Industrial Ecosystems 1: How Japan Built Its Supplier Base

Before we get to Toyota, it is worth pausing on this point.

This report isn’t about how to “build another Toyota”. But in Japan, the same small-factory pattern shows up in places that never had a giant anchor at all.

Sabae. A cold, snowbound farming district in Fukui prefecture, with no industry to speak of a century ago. In 1905 a local man brought in craftsmen from Osaka to teach spectacle-frame making as winter work for farmers who had no income for half the year.

A century later, that one district makes about 95% of Japan’s eyeglass frames and produced the world’s first titanium frames. It is a deep cluster of small family firms, and there is no giant at the top of it.

Higashi-Osaka. A single suburb of Osaka with roughly 6,000 factories, around 90% of which employ fewer than 20 people.

Locals say the district makes “everything from toothbrushes to satellites”, and in 2009 a group of these small firms built an actual satellite, the SOHLA Maido-1, to prove the point.

Panasonic and Sharp. The Kansai region grew its own dense web of small moulding, stamping and circuit-assembly shops around the consumer-electronics giants.

The shape was exactly the same as Toyota’s supplier network, but in a completely different sector.

The common thread here is not the anchor. It is a thick layer of small, capable, specialised firms, sometimes clustered under a giant and sometimes just clustered on their own.

But the best-documented example of how such a layer gets built, and gets tough enough to survive a fire, is Toyota’s home region of Aichi. That is where the story takes us.

Sakichi Toyoda was a carpenter’s son from a poor village on the coast near Nagoya. In the 1890s he started tinkering with weaving looms, because his mother and every other woman he knew wove cloth by hand for almost no money.

Over thirty years he built a loom that could spot a single broken thread and stop itself instantly, and one worker could then mind dozens of machines instead of one.

In 1929 the largest textile-machinery firm in the world, Platt Brothers of England, paid one million yen for the rights to that loom.

Sakichi’s son Kiichiro took the money and spent it studying a product Japan could not yet make: the motor car.

There was no plan for a national industry here.

There was a family, a windfall, and a hunch that a country which had just taught itself to build world-class looms could learn to build engines too.

Kiichiro opened a car department inside the family loom company in 1933.

Ford and General Motors already ran assembly plants in Japan, shipping in parts from Detroit and bolting them together with local labour. Every earlier Japanese attempt to build a car from scratch had died on the same problem.

There was nobody to buy parts from. No firm in Japan could supply a pressed-steel car body to a usable standard, or a proper crankshaft bearing, or brake components, or electrical parts, or rubber seals that held.

Kiichiro’s first engine, copied from a Chevrolet, kept cracking because no local foundry could pour the casting cleanly.

So Toyota did what a firm does when the supply base does not exist. It made everything itself: forging, casting, machining, glass, electrical, all in one place, all learned the hard way.

It was slow and expensive, and the cars were mediocre for years.

Assam today is in a milder version of this bind. India has deep supplier bases, but they sit two thousand kilometres away in Gujarat, Tamil Nadu and Maharashtra, and the Northeast has almost none of that capacity locally.

So the Japanese question, ported to Assam, is not “how do you build a supplier base from nothing?” It is a subtler one: how does a region stop importing every part and start making most of them nearby?

That is the question the Jagiroad anchor now puts to the state government.

In 1936 Japan’s government passed a law that would look familiar to anyone who has watched India’s electronics push.

Any large carmaker had to be Japanese-owned and licensed, and Ford and GM were capped and then squeezed out entirely by 1939.

Tariffs on imported parts went up, and Toyota and Nissan got the first two licences.

The military wanted trucks. The industry ministry wanted an industry. Protection was total, and it was pointed at two firms.

This policy wall gave Japan a domestic industry, and it worked because it came after two decades of small, high-cost producers making unremarkable vehicles. What changed things later was not the wall itself, but the four conditions Japan attached to sitting behind it:

• the licensed firms had to compete with each other, not carve up the market,
• they had to chase exports, so a foreign buyer’s judgement kept them honest,
• they had to hit quality numbers a government auditor could actually check,
• and the protection had a clock on it, not a permanent seat.

The wall without those conditions is just a subsidy. The wall with them is an industry.

In 1945 Japan lost the second World War, and Toyota lost its market a second time.

In 1949, an American adviser named Joseph Dodge put the Japanese economy through a brutal anti-inflation squeeze. Demand vanished, dealers stopped taking cars, and by mid-1949 Toyota could not pay its bills.

A syndicate of 24 banks bailed the company out on conditions: split off the sales arm, cut costs to the bone, and, over the founder’s objection, cut jobs.

Toyota moved to sack 1,600 workers, and the workforce struck for two months.

When it was over, more than 2,000 people were gone and Kiichiro Toyoda had resigned as president of the company his family had founded.

Toyota was made to spin off its electrical-parts division in 1949 as a separate company called Nippon Denso, now known simply as Denso- one of the largest auto-parts firms on earth today.

Aisin, the company whose factory would later burn down in 1997, was spun out of the group in the same wave. But Aisin, and Toyota and the ecosystem survived and thrived because of keiretsu.

Two kinds of keiretsu

There are two kinds of keiretsu in Japan, and only one is the subject of this report.

The horizontal keiretsu matters for Japanese corporate history, but it is not what built the small-factory base. This report is about the vertical keiretsu, built around a single manufacturing anchor like Toyota, and stacking its suppliers in tiers below it, each tier’s fate tied to the tier above.

Toyota sits at the centre. Denso and Aisin are on the ring immediately outside it, as former divisions turned into permanent suppliers. Below them sit a few hundred first-tier firms. Below those, a few thousand tier-two ones, and finally tens of thousands of tier-three workshops on the outermost ring.

They hold shares in each other, borrow from the same house bank, and stay linked for decades. None of them is fully free to walk away, and none of them is fully alone either.

That is the arrangement this report is about, and it is what made 200 strangers rebuild a brake valve in four days.

Toyota drew a hard lesson from nearly going bust: never again carry every cost yourself.

From then on, it deliberately kept itself smaller than it could have been, and pushed work outward to a network of suppliers it could guide, lean on in a boom, and share the pain with in a slump.

But that network did not yet exist.

It had to be built, and it was built slowly over the next fifteen years, through a combination of Toyota’s own pressure and technical help, new laws that we come to in the next section, and banks that were set up specifically to lend to firms no commercial bank would touch.

This is the moment in our story when small factories, the MSMEs, become the main characters.

The economist Michael Smitka, whose book Competitive Ties is the closest study of exactly this, argues that Japan’s supplier system was built as much to spread financial risk as to save cost.

The big firm stays lean and stable, while the smaller suppliers absorb the swings in the market. But in return, those suppliers get long relationships, technical help, and a real shot at growing with the customer. That bargain is what kept small firms willing to take on the risk in the first place.

The romance of this story is all about the big company.

But the useful bit is about a small factory, and about four unglamorous things a state can choose to build or not build, each of which the small factory needed in order to survive.

Picture a twelve-person stamping shop in Aichi in 1956, one that presses a bracket for a Toyota tier-one supplier. What does that shop need in order to stay alive, get better, and eventually be trusted enough to help save the anchor when a fire breaks out forty years later?

Four things, in roughly this order.

Alongside those four things, Toyota ran a suppliers’ association of its own called the kyohokai, which Toyota’s own historians describe as “a voluntary organization of parts suppliers”.

It ran shared training, quality circles and joint problem-solving for decades, among firms that were often rivals in each other’s product lines. Toyota now runs versions of it in a dozen countries.

None of this was fast, however. It took decades of sweat, blood and motor oil.

Now go back to the fire in Kariya in February 1997.

The reason 200 strangers could make a brake valve in four days was not improvisation nor a slick presentation on a business continuity plan, but fifty years of quiet groundwork bearing fruit. Every one of the five things a crisis needs was already in place:

• The standard was known. Those small firms already understood Toyota’s tolerances, because Toyota and its tier-ones had spent decades teaching them on quieter work.
• The trust was built. They took the job on a handshake because a law and a long relationship said they would be paid.
• The language matched. Their engineers already spoke the same production dialect, so drawings could be read without translation.
• The room existed. Many of them already sat in the same suppliers’ association, so a group call on day one could reach dozens of firms at once.
• The blueprints were open. The intellectual-property fights had been settled long ago in favour of keeping the group alive, so Aisin could fax its designs to rivals without a lawyer in the room.

That is what “depth” means in practice: not a long list of registered factories, but a web of relationships thick enough to route around a disaster.

First, the honest bit.

This ecosystem took Japan about fifty years and three brushes with collapse.

Even inside India the clock is long: the engineering cluster in Rajkot, Gujarat, which this series looks at next, took the better part of forty years to go from diesel-engine repair sheds to exporting castings. Nobody built that in five.

So a realistic Assam roadmap is measured in the working lives of the people who start it, which is twenty to thirty years to real depth.

This report is an aspirational map, not a promise. But what does transfer is method. Six things, all of which a state government can actually do.

The rest of this section unpacks each row.

What does not transfer, and Assam should stop wishing for it: a full-blown vertical keiretsu, with a captive main bank willing to fund the group through decades of losses and cross-shareholdings that lock suppliers in for a generation.

Nor a war economy that forced skills in-house; nor a US-run postwar occupation that could restructure Japanese industry by order; nor a central ministry with the staff and the political cover to run all of this for forty years.

The World Bank’s own study of East Asia warned that these tools misfire without strong institutions behind them, and that warning is the real subject of this series.

The honest promise at the end of it: Assam will not become Aichi. But over a generation, it can build a supplier base deep enough that the day the Jagiroad plant has a bad week, there are two hundred small firms in the Brahmaputra valley who know how to help.

Every one of the six levers in the last section already has an Indian scheme sitting on the shelf.

The problem is not a missing scheme. It is that policy conversation in Delhi and Dispur has, for a decade, reached for the same tool for every job: the Production-Linked Incentive, or PLI. PLI pays cash for output above a target, and it has worked well for what it was built for: mobile phones, and a handful of large assembly units.

But it is a hammer, and most of what Assam needs is not a nail.

Why PLI alone will not build a supplier base

PLI has three quiet limits that matter for the tier-three and tier-four layer:

• It rewards the anchor, not the layer below. The cheque goes to the firm that hits the output number, which is almost always a large one. A twelve-person stamping shop rarely qualifies, and rarely sees the money.
• It is time-limited, and outcome-blind. The support ends on a fixed date, whether or not the firm has become better at what it does. There is no upgrade condition attached, only a volume one.
• It is manufacturing-only. Design, IP, tooling, testing, calibration and repair, which are the services that surround a supplier base and often start it, are not eligible.

The lesson from Japan is that a supplier base is built by a stack of instruments, not a single one.

The Indian stack, and what each piece is really for

Here is the honest matching, lever by lever, with the tool that already exists.

The rest of this section reads each row.

The imagination bit

Put those six together and it starts to look less like a wish list and more like a design brief. A workable Assam stack, in one line each:

• one anchor MoU per major cluster, with published local-content and quality conditions;
• one enforced payment law, routed through TReDS, with public dashboards;
• one small-firm lending pipeline that stitches NEDFi, SIDBI and CGTMSE into a single door;
• one design-linked route for services and IP, alongside PLI for hardware;
• one suppliers’ council per cluster, seeded from SFURTI and MSE-CDP;
• one export top-up on RoDTEP, with the biggest subsidies reserved for firms that ship out of India.

None of these needs a new central scheme. Every one of them needs a state that treats industrial policy as engineering, not as a press release.

That is the real transfer from Japan.

• Toyota Motor Corporation, 75 Years of Toyota (the loom patent sale, the 1936 licensing law, the 1949 to 1950 reconstruction and Kiichiro Toyoda’s resignation, the kyohokai, and the Denso spin-off).
• Michael A. Cusumano, The Japanese Automobile Industry: Technology and Management at Nissan and Toyota, Harvard University Press, 1985. Publisher listing.
• Michael J. Smitka, Competitive Ties: Subcontracting in the Japanese Automotive Industry, Columbia University Press, 1991 (the risk-sharing argument for why the supplier system formed).
• Toshihiro Nishiguchi and Alexandre Beaudet, The Toyota Group and the Aisin Fire, MIT Sloan Management Review, 1998, and the 1997 Aisin fire.
• Chalmers Johnson, MITI and the Japanese Miracle: The Growth of Industrial Policy, 1925 to 1975, Stanford University Press, 1982. Full text at the Internet Archive.
• The World Bank, The East Asian Miracle: Economic Growth and Public Policy, 1993.
• Japan’s Subcontract Act, 1956, and Yoshimura and Kato, The Policy Environment for Promoting SMEs in Japan, PECC, 2007.
• Fukui Prefecture, A Global Eyewear Hub (Sabae), and JETRO, Higashi-Osaka City data.

Related Nitisagar work:

• The Japan Industrial Resilience Playbook for Assam, the sibling report on how Japanese factories are engineered to survive earthquakes and typhoons.
• What the Japan summit means for Assam’s semiconductor cluster.
• The incentive estimator for Assam’s fifteen-year policy.