What happened?
In October 2018, KYB disclosed that inspection-record data had been rewritten for some seismic isolation and vibration-control oil dampers manufactured by KYB and its subsidiary Kayaba System Machinery (KSM), and that products outside the ministerial-certification standards or customer-agreed standards had been installed in buildings.
KYB appointed an External Investigation Committee to examine when and how practices such as multiplying inspection results by coefficients and moving measured values through test-machine functions had begun, how they continued, and who knew about them. The committee submitted its report dated February 4, 2019, and KYB released the disclosed version on February 13.
This article uses only that External Investigation Committee report. It does not mix in later reports or separate remediation materials. The report contains detailed records of test-machine operations, sales emails, responses after the issue surfaced, and why internal reviews did not detect the manipulation. The report is in Japanese; this English article is From Reports’ translation and summary and is not an official English translation.
12 passages that stood out
Page numbers below refer to the printed pages of the report.
Interview accounts cited delivery concerns; test-machine keys were assigned coefficients of about 0.85 to 1.15
When seismic isolation oil dampers were produced in small numbers, a unit that failed a witnessed inspection would be disassembled and adjusted, such as by retightening valves or replacing parts.
As mass production began, interview accounts described concern that disassembling and adjusting every unit that failed performance inspection would make it impossible to meet delivery schedules. Some personnel in manufacturing, engineering and quality assurance discussed a method for rewriting values. The committee cautioned, however, that it found no objective evidence showing that reassembly would actually have made delivery impossible, and it could not rule out that the practice was simply used to avoid the work involved.
Under an early method, several keys on the test-machine PC were preset with different coefficients, generally around 0.85 to 1.15, and an operator could press a key selected from the actual result to apply the coefficient automatically.
Around March 2002, KYB commissioned an outside vendor to modify the program for coefficient rewriting, and the modified program was installed that July.
Adjusted values looked unnatural, so the test software was changed to add 1% only to the compression-side load
A 2,000 kN test machine introduced in 2013 had a “balance ON” function that moved the midpoint of extension-side and compression-side damping forces toward the origin. Using the function could turn measurements such as 450 kN on one side and 550 kN on the other into values near 500 kN on both sides.
That created a different problem: after the origin adjustment, the extension and compression values could look almost identical, raising concern that customers might find the figures unnatural.
In October 2014, after discussion within the manufacturing department, an outside vendor was asked to change the software so that when balance ON was used, 1% was added only to the compression-side load, preventing the two sides from displaying the same value.
“20→0.98,” “50→1.02,” “ON.” The Excel files preserved the operations themselves
On the 2,000 kN test machine, performance-inspection results could be recorded in Excel files on the computer. Sheets from September 2017 onward contained not only test results but also records of coefficient rewriting and origin adjustment.
In the remarks field, “20→0.98” meant that 0.98 had been applied during a 20-kine excitation, “50→1.02” meant that 1.02 had been applied at 50 kine, and products subjected to origin adjustment were marked in red with entries such as “バ ON” or “ON.”
The report reproduces part of an actual table, including an entry reading “20→0.98 50→1.02 ON 100→0.97 ON.” Interview accounts said these Excel files were periodically shared by manufacturing with engineering. Kine is a unit of excitation speed: 1 kine equals 1 cm per second (printed p.26).
A sales email said to apply a coefficient if the value fell outside the specification; several sales employees were copied
On February 3, 2015, a member of Sales Department 2 sent an email to an engineering manager concerning a particular project.
According to the report, the email described applying a coefficient if damping force did not fall within the required range, and several employees in Sales Department 2 were also recipients.
The committee said the evidence strongly supported the inference that the sender, the sales employees copied on the message, and the engineering employees receiving it were aware of coefficient rewriting no later than February 3, 2015.
One sales manager from the period said he had not been aware of the practice. The committee considered that explanation of low credibility in light of other testimony that the email had been sent at his direction.
The August 9 warning was recorded. The internal investigation did not begin until August 20
On August 9, 2018, employee AA directly raised concerns about coefficient rewriting with manufacturing personnel and, later that day, met BA, a director and the person responsible for the plant.
AA recorded the August 9 meetings, and the External Investigation Committee received the audio data.
BA said the internal investigation began on August 20, pointing to the August 11–19 Obon holiday period. The committee said that, given the seriousness and urgency of the issue, this did not reasonably explain why about ten days passed before the investigation began.
A customer-witnessed inspection failed when the rewriting stopped
BA said he had become convinced by late August that coefficient rewriting was taking place, but did not stop it immediately, citing concerns including the possibility that the employee who raised the issue might take the matter to the media if the practice were suddenly halted. The committee, however, rejected BA’s testimony that he first suspected rewriting in August 2018 and found that he had known of it earlier (printed p.58, note 53).
After an instruction around September 11 to stop rewriting in performance inspections, the report records testimony that rewriting still occurred in customer-witnessed inspections around September 13 and 14, and that on September 17 there was an instruction to continue it.
On September 19, an instruction was given to stop rewriting in customer-witnessed inspections as well. When no coefficient rewriting was performed, that day’s inspection result was a failure.
The committee also said it did not find objective evidence proving that all rewriting had ceased in every performance and witnessed inspection after September 19.
After the problem was known, the committee found no objective evidence of an explicit, formal shipping-stop order
The report records that several KSM officials with authority over shipment said that no one, including themselves, had formally ordered shipments stopped. They also said KYB had not ordered KSM to halt shipments.
The External Investigation Committee concluded that it could not confirm objective evidence showing that KYB or KSM issued an explicit and formal shipping-stop order after the problem was recognized.
KYB, however, explained that on September 22 it began re-testing unshipped inventory in KYB’s presence and that from then on it stopped shipping products in which improper conduct had been involved.
“ON” was removed before a list was sent to a customer, while original files and emails were also deleted
In early September 2018, after the problem had become known internally, a KSM quality-assurance manager instructed staff to remove from the test-machine computer the Excel files that recorded coefficient rewriting. An employee copied the files to a USB drive and deleted them from the test machine.
In October, after a KYB manager asked what the red “ON” entries in the Excel file meant, KSM personnel explained that they reflected a legitimate “zero-point adjustment,” not misconduct. An instruction was then given to remove “ON” before submitting the material because it might confuse customers or prompt inquiries.
A list with “ON” removed was in fact submitted to a customer, and a later instruction was given to delete the list itself so that the customer would not learn that “ON” had appeared in the original.
FB deleted the copy of the list in his possession, but he had already sent it to KYB on September 30, so the list itself was not lost and was provided to the committee. KYB quality executive TB also deleted an email with the original list attached. The committee said no reasonable explanation had been given for deleting that particular email.
The committee did not establish that TB and UB, the KYB officials who instructed staff to remove “ON” before submission, understood it to be a record of misconduct. At the same time, it said the fact-checking was inadequate and warned that the series of data deletions posed a serious impediment to the investigation, prompting a request for stronger evidence preservation.
For one model of the main BDS type, interviews indicated coefficient rewriting on roughly 90% or more of the units
In its analysis of causes, the committee said one factor was that KYB and KSM accepted business beyond what their own technical and production capabilities could readily support.
According to interviews and related information, for one model of KSM’s main BDS-type seismic isolation oil damper, coefficient rewriting was said to have been performed on roughly 90% or more of the products.
This figure is presented in the report as an interview-based result, not as an audited count of every product. The committee said it suggested that mass production without data manipulation had itself become difficult.
The manufacturing department that had to rebuild failed units also performed the inspections, and it picked better-performing seismic isolation units for witnessed tests
At KSM, the manufacturing department itself performed performance inspections even though that department would bear the burden of rebuilding products that failed. The committee identified this as a structural problem: the department with the greatest stake in the outcome was also conducting the inspection.
There was also no system for someone outside the manufacturing inspection staff to check the inspection process or results.
In most customer-witnessed inspections of seismic isolation oil dampers, customers did not designate which unit should be tested, so KSM selected dampers that had performed well in the earlier performance inspection.
The report also says it became routine to excite the dampers immediately before the witnessed inspection so that the waveform would look cleaner and avoid customer questions or complaints.
One employee performed the same testing work for about a decade, at times almost alone; after promotion, he taught the manipulation method to subordinates
Employee EA handled performance inspections for seismic isolation and vibration-control oil dampers from 2007 through at least the discovery of the problem in August 2018, a period of about ten years.
From roughly 2007 through 2011, EA was generally the only person conducting the performance inspections. After being promoted around 2013, he became the person who taught subordinates how to manipulate inspection data and instructed them to do so.
The committee said long-term assignment of the same people to the same work weakened checks by supervisors and colleagues, and that promotion within the same work area helped the practice become inherited and entrenched.
After the Toyo Rubber scandal, an internal review concluded that automatic output meant the data could not be altered
After the Toyo Rubber Industries seismic-isolation data scandal was announced in March 2015, KYB’s quality organization reviewed whether KSM’s seismic isolation and vibration-control oil damper system prevented performance data from being altered.
The review concluded that the inspection quality data could not be altered because the results were output automatically.
The committee said the review had focused on the output stage and failed to consider the risk that data could be manipulated while operating the test machine.
Later special internal audits prompted by the Nissan final-inspection problem and the Kobe Steel data-falsification problem likewise concluded that KSM’s quality records showed no intentional data alteration or fabrication. The committee said those audits, too, had remained formalistic and failed to identify the actual manipulation.
What the report says about the background
The External Investigation Committee identified several factors behind the long duration of the manipulation: weak professional norms among people involved in manufacturing, a corporate culture that did not face inconvenient facts directly, taking orders without sufficient regard to technical and production capacity, weaknesses in inspection structure and methods, poor control of test machines, failures in information sharing, fixed staffing, and shortcomings in internal audit.
The test machines themselves were particularly important. Software made coefficient operations easy, and later software changes made adjusted results look less conspicuous. The committee said this made misconduct possible with effectively “one button,” turned it into routine work, and weakened operators’ psychological resistance.
The report was based on work conducted through January 31, 2019. At that time, investigation of similar misconduct across the broader KYB group was still continuing. This article does not extend factual claims beyond the scope of this report.
Corrections to this article
Note 01 identifies delivery concerns as interview accounts, rather than an established inability to meet deadlines (pp.36–37). Note 03 attributes Excel sharing to interview accounts, and note 06 adds the committee’s finding that BA knew of rewriting earlier (p.48; p.58, note 53). Note 08 limits the unestablished awareness of ON as a misconduct record to TB and UB (p.62). Note 10 limits the selection of better-performing units to witnessed inspections of seismic isolation dampers. Note 12 identifies March 2015 as the Toyo Rubber announcement date, rather than the date of KYB’s review. Source references for notes 02, 03, 06 and 08 were expanded, and note 11 retains the source’s at-least qualification for the testing assignment.