the boringwork.

Matsushita’s “Not for Bread Alone”: A Checked Review

by

TL;DR

  • The strongest evidence in this whole area supports Matsushita’s least mystical idea: management practice is teachable and it moves output. A randomized trial in Indian textile firms found 17% higher productivity in the first year after free consulting on standard practices.
  • The divisional structure Matsushita built in 1933 is the testable part of the legend, and the evidence is contingent, not triumphant. M-form adopters in US petroleum earned about two percentage points more on equity — until everyone copied it, after which the edge vanished.
  • «Employees are the real asset» survives as a correlation of r = .20 across 92 studies, but the causal arrow is bad: controlling for past performance almost wipes out the link between HR practices and future performance.
  • «Contribution to society» is not empty talk. Corporate-responsibility proposals that pass by a hair produce positive returns and higher labour productivity compared with ones that fail by a hair.
  • The book’s structural flaw is not that it is wrong. It is that one firm, one country, one era, narrated by admirers, cannot separate the principles from the luck — which is exactly the failure mode Denrell named in 2003.
Not for Bread Alone by Konosuke Matsushita — cover

Verdict

Recommend with caveats. Read the notes, not the book. Confidence: moderate.

Konosuke Matsushita ran a company for roughly sixty years and left behind a body of maxims that later got compiled, translated and repackaged many times over. The English catalogue of PHP Institute, the publisher he founded, lists no volume by this exact title — what circulates as «Matsushita’s Principles of Success» is a compilation of the same collected philosophy that appears under a dozen covers.

Some of what he asserted has since been tested by people with no stake in his reputation, and the parts that hold are the boring, procedural ones. The parts that made him famous — the water philosophy, the 250-year plan, business as service to humanity — are either untested or true in a much weaker form than the retelling suggests.

The claim on trial

Stated in falsifiable form, the book claims: a specific bundle of practices — long-term thinking, treating employees as the firm’s real asset, pricing goods down until they are as abundant as tap water, autonomous divisions, and contribution to society as the purpose of the business — explains Matsushita’s results and will produce comparable results elsewhere.

Two halves, and they fail differently. The first half — that these practices explain his results — is close to untestable, because there is one Matsushita. The second half — that they generalize — is testable, and has been tested piece by piece, mostly by economists and organizational researchers who had never heard of him.

One piece has not been tested at all. I searched Crossref and Google Scholar on 11 August 2026 for empirical work on the «water philosophy» as a pricing doctrine — deliberately driving price down ahead of cost to make a good universal — and found no study that isolates it. That is a checked negative, not a refutation. Untested is not refuted. But nobody should cite the water philosophy as evidence-backed, because there is no evidence either way.

The divisional structure is the testable part

In May 1933 Matsushita split his company into three self-contained divisions — radios; lamps and dry cells; wiring devices, resins and electrothermal products — each running its own factories, sales and profit and loss. Panasonic’s own published corporate chronicle states it was «the first single proprietorship in Japan to be organized along such lines». That is a company claiming primacy about itself, so treat it as a company claim. The date and the three-division design are not in dispute.

Does divisionalization pay? Armour and Teece looked at 28 petroleum firms from 1955 to 1973 and found that during the period when the structure was still spreading, M-form firms earned roughly two percentage points more on shareholders’ equity against a base of about 7.5%. Then the advantage disappeared in 1969–1973, once nearly everyone had adopted it. That is the honest shape of the finding: divisionalization was worth something while it was rare, and worth nothing once it was standard.

Hoskisson’s 1987 study in the Academy of Management Journal put the qualifier in its title: the performance effect of the multidivisional structure is contingent on diversification strategy. It is not a universal good. By 1993 the review literature had reached open disagreement — Hoskisson, Hill and Kim opened their survey by noting the same structure had been called both «the most significant organizational innovation in the twentieth century» and «an organizational fossil that is increasingly irrelevant».

The most awkward evidence comes from Panasonic. Its published corporate chronicle records a Groupwide reorganization effective 1 January 2003 into 14 independent business domains, and a further reorganization in 2013 in which 88 business units were consolidated into 49 units renamed «business divisions», managed by four in-house companies. The founder’s signature structure has been dismantled and rebuilt repeatedly by his own successors. A principle that gets re-cut every decade at the firm that invented it is a tool, not a law.

One note on sourcing. Those 1933, 2003 and 2013 details come from Panasonic’s own published corporate chronicle, which we could not reach at a stable address to re-check at the time of writing, so they are stated unlinked. They are uncontested corporate history, and the load-bearing finding is Armour and Teece’s anyway.

Employees as the real asset

Matsushita’s line that the company makes people before it makes appliances maps cleanly onto what researchers call high-performance work systems: selective hiring, training, job security, information sharing, pay tied to performance. Combs, Liu, Hall and Ketchen aggregated 92 studies and estimated the overall correlation with organizational performance at .20. The relationship was stronger when researchers studied whole systems rather than single practices, and stronger in manufacturing than in services — which is Matsushita’s exact setting.

Then the problem. Wright, Gardner, Moynihan and Allen measured HR practices in 45 business units against past, concurrent and future performance. The correlations were high and essentially identical at all three time points, and controlling for past or concurrent performance virtually eliminated the correlation with future performance. Profitable firms can afford good practices. That is not the same as good practices making firms profitable, and cross-sectional studies cannot tell the two apart.

There is one clean piece of evidence pointing the other way. Edmans tracked a portfolio of the «100 Best Companies to Work For in America» and found a four-factor alpha of 0.29% per month from 1984 to 2009 — about 3.5% a year. That is not causal proof that satisfaction creates value; it is evidence that the market underprices it, which is a different and slightly more interesting claim.

The long horizon

Matsushita’s 250-year plan is a rhetorical device. The underlying claim — that short horizons destroy value — has real support. Graham, Harvey and Rajgopal surveyed 401 financial executives and interviewed 20 more. 78% said they would sacrifice long-term economic value to deliver a smooth earnings path, and 55% would delay a highly profitable project rather than miss a quarterly consensus number. Executives report doing the thing Matsushita warned against.

Asker, Farre-Mensa and Ljungqvist compared listed firms with matched private ones and found public firms «invest substantially less and are less responsive to changes in investment opportunities», with the gap widest in industries where share prices react hardest to earnings news. Long horizons are worth something, and public-market pressure erodes them.

Founder and family control is where the story gets complicated. Anderson and Reeb found founding families held stakes in about a third of the S&P 500, roughly 18% of equity, and that family firms outperformed non-family firms, with family CEOs outperforming outside hires. But Bennedsen, Nielsen, Pérez-González and Wolfenzon used the gender of a departing CEO’s firstborn child as an instrument in Danish data and found that family successions cut operating profitability on assets by at least four percentage points. And Bloom and Van Reenen, surveying 732 medium-sized firms across the US, France, Germany and the UK, identified family succession by primogeniture as one of the two main causes of badly run firms, alongside weak competition.

So: family ownership can support a long horizon; family succession tends to destroy value. Matsushita handed the presidency to his son-in-law. The book does not treat this as a risk. It should.

What actually transfers

Here is the finding that partly vindicates the whole enterprise. Bloom, Eifert, Mahajan, McKenzie and Roberts ran a field experiment on large Indian textile firms, giving randomly chosen plants free consulting on standard practices — quality control, inventory, machine maintenance, order tracking. Productivity rose 17% in the first year, and within three years treated firms opened more plants. The reason the practices had not been adopted already was informational: managers did not know they existed or did not believe they worked.

That is Matsushita’s position, stated in the language of a randomized trial. Management is not a personality trait. It is a set of procedures you can teach to someone who does not have them.

Two qualifications, both from the same research programme. When the researchers went back nine years later, about half the adopted practices had been dropped, mostly because of managerial turnover and lack of director time — though a large gap between treated and control plants survived. And in US Census data covering 35,000 manufacturing plants in 2010 and 2015, management practices accounted for more than 20% of the variation in productivity, but 40% of the variation in practices was within the same firm. Even well-run companies run their own plants inconsistently. Founder philosophy does not propagate on its own.

The halo problem you cannot argue around

Every founder-philosophy book has the same defect, and naming it precisely matters more than any individual citation. The sample is one firm. The observation window starts after the firm succeeded. The narrators are people who admire the subject. Under those conditions, any set of practices the founder happened to use will look like the cause of the outcome, because the failures that used the same practices were never written up.

Denrell formalized this. Organizations learn by watching other organizations, but the ones available to watch are survivors, and «there is a strong tendency to focus on successful organizations in books and the business press». His result: undersampling failure makes risky practices look reliably good even when, across the full population, they have no relationship to performance at all. Matsushita’s decision to keep all employees on payroll through the 1929 collapse reads as wisdom because the firm survived. Run that decision a thousand times in a thousand firms and you would get a distribution, not a maxim.

The same distortion shows up in formal ratings, not just in books. Brown and Perry’s 1994 paper in the Academy of Management Journal is titled «Removing the Financial Performance Halo from Fortune’s Most Admired Companies» — a whole methodology built because reputation scores track prior financial results closely enough to contaminate them. If trained analysts scoring a standardized survey cannot keep results out of their judgement of quality, a biographer certainly cannot.

Where the book is genuinely right

Three things, each with the same citation standard applied.

First, teachable practice beats charisma, and the Indian trial’s 17% figure is the best number anyone has on it. Matsushita ran a management training institute from 1934 because he thought this. He was right, decades early.

Second, «contribution to society» has measurable content. Flammer compared corporate-responsibility shareholder proposals that passed by a razor-thin margin with ones that failed by the same margin — a near-random assignment. Passing proposals produced positive announcement returns, better accounting performance, and increases in labour productivity and sales growth. The effect is specific to close-call proposals and does not license every such programme, but it is a real causal estimate.

Third, the divisional structure in 1933 was a genuine innovation, whatever its later record. Armour and Teece’s finding — that the advantage existed while adoption was still spreading — actually flatters Matsushita, since he was decades ahead of the diffusion curve in Japan.

Who should actually read it

Read the collected philosophy if you are building an operating culture from scratch and want a coherent worked example, or if you are interested in Japanese industrial history. The material is short, plain and occasionally sharp.

Skip it if you want to know which of these practices will work in your firm. It cannot tell you, because it never compares itself to anything. For that, read Bloom and Van Reenen’s survey work and the Indian trial, which measure the same claims against controls. Our other book reviews apply the same test.

For observational evidence about what managers actually do, see «Managing»; for a founder philosophy with no evidence at all, «The 8th Habit».

One thing to try

Pick the single most mundane practice in the Indian trial’s list and check whether your unit actually does it: does every machine have a written, dated preventive-maintenance record, and does someone review those records weekly? Not a value. Not a philosophy. A logbook. Plants that adopted this class of practice gained 17% in productivity within a year, and had skipped it only because nobody told them it mattered. Run it eight weeks and compare defect rates and downtime against the eight weeks before.

Get the book

Find «Not for Bread Alone» on Amazon — as an Amazon Associate, The Boring Work earns from qualifying purchases (disclosure).

When to see a professional

This page is general information about management research, not financial, investment or legal advice. The equity-return findings cited here describe historical patterns in academic datasets and are not a recommendation to buy or sell anything. If you are restructuring a company, changing ownership or succession arrangements, or making capital-allocation decisions, talk to a qualified accountant, corporate lawyer or licensed financial adviser.

The boring bottom line

Matsushita’s central claim is partly right, and right for reasons he could not have known. Standard management practice really is teachable and really does raise output — 17% in a randomized trial. Employee-centred systems really do correlate with performance, at r = .20, though the causal arrow is unresolved. Long horizons really are worth defending, and executives really do admit to trading them away. Stakeholder orientation really does show up in returns when you can find a near-random experiment.

What does not survive is the packaging: that these practices, as a bundle, explain his particular firm’s rise. One firm, one country, one era, told by admirers, with the failures unrecorded. Take the logbook. Leave the legend.

Sources