What Pension Funding Means for Cash at IBM, CPKC and Lockheed

The signal

The investable cash benefit is documented contribution relief over time. A pension asset or accounting income figure needs a separate legal and cash-flow bridge.

IBM shows why a surplus can coexist with expense

IBM’s 2025 annual report records US$18.073 billion of assets against US$13.207 billion of benefit obligations in its qualified U.S. Personal Pension Plan. The US$4.866 billion surplus is a year-end accounting measurement; the implied asset-to-obligation ratio is approximately 136.8%. It is not a September 2026 valuation or an amount shown as unrestricted corporate cash. [2]

The September 2026 filing says pension plans remain well funded and reiterates that no U.S. defined-benefit contributions are legally required for the year. It nevertheless reports U.S. pension cost of US$122 million in Q3, up from US$77 million a year earlier. Its expense table includes service and interest costs, expected asset returns and amortization of prior actuarial losses. [1]

The company can therefore incur accounting expense without a corresponding required contribution in that year. An analyst who substitutes expense for funding would misstate the cash requirement. The reverse error is to assume no required contribution means no pension cost, no voluntary payment, or no future exposure. Each conclusion goes beyond the disclosure.

The assets belong to a defined pension arrangement

IBM describes qualified-plan contributions as going to an irrevocable trust for participants and beneficiaries. Its annual report separately presents qualified and nonqualified arrangements, U.S. and non-U.S. plans, and nonpension postretirement obligations. A surplus in the qualified U.S. plan does not establish unrestricted access to cash or automatically offset every other benefit obligation. [2]

This is the key valuation distinction. Reduced future contributions can leave more operating cash available inside the business. That is an economic benefit even when the company cannot withdraw a dollar from the plan. But it must be valued through a defensible path of future funding requirements, rather than by adding the entire accounting surplus to cash at face value.

Jurisdiction and plan boundaries matter. A group can sponsor several legally distinct plans with different funding rules and different maturity profiles. Netting all their accounting positions may be useful for one disclosure purpose while obscuring where the sponsor must actually send money. A corporate cash model should preserve those boundaries until the contribution requirements are reconciled.

CPKC: accounting income is larger than the cash movement

For the first nine months of 2026, CPKC reported C$290 million of net periodic pension recovery, compared with C$272 million in the same period of 2025. Contributions net of refunds were C$5 million, against C$11 million. The pension figures exclude the separately disclosed other-benefit cost. [3]

The improvement in accounting pension income was C$18 million. The reduction in net contributions was C$6 million. Those changes should not be presented as the same cash saving. Expected investment returns and other accounting components influence the recognized recovery, while contributions reflect a different funding process.

Using reported nine-month operating cash of C$4.252 billion, the net contribution amount represents approximately 0.12% of that cash flow. This is a historical scale comparison, not an estimate of future funding or a measure of the pension plan’s return. It also uses contributions net of refunds; a gross-contribution ratio would need gross payments. The small observed burden is favorable evidence, but its persistence requires the funding position and applicable rules to remain supportive. [3]

Lockheed: an accounting deficit does not dictate this year’s payment

Lockheed’s 2025 annual report shows US$22.850 billion of qualified pension plan assets against US$26.735 billion of projected benefit obligations, a US$3.885 billion accounting deficit. It also states that ERISA funded status is calculated differently from GAAP, records US$860 million of cash contributions during 2025, and expects no required contributions in 2026. [4]

The Q3 2026 filing reports US$277 million of nine-month qualified-plan FAS pension expense and no contributions during either the first nine months of 2026 or the comparable 2025 period. The current expense is lower than the prior US$334 million, but the contribution comparison for those matched interim periods is zero versus zero. [5]

This timing detail prevents a false cash-flow attribution. The annual 2025 contribution cannot be subtracted from a nine-month 2026 figure and called a like-for-like operating improvement. Nor does a GAAP deficit automatically imply an immediate statutory payment of the deficit. Funding calendars, prior contributions and the applicable calculation need to be read together.

Government-contract accounting adds another layer

Lockheed also explains that pension cost recoveries under government-contract Cost Accounting Standards can occur in different periods from actual contributions. Its reporting reconciles FAS pension expense with CAS costs. These recoveries and adjustments are another reason not to compare a contractor’s pension-related operating-profit line mechanically with an ordinary industrial company’s pension cost. [4] [5]

The analytical sequence should follow actual cash: contributions paid by the sponsor, contractual recoveries where relevant, and the operating cash-flow reconciliation. Accounting adjustments help explain earnings but do not create a second cash receipt merely because they appear in a profit bridge.

The same caution applies to free-cash-flow calculations. Reported operating cash already incorporates its accounting-to-cash reconciliation. Subtracting pension expense again, or adding accounting pension income again, can double count the item. A clearly labelled sensitivity to a different future contribution requirement is useful; silently rewriting historical cash flow is not.

Funding rules and data dates deserve separate treatment

The U.S. single-employer funding statute defines required contributions using target normal cost, funding targets, plan assets and applicable amortization or balance adjustments. That framework is distinct from simply subtracting accounting assets from projected obligations. Our model does not calculate statutory minimum contributions, and the U.S. rule should not be applied indiscriminately to foreign plans. [7]

PBGC’s data documentation supplies a second caution. Its August 2026 release is labelled the first installment of 2024 pension insurance data, and it explains that tables can have different end years because their sources become available at different times. Form 5500 deadlines and extensions create a further lag. A newly published table is not necessarily a current funded-status snapshot. [6]

The insurer’s financial position is also a different entity from the corporate plans it covers. An improvement in PBGC’s own finances cannot be used as a substitute for an issuer’s contribution forecast. The relevant evidence is the sponsor’s plan-specific funding information, its latest cash update and the rules governing each obligation.

A sensitivity that keeps assets and liabilities together

Our calculator begins with plan assets minus the benefit obligation. It then applies an assumed portfolio valuation change to assets and a first-order duration response to the obligation. The liability approximation is obligation multiplied by one minus liability duration times the discount-rate change. A 50-basis-point decline is entered as −0.005, not −0.5.

Consider hypothetical assets of 110 and an obligation of 100, measured in the same currency units. An assumed liability duration of 12 and a 50-basis-point discount-rate decline increase the approximated obligation to 106. If the portfolio falls 5% under the same shock, assets become 104.5 and the initial surplus of ten becomes a deficit of 1.5. This is a stress illustration, not a forecast for any company in the study.

A more closely matched portfolio is the countercase. If its value instead increases 5.5% under that rate decline, assets become 116.05 and the modeled surplus is 10.05. The balance-sheet benefit largely survives. Asset and liability assumptions must describe the same scenario; changing the discount rate while ignoring the portfolio’s rate exposure can manufacture a misleading result.

The cash decision comes after the funding sensitivity

The tool separately displays actual cash contributions divided by operating cash for a matching period. In its hypothetical default, contributions of two and operating cash of 50 produce a 4% burden. That historical ratio is not derived from the modeled surplus or deficit. If operating cash is nonpositive or effectively zero, the ratio is suppressed because it would not provide a meaningful coverage measure.

Duration is an explicit user assumption when a sponsor has not disclosed it. The instantaneous approximation omits convexity, benefit payments, new service accrual, contributions and the detailed shape of the discount curve. It can reveal exposure, but it cannot establish a contribution date, the amount legally payable or the amount shareholders may recover.

The evidence supports a conditional positive case: sustained funding strength can reduce required cash payments and improve flexibility. It also identifies the falsifier for an easy surplus story: accounting gains without contribution relief, or a funding position that deteriorates materially under coherent asset-and-liability shocks. For investors, the next step is a multiyear contribution schedule reconciled to cash and legal funding bases. IBM, CPKC and Lockheed show why that schedule is more informative than either a headline pension asset or a single year’s expense.

The evidence, side by side

Pension Funding and Cash Bridge

Hypothetical instantaneous funding sensitivity; not a statutory contribution estimate, year-ahead forecast or estimate of cash shareholders may withdraw.

Finite inputs; assets≥0; obligation>0; duration≥0; portfolio return≥−100%; modeled liability factor must remain positive. Suppress contribution ratio for CFO≤1e−9 in selected units. Negative net contributions must be explicitly described as refunds. Duration must be disclosed or labelled as a user assumption.

First-order duration only; no convexity, benefit payments, service accrual or contributions in the valuation shock. Asset return must be coherent with the discount-rate scenario. Historical contribution ratio is a separate observed cash comparison.

Scenarios to watch

Asset loss plus lower discount rate

Assets 110 fall 5% to 104.5; obligation 100 rises to 106 with duration 12 and a 50bp rate decline; funded status moves from+10to−1.5.

Portfolio responds with liabilities

With a 5.5% portfolio gain under the same 50bp discount decline, assets 116.05 less obligation 106 leave 10.05. This is a hypothetical countercase, not a measured hedge.

Contribution relief

Observed or documented reductions in required contributions improve cash flexibility; the amount must come from funding disclosures, not the sensitivity’s surplus.

Method and limitations

Three-sponsor study of IBM, CPKC and Lockheed Martin using five issuer filings plus PBGC documentation and the U.S. funding statute, accessed by 13 November 2026. Annual measurement dates and interim cash periods are retained separately. Monetary amounts are not pooled across U.S. and Canadian dollars. CPKC figures use pensions only and net contributions. IBM’s legal requirement is not presented as actual cash paid. Lockheed’s matched nine-month cash comparison is zero versus zero. No 30-sponsor panel or statutory actuarial model is claimed.

A surplus can be restricted by plan terms and law; contribution relief differs from cash recovery. No assessment of lawful withdrawals is made.

Accounting funded status, statutory funding and PBGC insurer finances are separate measures; foreign plans require their own rules.

The duration sensitivity is a simplified instantaneous scenario. No company-specific liability duration is invented, and no scenario probability is assigned.

Primary sources and evidence register

Source dates and reporting periods differ. Several documents from one issuer are one source family, not independent confirmations. The evidence download contains the observations used in this article.

Frequently asked questions

Can a company have pension expense without required contributions?

Yes. Accounting expense and statutory funding use different processes. IBM’s current disclosure illustrates this distinction.

Does an accounting deficit mean an immediate cash payment?

No. Required contributions depend on the applicable funding basis, prior funding, balances and timing. Lockheed’s disclosure provides a concrete counterexample.

Can pension surplus be added to corporate cash?

Not automatically. Plan assets may be restricted for participants. The more defensible cash benefit is a documented reduction in future contributions.

Does the calculator estimate next year’s legally required contribution?

No. It illustrates funded-status sensitivity and separately scales historical cash contributions. Statutory funding requires a plan-specific actuarial calculation.