greater efficiency by insisting that the layers beneath her remain "lean", with a minimum of surplus cash and a maximum of financial pressure.

Her indiscriminate austerity can in some cases lead to real efficiency improvements. But she has no way to titrate the squeeze. She cannot even be aware of the subtleties of the core business process, and the relationships that support it. So it becomes likely that the austerity she imposes will lead to foolish economies, forcing the underlying business to prioritize short-term, "hard" cash flow over "soft" intangibles like relationships, reputation, maintenance and development of skills, standards of product quality. The net effect is to degrade the capabilities of the business, and over time, its economic value as well.

This pathological outcome is often rewarded rather than punished by market forces, even over long periods of time, despite the clear impairment of the core business that it provokes.

Let's understand how this works. We began with an economic unit, M-C-M, and we posited a buyer, so it became M-(M-C-M')-M'. But often a purchaser of enterprises doesn't purchase just one! A "holding company" may have multiple units. Suppose our buyer purchases three units. Then we have a structure that looks like this:

            /(M-C-M')\
           M-(M-C-M')-M'
            \(M-C-M')/

Of course, we can have more than one level of indirection:

            /(M-C-M')\
           M-(M-C-M')-M'
          / \(M-C-M')/ \
         /  /(M-C-M')\  \
        M--M-(M-C-M')-M'-M'
         \  \(M-C-M')/  /
          \ /(M-C-M')\ /
           M-(M-C-M')-M'
            \(M-C-M')/

Maybe the first diagram is a regional supplier that holds three business units (each one its own geographically distinct shop with separate management). The second diagram is what happens when a national firm combines three such regional suppliers.

We might further imagine a PE firm "rolling up" three such national firms, hoping to create an industry powerhouse. The level of indirection would increase by one, and the breadth would grow from 9 once-distinct commodity-transforming business units to 27.

             /(M-C-M')\
            M-(M-C-M')-M'
           / \(M-C-M')/ \
          /  /(M-C-M')\  \
         M--M-(M-C-M')-M'-M'
        / \  \(M-C-M')/  / \
       /   \ /(M-C-M')\ /   \
      /     M-(M-C-M')-M'    \
     /       \(M-C-M')/       \
    /        /(M-C-M')\        \
   /        M-(M-C-M')-M'       \
  /        / \(M-C-M')/ \        \
 /        /  /(M-C-M')\  \        \
M--------M--M-(M-C-M')-M'-M'-------M'
 \        \  \(M-C-M')/  /        /
  \        \ /(M-C-M')\ /        /
   \        M-(M-C-M')-M'       /
    \        \(M-C-M')/        /
     \       /(M-C-M')\       /  
      \     M-(M-C-M')-M'    /
       \   / \(M-C-M')/ \   /
        \ /  /(M-C-M')\  \ /
         M--M-(M-C-M')-M'-M'
          \  \(M-C-M')/  /
           \ /(M-C-M')\ /
            M-(M-C-M')-M'
             \(M-C-M')/

Obviously, in real life, actual bundlers or acquirers of businesses don't always purchase precisely three firms. We've chosen that number for convenience. Purchasers in real life aren't neatly stacked into levels. Very big firms can acquire tiny assets directly.

But the broad point holds — each degree of indirection magnifies potential scale. One CEO who can handle three business-unit managers as direct reports can indirectly manage many, many units, if her direct reports can also handle three subsidiaries.

It remains the case, as it was in our earlier discussion, that each with each level of enclosure, information is lost about the units that are subsume. It also remains true that the controlling interest has basically one lever by which to try to improve the technical efficiency of inner units, the squeeze.

If the internal capabailities of businesses — how well they produce their product or provide their service — were what mattered most, these behemoths of agglomeration would be stumbling dinosaurs. They would fail in the marketplace. Indeed, during a more enlightened time (the 1970s), it became conventional wisdom that conglomeration was inherently stupid, for exactly the reason sketched here.

But firms have external capabilities as well. Rather than improve business processes internally, they can alter the environment within which their business units compete, and find ways to help those units gain advantage even despite diminished capabilities and product quality. Instead of "focusing on their core technology", in the old business lingo, controlling interests of highly indirected units can exploit market power and political influence to enhance profitability.

Indeed doing so becomes the core technology of the holding company. As the degree of indirection grows, the holding company becomes ill-suited to compete on the basis of technical excellence with smaller firms whose management sits much closer to the technical process. They would be outcompeted in that niche.

Further indirected firms are instead specialists in exploiting market and business power. As an economy tolerates greater degrees of indirection, one should expect technical quality to decrease while strategies based on exploiting market and political power come to dominate.


BOEING 777                          |
by Jon Hyatt                        |
whatfer@u.washington.edu          .-'-.
                                 ' ___ '
                       ---------'  .-.  '---------
       _________________________'  '-'  '_________________________
        ''''''-|---|--/    \==][^',_m_,'^][==/    \--|---|-''''''
                      \    /  ||/   H   \||  \    /
                       '--'   OO   O|O   OO   '--'

via https://www.asciiart.eu/vehicles/airplanes

There are, of course, a lot of caveats.

Business processes may have technical economies of scale and scope. The argument above tacitly assumes that each business unit is already at the scale that exhausts technical economies of scale. If that's not the case, perhaps purchasing equipment and talent exploited too artisanally, and then bundling them into an enterprise that operates at scale, could yield efficiencies that more than compensate for any information problems or the social costs of market power.

Of course nearly all business combinations tout supposed "synergies" by which greater technical economies of scale will be exploited. Both logic and experience suggest we should be wary of those claims.

Logically, there is no need for a firm operating capably but at an inefficient scale to take on the information costs associated with less direct ownership. To improve efficiency, a more direct controlling interest can just raise external capital and grow to scale. This way, control remains at the level of the business unit, where managers understand the technical process, and are drenched in soft information about relationships with workers, vendors, and customers.

If — but only if — firms are expected to thrive or die primarily based on technical prowess, then capable 19th-Century-style capitalists, hands dirty from the processes they manage, should be the ideal parties to raise external capital. Finance would mostly be a competition to find mid-scale business (most technical efficiencies are exhausted by moderate scale) whose principals are hip-deep in managing their firms.

But if financiers see market power as the great determinant of success or failure in an industry, capital will flow to bundlers rather than do-ers, to important people rather than talented builders. Once some firms enjoy outsize market power, it becomes implausible for technical excellence alone to compete. A bad equilibrium takes hold. Capital providers expect winners to win, and become reluctant to finance smaller units, unless their exit strategy is to be acquired by a bigger fish. Investors understandably perceive a "kill zone" in going against "market leaders", and don't want to risk capital there.

"Market leaders" then substitute predatory pricing for technical acumen, making quick work even of superior competitors who can't raise external capital to outlast the assault. And they can't. They would have raise a tremendous amount of capital, on incredible faith by investors, to last long enough that their less capable but better resourced competitors would even come under pressure to relent.

Beyond the borders of Hyde Park, South Chicago, tactics like predatory pricing, payola, and construction of advantage by exerting political influence are effective. Capital providers understand this, so are disinclined to invest in quixotic new entrants. Then high cost of capital becomes yet another reason why smaller, more capable firms cannot compete.

Any hope of a diverse "ecosystem" collapses. Only the consolidated survive.

Ironically, given the hype bundlers deploy to justify acquisitions, even when there really are unexploited technical economies of scale across subunits, bundlers may lack the information and skill needed to recognize and exploit them. It doesn't matter. "Exploiting synergies" was always at best a secondary aspiration. Mere acquisition is enough to increase the bundler's market power. If there is "efficiency improvement" available, it is usually to shut down elements of the acquired business unit, eliminating competition while letting other units exploit the zombie brand.

This may indeed be business-efficient! Revenue goes up! (We sell more brands, against less competition.) Costs go down! (Workers are laid off.) But no production efficiency is introduced. No business unit is transformed to operate at a more efficient scale. The only talent actually deployed is keeping a straight face while you gush to the world about the synergies.

Actually exploitating economies of scale would involve neither retaining acquired business units, nor shutting them down, but combining them. It would require bringing the talents and assets of multiple business units together to hammer out a unified processes with thick information flows. This is hard, and risky. It requires deep knowledge of the various businesses, and still often fails. Indirect controlling interests know enough to know this. They don't usually try.

Berle and Means famously grappled with the separation of ownership and control inherent in modern corporations. In a public company, the controlling interest is really management and the Board of Directors, even though in principle the "owners" of the firm are shareholders. How do we prevent the managers of the firm, who are present and effectively control it, from putting their own interests before those of dispersed, distant, disorganized, passive, shareholders?

Indirection turns that logic on its head. With indirection, owners remain distant, but they are organized, unified and active. Instead of protecting owners from managers, the struggle becomes to protect productive business units and the broad public from the predations of owners whose competitive strategy is to accumulate and exploit market power and political influence.

And not necessarily owners.

In order to explore "indirection', we've started by imagining small firms getting acquired and bundled into bigger agglomerations, yielding a kind of syndrome where

  1. The controlling interest of a large enterprise is informationally distant from the units in which core business processes occur
  2. The controlling interest imposes dictates on the units in which core business processes occur, even though it lacks detailed information about their operations and circumstances. Usually this is a "squeeze" to maximize the short-term cashflows remitted to the controlling interest.
  3. The dictates imposed by the controlling interest successfully achieve short term objectives, but impair the longer-term technical capability of the agglomeration.
  4. The agglomeration enjoys competitive advantage despite impairment of its internal capabilities, by shifting its focus from internal process to external action. It deploys market power and political influence to manage the environment in which it competes, rather than pursue production excellences.

It is the syndrome, not the particular history, we are interested in. We claim this syndrome amounts to a niche or equilibrium that many industries find, and that the broader economy and financial system comes to accommodate or even demand, ensuring its reproduction.

The controlling interest need not be owners. The term chickenization is derived from the experience of chicken farmers who are notionally independent, but who work under contract to integrators. Integrators both supply the chicks and feed, and pay for bulked-up birds. They dictate nearly every aspect of "independent farmers'" production process, yet squeeze them (setting them against one another in tournaments) to somehow fatten their birds ever faster. The integrators are the controlling interest, by virtue of the monopsony power they wield, even though they never formally acquire the farms that raise the chicks. I don't think anyone could argue that Tyson Foods, Pilgrim's Pride, or Perdue produce the healthiest or tastiest or in any sense best chickens. To the degree those leviathans seem to be "efficient" it is largly by virtue of shifting costs and burdens to workers, farmers, and consumers' long-term health in exchange for keeping prices at the register cheap. The chicken farmers they integrate have little scope in which to innovate. Product quality simply is not the basis on which these firms compete.

Boeing is an interesting mix. It has grown informationally distant from the units where core business processes occur. However, this was not because it acquired but failed to integrate new firms, but simply due to a change in management in 2003. The new management had no visibility into Boeing's traditional, very excellent, technical processes. What they could not measure, they could not value, discipline, or control. So they undertook to reorganize the firm to be more legible and more "bottom-line focused".

The reorganization succeeded, in the sense that the old technical processes were diminished, and replaced by contracts with outside vendors, which the new managers were capable of negotiating, understanding, and monitoring. Sometimes the outside vendors were spun out of Boeing itself.

Unfortunately, this new regime is technically far inferior to the one that it replaced. Even competently outsourced work is more informationally distant than the well-integrated in-house process which Boeing had enjoyed when its leadership had been promoted from within its own ranks. And Boeing's outsourcing hasn't always been very competent. "Negotiating great pricing" feels like a competence to a manager. But when you negotiate from a firm you just spun out better pricing than the costs you would have borne if the work were still in-house, probably you haven't achieved some magical market efficiency. Probably you've just bullied a supplier into cutting corners, and hoping you won't notice.

Boeing has a great deal of market power and political influence. The firm is essential to the United States in both commercial aviation and defense markets. Not many firms can get away with killing 346 people through incompetence and gross negligence in spectacular mediagenic accidents, yet face almost no meaningful sanction.

The new, more outsourced Boeing is substantially a controlling interest of its smaller suppliers, over whom it wields tremendous monopsony power. Those great prices it negotiated, all those efficiencies that eventually turned into catastrophies, were just the squeeze.


Indirection and the character of capitalism (Part I)

Remember when the byproduct of capitalists making money was

more massive and more colossal productive forces than have all preceding generations together. Subjection of Nature’s forces to man, machinery, application of chemistry to industry and agriculture, steam-navigation, railways, electric telegraphs, clearing of whole continents for cultivation, canalisation of rivers, whole populations conjured out of the ground

Capitalism now, at least in America, feels a lot more... flaccid.

To be sure, money is being turned into more money more rapidly than ever before. But the gearing between that process and actual production of real-world goods and services seems weak.

This is admittedly hard to measure. So many goods and services are digital now, and we can only really quantify them by "value", by how much people are willing to pay for them. If people are willing to pay half a trillion dollars to Google and Facebook (er, "Alphabet" and "Meta") for ads, should we hail that as our generation's equivalent of the application of chemistry to industry and agriculture?

You can be as apologist was you wanna be. But I'll go with "no". In fields that require mastery over the forces of nature — rather than just mastery over one another's wallets — Boeing really is the iconic American company.

Here's a simple theory.

Marx famously described the general formula of capital as M-C-M'. That stands for MoneyCommodityMo' Money.

Ordinarily economists think of money as a mere intermediary in, or perhaps a veil over, the exchange of real goods and services (C-M-C), But for the capitalist, real goods and services are the mere intermediary, in a "circuit" that turns money into more money.

the money itself is only one of the two forms of value. Unless it takes the form of some commodity, it does not become capital. There is here no antagonism, as in the case of hoarding, between the money and commodities. The capitalist knows that all commodities, however scurvy they may look, or however badly they may smell, are in faith and in truth money, inwardly circumcised Jews, and what is more, a wonderful means whereby out of money to make more money.

Setting aside the gratuitous antisemitism, in Marx's time the capitalist was typically acquainted with "commodities, however scurvy they may look, or however badly they may smell".

The capitalist purchased the odiferous commodities himself. He managed, understood, and oversaw highly technical processes by which stinky raw materials were transformed into items suitable for sale to yield that sweet M'.

Now let's think about what happens when capitalism evolves to a more indirect form. Consider the following circuit: M-(M-C-M')-M'

Here we describe a new breed of capitalist, who treats the old capitalist's enterprise as itself a quasicommodity by which an initial investment M might be converted to a greater sum of money M'. In the course of this transaction, the initial capitalist is converted to a mere manager, or else retires and is replaced by a manager.

The new purchaser does not herself take on the role of our first capitalist. Perhaps she tours the facilities occasionally, but in general she need never smell the stink of the commodities before they are reconverted to sell and smell like roses. She engages in a purely financial transaction. She pays M, receives control rights, and hopes by virtue of this process to earn a juicy M'.

There are two things to notice. First, our new purchaser has little ability to contribute directly to the technical quality or efficiency of the production process. From her corner office in a distant skyscraper, she can only ever understand that process in the most abstract terms. Perhaps "on the business side" she can help a bit. Perhaps the accounting systems can be replaced with something more modern and shared across her other holdings, yielding a genuine efficiency.

But to the core technical competence of the firm, she can contribute nothing, directly. Its work is simply outside of her domain.

Perhaps she can contribute indirectly, however. Perhaps at the original firm, the original owner — 19th-Century capitalist bastard though he was — had personal relationships with his most skilled employees, his most loyal customers, the ecosystem of vendors from whom he purchased what to Marx becomes only C. Perhaps because of those relationships, the old man was paying more to his employees than their replacement cost in the labor market. Perhaps, recalling a time when his vendors kept him afloat through a cash-flow crisis, our capitalist did not always aggressively bid contracts for his inputs and go with any lowest cost provider. Perhaps work conditions for employees, many of whom he personally knew, were expensively pretty nice. Perhaps he kept prices stable for long-time customers even when the market would bear more, increasing production and delaying opportunities to win new customers in order to prevent stockouts. Perhaps he maintained a level of quality greater than was strictly necessary to make his sales.

Our 19th Century capitalist was no altruist. He didn't do these things because he was a nice guy. He believed that by treating his employees, vendors, and most loyal customers well, he was being "long-term greedy".

He might have been right, but he might have been wrong too. The optimal give in the give-and-take of positive-sum relationships is difficult to compute objectively. The milk of human kindness, especially towards people one knows and has known for a very long time, might become a source of cognitive bias.

So, our purchaser in her corner skyscraper might actually contribute to the financial efficiency of the core business simply by a squeeze. If, by hypothesis, the enterprise was overpaying its workers, undercharging its customers, overpaying its vendors, and overspending on quality, then draining the enterprise of cash and keeping it under financial pressure might remedy that. Our new purchaser might even contribute to the technical efficiency of the core business process. What if the work environment had previously been so comfortable that workers had little incentive to innovate labor-saving improvements? Reduce the head count, and perhaps the remaining workers will develop ways of getting the same work done. That would be a real efficiency for the firm, and for the economy as a whole, as the workers who were laid off can be reemployed to produce new goods elsewhere.

So, the M-(M-C-M')-M' process may genuinely yield efficiencies. But it remains the case that the controlling party is more removed from the technical details of production than the controlling party in a traditional M-C-M' operation. And beyond quickly exhausted one-offs like accounting synergies, she has only one lever by which to use her control to improve the business: Squeeze.

Now consider the following circuit: M-(M-(M-(M-C-M')-M')-M')-M'

Perhaps what initially was a small firm with its own factory is purchased by a regional supplier which is purchased by a national firm which is purchased by a private equity firm.

With every level of indirection, the controlling interest grows informationally more distant from the technology of production. It remains capable of affecting the core production process via only the same single lever. Squeeze.

In the one-level M-(M-C-M')-M' case, we hypothesized that perhaps the initial entrepreneur really was overpaying, undercharging, and underinnovating, out of cognitive biases derived from being too close to the nexus of relationships that constitute the firm. But the entrepreneur's "long-term greedy" theory was never wrong in principle. It is possible to underpay ones workers and vendors, overcharge ones customers, skimp on quality, in ways that improve short-term profits but impair the long-term value of the business.

In the one-level case, the controlling interest may retain some capacity to judge and balance. The former owner, now the manager, of the enterprise can make his case for why it'd be foolish to sharply hike a big loyal customer's price to whatever the market will bear, for why accepting a bit of opportunity cost might be better for the firm over time. Shop-level management still has the ear of the controlling interest, who may be convinced that some indulgence of customers, vendors, and workers is in fact worthwhile. Sure, the buyer should squeeze. But she should squeeze only so hard.

But once you get to M-(M-(M-(M-C-M')-M')-M')-M', that informational connection is entirely lost. If all you have is a hammer, every problem is a nail. If all you have is the squeeze, every problem must be inefficient slack. With each new level of indirection comes an increased bias to squeeze, and decreased information about intangible, forward-looking costs of the austerity you are imposing.

Where our 19th Century entrepreneur may have suffered a bias towards overindulgence, our highly indirect owner suffers a bias towards "penny wise pound foolish".


If this were the end of our story, capitalism red in tooth and claw might eventually sort things out for us. Enterprises too indirectly owned would make choices that, over time, would work out more poorly than "less efficient", less indirectly owned firms. The market might settle on some optimal degree of indirection.

But another factor interferes.

Indirection comes not only with depth, but with breadth and scale. Breadth and scale confer capabilities that, to the controlling interest, are real opportunities. But to society at large, they must be scored as costs.

For more on that we must wait until Part II.


How to understand approval voting

The electoral system I favor for most elected positions is approval voting.

Under approval voting, voters are handed ballots identical to the ones they currently receive. The only difference is that instead of being asked to select our one most favored candidate, we are asked to select one or more candidates of whom we would "approve" for each position.

Votes are tallied just like they are now. Every votes is counted, and the candidate with the most votes wins. Every candidate is allocated all of the votes they received, regardless of whether a voter selected one or several candidates. Approval voting is extremely simple.

Approval voting addresses what voters most detest about the current system, that often we cannot express our true views without empowering those we most oppose.

Suppose you are some neocon, neoliberal, never-Trump person. You don't like Joe Biden. He's been veering towards social democracy, which you call "communism". But Trump, Trump is an idiot moral catastrophe who will destroy America. Under the current system, your only choice is to hold your nose and vote for the communist. Even if your true heartthrob — Mitt Romney! — could be persuaded to run as a third-party candidate, you couldn't vote for him. It would split the anti-Trump coalition. Given the dominance of the two big parties in America, a vote for a third party means a vote withheld from one of the candidates that might actually win. So if you care about the consequences of your vote rather than mere narcissistic self-expression, you have to vote for the one of the two major parties you think least bad, rather than for the third-party candidate you love. But then the fact that you do this, and that everybody else does this, cements in place the dominance of two parties few of us are very happy with.

Approval voting lets us break the cycle. Under approval voting, our beleaguered neolib neocon can put a check next to Mitt's name on her ballot, and also next to Comrade Joseph's. The first time around, while the two-party habit is still strong among us, the vote for Mitt is probably just self-expression. But it's no longer narcissistic self-expression, because now it does no harm. The vote for Biden still sets Trump back just as fully as it would have under our current system. And the self-expression is productively public. When votes are tallied, if a silent near-majority has expressed its love of Mitt and his new Old-Tyme Republican Party, voters will observe that, and next time around OTRP candidates will be real contenders rather than mere messaging.

Under approval voting, voters can always vote their true preference without splitting their broader coalition and handing an advantage to the worst candidate. And because voters' actual preferences become publicly visible, the self-fulfilling dynamic that stabilizes our current two-party duopoly is undermined. Third-parties can emerge as contenders. Once major parties can submerge.

"Breaking the two-party doom loop" is the first thing most discontents look for in electoral reform, and approval voting does that. But it does it in a way that has some really useful characteristics. In particular, it privileges candidates who can reach across factional lines over candidates who appeal exclusively to their own factional base.

In single-winner elections, where one person is going to have to serve a whole public, that person should be someone nearly everyone can live with, rather than a person adored by her own faction and hated by others. Elections ought not be contests over who will dominate whom. They are means by which we establish a government under which we all must thrive together.

Not all elections should be single-winner elections! We absolutely do want the purest, strongest, and best expressions of all of our political factions to participate in government. We just don't want the purest, strongest, and best of only one faction dominating the government and usurping control of the state. At the Federal level, members of the US House of Representatives should be elected not by any single-winner voting scheme, but by one of the many electoral systems that yield proportional representation, including MMP, multi-winner single transferable vote, or random ballot. A deliberative body meant to represent and include the full diversity and cacophony of the polity is quite different from a single official who must serve the whole public. The two demand different kinds of elections.

The US House should be assembled by proportional representation. The Senate, in which two at-large electeds must represent the whole publics of their states, should be selected by approval vote. That reform alone would dramatically alter the character of the body, and help render it a cooling saucer rather than a veto point for ideologues.

Approval voting represents a very incremental reform. It is approximately the current system. The most obvious strategy for voters is simply to cast one vote for their most preferred candidate. Most people will likely do just that. If everybody did it, then nothing would have changed.

But not everybody will vote only for their one favorite candidate. Some fraction of voters will "approve" two or more candidates.

Contrary to some very mistaken presentations, people who select multiple candidates are not "casting multiple votes" and thereby usurping voting power, violating the principle of one-person-one-vote. On the contrary. Selecting multiple candidates under approval voting is an act of generosity. You have selected your first choice, but then you add a candidate that is someone else's first choice, that is lesser from your perspective, and put that candidate on an equal basis to your own. The more candidates you select, the less you are insisting "my way or the highway". You are supporting your way, but you are also assenting to other ways that are not yours, but you can live with. This inverse relationship between preference imposition and number selected becomes obvious when you consider what happens if a voter approves every candidate. That has the same effect on the outcome as simply not voting at all.

So multiple selectors are generous voters. They are voters looking to find bridges and overlaps between their own preferences and those of other factions. And in what would be a close election if everybody chose only their fave — or under the current system — it is these generous voters, willing to make common cause with people whose values and interests differ somewhat from their own, who become kingmakers. Ironically, individually each voter cedes power by selecting more than one candidate. But as a class, approval voting elevates this group of less narrowly insistent voters to a position of influence.

Approval voting disfavors "mobilize the base" electoral strategies in favor of outreach and persuasion. One shouldn't evaluate electoral systems only by the electoral outcomes they tend to produce, but also by how they shape the behavior of candidates and the electorate. Our current system, in which primaries favor factional extremists, who are then mostly either shoe-ins for their position or subject to coin-flip general elections, does more even than Elon Musk to turn politics into a carnival of performative whack. Approval voting would dramatically change candidate incentives, and by doing so, would dramatically change our society for the better, regardless of who wins the various contests. Approval voting promotes social cohesion.

But wouldn't approval voting just elevate squish moderates, or even worse, "centrists"? I know, dear reader, you don't like "centrists". I detest "centrists". But what "centrism" has come to mean in American political discourse is the set of values and interests well served by the status quo, or at least the status quo as it prevailed until 2016. "Centrists" are socially liberal but fiscally conservative. "Centrists" don't want to tax billionaires, because they worry about incentives and "supply side effects" and whether soaking the rich would do "institutional violence" somehow to the country.

The actual center of American opinion looks nothing at all like this. Across factional lines, taxing extraordinary wealth more heavily is close to universally popular. Only the very rich themselves, and the weird sliver of "centrists" that serves them, object. The actual center in the United States is less socially liberal on some causes than "centrists" (which is why "anti-woke" can be an effective populist strategy), but very liberal on reproductive rights, and is overwhelmingly live-and-let-live. Approval voting would give us politicians whose agendas would be better aligned with the actual center of the electorate. This would look nothing like the agenda of contemporary so-called "centrists" in politics and media.

Approval voting would tend to yield candidates not perfectly aligned to my own socially liberal, social democratic views, but closer than what either party gives us now. And that would be the experience of most voters. From whatever direction we come, we do tend towards a common center when the politico-plutocratic freak show is not actively dividing us. Approval voting yields winners who deliver to no one everything they want. But its winners will be much closer to the vast majority than what's on offer from a duopoly that performatively chisels open social and cultural fissures in order to distract from a plutocratically constrained "consensus". And its winners will enjoy broader mandates than the 50% + one voter that the current system tends to, making it easier for elected officials to actually get things done.

Approval voting improves social cohesion. Social cohesion improves state capacity. State capacity deployed on behalf of a widely shared consensus improves all of our lives, in the ultimate virtuous circle.


Why does wage compression underwhelm?

Here's a graph from Autor, Dube, & McGrew on "The Unexpected Compression":

I agree with many commentators on the (economic) left and center-left that economic policy under the Biden Administration has been extraordinarily — and unexpectedly! — good. But that policy excellence has not, um, trickled down, to satisfaction and enthusiasm among its supposed working class beneficiaries.

Among so-called centrist pundits, as I (uncharitably) tend to read them, I detect a hint of outrage. The Biden Administration has been working for the plebes like no other administration in decades, yet the ingrates don't give the administration any credit. There are shades of the old "economic anxiety" debate. Really, it's pointless to try to do anything for downscale Americans. They are just a bunch of racists or (in the case of Black and Latino men) chauvinists and traditionalists.

I want to emphasize, the Biden Administration really has been working to serve the working class like no administration in decades!

But so far, from the perspective of regular humans, the results have been underwhelming. The trends are meaningfully positive, and given four more years — or better yet an electoral trifecta ‐ I think the Democrats might enjoy electoral dividends. But November 2024 is probably too early.

The graph above is the center of the case for working-class ingratitude. Since Biden took office, wage growth at the bottom of the distribution has dramatically outstripped growth at the middle and the top. If you check out the paper, you'll find similar graphs showing that non-college-graduates have done better than college graduates, young workers have done better than older workers, low-wage occupations have done better than mid- or high-wage occupations. Looking only at these wage stats, you might argue that Biden economy has served youth and Trump's base especially well.

However, if you were to look at this triumphalist post by Mark Zandi, you might conclude that the Biden Administration has been working for rich assetholders:

It's a rising tide lifting all boats, right?!

Well, maybe. But in relative terms, or as an expression of priorities, the administration can't be working for working people and for the well-to-do asset-holding class.

In dollar-weighted terms, the wage gains at the lower-end of the working class represent a small fraction of the $300K per household that their betters have effortlessly enjoyed. As Zandi points out, the less-affluent 40% earns nearly none of that, so those wage gains are swimming to keep up with $500K among asset-holding households. More realistically, the middle of the distribution experienced much more modest gains, while households at the top earned millions and billions. Working people continue to fall behind the asset-holding class under Joe Biden, even as the wage component of income has compressed.

In general, wage inequality is declining at the same time as the labor share of income — when measured correctly for distributional purposes, including capital gains — has declined precipitously.

But it's worse than that. Asset wealth, at least in theory, represents claims on future cash flows that are going to be paid by someone. The appreciating "nonfinancial assets" in Zandi's graph represents home ownership becoming ever more inconceivable to much of the public. Institutional investors now pay those flows, in order to transform them into exorbitant rents that working people will pay without accumulating home equity. Soaring equity prices, if equity prices are growing faster than unit growth in goods and services, represent expectations of increased profit margins that working people will find embedded either in prices or subdued wages.

It's true that asset price collapses tend to be bad for "Main Street" (if the government fails to do its job) because the investor class panics and fires everybody and generally throws a temper tantrum to remind the world to be solicitous of "business confidence".

But that asset price collapses can hurt working people does not imply that asset price booms are good for them. A great moderation in asset prices — in the cost of housing, in the profitability of firms — is what best serves working class people.

Never mistake applause from The Economist for quality economic management. The correlation coefficient is quite negative.

Finally, it's worth thinking about the lingering hangover from the past few years' inflation. Among the commentariat, the explanation I often hear (with a suppressed condescending tone of "oh those idiots") is that even though inflation has declined substantially, the price level has not, so people are mad that their groceries still cost more than prices they anchored on a few years ago. (The condescension comes from sophisticates' understanding that, although the policy apparatus strives to bring inflation down to target, it does not seek, and would not tolerate, outright deflation to reverse an increase in the overall price level.)

I can condescend with the best of 'em, and I don't disagree that this is part of the story. But I think economically sophisticated commentators miss something much more obvious, which is that constant nominal wages at a constant price level feel to an individual worker very different from getting a 4% raise that fully covers 4% inflation.

Obviously, in the second case, there is an issue of lags. Inflation tends to mount continuously, while compensatory raises happen in discrete steps. Workers suffer losses in purchasing power for some time, even if raises eventually undo them.

But more fundamentally, workers often experience raises as rewards, as the fruit of their personal merit. If a worker is given a big raise, she usually doesn't attribute it to macroeconomic circumstances, but to her own hard work and savvy.

If a worker earns a constant nominal salary at a constant price level, she just never got a raise. But if a person gets a 4% raise and the price level rises by 4%, her experience is she earned a 4% raise through sweat and skill and staying late, ginning up the nerve to ask, demanding, holding firm. But then Joe Biden came along and took it away from her with his inflation.

Our graph shows low wage workers enjoying a "real wage increase" of roughly 4% from when Joe Biden took office in January 2021 until July 2023, adjusting for CPI-U inflation. In nominal terms, that translates to our worker struggling and striving and wheedling from her asshole boss raises of 21% over the period, and then watching inflation eat away nearly 80% of the gain. No wonder she's maybe a bit bitter.

If she had some stock, or a home whose value jumped from $400K to $600K over the period, she might feel a bit better.

But people like that, you know. They just never plan.